Mouse Models of Mineral Bone Disorders Associated with Chronic Kidney Disease

Patients with chronic kidney disease (CKD) inevitably develop mineral and bone disorders (CKD–MBD), which negatively impact their survival and quality of life. For a better understanding of underlying pathophysiology and identification of novel therapeutic approaches, mouse models are essential. CKD can be induced by surgical reduction of a functional kidney mass, by nephrotoxic compounds and by genetic engineering specifically interfering with kidney development. These models develop a large range of bone diseases, recapitulating different types of human CKD–MBD and associated sequelae, including vascular calcifications. Bones are usually studied by quantitative histomorphometry, immunohistochemistry and micro-CT, but alternative strategies have emerged, such as longitudinal in vivo osteoblast activity quantification by tracer scintigraphy. The results gained from the CKD–MBD mouse models are consistent with clinical observations and have provided significant knowledge on specific pathomechanisms, bone properties and potential novel therapeutic strategies. This review discusses available mouse models to study bone disease in CKD.


Introduction and Rational for Mouse Models of CKD-MBD
Chronic kidney disease (CKD) is highly prevalent and induces mineral and bone metabolism disorders (MBD), even in the early stages of CKD [1,2]. CKD-MBD encompasses abnormalities of calcium and phosphate metabolism induced by the deregulation of parathyroid hormone (PTH), fibroblast growth factor 23 (FGF23) and vitamin D homeostasis that lead to renal osteodystrophy (ROD), a term used to describe the different patterns of skeletal abnormalities in patients with CKD and extra-osseous calcifications. The latter comprises the rare but life-threatening soft tissue calcification of the skin (namely calciphylaxis), as well as the highly prevalent vascular calcifications, essentially contributing to an exceedingly high cardiovascular morbi-mortality [2][3][4][5][6][7][8].
Patients with CKD present a broad spectrum of bone histological changes that include mineralization and bone turnover abnormalities that reduce bone mass and strength; in contrast, bone volume is usually preserved [9][10][11][12]. Histological bone alterations are driven by the stage of CKD and the associated abnormal hormone and mineral metabolism, as well as by the therapeutic interventions, with mineralization abnormalities being more prevalent in children than in adults [11]. Thus, the type of bone disorder evolves over time, and the disease pattern may change substantially, e.g., from a PTH-induced high turnover bone disease to suppression of PTH by high dose active vitamin D treatment and/or high calcium intakes, resulting in the histological phenotype of an adynamic bone, which confers a particularly high risk of vascular calcification since calcium binding capacity of the bone is suppressed [13].
The pathophysiology of CKD-MBD is complex and still not fully understood; studies for the in-depth pathophysiological understanding of disease development and progression and for improved clinical management are urgently needed. However, randomized controlled trials in the field of CKD-MBD are costly and challenging due to the high heterogeneity of bone disorders, the variety of treatments with limited standardization, the underlying diseases and associated treatments affecting bone health per se, and the fact that bone evaluation is not well standardized [14]. Metabolic diseases such as oxalosis or cystinosis have a direct toxic effect on bone [15,16]. Moreover, clinical studies, at best, require repeated bone biopsies but are hardly performed in clinical routines due to their invasiveness [17].
Thus, preclinical animal models represent a valuable tool to overcome these major barriers and limitations. In a controlled environment with standardized conditions of CKD and mineral supply, animal models help to identify specific underlying molecular mechanisms, characterize the role of various contributing factors and determine the efficiency of specific treatments of CKD-MBD. Several CKD-MBD mouse models have been developed that exhibit vascular calcifications [18]. The present review gives an overview of these models and discusses their contribution to the understanding of ROD in CKD-MBD.

The Mouse Model to Mimic Human CKD-MBD
In 1987, Gagnon et al. were the first to establish stable CKD conditions in mice by nephrectomy and contralateral kidney electrocauterization, which resulted in similar biological and bone alterations to those observed in humans with CKD-MBD [19]. Mice and humans develop similar hormone dysregulation, and toxin accumulation, usually characterized by increased serum creatinine and urea concentrations, hyperphosphatemia and increased circulating PTH and FGF23 levels. In addition, some mouse models develop vascular calcifications and display a large spectrum of bone abnormalities encountered in humans. A bone/vascular axis has been described in two surgery-induced CKD mouse models with high bone turnover, i.e., a relationship between aortic mineral elements and calcifications and the type of osteodystrophy [20,21].
Since the first description, many histological and/or micro-CT bone analyses have been performed in mice with CKD. The objectives of these studies were diverse, including the characterization of CKD-associated bone disease, the relationship between bone and vascular disease, and the impact of existing or new therapies. The phenotype and severity of ROD are highly variable and depend on age, gender, the genetic background of the mice, the degree of CKD and secondary hyperparathyroidism (SHPT) achieved and the dietary regime (and notably the calcium and phosphate contents). In addition, the time interval between CKD induction and bone analysis is critical. Moreover, there is a lack of consistency in the parameters assessed, especially by histology and micro-CT.
Other species to study CKD-MBD were mainly rats, applying a 5/6 nephrectomy or adenine diet. This, however, precludes genetic modifications; only spontaneous development of autosomal dominant polycystic kidney disease could be used in rats to study CKD-MBD [18,22]. Other animal models include dogs, cats and rabbits. Substantial species-specific variations have to be considered, e.g., in vitamin D metabolism. Rabbit kidneys are resistant to adenine toxicity [23]. CKD-related osteodystrophy hardly develops in cats [24].

Induction of CKD
Since this first description of a CKD-mouse model, several methods of CKD induction have been used. Here, we summarize the various models used to study ROD, as extensive reviews on CKD in mice were previously performed [25,26].
Four main categories of CKD-MBD mouse models were identified, namely CKD induced by surgery, nephrotoxic compounds, spontaneous CKD and genetic engineering [26]; they are summarized in Table 1. In these models, the age of CKD induction was highly variable, ranging from 5 to 38 weeks. CKD degree was progressively increasing over time, mainly in genetic models of CKD. Surgical models were the most frequently used [19][20][21], as they offered the possibility to compare the phenotype before and after surgery and were described as stable over time. Subtotal nephrectomy, so-called 5/6th nephrectomy, mimics quite well the process of reduction in nephron mass occurring in humans during renal failure. The surgical models followed the original description from Gagnon et al. [19], with a unilateral nephrectomy and an electrocauterization of the remnant kidney to reduce the total renal mass to 1/6th [21,[28][29][30][31][32]37,38,[40][41][42][43][44][46][47][48][49][50][51]. The degree of cauterization is critical as the amount of remnant kidney tissue defines the CKD severity [30]. Alternative surgical methods following the unilateral nephrectomy were polar excision of the remnant kidney [20,27,[33][34][35][36]39] or ligation of branches of the renal artery [45]. The surgery was mostly done in two steps with a time interval of one to two weeks but was occasionally performed in one step [36,45]. However, 5/6th nephrectomy is challenging in mouse, due to the small size and the low blood volume, and may lead to a high mortality rate. Consequently, other models were developed.    Adenine feeding was first used in rats and leads to CKD due to the crystallization of adenine metabolites in proximal tubules [25,69]. The major advantage of the adenine model is the avoidance of surgery and its associated complications. The protocol was then adapted to mice but required a casein-containing diet for the appetence. In CKD-MBD studies in mice, a 0.2% adenine diet was used for 2 [64], 5 [65], 6 [52,66,67], 8 [46], 12 weeks [53] or between 6 and 12 weeks [54].
Frausher et al. reported that Brown agouti/2 (DBA/2) mice fed with a high phosphate diet develop progressive calcifications and CKD [55]. Some authors used both groups of mice with surgically induced CKD and adenine-induced CKD [46,64], with the degree of CKD being higher in the latter in one study [46].
Genetically engineered mouse models comprise the Juvenile Cystic Kidneys (JCK) mice (a model of human Autosomal Dominant Polycystic Kidney Disease) [57,58], the Col4a3 −/− and Col4a5 y/− mice (models of Alport disease) [59][60][61][62] and the iCTCFpod −/− mice (a model of nephrotic syndrome with kidney failure) [63]. LDLR −/− mice fed a highfat diet are prone to develop the metabolic disease, characterized by insulin resistance, type 2 diabetes and atherosclerosis, and can be considered as an early CKD model. This model was generally used with complementary surgery to increase the degree of CKD [29][30][31]41]. ApoE −/− mice develop hypercholesterolemia and atherosclerosis, and following surgical kidney mass reduction, were used as a combined CKD-and hypercholesterolemia-induced vascular disease model, i.e., a model of advanced metabolic syndrome [21,42,48].
To demonstrate specific CKD-induced effects on bone independent of the dysregulation of PTH, active vitamin D, calcium and phosphate, Lund et al. applied a low phosphate diet (0.2%) in CKD mice and supplemented calcitriol to compensate for the lack of renal synthesis of the bioactive form of vitamin D and to prevent HPT and by this was able to also maintain serum phosphate and calcium levels in the normal range [40]. These mice developed very low turnover bone, i.e., an adynamic bone disease with depressions in osteoblast number, bone formation and mineral apposition rate. Zhang et al. used 0.2% and 0.02% phosphate diets in Alport mice to study the relation between phosphate supply and FGF23. Fractional excretion of phosphate was independent of serum FGF23 levels [62].

Impact of Strain, Gender, Age
The susceptibility to CKD and to renal osteodystrophy depends on the mouse strain and gender. The C57BL/6 background has been used most frequently but is relatively resistant to the development of glomerulosclerosis, proteinuria and hypertension [18,25,26]. DBA/2 [37,38] and Crlj:CD1 mice [35] develop CKD-MBD, with a full spectrum of high turnover ROD in mice fed with an HP diet.
Bone analyses were performed after highly variable time intervals following the induction of renal injury, ranging from 3 to 20 weeks [20,32,35]. CKD duration impacts the bone, with longer CKD duration substantially worsening bone disease [35,58]. Kadokawa et al. described an age-related regression of the trabecular architecture that is accelerated by CKD [35].
In recent years, new methods to assess GFR have been developed in laboratory animals, and their principle strengths and limits have already been reviewed [71]. Alternative methods, such as the transcutaneous measurement of GFR using the fluorescent renal marker FITC-sinistrin [72,73], radiolabeled markers or unlabeled radiocontrast agents have been described in mice and canines [74], but were not used in the reported CKD-MBD studies [71,75,76]. Cystatin C is an endogenous marker frequently used in humans for GFR estimation and seems to be a sensitive marker in mice, but its use remains scarce in mice [71,77].
Serum PTH levels showed large variations. Mouse models with severe HPT were expected to present with a high turnover bone disease, but due to the plethora of different models, an array of different bone findings have been described. An increased PTH was found in some low-turnover bone disease models [55,63], and LDRL −/− mice fed with a high-fat/cholesterol diet are described as resistant to the bone remodeling effect of SHPT [29,41].

Histomorphometry
The gold standard technique to determine trabecular bone microarchitecture and bone remodeling dynamics remains bone histomorphometry, performed ex vivo in mice. Different from humans, where a small iliac crest biopsy is available, the analysis in a mouse's whole bone can be performed after the sacrifice. Some specificities in rodents, such as difficulty in recognizing osteoclasts, exist, and specific embedding and staining have to be used [78]. Both human and mouse bone analysis by histomorphometry require specialists able to interpret the samples [12,79] and only allow a two-dimensional (2D) analysis that requires "calculations" to move into 3D.
The same nomenclature and classification as in humans were frequently used in rodent studies in the absence of a specific nomenclature for rodents. The current classification and treatment strategy for ROD in humans is based on changes in bone turnover, mineralization and volumes (Table 2) [6,13] and requires a double tetracycline administration prior to the biopsy of the iliac crest to obtain dynamic parameters [7,80]. Parfitt et al. described and developed a standardization and nomenclature for bone histology in 1985 that was updated in 2012 [79,81] and is still in use. Table 2. Classification of renal osteodystrophy based on turnover and mineralization. The bone volume can be low, normal or high in the various forms of ROD.

Type of Renal Osteodystrophy Histomorphometric Description
Osteitis fibrosa Increased turnover, normal mineralization Osteomalacia Decreased turnover, abnormal mineralization Adynamic bone disorder Decreased turnover, normal mineralization (decreased cellularity) Mixed osteopathy Increased turnover, abnormal mineralization The turnover reflects the rate of skeletal remodeling resulting from the balance between bone resorption and formation and is assessed by double labeling and corresponding bone formation rate or activation frequency. Measurements of osteoblasts surface and number, osteoid surface, osteoclasts number and bone eroded surfaces are related to bone turnover and are indicative, even though they are less accurate than the double labeling. Bone turnover is affected by different parameters, including PTH.
Osteitis fibrosa is a high-turnover bone disease secondary to SHPT, and mineralization defects in mixed uremic osteodystrophy are most often attributed to vitamin D deficiency [2,7].
The mineralization reflects the amount of unmineralized osteoid and is assessed by static measurements, such as osteoid volume and thickness, and by dynamic parameters (e.g., mineralization lag time) [7].
Volume indicates the amount of bone per unit volume of tissue and is assessed by measurements of bone volume on cancellous bone. Cortical and cancellous bone volumes can be differently affected by CKD.
Bone volume is not classically used to stratify bone diseases. Bones with a low turnover and normal mineralization are classified as adynamic bone disease, whereas those with a high bone turnover, especially when they exhibit other signs of high PTH, have osteitis fibrosa. Osteomalacia is diagnosed in case of abnormal mineralization and low bone formation, whilst bones with abnormal mineralization and a low bone formation have mixed disease [7,82], as illustrated in Table 2.
Some studies only give some qualitative analysis. The number of analyzed parameters was highly variable from one study to another, and in Table 3, we report the most frequently used. Some studies analyzed one or more formation parameters, such as osteoid surface, volume and osteoblast quantification. Osteoid volume and surface were frequently increased in CKD. Other also analyzed resorption parameters such as osteoclasts quantification and eroded surface, and an increase of the latter was often observed. Trap staining can be used to quantify the osteoclasts [39,52]. Structure parameters such as bone volume/trabecular volume, trabecular thickness, numbers or spacing were frequently reported, and these parameters were variably affected by CKD.
Tetracycline-derived labels are the gold standard in humans to describe bone turnover. Double labeling was also performed in mice with various protocols and various parameters. However, differences in the formation parameters assessed by double labeling were not always significant, and authors used either formation or resorption parameters to classify into a high or a low turnover disease. In only a few studies, dynamic parameters in histomorphometry analysis were not considered [19,[37][38][39]47,49,51,54,59,68]. In contrast, one or more dynamic parameters, such as mineral apposition rate, mineralization lag time, adjusted apposition rate and bone formation rate assessed thanks to the labeling, were performed in the majority of the studies with intraperitoneal or subcutaneous injections of fluorochromes [21,[27][28][29][30][31][32][33]36,40,41,43,44,46,48,52,53,55,57,58,60,61,63,64,66] (Table 3). The time of the first fluorochrome injection (4 to 14 days before sacrifice), the second injection (2 to 3), as well as the delay between injections, vary (2 to 5 days). One study used triple labeling with injections 5 weeks, 2 weeks and 2 days before necropsy [66]. The main fluorochromes used were calcein, tetracyclin, alizarin, demeclocycline and xylene orange with various combinations in the studies.
The expression of specific markers from bone cell subtypes (for example, klotho, FGF23, RANKL and OPG, TNAP, osteocalcin, Phospho1) was frequently determined by immunohistochemical staining and quantitative PCR [27,33,39,46,[57][58][59]65,68,84]. These analyses provide insight into the molecular machinery of bone and into the regulation of systemically active hormones such as FGF23, secreted by osteocytes and regulated by osteocyte protein dentin matrix protein 1, which is downregulated in CKD [61]. Table 3. Histomorphometric findings based on the articles reviewed and presented in Table 1. Only findings of mice with CKD on normal and high phosphate diets are depicted, but no findings from specific intervention groups.
Images of whole mouse bones can be obtained by micro-CT, from X-rays or more sophisticated from a synchrotron [85,86]. Micro-CT allows for a higher spatial resolution (6 to 20 µm) and the analysis of trabecular microarchitecture in three dimensions and informs on the cortical structure. These results showed good correlations with histological findings in animals and humans [89][90][91]. Specific micro-CT analysis guidelines were described for rodents [92].

Cortical Parameters Abbreviations Main Finding
Average cortical thickness Ct.Th decreased (13)

Other Bone Exploration Methods Used
We recently proposed the use of in vivo quantitative bone planar scintigraphy to investigate bone remodeling in mice with or without CKD [44,93]. We designed a quantitative evaluation of bone uptake of phosphonate tracer on the knee's regions of interest (ROI) at the epiphyseal plate regions, drawn on bone planar scintigraphic images, as a measure of osteoblast activity. An index was calculated from the counts in the knee's ROI (normalized by pixels and seconds), corrected for activity administered, decay between administration and imaging and individual animal weights. This index was applied in healthy and CKD mice.
Titanium implant resistance was evaluated by a biomechanical push-in resistance test and by an evaluation of bone-implant contact [47,49,50]. Bone calcium content was quantified by ash studies [33,34,62], and pyrophosphate content by fluorimetric pyrophosphate assay [27].

Conclusions
In conclusion, an array of CKD-MBD mouse models has been developed over the past 20 years. These models recapitulate the bone phenotypic spectrum of ROD as present in humans with CKD, and pathophysiological mechanisms are largely concordant. Due to the complex network of underlying pathomechanisms of MBD in CKD and the various phenotypes of ROD, it is, however, still challenging to choose the most appropriate mouse model to address a specific scientific question. Research teams have to consider the strengths and limitations of each mouse model carefully. Due to the great heterogeneity in the protocols and the parameters studied, careful interpretation of the findings is required. On the other hand, experimental mouse models allow controlling, and thus avoiding confounding factors, to provide valuable insights into specific pathomechanisms of CKD-MBD and the therapeutic potential of respective interventions.