Bone remodeling in the longest living rodent, the naked mole‐rat: Interelement variation and the effects of reproduction

Abstract

The pattern of bone remodeling of one of the most peculiar mammals in the world, the naked mole‐rat (NMR), was assessed. NMRs are known for their long lifespans among rodents and for having low metabolic rates. We assessed long‐term in vivo bone labeling of subordinate individuals, as well as the patterns of bone resorption and bone remodeling in a large sample including reproductive and non‐reproductive individuals (n = 70). Over 268 undecalcified thin cross‐sections from the midshaft of humerus, ulna, femur and tibia were analyzed with confocal fluorescence and polarized light microscopy. Fluorochrome analysis revealed low osteogenesis, scarce bone resorption and infrequent formation of secondary osteons (Haversian systems) (i.e., slow bone turnover), thus most likely reflecting the low metabolic rates of this species. Secondary osteons occurred regardless of reproductive status. However, considerable differences in the degree of bone remodeling were found between breeders and non‐breeders. Pre‐reproductive stages (subordinates) exhibited quite stable skeletal homeostasis and bone structure, although the attainment of sexual maturity and beginning of reproductive cycles in female breeders triggered a series of anabolic and catabolic processes that up‐regulate bone turnover, most likely associated with the increased metabolic rates of reproduction. Furthermore, bone remodeling was more frequently found in stylopodial elements compared to zeugopodial elements. Despite the limited bone remodeling observed in NMRs, the variation in the pattern of skeletal homeostasis (interelement variation) reported here represents an important aspect to understand the skeletal dynamics of a small mammal with low metabolic rates. Given the relevance of the remodeling process among mammals, this study also permitted the comparison of such process with the well‐documented histomorphology of extinct therapsids (i.e., mammalian precursors), thus evidencing that bone remodeling and its endocortical compartmentalization represent ancestral features among the lineage that gave rise to mammals. It is concluded that other factors associated with development (and not uniquely related to biomechanical loading) can also have an important role in the development of bone remodeling.

Naked mole‐rats have scarce bone resorption and remodeling. Development of secondary osteons occurs regardless of reproductive status. Breeders had higher bone resorption and secondary reconstruction than subordinates. Zeugopodial bones had less bone remodeling than stylopodial bones.

1. INTRODUCTION

Bone remodeling is the mechanism by which old bone is replaced by new bone by the sequential and coupled action of bone resorbing cells (osteoclasts) and bone forming cells (osteoblasts) at single bone sites (Frost, 1969; 1987(a), 1987(b)a; Jaworski, 1992; Parfitt, 2010; Seeman, 2008; Allen & Burr, 2014; Currey et al., 2017). Due to the coordinated activity of osteoclasts and osteoblasts, this process results in the formation of secondary osteons (=Haversian Systems) (Currey & Dean, 2017; Enlow, 1962; Jaworski, 1992) and is usually referred to as secondary reconstruction (Amprino, 1948; Chinsamy‐Turan, 2005; 2012a; Enlow, 1962). Bone remodeling and secondary osteons (SOs) occur principally along endosteal, intracortical, and trabecular bone surfaces (Allen & Burr, 2014; Enlow, 1963; McFarlin et al., 2008; Parfitt, 2010; Seeman, 2008; Singh & Gunberg, 1971). SOs can be found isolated, forming small groups or forming dense groups (Hillier & Bell, 2007; Jaworski, 1992). When dense aggregations of SOs accumulate through several generations and cover a defined bone surface, this is called dense Haversian bone tissue and represents a specific subtype of—secondary—bone, usually present in large vertebrates (Chinsamy‐Turan, 2005; Enlow, 1963; Francillon‐Vieillot et al., 1990; Jaworski, 1992; de Ricqlès et al., 1991). SOs have been described for a wide range of vertebrates (Enlow, 1969; Enlow & Brown, 1956, 1958; Foote, 1916), as well as for all the main mammalian lineages including the most diverse of them, Rodentia (Currey et al., 2017; Enlow, 1962; Enlow & Brown, 1958; Felder et al., 2017); Foote, 1916; Jaworski, 1992; Jowsey, 1966; Locke, 2004; Montoya‐Sanhueza, 2010; Ruth, 1953; Singh & Gunberg, 1971; Singh et al., 1974; Straehl et al., 2013.

The functions of bone remodeling have been extensively studied and can be grouped into two main roles, which are not mutually exclusive: (i) a biomechanical role, as a process of self‐repair responsible for the removing of microdamage accumulated in bone material and which is subsequently replaced with new material to avoid fatigue fracture (Currey, 2002; Currey et al., 2017; Frost, 1987a; Pearson & Lieberman, 2004); and (ii) as a metabolic regulator of the skeleton to assist with mineralization levels and calcium homeostasis (Amprino, 1948; Currey, 2002; Doherty et al., 2015; Parfitt, 2010). An additional function of SOs has also been associated with the development of bone tuberosities and sites of muscle attachment during the process of bone growth and relocation (Enlow, 1963; McFarlin et al, 2008; Parfitt, 2010). Thus, bone remodeling is a fundamental process for the postnatal development of the skeleton aiding in keeping material properties and balanced mineral homeostasis of the adult skeleton. Eventually, altering bone remodeling's normal function and activity may lead to several osteopathies including osteoporosis, whereby understanding the causes (etiology) and mechanisms of bone remodeling represents a major goal in bone biology research (e.g., Agarwal & Stout, 2003; Parfitt, 2010; Bonucci & Ballanti, 2014; Doherty et al., 2015; Currey et al., 2017; Piemontese et al., 2017).

Most of our knowledge on bone remodeling comes from investigations on laboratory rodents and humans, both of which exhibit a distinct pattern of remodeling. Humans develop highly remodeled bones forming dense Haversian tissue in adulthood (Cambra‐Moo et al., 2014; Goldman et al., 2009), whereas mice (Mus musculus) and rats (Rattus norvegicus) develop scarce and isolated SOs during their relatively shorter lives (Enlow & Brown, 1958; Forwood & Parker, 1986; Singh & Gunberg, 1971). Such disparate patterns have obscured our attempts to understand the functions of bone remodeling among mammals, probably because other factors such as body size and phylogenetic signal (which are still poorly understood) are also relevant in explaining this process. Recently, a series of studies described the complete lack of SOs and bone remodeling in the naked mole‐rat (NMR), Heterocephalus glaber (Currey et al., 2017; Carmeli‐Ligati et al., 2019), a small mammal that can live for up to 30 years in captivity and exhibits the longest lifespan of any rodent known to science (Dammann & Burda, 2007; Sherman & Jarvis, 2002). However, Montoya‐Sanhueza et al. (2020) reported and illustrated conclusive evidence for the development of SOs in H. glaber.

Montoya‐Sanhueza et al. (2020) also reported that bone remodeling comprised a limited process among subordinates and that SOs appeared mainly in the humerus and femur and to a lesser extent in the ulna and tibia, thus suggesting a differential pattern of bone remodeling between stylopodial and zeugopodial bones. Information documenting similar patterns in extant vertebrates is scarce and dispersed, for example, testudines (Bhat et al., 2019), equids (Nacarino‐Meneses et al., 2016), caprinids (Cambra‐Moo et al., 2015), and primates (Warshaw, 2008). Surprisingly, additional reports come from archeological and paleontological studies describing differences between bones in past human populations (Cho & Stout, 2011; Mulhern, 2000), as well as in a diverse range of extinct vertebrates such as non‐mammalian therapsids, equids, elephant birds, and dinosaurs (e.g., Chinsamy et al., 2020; Cullen et al., 2014; Martinez‐Maza et al., 2014; Padian et al., 2016). Despite the incidence of this phenomenon in a diverse group of vertebrates, there is a considerable gap in our understanding of the causes leading to differential skeletal homeostasis among these taxa.

In order to increase our knowledge on this process among small mammals, we assessed the extension and amount of bone remodeling (i.e., formation of secondary osteons) in the long bones of NMRs. We assessed prolonged in vivo bone labeling in non‐reproductive individuals, as well as analyzed a large sample of reproductive and non‐reproductive individuals to assess the effects of reproduction on their bone microstructure. It is known that mammalian reproduction encompasses strong metabolic effects on the endochondral and intramembranous osteogenesis of females (Miller & Bowman, 2004; Miller et al., 1986; Redd et al., 1984; Vajda et al., 2001). In NMRs, it has been reported that male and female subordinates are sexually monomorphic with no significant differences in skeletal phenotype (Pinto et al., 2010), although the cortical and trabecular bone structure as well as its bone quality are significantly lower in subordinate females than in female breeders (Dengler‐Crish & Catania, 2007; Pinto et al., 2010). Similarly, anabolic changes (bone formation) in trabecular bone (lumbar spine) of NMRs have been reported across pregnancy and lactation (Dengler‐Crish & Catania, 2007, 2009; Henry et al., 2007), while Pinto et al. (2010) also mentioned extensive catabolic activity (i.e., bone resorption) in females during late pregnancy and lactation. These studies demonstrate the female breeders of NMRs experience considerable changes in their skeletal homeostasis, although none of these studies have assessed the pattern of mineral mobilization, bone resorption, and bone remodeling at the microstructural level. Therefore, additional studies on the bone dynamics and bone histology of NMRs during reproduction are also needed. This study allowed us to assess the process of bone remodeling in NMRs, thus expanding our knowledge on the function and causes of this process in small mammals.

1.1. Naked mole‐rats: A novel model to study bone dynamics

NMRs are small (~34 g) subterranean mammals characterized by having slow somatic growth, low metabolic rates, and low basal body temperatures as compared to other mammals (Bennett et al., 1991; Buffenstein et al., 2020; Buffenstein & Yahav, 1991; Jarvis & Sherman, 2002; O’Riain, 1996; Šumbera, 2019; Zelová et al., 2007). They live in large eusocial colonies (averaging 78 individuals) typically composed of one breeding female (queen) and 1–2 reproductive males, while the rest of the colony members (subordinates) remain reproductively suppressed and therefore having a hypogonadic condition (Jarvis, 1981; Jarvis & Sherman, 2002; Sherman et al., 1992). They excavate extensive burrow systems in arid habitats with hard and solidified soils (Holtze et al., 2008; Jarvis & Sherman, 2002), principally aided with their chisel‐like incisors and secondarily by their fore‐ and hindlimbs, so that their long bones are expected to experience high strains during tunnel excavation. However, several studies have revealed that despite these factors (i.e., energetically demanding foraging/locomotor strategy and hypogonadic condition of subordinates), their femora maintains high bone structure and mechanical properties for most of their lifespan (e.g., Carmeli‐Ligati et al., 2019; Edrey et al., 2011; Montoya‐Sanhueza et al., 2020; Pinto et al., 2010). This was also confirmed for the humerus, ulna, and tibia, thus evidencing a systemic functional adaptation to principally withstand the mechanical strains imposed during digging behavior (Montoya‐Sanhueza et al., 2020).

Additionally, the breeding female can have multiple litters in 1 year, comprising of up to 27 pups per litter (average of 12 young per litter) (Bennett et al., 1991; Hood et al., 2014; Jarvis, 1991). This suggests that the patterns of mineral mobilization during reproduction may considerably affect the skeletal homeostasis of female breeders in order to ensure sufficient mineral resources destined for skeletal development of young. Hood et al. (2014) found that the milk of the breeding female is rather diluted (i.e., having a high water content) in comparison to other rodents, although they have higher calcium‐to‐phosphorus (Ca: P) ratios relative to other rodent milks, probably associated with high levels of mineral mobilization from bones. For these reasons, NMRs represent a unique model to assess the mechanisms involved in the delayed senescence of their skeletal system, particularly their age‐related bone loss and reproductive‐related skeletal homeostasis (e.g., Buffenstein et al., 2012; O’Connor et al., 2002; Pinto et al., 2010).

2. MATERIAL AND METHODS

Reproductive (n = 6) and non‐reproductive (n = 64) adults of both sexes were studied (Table 1). Specimens were derived from captive colonies kept at the University of Cape Town (UCT), University of Pretoria (UP), and Northeast Ohio Medical University (NOMU). For complete ontogenetic information and housing details of individuals, see Montoya‐Sanhueza et al. (2020). Ethical approval for the specimens #156‐164 from the NEOMED Institutional Animal Care and Use Committee in accordance with the Guide for Care and Use of Laboratory Animals published by the National Institutes of Health. For further identification of the individuals of this study, we used the last three numbers of their ID codes (Table 1).

TABLE 1.

Adult naked mole‐rats analyzed in this study, indicating breeding status (breeder or subordinate), ID, institution, sex, age (years), body mass (BM) and femoral length (FL). The asterisk (*) indicates individuals injected with fluorochrome labels. Individuals with (+) symbol in BM column indicate specimens with incomplete BM (see text).

Colony status	ID GM	Institution	Sex	Age (years)	Body mass (BM)	Femoral length (FL)
Subordinate	JJ−008	UCT	F	~10	‐	16.17
Subordinate	JJ−009	UCT	M	‐	‐	15.57
Subordinate	1200‐048	UCT	M	‐	29.84	14.14
Subordinate	1200‐049	UCT	F	‐	19.82	13.86
Breeder	1200‐051	UCT	F	‐	22.41+	13.95
Subordinate	1200‐052	UCT	M	‐	16.37+	12.99
Subordinate	1200‐053	UCT	M	‐	32.14	13.97
Subordinate	1200‐054	UCT	F	‐	31.50	14.28
Subordinate	1200‐055	UCT	M	‐	26.82+	13.68
Subordinate	1200‐056	UCT	‐	‐	24.03+	14.57
Subordinate	1200‐057	UCT	M	‐	32.47	14.58
Subordinate	1200‐058	UCT	M	‐	30.24	13.98
Subordinate	1200‐059	UCT	F	‐	17.72	12.17
Subordinate	1200‐060	UCT	M	‐	35.81	14.51
Subordinate	1200‐061	UCT	M	‐	39.91	14.64
Subordinate	1200‐063	UCT	F	‐	22.85	12.89
Subordinate	1200‐064	UCT	F	‐	31.29	14.1
Subordinate	1200‐065	UCT	F	‐	32.71	14.86
Subordinate	1200‐068	UCT	F	‐	23.45	13.97
Subordinate	1000‐070	UCT	F	‐	31.94	13.46
Subordinate	1000‐071	UCT	F	‐	34.58	14.77
Breeder	1000‐072	UCT	F	‐	26.77	14.41
Subordinate	1000‐073	UCT	M	‐	28.56	14.84
Subordinate	1000‐074	UCT	M	‐	26.44	14.16
Subordinate	1000‐075	UCT	‐	‐	23.95	14.61
Subordinate	1000‐076	UCT	F	‐	29.54	14.74
Subordinate	1000‐077	UCT	F	‐	44.58	15.75
Subordinate	1000‐078	UCT	F	‐	36.08	14.75
Subordinate	1000‐080	UCT	F	‐	41.44	15.3
Subordinate	1000‐081	UCT	M	‐	34.45	15.6
Subordinate	1000‐082	UCT	M	‐	43.97	16.02
Subordinate	1000‐083	UCT	F	‐	23.43	13.76
Subordinate	1000‐084	UCT	F	‐	34.49	15.2
Subordinate	1000‐085	UCT	M	‐	27.19	14.56
Subordinate	1000‐086	UCT	F	‐	28.72	14.01
Subordinate	5000‐087	UCT	M	‐	31.06	15.8
Breeder	5000‐089	UCT	F	‐	22.30+	15.2
Subordinate	5000‐090	UCT	F	‐	35.17	14.14
Subordinate	5000‐091	UCT	F	‐	28.51	13.7
Subordinate	5000‐092	UCT	M	‐	34.40	15.83
Subordinate	5000‐093	UCT	F	‐	30.40	13.76
Subordinate	5000‐094	UCT	M	‐	33.83	14.94
Subordinate	5000‐096	UCT	F	‐	24.63+	13.9
Subordinate	5000‐097	UCT	F	‐	31.13+	14.61
Subordinate	5000‐098	UCT	F	‐	29.36	14.46
Subordinate	5000‐099	UCT	M	‐	29.82	15.55
Subordinate	5000‐100	UCT	F	‐	21.64	13.79
Subordinate	5000‐101	UCT	M	‐	19.21+	13.03
Subordinate	NB−420	UP	M	‐	22.92	14.29
Breeder	NB−498	UP	F	‐	47.00	15.72
Subordinate	NB−499	UP	M	‐	66.00	15.88
Subordinate	NB−500	UP	‐	‐	17.93+	11.6
Subordinate	NB−501	UP	F	‐	63.00	16.47
Subordinate	NB−505	UP	M	‐	19.58	13.86
Breeder	NB−506*	UP	F	‐	16.04	14.72
Subordinate	NB−507*	UP	F	‐	41.00	14.32
Subordinate	NB−508*	UP	F	‐	35.00	13.86
Subordinate	NB−509*	UP	F	‐	39.00	14.48
Subordinate	NB−510*	UP	F	‐	30.00	12.97
Subordinate	NB−511*	UP	F	‐	33.00	14.43
Breeder	CDC−155	NOMU	F	2.83	56.60	15.2
Subordinate	CDC−156	NOMU	M	2.67	47.50	13.68
Subordinate	CDC−157	NOMU	M	2.67	49.20	14.31
Subordinate	CDC−158	NOMU	M	2.50	59.20	15.59
Subordinate	CDC−159	NOMU	M	2.83	34.10	12.31
Subordinate	CDC−160	NOMU	M	2.75	34.80	13.45
Subordinate	CDC−161	NOMU	M	2.17	39.60	14.57
Subordinate	CDC−162	NOMU	M	2.67	47.30	‐
Subordinate	CDC−163	NOMU	M	2.33	42.80	13.26
Subordinate	CDC−164	NOMU	M	2.33	45.00	13.77

Abbreviations: NOMU, Northeast Ohio Medical University; UCT, University of Cape Town; UP, University of Pretoria.

In order to illustrate the morphological variation within our sample, body mass (BM) and femoral length (FL) were measured. BM of specimens #155‐164 and #506‐511 were obtained immediately following death, while the BM of the rest of the specimens was obtained after defrosting them, so these data could be underestimated and therefore should be interpreted with caution (Table 1; Figure 1A). A standard electronic balance (0.01 g) was used to measure BM. Some specimens presented missing arms, legs, or open stomachs when collected from colonies, so these specimens were not included in Figure 1A. FL was measured using a Mitutoyo digital caliper (0.01 mm) and corresponds to the total length of the femur from its proximal articular surface to the distal one. This measure was obtained for almost all individuals and has been incorporated in previous studies of NMRs (e.g., Carmeli‐Ligati et al., 2019; Pinto et al., 2010), thus representing a good element for further comparisons. To visualize the distribution of body sizes within the sample, BM was plotted against FL using ordinary least square (OLS) regression (Figure 1A).

FIGURE 1.

Morphological features of adult naked mole‐rats. (A) Linear relationship between femoral length (FL) and body mass (BM) including non‐reproductive (subordinate) individuals and female breeders (queens). Females injected with fluorochromes (#506‐511) are indicated with a (+) symbol. Specimens of known‐age (#155‐164, between 2‐3 years) are enclosed within a polygon. The bones of the individuals showing double fusion of epiphyses are also indicated; non‐growing bone (nGB). Ordinary least square (OLS) parameters: R ² = 0.23, Slope =5.795, Intercept = −49.72 (n = 54). (B) Graph showing the changes in BM during the ~10 month in vivo bone labeling period. See the details of the fluorochrome experiment in Table 2. Abbreviations: alizarin red (Ar), calcein (Cn), femur (F), humerus (H), oxytetracycline (Ot), ulna (U), tibia (T)

Since the epiphyses of mice and rats typically remain unfused during their relatively short lives and even after somatic maturity is attained (Farnum, 2007; Geiger et al., 2014; Parfitt, 2002a; Roach et al., 2003), these are considered poor estimators of skeletal maturity among rodents. However, because the extreme longevity of NMRs, it would be interesting to assess if this species follows a similar pattern. More importantly, the estimation of epiphyseal fusion may contribute to the determination of endochondral (longitudinal) growth sequences and therefore potential patterns of interelement variation in this species. Both proximal and distal epiphyses were measured and classified into one of the three categories depending on the degree of epiphyseal fusion; non‐fused (NF), half‐fused (H), and fused (F) (Table S1). Thus, this parameter provided an estimate of growth plate closure and definitive cessation of endochondral ossification. When both epiphyses showed either a fused or half‐fused condition, this was referred to as double fusion and the bone was considered to have ceased its longitudinal growth and then classified as a non‐growing bone (nGB) (Table S1).

2.1. In vivo bone labeling

Nine randomly selected mature individuals from colonies kept at UP were injected with three different fluorochromes over a period of ~10 months (Table 2; Figure 1B). This sequence permitted tracking osteogenesis along almost an annual cycle, which to our knowledge represents the first study assessing long‐term bone dynamics in a small mammal. A previous study on a small primate, the grey mouse lemur, Microcebus murinus (60–80 g), assessed osteogenesis for a period of 20 days (Castanet et al., 2004). Rabey et al. (2015) assessed the effects of activity on bone osteogenesis in laboratory mice (~34 g) which were injected three times over a period of 28 days in an experiment that lasted 78 days. Individuals in our study were identified by using subcutaneous Passive Integrated Transponder (PIT) tags (except the queen #506). Of nine specimens tagged, four of them lost their PIT tags so those specimens were not retrieved.

TABLE 2.

Details of the in vivo bone labeling experiment in naked mole‐rats indicating body mass (g) at each injection, dosage and dates of administration and death (DOD). The observed labels (Obs.L) of Alizarin red (Ar) and Calcein (Cn) were also recorded for the humerus (H), ulna (U), femur (F) and tibia (T). Note that some specimens showed both fluorochromes while others only one. Oxytetracycline (Ot) was not recorded in individuals (see text). Fluorochrome injections were administrated every 6‐4 months over a period of 10 months. The queen (*) died before the finalization of the treatments.

Sex	ID	Body mass (g)			DOD Euthanization	Obs.L (H)	Obs.L (U)	Obs.L (F)	Obs.L (T)
		Injection 1 Ot	Injection 2 Ar	Injection 3 Cn
F	506*	26	‐	‐	11/06/2018	‐	‐	‐	‐
F	507	35	34	41	06/04/2018	Ar/Cn	Ar/Cn	Ar/Cn	Cn
F	508	35	32	35	06/04/2018	‐	‐	‐	‐
F	509	29	36	39	06/04/2018	Ar/Cn	‐	Ar/Cn	Cn
F	510	23	28	30	06/04/2018	Cn	Ar/Cn	Ar/Cn	Cn
F	511	27	31	33	06/04/2018	Ar/Cn	‐	Ar/Cn	Ar/Cn

Dosage (1.5 mg/kg) (80 mg/kg) (10 mg/kg)

Date of Injection 13/04/2017 25/10/2017 06/02/2018

The three fluorochromes were administrated intramuscularly in the leg after the 6^th and 4^th months from the first injection (Table 2; Figure 1B): Engemycin–Oxytetracycline (10%) (Ot), Alizarin (complexone) red (Ar), and Calcein (Cn), which, respectively, fluoresces yellow, red, and green (Table 2). Further details of the fluorochrome labeling procedure are presented in Table 2. Unfortunately, the queen died before injecting the last two fluorochromes. It is important to note that from the specimens successfully injected, none of them registered the first Ot label. It is likely that the low concentrations of this fluorochrome (Table 2) were not appropriately assimilated by the individuals. For this reason, the histological descriptions are focused only on the last two fluorochromes (Ar and Cn), thus covering a period of 7 months from the second injection (Ar) until the specimens were euthanized (Table 2). Fluorochrome labels were detected using confocal fluorescence microscopy (Zeiss LSM 880 Confocal—Fast AiryScan Technology) at the Confocal and Light Microscope Imaging Facility of the Faculty of Health Sciences at UCT. All experiments were approved by the Animal Ethics Committee of the University of Pretoria (AEC–UP: EC024‐17).

2.2. Osteohistological procedures and nomenclature

Undecalcified cross‐sections from the midshaft of the diaphysis, that is, ∼50% of the total bone length from the proximal articular surface were prepared at the Department of Biological Sciences at UCT. A total of 268 thin cross‐sections of 80–100 µm thickness were analyzed and high‐quality photomicrographs were taken with a Nikon Eclipse E200 Polarizing Microscope. The bone histomorphology described here was visualized using conventional transmitted light and polarized light microscopy with a gypsum (¼ lambda) filter and follows the nomenclature of Enlow (1963), Francillon‐Vieillot et al. (1990), de Ricqlès et al. (1991), Bromage et al. (2003), and Chinsamy‐Turan (2005, 2012a). Additional nomenclatural terminology regarding bone remodeling is provided in Section 2.3. Full description of the osteohistological procedures is also presented in Montoya‐Sanhueza et al. (2020).

2.3. Quantification of bone remodeling

The formation of secondary osteons (SO) in mammals is usually used to estimate their levels of bone remodeling and bone turnover (e.g., Forwood & Parker, 1986; Frost, 1987b; Ruth, 1953; Vajda et al., 1999). In larger mammals, these quantifications are based on the extension of bone surface covered by secondary bone (e.g., osteon population density, Frost, 1987b; Cho & Stout, 2011; Martinez‐Maza et al., 2014). Because SOs in subordinate NMRs are not highly abundant and they never form dense Haversian bone (Montoya‐Sanhueza et al., 2020), we quantified the total number of SOs per cross‐section in each bone element. This information allowed us to know which bones are most frequently remodeled, as well as compare the total density of SOs between bones.

The first phases of bone remodeling can be structurally identified by the formation of resorption cavities (RC), which occurs when osteoclasts remove both matrix and minerals from bone (Sissons et al., 1984). This cellular activity results in the formation of Howship's lacunae which are eroded surfaces with scalloped borders in the cortex (Sissons et al., 1984). Bone resorption is followed by osteoid formation, which is ultimately mineralized and consequently results in the formation of a SO. SOs can be recognized by a series of histological features including: (i) delimited margins or cement/reversal line; (ii) a Haversian canal; and (iii) concentric lamellae with osteocytes deposited around the Haversian canal (Currey, 2002; Jaworski, 1992; Martin et al., 1998; de Ricqlès et al., 1991). However, SOs exhibit high variation in their size, shape, and frequency among mammals. For example, Felder et al. (2017) recently reported that larger mammals have larger osteonal and Haversian canal areas as compared to smaller mammals (in absolute terms), although these parameters are smaller in larger mammals (in relative terms) and scale with negative allometry. This variation may be associated with the bone surface (endosteal or periosteal envelopes) and region (endocortical, intracortical, or pericortical) in which it develops (e.g., Kim et al., 2015; Thomas et al., 2005), as well as have an ontogenetic (Maggiano et al., 2016) and genetic basis (Havill et al., 2013). SOs can also exhibit incomplete remodeling associated with its development, that is, “secondary osteon's ontogeny” (Andersen et al., 2013; Jaworski, 1992; Kearns et al., 2008; Maggiano et al., 2016). Thus, one SO can contain both a resorptive surface and a mineralizing surface at the same time, and therefore its Haversian canal may appear not fully developed. This has been called as the reversal osteon, resorbing osteon, or simply immature osteon (e.g., Montoya‐Sanhueza & Chinsamy, 2017; Mori & Burr, 1993; Vajda et al., 1999). Given the morphological variation found within a complete SO in mammals, this study does not make a distinction between specific morphologies and consider a SO as having a cement line surrounding recently formed bone and a Haversian canal or an eroded canal, both of which indicate centripetal bone formed around a vascular canal. For these reasons, to estimate the degree of bone remodeling in this study we described and quantified the development of mature and immature SOs, regardless of shape and location (e.g., Montoya‐Sanhueza & Chinsamy, 2018). Significantly large RCs undergoing secondary reconstruction were not considered as secondary osteons per se, but were also included in the qualitative descriptions since they represent extensive catabolic activity followed by more recent anabolic activity (Montoya‐Sanhueza & Chinsamy, 2018, and references therein).

3. RESULTS

The mean body mass (BM) of the injected subordinates at the end of the experimental period, excluding the breeding female (#506), was 35.60 g (n = 5), a value which is between the lightest (17.72 g) and heaviest (66 g) subordinates of the sample (Table 1; Figure 1A). BM increased during the ~10 months of the labeling experiment, except in the individual #508, which maintained an unchanged BM (Table 2; Figure 1B). The mean femoral length (FL) of the injected specimens was 14.01 mm (n = 5), which is between the shorter (11.60 mm) and longest (16.47 mm) femora within the sample (Table 1; Figure 1A). The mean BM and FL of the rest of the subordinates (excluding known‐age specimens #156‐164) were 32.23 g (n = 27) and 14.30 mm (n = 30) for females and 33.26 g (n = 18) and 14.66 mm (n = 22) for males, respectively. In general, these data showed that adult subordinates varied considerably in BM and that the injected individuals (of intermediate size) were still growing (Figure 1B). The OLS between BM and FL presented a low coefficient of determination (R ² = 0.23, Figure 1A). Considerable variation in BM was still present (low R ² = 0.49), when only the subordinates from UCT colonies were plotted. In breeders, the mean BM and FL were 36.60 g (n = 4) and 14.87 mm (n = 6), respectively (Figure 1A). The min /max values for BM and FL of breeders were 16.04/56.60 and 13.95/15.72, respectively, thus also showing a high variation in BM (Table 1).

3.1. In vivo bone labeling

The five subordinates (#507‐511) injected with fluorochromes did not exhibit considerable differences in their bone matrix composition and/or main arrangement of bone tissues, and they did not differ considerably from the bone histology of the larger sample of subordinates.

3.1.1. Humerus

This bone showed thick cortical walls and varied degrees of intracortical resorption in the anterolateral side (Figure 2A). Fluorochrome labels indicating bone activity were few and appeared in the endocortical region, trabeculae, and to a lesser extent in the intracortical region (associated with osteonal formation), with no labels observed in the pericortical region (Table 2; Figure 2A). Thus, periosteal bone formation (growing in diameter) had apparently ceased during the experimental period (last 7 months) in this bone. The specimens #510 showed clear evidence of osteonal activity: the fluorochrome label was deposited centripetally in one side of a resorption cavity (RC) and its mineralized osteoid contained small osteocyte lacunae compared to the larger osteocyte lacunae of the surrounding lamellar bone (LB) matrix (Figure 2B). Most labels in the humerus were deposited within LB of endocortical regions and associated with resorption lines (RL), thus indicating secondary reconstruction of endosteally deposited tissues (Figure 2C).

FIGURE 2.

Fluorochrome bone labeling in the humerus of subordinate naked mole‐rats. (A) Distribution of double labels in endocortical and intracortical regions of four females indicated by red (Ar) and green (Cn) arrow‐heads. (B) Formation of a secondary osteon (SO) and detail of its Haversian canal (HC). The osteocyte lacunae (Os.Lc) of the SO are smaller than the Os.Lc of the surrounding lamellar bone (LB) matrix. Arrow‐heads indicate endosteal resorptive surfaces (lateral side). C) Fluorochrome labels were detected in areas of endosteal lamellar bone (ELB) (medial side). Abbreviations: anterior side (a), cement line (CL), isolated trabeculae (i.Tr), ateral side (la), medullary cavity (MC), mineralized osteoid (MO), resorption cavity (RC), resorption line (RL), trabeculae (Tr), vascular canal (VC)

One of the smaller specimens (#511) showed more endosteal activity as compared to the other specimens (Figure 2A, C). The bone histology of this specimen showed the clearest signs of cortical drift among all the labeled specimens, consisting of: (i) bone apposition in the endocortical region (medial side) and (ii) endosteal bone resorption in the opposite (lateral) side, thus indicating cortical drift toward the lateral side of the diaphysis.

In general, these data suggest that bone modeling in the humerus has mostly ceased and that most of the bone formation has been attained prior to the experimental period, although bone remodeling is still active at intracortical and endocortical surfaces. Additionally, the permanence of double labels in the trabeculae over a period of 7 months (Figure 1B) indicates low rates of bone resorption.

The quantification of SOs in subordinates showed that 50% of the humeri analyzed exhibited at least one SO (Table 3). The maximum number of SOs per individual was six, mostly associated with the anterolateral side of the humerus.

TABLE 3.

Details of the quantification of secondary osteons (SOs) in naked mole‐rats (subordinates and female breeders). Percentage (%) of individuals showing at least one SO, Mean of SOs per individual, minimum (min.) and maximum (max.) number of SOs per individual and total number of SOs quantified per bone element.

	Humerus	Ulna	Femur	Tibia
Subordinates
Subordinates (%)	50	5.36	14.52	4.92
Mean (SO)	1.23	0.09	0.16	0.05
max. (per ind.)	6	3	2	1
min. (per ind.)	0	0	0	0
Total (SO)	74	5	10	3
Breeders
Breeders (%)	20	0	20	0
Mean (SO)	3	‐	0	‐
max. (per ind.)	3	0	1	0
min. (per ind.)	0	0	0	0
Total (SO)	3	0	1	0

3.1.2. Ulna

In general, the ulna showed reduced intracortical and endosteal bone resorption, which often resulted in an almost completely occluded MC and a highly compacted cortex (Figure 3A). Only two individuals showed fluorochrome labels (Table 2), which were very limited in extension and mostly associated with endocortical LB (posterior side) (Figure 3A). Only one specimen (#507) showed a periosteal label on the lateral side, indicating cortical drift toward this region. This indicates that periosteal bone growth has mostly ceased in the ulna of the injected individuals and that most of the bone formation was attained prior to the experimental period. The low resorptive and osteogenic activity of this bone contrasts with the relatively higher bone turnover of the humerus.

FIGURE 3.

Fluorochrome bone labeling in the ulna (A) and tibia (B) of subordinate naked mole‐rats. Bone labels were less evident in these bones and were mostly restricted to endocortical regions. Abbreviations: anterior (a), lateral (la), medial (me), posterior (p)

SOs in the ulna of subordinates were considerably lower than in the humerus and only 5.36% of the individuals exhibited at least one SO (Table 3). The maximum number of SOs per individual was three, mostly associated with the anterior side of the bone.

3.1.3. Femur

This bone showed thick cortical walls and some trabeculae in the endocortical region (Figure 4A). Fluorochrome labels appeared associated with the trabeculae and endocortical regions (medial and lateral sides), as well as in the pericortical region, at the tip of the lateral side (Table 2; Figure 4A, B). The few bone labels found in the pericortical region indicated that radial bone growth (i.e., bone modeling) was still occurring when fluorochromes were injected. The limited activity observed in the trabeculae and endocortical regions evidenced low remodeling activity (and slow bone turnover) of these surfaces, since the labels were maintained for the last 7 months (Table 2). Labels were associated with lines of arrested growth (LAGs) and LB (Figure 4B, D, E), indicating cyclical and slow bone deposition, respectively. The femur also showed clear evidence of osteonal remodeling: fluorochrome labels were deposited centripetally on one side of an eroded vascular canal (VC), containing mineralized osteoid with small osteocyte lacunae compared to the osteocyte lacunae of the surrounding LB matrix (Figure 4C). The femur of the specimen #511 showed more endosteal bone formation as compared to the rest of the specimens (Figure 4D), similar to the observations made for the humerus of the same individual (Figure 2C). In general, the bone histology of this and other specimens showed bone apposition in the lateral side, in endocortical regions of the anterior, medial, and sometimes lateral sides, as well as endosteal resorption in the posterior side (Figure 4B, D‐E). This indicated cortical drift toward the lateral and posterior side of the diaphysis.

FIGURE 4.

Fluorochrome bone labeling in the femur of subordinate naked mole‐rats. (A) Four females showing endocortical, intracortical and pericortical distribution of double labels. Periosteal labels were recorded mostly in the tip of the lateral side, where lines of arrested growth (LAGs) were also observed. (B) Detail of LAGs in the lateral tip of the femur (#510). A fluorochrome label (FL) of Alizarin red (Ar) was deposited along with this growth mark. C) Remodeling of an eroded vascular canal (VC) (endocortical region). D) Calcein (Cn) labels were deposited in endosteal lamellar bone (ELB) (lateral side). E) Double labels were deposited in ELB (medial side). Abbreviations: anterior side (a), cement line (CL), isolated trabeculae (i.Tr), lateral side (la), medullary cavity (MC), mineralized osteoid (MO), parallel‐fibered bone (PFB), osteocyte lacunae (Os.Lc), resorption cavity (RC), trabeculae (Tr)

The number of SOs quantified in the femur of subordinates was considerably lower as compared to the humerus. Only the 14.52% of the femora analyzed exhibited at least one SO (Table 3). The maximum number of SOs per individual was two, mostly associated with the lateral side of the bone.

3.1.4. Tibia

This bone exhibited a highly compacted bone with a small MC (Figure 3B). Fewer fluorochrome labels were observed in this bone as compared to the femur (Table 2). This indicates that most of the bone growth of the tibia occurred before the experimental period. This bone recorded mostly the last Cn label in endocortical regions, although the specimen #511 also exhibited a thin Ar label in a localized area of the pericortical region (Figure 3B). The minimal periosteal activity of this bone indicates that radial bone growth has apparently ceased during the experimental period. Likewise, the minimal presence of RCs in the injected specimens indicates that this bone experience a slower bone turnover as compared to the femur.

The tibia of subordinates showed the lowest quantity of SOs and only the 4.92% of the tibiae exhibited at least one SO (Table 3). The maximum number of SOs per individual was one.

3.2. Bone remodeling in breeders

All breeders (6) exhibited similar bone matrix composition and bone tissue distribution as compared to non‐breeding subordinates. However, breeders exhibited a wider variation in bone microanatomy, cortical porosity, and secondary reconstruction (bone remodeling) of endocortical and intracortical surfaces as compared to subordinates. This resulted in some individuals with considerable thinning of cortical walls, as well as in extreme trabecularization and enlargement of MCs (Figure 5A). Three breeders showed high levels of bone resorption and secondary reconstruction (i.e., #072, #089, and #506), not comparable to the observed histomorphological pattern of subordinates. The other breeders (#051, #498, and #155) showed scarce‐to‐moderate bone resorption. Among all the bones analyzed, the humerus was the most affected by remodeling, specifically the anterolateral region (e.g., #506, Figure 5B). This region is normally remodeled in subordinates (e.g., #507, #511; Figure 2A), although this process is accentuated in some breeders, reaching high trabecularization (Figure 5A). This was evidenced by extensive endosteal and intracortical resorption forming enlarged RCs, often followed by secondary reconstruction (Figure 5B). The ulna of breeders also showed endosteal resorption, but this bone is less altered as compared to the humerus. The femur exhibited considerable thinning of their cortical walls (Figure 5A, D). Although the femur usually lacks trabecular development in subordinates (e.g., #507; Figure 4A), some breeders showed active endosteal resorption (e.g., #498) to considerable trabecularization of the MC (e.g. #072; Figure 5A). In the tibia, increased intracortical resorption, endocortical trabecularization, and expansion of the MC were also observed, but this bone also appeared less altered than the femur. In general, apart from the endocortical and intracortical regions, the pericortical region did not exhibit considerably alterations in breeders.

FIGURE 5.

Bone remodeling in female breeders. (A) Humeral and femoral cross‐sections of reproductive females showing high variation in cortical microanatomy, e.g. relatively thinner cortical walls, increased endosteal bone resorption (‐) and larger medullary cavities (MC) as compared to subordinates (compare with Figures 2A, 4A). The femur of the specimen #072 developed highly trabecularized bone (TrB) in the MC. (B) Detail of the anterolateral side of the humerus (#506) showing enlarged resorption cavities (RC) with secondary reconstruction. Yellow arrow‐heads indicate resorption lines. (C) Medial side of the humerus (#506) showing detail of secondary osteons (SO). The cement line is indicated by yellow arrow‐heads. Note the different orientation and size of the osteocyte lacunae within the SO with respect to the osteocyte lacunae of the surrounding lamellar bone (LB) matrix. This indicates a change in bone tissue arrangement and therefore secondary centripetal infilling. Both the Haversian canal and the endosteal margin of the bone are under resorption (‐), which is evidenced by the presence of Howship's lacunae (HL). D) Detail of a SO formed in the posterior side of the femur (#506). Endosteal resorption (‐) is also observed. For a better visualization, this image was vertically inverted

The development of SOs in breeders followed a similar pattern as in subordinates, although these were present only in two individuals (Table 3). Thus, 20% of the humeri and femora analyzed exhibited at least one SO (Table 3). The ulna and tibia of breeders did not present SOs. The bones of the specimen #072 showed highly resorbed bone sections, so no SOs were discernible.

3.3. Epiphyseal fusion

Most individuals presented unfused epiphyses and only a few of them of different body sizes showed simultaneous double fusion (of both proximal and distal epiphyses), thus indicating that such bones have completely ceased their longitudinal growth (Figure 1A; Table S1). Figure 1A illustrates the dispersed distribution of the individuals presenting non‐growing bones (nGB). In general, the forelimb presented more cases of double fusion (14) than the hindlimb (8). In the forelimb, double fusion was highly frequent in the ulna (12 cases) and least frequent in the humerus (2 cases), whereas in the hindlimb, this was slightly more frequent in the femur (5 cases) than the tibia (3 cases). Some individuals of known age showed double fusion, the breeder of 2.83 years (humerus, ulna, and femur, #155) and two subordinates of 2.67 (ulna, #157) and 2.50 (femur and tibia, #158) years, thus indicating that cessation of longitudinal growth for these bones has already occurred before the third year of life, regardless of attainment of sexual maturity. None of the injected individuals presented double fusion. However, half of the breeders exhibited at least one double fusion of their long bones, predominantly the ulna.

4. DISCUSSION

In this study, we described the process of bone remodeling of a large sample of subordinate and reproductive individuals of Heterocephalus glaber. Prior to this study, varied information existed about the pattern of bone remodeling of this species, principally regarding the assumptions that this species exhibited either a high level of bone remodeling or a complete lack thereof. Similarly, the bone microstructure and pattern of bone remodeling of reproductive individuals remained unknown. The multidisciplinary analysis of a large number of individuals including in vivo bone labeling of subordinates, histomorphological assessment of bone remodeling, and analysis of epiphyseal fusion allowed us to present conclusive evidence for (i) the development of secondary osteons (SOs) in H. glaber (regardless of sex and sexual maturity) (Figure 2B), (ii) the level of bone remodeling in different bone elements, as well as (iii) the presence of extensive reproduction‐related secondary remodeling in female breeders (Figure 5B). Interspecific comparisons with extant rodent models as well as with extinct taxa allowed us to hypothesize about the role and ancestrality of this process within the therapsid‐mammalian lineage.

The development of secondary osteons in all bones analyzed demonstrated that bone remodeling is indeed quite limited in NMRs (Table 3), but not non‐existent as previously suggested (Carmeli‐Ligati et al., 2019; Currey et al., 2017). Based on the findings of this and a previous study (Montoya‐Sanhueza et al., 2020), we discard the claims that NMRs possess either many SOs or completely lacks them. The fact that a large number of individuals did not present SOs in their midshaft does not preclude them from developing them in other regions of their skeletons, especially in sites or bones typically associated with higher rates of bone turnover as compared to cortical regions, such as the metaphysis of long bones and vertebrae (Parfitt, 2002b).

In African mole‐rats, the development of SOs is not rare and it was initially demonstrated for long bones of the largest bathyergid, the solitary Cape dune mole‐rat, Bathyergus suillus (Montoya‐Sanhueza & Chinsamy, 2017, 2018), and recently confirmed for all the genera within Bathyergidae (Phiomorpha) (Montoya‐Sanhueza, 2020). The pattern of SO formation and cortical distribution in NMRs is similar to that described for B. suillus, which showed high variation in size and shape and were generally scarce and randomly distributed in intracortical and endocortical regions, usually associated with woven bone and compacted coarse cancellous bone (Montoya‐Sanhueza & Chinsamy, 2017). SOs are also present in the Cape porcupine, Hystrix africaeaustralis (Hystricidae) (Montoya‐Sanhueza, 2020), the largest rodent in Africa, which is a sister group of the clade including NMRs, conformed by Caviomorpha +Phiomorpha (Patterson & Upham, 2014; Upham & Patterson, 2015). However, as also reported for B. suillus and other rodents in general (Enlow & Brown, 1958; Jaworski, 1992; Montoya‐Sanhueza & Chinsamy, 2017), NMRs do not develop dense Haversian tissue.

In general, rodents develop few SOs in their bones (Enlow, 1963; Enlow & Brown, 1958; Forwood & Parker, 1986; García‐Martínez et al., 2011). However, they still exhibit typical responses to changing biomechanical and metabolic conditions, thus initiating localized bone remodeling during increased locomotor activity (e.g., Forwood & Parker, 1986; Bentolila et al. 1998), diet‐related mineral deficiency (e.g., Ruth, 1953), or lactation (Miller et al., 1986; Ruth, 1953). These data and the observations presented in our study provide additional support to the hypothesis that bone remodeling in small‐sized mammalian lineages such as Rodentia may have an important role in calcium regulation and mineral homeostasis. Currey et al. (2017) suggested that the reduced number of SOs in rodents and other small mammals and birds may be due to their short lifespans, which would impede its development. However, the existence of few SOs in the long‐lived NMR refutes this hypothesis, thus indicating that reduced bone remodeling in small mammals occurs regardless of lifespan constrains (Currey et al., 2017). The possibility that other constraints such as phylogeny and body size may have more important roles on this phenomenon needs further assessment (e.g., Currey et al., 2017; Jaworski, 1992).

Nonetheless, an important factor explaining this phenomenon can be associated with the simple fact that the regions usually undergoing bone remodeling in small mammals (i.e., endocortical regions) are rapidly obliterated during ontogeny thus impeding the record of this process in the adults, which also have thin cortical walls. Based on our findings, African mole‐rats maintain thick cortical walls over their lives (Montoya‐Sanhueza, 2020; Montoya‐Sanhueza & Chinsamy, 2017), thus allowing the observation of both ontogenetic and metabolic processes in their bone microstructure. Similarly, several extinct taxa including mammalian precursors and other non‐mammalian cynodonts exhibit thick cortical walls with predominance of bone remodeling occurring in their inner endocortical regions (e.g., Botha & Chinsamy, 2004; Chinsamy & Hurum, 2006; Botha‐Brink & Angielczyk, 2010; Ray & Chinsamy, 2004; Ray et al., 2004; Shelton & Sander, 2017). This indicates the ancestrality of the remodeling process and its regionalization within the therapsid lineage.

4.1. Scarce bone remodeling and low osteopenia in Heterocephalus glaber

Remodeling in subordinates was minimal and mostly associated with the development of a few SOs (1‐6) per bone (Table 3). These were usually rounded and relatively small, mainly associated with woven bone or lamellar bone matrices (Figure 2B). Moreover, the presence of incompletely formed SOs (e.g. Figure 4C) indicated different stages of osteon maturity. The assessment of in vivo bone labeling in subordinates also showed a pattern of reduced bone remodeling and slow bone turnover among mature individuals. Injected specimens exhibited limited bone formation and fluorochrome labels are maintained over an extended period (~7 months), thus suggesting a limited resorption of early deposited bone matrices (Figure 2, 3, 4). It is most likely that most of the cortical bone formation in these individuals occurred before the experimental period, when individuals weighed less than 29.8 g on average (Table 2) and that bone modeling in these bones has almost ceased completely, especially in zeugopodial elements. It is possible that previous osteogenic activity at the beginning of the experimental period may not have been recorded due to the inability of recording the first oxytetracycline injection. However, it is unlikely that the osteogenic activity occurring during this period was responsible for the formation of the entire cortex.

The scarce bone loss in the long bones of adult NMRs indicates that osteopenia (i.e., intrinsic age‐related bone loss) is quite limited, as also noted for other bathyergids (Montoya‐Sanhueza, 2020; Montoya‐Sanhueza & Chinsamy, 2017). This pattern of mineral homeostasis contrasts with the observations made on surface‐dwelling mammals, which show more pronounced bone loss with aging, especially from endocortical regions (e.g., Bonucci & Ballanti, 2014; Cerroni et al., 2000; Duque & Watanabe, 2011; Frost & Jee, 1992). It has been suggested that osteopenia is in part due to the inability of the skeletal system to respond to mechanical loading in skeletally mature animals (Pearson & Lieberman, 2004). However, studies in both animal models and humans suggest that mechanical loading minimizes the extent of endosteal bone loss associated with aging (e.g., Jones et al., 1977; Bassey & Ramsdale, 1994; Fehling et al., 1995; Honda et al., 2001; Lee & Lanyon, 2004; Peck & Stout, 2007, and references therein; Maggiano et al., 2011). Montoya‐Sanhueza et al. (2020) reported that a large part of the cortical bone of NMRs is composed of endosteal bone. It is probable that endosteal surfaces maintain a high biomechanical responsiveness during ontogeny and/or that the high amounts of endosteal bone formed in NMRs are more resilient to resorption during ontogeny due to the increased and sustained physical activity of their fossorial habits, even in captivity (Montoya‐Sanhueza, 2020). Our study provides histomorphological evidence to support these hypotheses: (i) endosteal surfaces were still active even when most modeling processes have dropped their major formative functions and (ii) trabeculae and endosteal margins were scarcely resorbed during long period. Thus, the low bone remodeling and bone turnover of subordinates help explaining the cellular mechanisms involved in the maintenance of bone structure and bone quality observed during their ontogeny (Carmeli‐Ligati et al., 2019; Montoya‐Sanhueza et al., 2020; Pinto et al., 2010). In general, the results of this study support a (positive) systemic regulation of the cortical thickening of long bones of NMRs, where endosteal and intracortical resorption are considerably down‐regulated, probably because a lifestyle with sustained mechanical loading. This ultimately minimizes the chances of increasing intracortical bone porosity and thus reducing fracture risks during digging behavior. This may ultimately represent a functional adaptation not only present in NMR but also in other subterranean and fossorial mammals.

4.2. Interelement Variation in Bone Remodeling

The quantification of bone remodeling of all main long bones of NMRs allowed us to determine the presence of interelement variation between stylopodial (humerus and femur) and zeugopodial bones (ulna and tibia) (Table 3). Several lines of evidence presented in this study suggest a differential pattern of bone remodeling and skeletal homeostasis among these bones, which is ultimately proposed to be associated with their patterns of vascularization, bone growth (modeling), and skeletal maturity.

The formation of secondary osteons was found more often in stylopodial bones than in zeugopodial bones (Table 3). Among all bones, the humerus was the most common bone developing SOs (77 in total), followed by the femur (11) and much less frequently the ulna (5) and tibia (3) (Table 3). This variation is not surprising since substantial skeletal heterogeneity has been previously reported in extant and extinct vertebrates, including differences in bone remodeling and microstructural properties among different skeletal elements (e.g., Amling et al., 1996; Bhat et al., 2019; Botha & Chinsamy, 2004; Cambra‐Moo et al., 2015; Chinsamy et al., 2020; Chinsamy & Warburton, 2020; Cullen et al., 2014; Goldstein, 1987; Martinez‐Maza et al., 2014; Nacarino‐Meneses et al., 2016; Padian et al., 2016; Parfitt, 2010; Peck & Stout, 2007; Ray & Chisanmy, 2004). However, considerable variation on methodological procedures, sampling methods, and taxonomic biases among these studies have greatly obscured the real nature of such patterns (see details in Cho & Stout, 2011; Padian et al., 2016). For this reason, a complete overview of the causes of such variations is out of the scope of the present study, and we rather focus on the most relevant aspects discussed in the literature.

In general, interelement variation of bone remodeling has been attributed to differing mechanical loading histories between bones, so that high strain magnitudes and frequencies are positively correlated with increased bone remodeling (Cambra‐Moo et al., 2015; Cho & Stout, 2011; Pearson & Lieberman, 2004). However, in humans and many vertebrates, bone remodeling also tends to be relatively accelerated in ribs, spine, and pelvis, which are areas with high bone turnover (Cho & Stout, 2011; Currey et al., 2017; Foote, 1916; Parfitt, 2002b) and not necessarily experiencing high biomechanical strains (Currey et al., 2017; Lad et al., 2016; McFarlin et al., 2008). Thus, the causes of bone remodeling are not conclusive.

In the case of NMRs, it is expected that forelimb bones would experience higher strains (especially bending) as compared to hindlimb bones due to their direct involvement in scratch‐digging behavior. This would explain the fact that the cumulative amount of bone remodeling is indeed higher in the forelimb (Table 3), although mostly concentrating in the humerus. The actual biomechanical loading histories of these bones, especially between humerus and ulna, are unknown for fossorial animals, making it difficult to estimate the real strains experienced during locomotion and digging behavior. Further studies assessing the normal and conditioned loading strains among these bones in NMRs may help explaining the relationship between mechanical load and bone remodeling.

Some authors have hypothesized that the lower remodeling activity of small mammals is determined by their small body size and thin cortical walls (i.e., reduced bone surface), which would not accommodate the cavities produced by the remodeling process, making them prone to failure (Felder et al., 2017; Currey et al., 2017). Thus, the fact that Haversian remodeling in NMRs occurred less frequently in smaller and slender bones (with comparatively smaller cross‐sectional areas) may represent an adaptation to reduce fracture risk by reducing the formation of too large resorption cavities. In this sense, it is likely that the reduction of bone remodeling in fossorial species such as NMRs that experience high biomechanical strains throughout life may be the result of selective pressures for a less variable skeletal homeostasis, particularly in the smaller bones of their skeletons. Nevertheless, it is important to note that the reduced formation of SOs in these smaller bones (ulna and tibia) may actually represent the lower incidence of both vascularization and resorption cavities of these bones as compared to the humerus and femur (see Montoya‐Sanhueza et al., 2020), rather than differences in bone size per se.

The better vascularization, high formation of resorption cavities, and higher occurrence of SOs in stylopodial elements (in comparison to zeugopodial elements) are all co‐variants associated with increased growth rates and fast bone turnover in the humerus and femur. This is in agreement with the fact that most of the osteogenic activity observed in the fluorochrome analysis was associated with humeri and femora (Table 2). Overall, these data indicate comparatively low rates of bone turnover for the ulna and tibia and therefore a more stable skeletal homeostasis for these bones as compared to the humerus and femur.

In developmental terms, this information suggests that zeugopodial bones may have also decreased (or ceased) their main modeling (growth) functions, while stylopodial bones may have still experienced some levels of bone modeling during the labeling period. This latter is supported by the periosteal bone formation observed in the lateral side of the femur (Figure 4B. Some authors have proposed that the later attainment of skeletal maturity of some bones (i.e. relative time of growth—heterochrony) may explain the patterns of interelement variation by accumulating differential degrees of remodeling throughout life (Cho & Stout, 2011; Mulhern, 2000). To explain the differences in amount of bone remodeling between the femora and ribs of human archeological populations, Mulhern (2000) proposed that if the femur takes a longer time to mature than the rib, fewer SOs would have accumulated at any given chronological age. More recently, Padian et al. (2016) suggested a similar hypothesis to address the fact that small bones of large and fast growing vertebrates such as dinosaurs accumulate higher levels of intracortical remodeling in comparison to large bones. Contrary to our findings, these latter authors found higher levels of bone remodeling in smaller bones, thus suggesting that the degree of bone remodeling in such animals is a function of different growth trajectories among elements and whole‐body metabolic rates, so that larger bones use their energy to growth instead of developing SOs (Padian et al., 2016).

The analysis of epiphyseal fusion in NMRs provided important information to support the hypothesis of an earlier attainment of skeletal maturity in zeugopodial bones. The pattern of endochondral ossification and growth plate closure demonstrated that the ulna accumulated a substantial number of cases of double fusion, having more non‐growing bones as compared to the other long bones of the same individual (Figure 1A; Table S1). Certainly, this indicates that the ulna fuse its epiphyses before the humerus and therefore stops growing (elongating) earlier. Contrarily, the humerus showed the lowest number of epiphyseal double fusions among all bones analyzed (Figure 1A; Table S1), thus suggesting that its proximal epiphysis may still be growing during late ontogeny and probably maintains a more variable and active skeletal homeostasis throughout life. Regarding the hindlimb, the tibia showed only slightly lower levels of double fusion than the femur (Table S1), so it appears that the hindlimb bones have more equal timing of bone elongation and radial growth as compared to forelimbs, thus enabling the continuation of growth for longer time as compared to the forelimb bones. At this respect, Pinto et al. (2010) showed significant changes in the femoral length (FL), cortical area, and cortical thickness of captive individuals of NMRs ranging between 2 and 15 years old. This suggests that the femur of NMRs may continue growing in specimens older than 2 years old. Likewise, unfused femoral epiphyses in 2‐year‐old specimens were reported by Edrey et al. (2011), thus indicating that at this age, growth plates were still active. Moreover, Montoya‐Sanhueza et al. (2020) reported that during the early postnatal ontogeny of NMRs, the zeugopodial bones showed increased bone thickening (and smaller medullary cavities) as compared to stylopodial bones of the same individual, thus indicating that intramembranous (radial) ossification in zeugopodial elements attained maturity earlier than stylopodial elements (i.e., ulna and tibia reach peak bone mass before the humerus and femur). These data suggest an earlier attainment of relative skeletal maturity of distal elements in NMRs.

Another factor associated with the increased bone turnover of stylopodial elements is the development of bony protuberances. SOs in the humerus were associated with sites for muscle attachment, specifically with the anterolateral side of the bone where the deltoid crest develops (Montoya‐Sanhueza, 2020). This region of the cortex (in bathyergids and other mammals) is usually well vascularized and exhibits high intracortical porosity during ontogeny (Chinsamy & Warburton, 2020; Enlow, 1962, 1963; Montoya‐Sanhueza, 2020; Montoya‐Sanhueza et al., 2020; Montoya‐Sanhueza & Chinsamy, 2017), thus indicating that the development of SOs is directly associated with regions of high vascularization and high bone turnover. Enlow (1962, 1963) described this pattern and suggested that the sequential resorptive and depositional activities of remodeling serve as a mechanism for the relocation of muscles and other soft tissue attachments along growing bone surfaces (McFarlin et al, 2008, and references therein). The humerus of fossorial species including bathyergids and other non‐fossorial mammals undergoes a dynamic process of bone modeling for the relocation of the deltoid and pectoral tuberosities/crests during bone growth (Chinsamy & Warburton, 2020; Montoya‐Sanhueza, 2020; Montoya‐Sanhueza & Chinsamy, 2017). Most importantly, the humerus of mammals experience variable degrees of torsion during its morphogenesis (Maggiano et al., 2015; Montoya‐Sanhueza & Chinsamy, 2017, and references therein). In this sense, Montoya‐Sanhueza and Chinsamy (2018) suggested that the intracortical porosity of the humerus of B. suillus is expected to be higher than the one quantified for the femur, regardless the gender of the individual. Although we did not quantify the intracortical porosity of different bones, the humerus of NMRs seems to follow a similar pattern to that observed in B. suillus. It is likely that these morphogenetic factors contribute substantially to the increased bone turnover of this element. However, because the considerably reduced size of the deltoid tuberosity in NMRs in comparison to other bathyergids (Montoya‐Sanhueza, 2020), it is likely that this high bone turnover is rather associated with the intrinsic rotation of the humerus in this species. Additionally, the femur does not undergo diaphyseal rotation and does not exhibit a protuberant tuberosity, although the presence of the third trochanter in its proximal region may be associated with the more frequent development of SOs in comparison to the tibia, although much less frequent than the humerus.

Consequently, there does not appear to be a single causal factor explaining our observations on interelement variation. However, the pattern of vascularization and differential timing of growth sequences (heterochrony) between bones may represent an important set of co‐variants determining the pattern of bone remodeling in NMRs and possibly in other animals (Cho & Stout, 2011; Mulhern, 2000; Padian et al., 2016). Additionally, the finding of high variability in the pattern of skeletal homeostasis among different elements of early mammals and their ancestors (e.g., Botha & Chinsamy, 2004; Ray & Chinsamy, 2004) evidences that bone remodeling and its interelement variation represent an ancestral feature among therapsids, probably associated with the diverse heterochronic patterns of growth and maturity of the skeletal system described for this lineage (Ray et al., 2004; Chinsamy & Hurum, 2006; Chinsamy‐Turan, 2012b; Huttenlocker & Jennifer Botha‐Brink, 2014; O’Meara & Asher, 2016).

This information points to the fact that the selection of bone elements for histological analysis may be highly relevant for further assessments of different aspects of the bone biology of vertebrates such as bone matrix maturation, early morphogenesis, skeletochronology, periodic osteogenesis, and biomechanical function (e.g., Cho & Stout, 2011; Padian et al., 2016). This is even more pertinent when assessing different aspects of the skeletal homeostasis such as sex‐related bone loss, age‐related bone loss, and reproduction‐related mineral mobilization (e.g., Doherty et al., 2015; Parfitt, 2010).

4.3. Bone remodeling in breeders: The effects of reproduction

High bone remodeling was detected in some female breeders, although other females did not present considerable changes in their bone microstructure (Figure 5). This indicated a high variation in bone remodeling among breeders. Pinto et al. (2010) briefly mentioned endocortical bone resorption in late pregnancy and during lactation in NMRs, although they did not provide histomorphological evidence for this process. Dengler‐Crish and Catania (2009) reported reproduction‐related bone formation during spine elongation at the onset of reproductive activity and pregnancy. These studies suggest a high level of variation in skeletal homeostasis in reproductive NMRs, which is probably associated with the multiphase nature of female's reproductive cycle which involves pregnancy, lactation, and postlactation. Mammalian females are known to undergo marked fluctuations of their skeletal homeostasis, such as high mineral imbalances and skeletal deterioration, especially during lactation so that they can meet the needs of fetal development (Redd et al., 1984; Kovacs & Kronenberg, 1997; Kovacs, 2005; Vajda et al., 2001; Cerroni et al., 2003; Miller & Bowman, 2004). These effects are augmented when females experience food resource limitation. Experimental studies have reported how Sprague–Dawley rats on a low‐calcium diet have increased intracortical remodeling and developed a great number of SOs during lactation as compared to controls with normal diets (Ross & Sumner, 2017; Ruth, 1953). These data demonstrate a clear increase in bone dynamics (bone turnover) associated with reproductive cycles and indicate that the processes of gain and loss of mineral content from female skeletons are temporally regulated. For example, anabolic activity has been reported during some phases of reproduction in mammals, specifically during postlactation (Miller & Bowman, 2004; Vajda et al., 2001). Thus, the variation in the remodeling pattern observed among breeders in this study may be associated with the specific reproductive stages that these females experienced. Because NMR queens often become pregnant again during lactation (Jarvis, 1991), they show a minimal postlactation period, which is usually destined for skeletal recuperation (Miller & Bowman, 2004). However, Dengler‐Crish and Catania (2009) suggested that the anabolic effects of new pregnancies may lessen the impact of bone loss that commonly occurs in lactating females. Unfortunately, no information on the specific reproductive stages that the breeders of this study were experiencing is available.

A similar generalized catabolic activity in the appendicular skeleton of females of the solitary B. suillus had been reported, most likely as an effect of reproduction (Montoya‐Sanhueza & Chinsamy, 2017). This demonstrate that NMRs and African mole‐rats may share a similar mechanism of mineral mobilization regulated by sex steroids during reproduction as compared to other mammals (Dengler‐Crish & Catania, 2009; Montoya‐Sanhueza & Chinsamy, 2018). Nevertheless, additional studies assessing the extension and magnitude of different reproductive phases, as well as its effects on the female skeleton are further required, especially considering that female breeders in NMRs can experience successive pregnancies over a short period.

Overall, it is likely that the increased remodeling observed in female breeders is associated with the increased metabolism that females experience during reproduction. Urison and Buffenstein (1995) showed that the metabolic rate of pregnant NMRs is 1.4‐fold higher as compared to subordinate individuals. Similar findings have been reported for the female breeders of the social Ansell's mole‐rat, F. anselli (Schielke et al., 2017). In this sense, increased metabolic rates may trigger not only skeletal anabolism but also bone remodeling, thus increasing the levels of bone turnover and mineral mobilization for quick release of calcium destined for pup development. Likewise, the scarce bone remodeling observed in subordinate NMRs may also be associated with their generalized lower metabolic rates (Montoya‐Sanhueza et al., 2020). This would be in agreement with recent evidence showing that bone tissue matrices in NMRs are predominantly characterized by slowly deposited bone tissues and scarce vascularization (Carmeli‐Ligati et al., 2019; Montoya‐Sanhueza et al., 2020; Montoya‐Sanhueza & Chinsamy, 2016).

5. CONCLUSIONS

In general, NMRs exhibited low levels of bone resorption and bone remodeling, most likely linked to their low vascularization, low metabolic rates, and generalized slow growth rates. The reduced activity of this process represents one of the cellular‐level mechanisms explaining the maintenance of high bone quality and bone structure during the ontogeny of NMRs. The low remodeling activity seems to represent a generalized process in AMs and probably other subterranean and fossorial mammals. Bone remodeling was mainly observed in stylopodial (larger) bones as compared to zeugopodial (smaller) bones, thus suggesting a higher degree of bone turnover in these larger elements. Zeugopodial bones appeared to have attained maturity before stylopodial elements, thus showing reduced cellular activity and bone remodeling in adults. It is suggested that higher levels of vascularization, longer modeling activity (delayed skeletal maturity), and the development of bone tuberosities of stylopodial bones are important determinants of the degree of bone remodeling. Nevertheless, bone remodeling increased in some productive females, thus following a similar pattern to that found in other mammals during reproduction. This was associated with a higher metabolism of breeders as compared to subordinates, so that increments in metabolic rate are linked to increased remodeling activity. Consequently, it is suggested that bone remodeling in NMRs appeared to be associated with activities involving high metabolism, such as relocation of diaphyseal structures and reproduction. This information represents an important aspect of the skeletal homeostasis of NMRs, which help understanding the tissue‐ and cellular‐level processes involved in the maintenance of high bone quality during a species with prolonged lifespan. Moreover, based on histomorphological comparisons with other mammals and extinct vertebrates such as non‐mammalian therapsids, the lineage that gave rise to mammals, it was observed that both the endocortical compartmentalization of bone remodeling and its interelement variability represent ancestral features, probably associated with the early evolution of differential patterns of bone (re)modeling among these extinct taxa.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTION

GMS and AC designed the study; AC supervised GMS's doctoral thesis and supported the experimental procedures; NB supported the experimental procedures; NB and CDC provided NMR specimens; MO and GMS quantified data and carried out bone labeling procedures; GMS analyzed data, prepared the manuscript, created figures, and acquired microscopy images; all authors read, edited, and approved the manuscript.

Supporting information

Table‐S1