Decreased Sensory Responses in Osteocalcin Null Mutant Mice Imply Neuropeptide Function

Abstract

Osteocalcin, the most abundant member of the family of extracellular mineral binding gamma-carboxyglutamic acid proteins is synthesized primarily by osteoblasts. Its affinity for calcium ions is believed to limit bone mineralization. Several of the numerous hormones that regulate synthesis of osteocalcin, including glucocorticoids and parathyroid hormone, are also affected by stressful stimuli that require energy for an appropriate response. Based on our observations of OC responding to stressful sensory stimuli, the expression of OC in mouse and rat sensory ganglia was confirmed. It was thus hypothesized that the behavioral responses of the OC knockout mouse to stressful sensory stimuli would be abnormal. To test this hypothesis, behaviors related to sensory aspects of the stress response were quantified in nine groups of mice, aged 4–14 months, comparing knockout with their wild-type counterparts in six distinctly different behavioral tests. Resulting data indicated the following statistically significant differences: open field grooming frequency following saline injection, wild-type > knockout; paw stimulation with Von Frey fibers, knockout < wild-type; balance beam, knockout mobility < WT; thermal sensitivity to heat (tail flick), knockout < wild-type; and cold, knockout < wild-type. Insignificant differences in hanging wire test indicate that these responses are unrelated to reduced muscle strength. Each of these disparate environmental stimuli provided data indicating alterations of responses in knockout mice that suggest participation of osteocalcin in transmission of information about those sensory stimuli.

Introduction

Osteocalcin (OC, also called bone Gla protein or BGP) is an extracellular calcium-binding protein synthesized primarily by bone forming osteoblasts. Vitamin K-dependent post-translational modification of three glutamic acid residues to gamma-carboxyglutamic acid (Gla) enable the specific binding of OC to the surface of the bone mineral crystal, potentially regulating mineralization in vivo (Hauschka et al. 1989; Ducy et al. 1996). Although most OC is integrated as a stable component of bone, about 10% of newly synthesized protein is released to circulation (pOC) (Hauschka et al. 1989) and is thought to reflect bone formation or turnover.

A regulatory role for pOC in extra-skeletal physiology was suggested by the stimulus specific alteration in pOC documented in humans (Fujita et al. 1999; Napal et al. 1993), monkeys (Hotchkiss et al. 1998), and rodents (Fujita et al. 1999; Hotchkiss et al. 1998; Napal et al. 1993; Patterson-Buckendahl et al. 1995, 1988, 2005, 2007). These stimuli include those that also increase synthesis and secretion of glucocorticoids (cortisol or corticosterone; CS) and the catecholamine neurotransmitters norepinephrine (NE) and epinephrine (Epi). For example, humans (Napal et al. 1993) or rats (Patterson-Buckendahl et al. 1988) experiencing anxiety related stimuli had both an increase in plasma cortisol and a decrease in pOC.

In contrast, severe acute stressors of the “fight-or-flight” variety, including foot restraint immobilization (Immo) activate the hypothalamic–pituitary–adrenal and sympathetic nervous system (SNS) resulting in the rapid increase in several plasma hormones, including pCS, NE, and Epi (Kvetnansky and Mikulaj 1970). Immo of rats also elicits an increase in pOC by up to 50% within 5 min (Patterson-Buckendahl et al. 1995). Humans viewing an intensely emotional video experienced a similar acutely elevated pOC that was accompanied by a decrease in plasma ionized calcium (pCa++) (Fujita et al. 1999). This suggested a possible link between OC and activation of stress hormone secretion that could also involve OC modulation of plasma calcium concentrations.

The observed changes in pOC related to physical and/or mental stimuli must be detected and transduced to physiological responses. The localization of OC protein in sensory ganglia of rats is consistent with the possibility that it functions as a hormone or neuropeptide to transmit information regarding such stimuli (Ichikawa et al. 1999a, 2005; Ichikawa and Sugimoto 2002). To confirm a hormonal function requires either a naturally occurring mutation or genetically modified model that lacks the molecule of interest. To date, there is no known naturally occurring vertebrate species, other than the cartilaginous fish, that does not make OC. Whereas humans have a single gene (Bglap) encoding OC (Puchacz et al. 1989), mice have a cluster of three highly homologous OC-related genes (Desbois et al. 1994). In mice, two of the OC genes, termed OG1 and OG2 or bglap1 and bglap2, are expressed primarily in bone, and both have been deleted from an OC null mutant strain (KO, OC−/−), removing all OC from bone and circulation (Desbois et al. 1994; Ducy et al. 1996). The third gene, termed osteocalcin-related gene (bglap-rs1), remains, expressed primarily in kidney and does not contribute to pOC. The OC KO mice, although generally phenotypically normal, have stronger bones of greater mineral density than their wild-type (WT) counterparts when fully mature (Ducy et al. 1996).

Based on our observations of OC stress responsiveness, we first confirmed the expression of OC in mouse and rat sensory ganglia. We then hypothesized that the behavioral responses of the OC knockout mouse to stressful sensory stimuli would be abnormal. To test this hypothesis, we compared the behavior of OC−/− mice and their wild-type C57BL/6 counterparts in a battery of tests as recommended for behavioral phenotyping of mutant mice (Crawley 1999; Crawley and Paylor 1997). These included the open field novel environment, mechanical allodynia, thermal sensitivity, and balance (proprioception). Compared to wild-type controls, OC knockout mice showed diminished responses to acute mildly stressful stimuli. We generated data in mice of differing ages to show that the observed differences in sensory responses were not age specific.

Materials and Methods

All protocols were reviewed and approved by the Rutgers Institutional Animal Care and Use Committee and were consistent with guidelines specified by the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Initial breeding stock of osteocalcin null mutant mice (OC−/−), from which OG1 and OG2 have been deleted, were a generous gift from Drs. Gerard Karsenty and Patricia Ducy, Baylor University Medical Center (Ducy et al. 1996). These mice, back-crossed to C57BL/6J for more than 15 generations, are considered to be pure C57BL/6 mice. The OC−/− mice males were mated at Rutgers with C57BL/6J females (Jackson Laboratories) to obtain heterozygous breeding stock maintained in the Rutgers laboratory animal facility. Descendants of those heterozygous pairs, genotyped according to standard protocols, provided OC+/+ and OC−/− subjects for the present experiments. Prior to experiments, animals were housed for at least one week in the vivarium of the Center of Alcohol Studies under controlled temperature (22 ± 2°C) and light (12 h dark/12 h light, lights on at 6 AM) conditions.

Gene Expression and Protein Content of Neural Tissues

Because detection of immunoreactive protein was previously reported in sensory ganglia (Ichikawa et al. 1999a), we obtained sensory ganglia from a retired breeder female rat and WT and KO mice to confirm OC gene expression. Total RNA was extracted with Qiagen RNeasy reagents and on-column DNase treatment to remove any remaining genomic DNA according to manufacturers instructions. RNA was converted to cDNA with Multi-Scribe reverse transcriptase reagents (Applied Biosystems). We determined expression of rat OC by semi-quantitative RT-PCR with probes from Applied Biosystems (Bglap, Rn01455285_g1). For determination of OC gene expression in mice, we designed custom primers and a FAM dye labeled probe specific to the OG1/2 exon 3&4 boundary sequence (Applied Biosystems).

• Forward primer: GGGCAATAAGGTAGTGAACAGACT

• Reverse primer: CAAGCAGGGTTAAGCTCACACT

• Reporter sequence: ACTGAGGCTCCAAGGTAG

For protein determinations, we obtained tissues including DRG and TG from male and female retired breeder rats and DRG, TG, kidney, lung, muscle, and femur midshaft from 4 mo old male (1 WT and 1 KO) for determination of OC protein. Animals were killed by CO₂ asphyxiation, and rapidly perfused transcardially with heparinized isotonic saline solution to remove blood. Ganglia and small amounts of tissues were removed and frozen at −20°C pending protein extraction. Tissues were homogenized in 100 μl of 0.1 N HCl, centrifuged and supernatants lyophilized to remove acid. Tissues were reconstituted in 100 μl of water and protein concentration was quantified by UV absorbance at 280 nm on a NanoDrop^™ spectrophotometer. Osteocalcin concentration was determined in rat extracts by the method or Patterson-Allen et al. (Patterson-Allen et al. 1982), for which the detection limit is <3 ng/ml, with inter and intra-assay variations less than 6%. Mouse OC concentrations were determined by the method of Gundberg et al. (1992) for which the detection limit is 2.2 ng/ml, and intra and inter-assay variations are less than 6.4 and 12%, respectively.

Behavioral Experiments

Open Field Behavior

A Plexiglas^™ hole-board box 40 × 40 × 30 cm with 2 cm holes drilled in the floor in alternate squares around the periphery was elevated 2.4 cm above the floor by resting the corners on rubber stoppers. Mice were placed in the center of the field at the start of each 10-min session. Input from a Canon digital video camera to ANY-maze^™ tracking software (Stoelting Corporation, Wood Dale, IL) monitored the mouse’s position and mobility. A trained observer positioned near one corner simultaneously used specific computer keys to enter rearing and grooming activity. The field was illuminated by an overhead floodlight positioned 166 cm above the floor of the cage providing an average light intensity of 908 lux. Between each test, the apparatus was cleaned and wiped with dilute acetic acid, then wiped with water and dried. Seven WT and seven KO male mice, approximately four months of age at the beginning of testing were observed in the hole-board arena. WT mice weighed an average of 28.31 ± 1.12 g and KO mice weighed 29 ± 1.36 g (not significantly different). Each mouse underwent three trials spaced 3 to 5 days apart: novel, repeat, and response to intraperitoneal injection of 150 μl of sterile saline immediately before placement into the arena.

Von Frey Test

Mechanosensory responses were tested in 9 KO and 9 WT matched littermate male pairs from our colony, aged 9 months, using the Touch Test Sensory Evaluator (Von Frey Fibers, Stoelting Co.; Wood Dale, Illinois). The subjects were habituated for 15 min before testing in separate chambers (6 × 6.3 × 6 cm) constructed of Plexiglas^™ walls, wire mesh floors (1/4 in. hardware cloth), and hinged mesh lids. A series of fibers designed to impose forces of 0.16, 0.4, 0.6, 1, 1.4, and 2.0 g was used manually to touch each hind-paw three times consecutively. Two trained individuals, blinded to the genotype of the subjects, observed each mouse and assigned a subjective value using the following numerical criteria: 0, no response; 0.5, lifting paw slowly, slightly, and 1–2 times during each series of three touches; 1, lifting paw every time touched; 1.5, lifting whole leg every time touched, tending to bite or lick paw; 2, flinching, jumping a little, and licking paw. Values were averaged if observers reported different numbers. Data presented for each mouse are the average of three trials spaced a minimum of 3 days apart.

Balance Beam Test

The ability of two groups of male mice, one aged 6 months (15 WT, 15 KO), and a second group aged approximately 14 months (11 WT, 12 KO) to balance on a PVC tube was evaluated. This test can be used to assess balance and motor control (Kalueff and Tuohimaa 2005). The tube, 3.3 cm diameter and 88 cm long, was suspended between two IV poles 40 cm above a rectangular containment structure. A lengthwise strip approximately 2 cm wide was lightly sanded with emery cloth to provide slight traction. Thin corrugated cardboard was placed on the floor of the structure to protect an animal from injury if it fell off the beam. A digital video camera placed above the beam allowed video input of locomotor behavior into ANY-maze^™ software for 300 s. Each mouse was placed in the center of the beam and the program started. Behaviors including looking over the side of the bar (head dips), grooming, rearing, and slipping were manually input to the computer by keyboard. The software automatically recorded distance traveled, speed, right and left end entries, and time immobile. If a mouse fell from the beam within 10 s, it was placed back onto the beam and the recording restarted. If it fell later during the test, the recording was stopped and the data recorded to that point were saved. The beam was cleaned after each test. The percentage of time each mouse was immobile was calculated to indicate overall locomotion activity on the bar.

Cold Plate Test

One week following the balance test, a Columbus Instruments Cold/Hot Plate Analgesiometer (20.32 cm², temperature range 5–79.9°C) was used to determine sensitivity of the same 6 mo old mice previously tested on the balance beam to a thermostatically controlled metal surface. The plate was enclosed by a covered transparent Plexiglas^™ box 22.9 cm high. Behaviors were digitally recorded by ANY-maze^™ software with manual keyboard input of paw licking or shaking, rearing, jumping, grooming, or latency to forepaw withdrawal during four 60 s trials spaced 3–4 days apart. The plate and surrounding box were cleaned with Clidox^®, wiped with a wet cloth and dried between animals. To minimize stress responses, the mice were first habituated to the apparatus during two trials during which normal behaviors were evaluated. Observations were repeated twice more with the plate cooled to 10 or 5°C and behaviors that appeared specific to the plate temperature were evaluated.

Tail-Flick Test

Thermal response to heat was quantified by the time the mouse took to lift its tail out of heated water, a method commonly used to test analgesic drug efficiency (Caterina et al. 2000; Ledent et al. 1999). Two groups of male mice, aged 6 (16 KO, 14 WT) and 11 (9 KO, 11 WT) months, were tested. One observer wrapped a mouse loosely in a soft cloth, leaving its tail exposed, and dipped its tail into a water bath for a maximum of 30 s. Response time was determined by a second observer with a stopwatch. To habituate the mouse to the handling and dipping procedure and control for possible stress responses, latency to respond to room temperature water was quantified, followed by a succession of water baths heated to 47.5, 49.5 (6 months old) or 51°C (11 months old).

Hanging Wire Test

The hanging wire test can be used to determine whether there is any effect of gene mutations on neuromuscular function (Gomez et al. 1997; Rafael et al. 2000). For this test, a 1.3 mm thick wire was fastened to two ring stands placed 48 cm apart and elevated 51 cm above a padded surface. Each animal was held by tail suspension above the wire and allowed to grasp the wire with its forepaws. It was then released and its ability to remain hanging onto the wire was monitored for up to 3 min by digital camera input to Any-maze software. In addition to recording falls, the software also recorded reaching the ends of the wire and immobility. Male mice approximately 4 months old (10 KO, 9 WT) were tested on two occasions 1 week apart. Data were analyzed by repeated measures ANOVA for effect of genotype and repetition.

Statistical Analysis

Statistical analysis for all experiments was performed with Statview^™ (SAS Inc., Cary, NC) for Windows using ANOVA or ANOVA with repeated measures as appropriate. Post hoc comparisons were made using Fisher’s Least Significant Difference (LSD). Data yielding p-values less than 0.05 were considered significantly different.

Results

Gene Expression and Protein Measurements

Because the presence of OC protein in rat sensory ganglia had been previously reported (Ichikawa et al. 2000; Ichikawa et al. 1999a), we first confirmed both gene expression and protein in rat ganglia. Qualitative determination of gene expression and protein concentration in rat ganglia are given in Table 1. We then took tissues from WT and KO mice to determine whether the OG1/2 genes were expressed in ganglia as well as bone and whether OC protein could be detected in non-neural tissues as well as in ganglia and bone. Results of gene expression in ganglia and bone of the mice are shown in Table 2. Tissue OC concentrations relative to soluble protein from selected mouse neural and non-neural tissues are given in Table 3. It should be noted that although the present data are from individual mice, we have measured gene expression or protein concentrations in mouse DRG and TG in numerous other experiments yielding similar results. We have also cultured isolated WT DRG and determined that the cells express OG1/2, thus confirming that expression detected in excised tissue was specifically of neural origin (unpublished results). Collectively, these results confirm that OC gene is expressed and transcribed to protein in ganglia of both species, but not in the KO mouse.

Table 1.

Osteocalcin gene expression and protein concentration in ganglia of adult rat

Tissue	Gene expression relative to Trigeminal Ganglia	Protein concentration, ng/mg protein
Trigeminal ganglia	1	Not determined
Dorsal root ganglia	0.1472	10

Each column represents tissues from one animal

Table 2.

Real-time RT-PCR analysis of bone and ganglia for expression of OG1&2 is expressed as fold difference relative to WT bone

Tissue	WT	KO
DRG	0.0015	Undetectable
TG	0.017	Undetectable
Lumbar vertebra	1.00	Undetectable

Table 3.

Osteocalcin, ng/mg protein, in selected tissues of perfused adult wild type and knockout mice

Tissue	Wild-type	Knockout
DRG	59	Undetectable
TG	42	Undetectable
Femur	214	Undetectable
Kidney	Undetectable	Undetectable
Muscle	Undetectable	Undetectable
Lung	Undetectable	Undetectable

Open Field Arena

Locomotion and behaviors of 7 KO and 7 WT from four litters of 4 month old mice were analyzed during three trials in an open field/hole board arena with digital video camera input and manual keying of rearing, grooming, and poking to ANY-maze^™ automated tracking software. Repeated measures ANOVA indicated non-significant trend for genotypic differences in total distance traveled by these mice (Fig. 1a) during 10-min trials (F[1,12] = 2.53; p = 0.138), significant habituation to the field by all mice (F[2,12] = 115; p < 0.0001), and a significant interaction between genotype and distance traveled (F[2,12] = 3.50, p < 0.05). During both the Novel trial and the Saline injection trial, KO mice traveled significantly farther than WT mice (p < 0.05). Repeated measures ANOVA of the number of grooming events over all three trials (Fig. 1b) showed a trend for genotypic difference (F[1,12] = 4.51; p = 0.055), a significant effect of repeated testing ((F[2,12] = 9.48; p < 0.001), and a non-significant trend for interaction (F[2,12] = 3.02; p = 0.068). Because injection of saline immediately before placing the animal into the arena constituted a potentially painful or irritating sensory stimulus that might be expected to alter behavior, we focused specifically on a comparison of the second (Habituated) and third (Saline injection) trials following their initial (Novel) exposure to the arena. Data indicated a significantly lower response to injection in KO mice, as indicated by their lower number of grooming events compared to WT counterparts (F[1,12] = 4.91, p < 0.05). The KO mice increased grooming activity by only 71% compared to a 360% increase in this behavior by WT mice (p < 0.01). Other behaviors (rearing and hole-poking) did not differ significantly (data not shown).

Fig. 1.

Distance traveled (a) and grooming behavior (b) by 7 WT and 7 KO male mice in the open field arena during three trials. Values are mean ± SEM. A significant genotypic difference within trial is indicated by *p < 0.05 and **p < 0.01

Von Frey Test

Sensitivity to touch of the hind paws is another mechanism that depends on the sensory nervous system, conveyed through lumbar dorsal root ganglia (Caterina et al. 2000). We quantified responses to pressure exerted by Von Frey fibers on hind paws of nine matched littermate male KO and WT pairs aged 9 months in three trials spaced at least 3 days apart (Fig. 2). Repeated measures ANOVA indicated a significant effect of genotype on response (F[1,16] = 4.529, p < 0.05), a significant effect of force (F[1,5] = 57.116, p < 0.001) and a significant interaction between force and genotype (F[1,5] = 4.103, p < 0.01). Post hoc analysis of individual force application indicated that WT mice had a significantly more frequent response than KO when 0.6 g of force (F[1,16] = 9.924, p = 0.0062) or 1.4 g of force (F[1,16] = 10.182, p = 0.0057) was applied.

Fig. 2.

Mean number of responses (±SEM) of nine matched littermate pairs of male WT and KO mice aged 9 months to hind paw touch using von Frey fibers of increasing force. Each mouse was touched three times with each fiber in each of three separate trial series. Repeated measures ANOVA indicated a significant genotypic difference in response to increasing force (p < 0.05). A significant genotypic difference within individual force measurement is indicated by *p < 0.05

Balance Beam

The ability to maneuver on a balance beam is dependent on sensing by proprioceptive neurons in the dorsal root ganglia. Two groups of mice, 6 and 14 months of age were observed on a balance beam. There were no differences in head dips, grooming, rearing, or slipping behaviors (data not shown). Because the amount of time each mouse spent on the beam before falling off varied, the amount of time it spent immobile was calculated as a percentage of the time spent on the beam. As shown in Fig. 3, both 6 and 14 month old mice differed significantly with regard to genotype. Six-month old KO mice spent 2.4 times longer (133 ± 33 vs. 55 ± 13 s) on the beam before falling (F[1,26] = 4.34, p < 0.05), but were immobile a greater percentage of their time on the beam than WT (Fig. 3a). Similarly, 14 months old KO mice spent about 60% longer than WT on the beam before falling (103 ± 24 vs. 64 ± 14 s); however, these data did not achieve significance. As a percentage of time on the beam, however, KO mice were immobile significantly longer than WT mice (F[1,22] = 13.5, p < 0.005) (Fig. 3b). In each group, the KO mice tended to freeze and did not negotiate the length of the beam as often as WT mice, suggesting deficient motor control.

Fig. 3.

Time spent motionless on the balance beam by 6 months old male mice (a), 14 mo old male mice (b). Numbers in parentheses in bars indicate number of mice per group. Values are mean ± SEM, expressed as a percent of total time on the beam. **Indicates significant genotypic difference of p < 0.01

Thermal Response—Cold

The same young mice that were tested previously on the balance beam were then evaluated for their responses to noxious cold. Mice were placed on a thermostatically controlled metal plate, beginning with two trials at room temperature to habituate the animals and to evaluate normal behavior in that environment. Their behavior was then documented with the plate chilled to 10 and 5°C. The characteristic behavior that differed from that displayed on the room temperature plate was to sit on their hindquarters and withdraw the forepaws toward the chest. When the plate was chilled to either 10 or 5°C, both genotypes responded by significantly increasing the time spent motionless (Fig. 4a). The WT and KO mice differed in the latency to withdraw the paws and remain motionless after being placed on the cold surface (Fig. 4b). At 10°C, KO mouse latency to respond was not significantly different from WT (F[1,23] = 0.427, p = 0.52). However, KO mice had a significantly slower response than WT at 5°C (F[1,25] = 5.8, p < 0.05) suggesting a higher threshold of the cold response in KO mice.

Fig. 4.

Behavior of 11 WT and 14 KO male mice aged 6 months when placed for 60 s on a metal plate at room temperature (H) or cooled to 10 or 5°C. a Time immobile, b Latency to forepaw withdrawal. Data are mean ± SEM. *, **, and ***Indicate significant response to cold surface within genotype of p < 0.05, <0.01 or p < 0.001. †Indicates significant effect of genotype, p < 0.05

Thermal Response—Heat

Two groups of mice, aged 7 and 11 months, were tested for response to heat by dipping their tails in water, beginning with room temperature to habituate them to the procedure. This was followed by successively warmer water at 47.5, 49.5, and 51°C to determine if there would be a threshold for divergence of response. There were no differences between genotypes at room temperature or at 47.5°C. As shown in Fig. 5, 7 month old KO mice showed a significantly greater latency than WT to remove their tails from 49.5° water (p < 0.05). These mice were not tested at 51° because their responses were too rapid to provide accurate timing; however, 11 month old mice showed a similar latency at the higher temperature of 51°C (p < 0.05). These data suggest an effect on the sensory reflex arc that stimulates tail flick at high temperatures.

Fig. 5.

Latency for mice to remove tails from a heated water bath. Mean ± SEM are for 14 WT and 16 KO male mice aged 7 mo tested at 49.5°C, and for 11 WT and 9 KO aged 11 mo at 51°C. *Indicates significant effect of genotype, p < 0.05

Hanging Wire Test

Because some of the responses we observed could have been due to impaired neuromuscular function, we assessed forelimb muscle strength by timing the ability to hang onto a wire before falling. As depicted in Fig. 6, there was no significant genotypic difference in ability to hang onto the wire (F[1,17] = 0.37, p = 0.55), nor was there any difference between the two trials (F[1,1] = 0.007, p = 0.92). Likewise, there were no differences in the ability to reach the ends of the wire or in the amount of time immobile (data not shown).

Fig. 6.

Ability of 9 WT and 10 KO male mice aged 4 months to hang from a wire by their forepaws. Mice were evaluated in two trials for a maximum of 3 min each, spaced 1 week apart. Values are mean ± SEM seconds until fall from the wire. There was no significant effect of genotype or repetition of the task

Discussion

An animal’s ability to obtain information about its surroundings is vital for its survival. Mechanical allodynia (paw sensitivity), light sensitivity, thermal sensitivity, and pain are ethologically relevant sensory modalities that can be accurately evaluated in laboratory tests in the mouse. These external stimuli are transduced through sensory ganglia, including dorsal root and trigeminal ganglia, and convey important information to the brain regarding the animal’s surroundings. Ichikawa and coworkers (1999a, b) found that the rat dorsal root ganglia, which provide sensory innervations throughout the body, and the trigeminal ganglion, which provide sensory input to the vibrissae, lips, and jaws, exhibit osteocalcin immunoreactivity. This finding was of interest because plasma OC levels responded to environmental stress and physical stimuli (Patterson-Buckendahl et al. 1995, 1988, 2001). We had previously hypothesized that OC is involved in the response to stressful stimuli (Patterson-Buckendahl et al. 1996).

An indication that OC may have more general sensory functions was found in the “Gene atlas of human protein-encoding transcriptome,” part of a broad investigation of mouse and human gene expression (Su et al. 2004). The array providing information on human bglap expression (available as GDS596/206956_at/BGLAP/Homo sapiens on the NCBI GEO website) shows greatest expression in sensory ganglia and in parts of the brain that involve interpretation of sensory input. Our present data are focused on the sensory ganglia, but we have done some preliminary investigation of gene expression in parts of the mouse brain and found results consistent with the human data represented in the Su submission. It is important to note that pancreas (Schwartz et al. 2011), adipocytes (Bartness and Bamshad 1998; Osaka et al. 1998), and testes (Sienkiewicz 2010) have significant interaction with dorsal root ganglia. These tissues in the mouse are reportedly regulated by OC that presumably originates from osteoclastic bone resorption (Ferron et al. 2008; Ferron et al. 2010; Lee et al. 2007; Oury et al. 2011). Is it possible that the OC comes from sensory output rather than from bone? The short time (seconds to minutes) required for responses to many of the sensory stimuli we have tested is inconsistent with the slower release by osteoclastic bone resorption, which takes hours to days (Oury et al. 2011; Ferron et al. 2010).

The initial descriptions of the OC−/− mice indicated that they were phenotypically normal, with the exception that as fully adult animals, they had stronger, more densely mineralized bone. This suggested that OC might be exerting autocrine inhibition of osteoblast function and thus bone mineralization. More recent data suggest that OC−/− mice may have characteristics of type 2 diabetes, including obesity and hyperglycemia (Lee et al. 2007), and lower male fertility (Oury et al. 2011); however, we have not observed these in the mice bred in our colony. Early casual observations of the behavior of OC−/− mice indicated that they might have some impairment of sensory responses, possibly related to the absence of OC. The present data on behavioral responses to a variety of sensory stimuli demonstrate that mice lacking the genes for osteocalcin differ from their wild-type counterparts and support a role of OC in sensory neuronal systems.

The background strain, C57BL/6 is known to be highly active, with a high degree of locomotion in an open field setting (Crawley 1999). There was increased locomotion by KO mice during the novel exposure and following the stressful stimulus of i.p. saline injection. This might be interpreted as a blunted response to the stressors. Grooming behavior during novel or habituation trials did not differ; however, when the mice were additionally stimulated by i.p. injection, grooming frequency by WT nearly tripled compared to the previous trial, whereas KO increased a non-significant 68%. This suggests that nociceptors in the skin of WT might be more sensitive to the injection stimulus than in KO, causing them to respond more vigorously. Painful or irritating stimuli, such as i.p. injection, are known to increase grooming behavior (Spruijt et al. 1992; van Erp et al. 1994).

Two additional experiments evaluated potential differences in mechanosensory responses. Von Frey filaments are widely used to evaluate the threshold of touch or pain response, especially in cases of suspected spinal or peripheral nerve damage. In mice this is evaluated by touching the glabrous surface of the hind paw with fibers of increasing force and watching for paw withdrawal or other responses that might indicate the animal felt irritation or even pain from the touch. Our KO mice exhibited a significantly lower number of responses, especially to fibers exerting forces from 0.16 to 2 g than did their WT littermates. This implies a higher mechanosensory threshold in the absence of OC in the KO mice.

Balance (proprioception) requires a complex integration of position, muscle stretch, and gravitational orientation, all transduced through sensory ganglia. Two groups of male WT and KO mice of different ages were evaluated for their ability to traverse a round plastic beam. In both cases, KO mice spent a significantly longer time on the beam, but also a greater percentage of that time immobile. This test was repeated with female mice and although time on the beam did not differ with genotype, the percent of time immobile was significantly greater for KO mice (data not shown). The immobility on the balance beam is attributed to impaired motor coordination, making them more cautious (Kalueff and Tuohimaa 2005). Had the mice moved out onto the beam, they might have fallen off more quickly. This seems a clear indication that some sensory mechanism is at work to prevent them from a potentially dangerous action. Although this test can also indicate increased anxiety, that is unlikely to be the cause of the impaired mobility here, because other tests of KO and WT mice commonly used to evaluate trait anxiety (elevated plus maze and light/dark box behavior) showed no genotypic differences in behavior (data not shown). Impairment of muscle strength also seems unlikely, because there was no genotypic difference in behavior during the hanging wire test (Fig. 6), which is commonly used to evaluate models of known muscle pathology such as muscular dystrophy (Gomez et al. 1997; Rafael et al. 2000; van Putten et al. 2010).

Three separate experiments documented genotypic differences in thermal responses. One tested the reflex arc that stimulates withdrawal of the tail from noxious heat, in this case hot water. Two groups of male mice of different ages indicated that KO mice were slower to respond to temperatures around 50°C, with an additional age-related decrease in sensitivity in the 11 mo old mice. The significantly elevated tail-flick latency in the KO mouse supports a role of DRG or spinal interneurons, rather than brain mechanisms in the phenotypic sensory deficits observed in the KO mouse. It is well established that tail-flick reflex arc is intrinsic to the spinal cord (Advokat 1989). Motoneuron deficits were also ruled out because OC is not present in the motoneurons (unpublished observation), and no differences were observed in the hanging wire test. Therefore a decreased response of the sensory neuron on the OC knockout mouse is sufficient to explain the elevated pain threshold. When other mice were exposed to a cold surface, behaviors differed with temperature. We had previously observed mice exhibiting the paw withdrawal behavior on the cold plate, but had not seen this described in the literature. The usual response to either hot or cold surface exposure is reported as latency to lift or shake a hind paw. To establish that the forepaw responses were specific to cold exposure, we compared behavior in the apparatus at room temperature. Neither paw withdrawal nor paw shake occurred under this condition, thus verifying that the responses we have reported are unique to cold exposure. The latency for paw withdrawal was greater for KO mice compared to WT mice at 5°C than at 10°C. This difference suggests the existence of a threshold for temperature sensing that discriminates between 5 and 10°C.

Many of the behaviors we observed are related to sensory input transduced at least in part by a transient receptor potential (TRP) mechanisms found either in dorsal root (thermal, balance, von Frey, or injection stimuli) or trigeminal ganglia (whisker sensation, facial pain). Among the most prominent thermally responsive channels is Trpv1 (VR1), which is sensitive to temperatures in excess of 42°C (Patapoutian et al. 2003). The tail-flick data suggested an age-related decrease in sensitivity that could involve VR1, which has been shown to change with age (Wang and Albers 2009). More extensive experiments focused on comparison of WT and KO DRG gene expression of OG1/2 and the thermal receptors in response to specific thermal challenges are clearly needed to determine potential interactions between OC and these receptors.

Results from Von Frey testing indicated a significant effect of genotype on another mechanosensory response (Fig. 2). Large neurons in dorsal root ganglia, known to express OC, also express parvalbumin and are considered muscular proprioceptors (Ichikawa et al. 1999a, b). Thus, the decreased sensitivity of the KO mice to Von Frey fibers suggests that OC may be directly or indirectly interrelated with mechanoreceptors as well as proprioceptors.

In summary, we have quantified behavioral and physiological differences in OC null mutant mice. Although OC is primarily known for its presumed skeletal function, we have demonstrated that it has a role in regulating sensory input and thus affecting behavior. Differences between mutant and wild-type mice occurred in open field tests, both hot and cold thermal responses, balance beam behavior, and Von Frey fiber testing. Furthermore, differences between KO and WT were observed in mice from 3 months to more than 14 months of age. Because behavior is ultimately related to the function of the nervous system, it will be important to investigate the extra-skeletal localization of OC and to test for potential receptors for OC in peripheral or central nervous system and for changes in those receptors in response to intense, prolonged or stressful sensory stimuli. We conclude that OC serves to increase responsiveness of specific sensory modalities in times of stress or discomfort, presumably through a neuronal mechanism.