Department of Restorative Dentistry and Biomaterial Sciences, Harvard School of Dental Medicine, 188 Longwood Ave., Boston, MA
Department of Vertebrate Zoology, Smithsonian Institution, 1000 Jefferson Drive SW, Washington, DC
Department of Organismic and Evolutionary Biology, Museum of Comparative Zoology, Harvard University, 26 Oxford Street, Cambridge, MA
Correspondence to: Martin T. Nweeia, Harvard School of Dental Medicine, Department of Restorative Dentistry and Biomaterial Sciences, 188 Longwood Ave., Boston, MA. E-mail: Martin_nweeia@hsdm.harvard.edu
The functional significance of the erupted narwhal tusk has been the subject of conjecture and theory since the writings of Albertus Magnus in 1495 (Magnus, 1495). It has been thought to serve as an acoustic probe (Best, 1981; Reeves and Mitchell, 1981), possibly associated with sound transmission (Ford and Fisher, 1978; Best, unpublished B.Sc. Thesis, University of British Columbia, 1972); a thermal regulator (Dow and Hollenberg, 1977); a swimming rudder (Kingsley and Ramsay, 1988); a breathing organ; a spear for hunting or finding food (Vibe, 1950; Harrison and King, 1965; Ellis, 1980; Bruemmer, 1993); an aggressive weapon in interspecific fighting (Brown, 1868; Beddard, 1900; Lowe, 1906; Geist et al., 1960) or self-defense against predators (Buckland, 1882; Gray, 1889: Freuchen, 1935); and a tool used for breaking the ice (Scoresby, 1820; Tomlin, 1967), digging (Freuchen, 1935; Pederson, 1960; Newman, 1971), or resting on ice (Porsild, 1918). Many studies describe the tusk as a secondary sexual characteristic used in aggressive encounters or intraspecific display (Lowe, 1906; Norman and Fraser, 1949; Miller, 1955; Silverman, 1979) associated with tusk fracture and head scarring (Porsild, 1922; Silverman and Dunbar, 1980), and to establish social hierarchy amongst males (Scoresby, 1820; Hartwig, 1874; Mansfield et al., 1975; Silverman and Dunbar, 1980; Gerson and Hickie, 1985).
Examining the narwhal tooth organ system through a multidisciplinary approach that combines studies of anatomy, morphology, histology, neurophysiology, genetics, and diet gives a more comprehensive view of its functional significance and highlights sensory ability as an added functional attribute.
Narwhal teeth have unusual anatomical features. Among them are (1) a sinistral spiral morphology (Worm, 1655; Scoresby, 1823); (2) an extreme degree of tooth asymmetry in males, with a single left tusk expression and embedded right tusk (Sonnini and Buffon, 1804; Home, 1813); (3) an extreme expression of sexual dimorphism, with the male having an erupted left canine tusk reaching 2.6 m and the female commonly with right and left embedded tusks, often less than 33 cm (Sonnini and Buffon, 1804; Home, 1813); (4) a unique form of tooth asymmetry in a double-tusked expression, since the morphology of the spiral remains sinistral for both left and right antemeres, and the left tusk is often slightly longer than the right (Worm, 1655; Sonnini and Buffon, 1804; Home, 1813); (5) a horizontal direction of eruption in both erupted and unerupted tusk forms (Linné et al., 1792: Egede and Wood, 1818); and (6) perforation of the tooth through the upper lip (Hampe, 1737; Brisson, 1756; Crantz, 1767; Donndorff, 1792).
Though narwhal teeth share many anatomical characteristics with other tusked animals including a lack of enamel (Ishiyama, 1987), the presence of cementum, dentin, pulpal tissue and their associated structures (Seltzer and Bender, 2002; Berkovitz et al., 2002), and the presence of the maxillary division of the fifth cranial nerve associated with tooth innervation (Nweeia et al., 2009), they have many distinguishing features. Among them are a cementum layer overlying over the erupted canine (Nweeia et al., 2012) tusk, a patent network of dentinal tubules through the full thickness of dentin (Nweeia et al., 2009; Boyde, 1980; Locke, 2008), and pulpal soft tissues, extending the full length of the tooth, diminishing only in diameter with age (Pederson, 1931; Dow and Hollenberg, 1977; Nweeia et al., 2009). Initial scanning electron micrographs of the erupted male narwhal tusk reveal patent dentinal tubules that extend the full thickness of the dentin and correspond to lumina on the tusk surface. The tubules radiating outward from the dentin-pulpal wall are similar to those of humans and other mammals in diameter, though the spacing is three to five times wider. Limited scanning electron microscopy (SEM) has been completed for other odontocetes, but suggests dentinal tubules are well occluded within the erupted portion of the dentinal layer for most odontocetes (Boyde, 1980). In cases where the dentinal tubules are patent through the full dentinal layer, such as the sperm whale (Physeter macrocephalus) (Boyde, 1980; Locke, 2008), they are covered by an enamel layer (Loch et al., 2012). Dentinal tubules have been described for other marine mammals, including the rough-toothed dolphin (Steno bredanensis) (Miyazaki, 1977), pantropical spotted dolphin (Stenella attenuata) (Myrick, 1980), grey seal (Halichoerus grypus) (Hewer, 1964), hooded seal (Cystophora cristata) (Mohr, 1966) harbor porpoise (Phoecoena phoecoena) (Perrin and Myrick, 1980), short-beaked dolphin (Delphinus delphis) (Gurevich et al., 1980), bottlenose dolphin (Tursiops truncatus), and pilot whale (Globicephala melaena) (Boyde, 1980; Locke, 2008), and the walrus (Odobenus rosmarus), hippopotamus (Hippopotamus amphibious), and killer whale (Orcinus orca) (Locke, 2008). Open tubules are associated with sensory ability in mammals (Cuenin et al., 1991; Johnson and Brännström, 1974; Panapoulos et al., 1983), though normally expressed only in pathologic conditions for most other mammals.
Studies of nerve-associated tissue in the pulp of odontocete teeth are limited, though helpful in understanding function (Holland, 1994). Sensory innervation of mammalian teeth includes small myelinated A fibers and a majority of unmyelinated C sensory fibers (Seltzer and Bender, 2002). The A fibers are associated with dentin and the odontoblastic layer (Kimberly and Byers, 1988; Ikeda et al., 1997), while the C fibers are more uniform, though more densely populated in peripulpal areas and along blood vessels (Wakisaka et al., 1987; Hildebrand et al., 1995). Both A and C fibers contain the neuropeptide calcitonin gene-related peptide, or CGRP (Silverman and Kruger, 1987; Fristad et al., 1994). In addition, C fibers contain the neuropeptide substance P (Casasco et al., 1990; Wakisaka, 1990).
The presence of sensory-associated genes in the pulp is an indicator for sensory function in mammalian teeth. Though there is a lack of genomic information for the narwhal, techniques are available to measure the gene expression profiles using a universal array platform to de novo sequencing that are informative to identify sensory-associated genes associated with the pulp (Velculescu et al., 1995; Unneberg, 2003; Roth et al., 2004). Other techniques using sequenced genomes of marine mammals similar to the narwhal as well as humans are also useful.
Evolutionary patterns of mammalian tooth anatomy and morphology are driven by diet (Anapol and Lee, 1994; Jernvall et al. 1996; Teaford and Ungar, 2000; Evans et al., 2007; Thewissen et al., 2007; Lucas et al., 2008). Thus, comparing the foraging habits of male and female narwhals can potentially provide useful information about the erupted male tusk. Previous dietary studies comparing male and female narwhals from the Baffin Bay population have found no difference in stomach contents between the sexes (Finley and Gibb, 1982; Laidre and Heide-Jørgensen, 2005); however, smaller sample size of female narwhals has limited these analyses, and stomach contents can be biased since they only provide information on the most recent meal from a specific foraging area. Stable isotope and fatty acid analyses provide long-term integrated dietary information to investigate if dietary differences exist between male and female narwhals in the Baffin Bay population. Both techniques investigate chemical signals in animal tissues which have incorporated isotopic and fatty acid values from their prey over various time frames, depending on the tissue. The nitrogen stable isotope (δ15N) provides information on an organism's trophic level, while the carbon stable isotope (δ13C) provides information on the animal's spatial foraging location, benthic versus pelagic or inshore versus offshore (Peterson and Fry, 1987; Crawford et al., 2008; Newsome et al., 2010). Fatty acids are transferred relatively unmodified from prey tissues to predator tissues, and thus can also be used to determine and compare diet among groups of organisms (Iverson et al., 2004). Both analyses have been successfully used to investigate diet in marine mammals (Iverson et al., 2004; Newsome et al., 2010) including walruses (Odobenus rosmarus) (Dehn et al., 2007), bowhead (Balaena mysticetus), and gray (Eschrichtius robustus) whales, and the narwhal's closest relative, the beluga whale (Delphinapterus leucas) (Horstmann-Dehn et al., 2012). When the two analyses are used together and provide complementary results they can verify and confirm dietary interpretations, as was done in determining the primary prey for the beluga whale as arctic cod (Loseto et al., 2008, 2009) and for the bottlenose whale as Gonatus (Hooker et al., 2001).
Brännström's theory of dental sensitivity is the most widely accepted mammalian model to explain this sensory ability and function for teeth. Changes to the interstitial fluid flow within dentinal tubules changes the conformation of odontoblastic cells connected to a pulpal nerve plexus which sends signals of sensory perception to the brain (Brännström, 1966). This theory explains the known ability of teeth to sense environmental stimuli, and supports the evolutionary descriptions of tooth precursors as sensory organs (Lumsden, 1987). Human and other mammalian teeth are known to be capable of sensing external stimuli (Anderson, 1968; Haegerstam, 1976; Byers and Dong, 1983; Byers, 1984; Balam et al., 2005) such as temperature (Yamada et al., 1968; Jyväsjärvi and Kniffki, 1987; Ahn et al., 2012), pressure (Mengel et al., 1992), proprioception (Hassanali, 1997; Catania and Remple, 2002; Ozer et al., 2002), osmotic gradients, galvanic potential (Ramirez and Vanegas, 1989; Heyeraas et al., 1994), nocioception (Hu et al., 1978; Lisney, 1983; Narhi et al., 1984; Shigenaga et al., 1986; Byers et al., 1988; Iwata et al., 1998; Kawarada et al., 1999; Andrew and Matthews, 2000), and percussion (Ogawa et al., 2002; Watanabe et al., 2003). Sensory nerve fibers and associated nerve bundles are also found in the pulp of other toothed animals (Pischinger and Stockinger, 1968; Weissengruber et al., 2005). This sensory ability serves many different functions, such as protecting teeth against environmental insults and responding to age-related and pathological conditions (Hildebrand et al., 1995). Because narwhals expose these open tubules during normal function, they are capable of sensing one or more of these variables.
The erupted male narwhal tusk exhibits sensory ability based on physiologic response of changing heart rate associated with tusk exposure to alternating solutions of fresh and salt water, and the presence of sensory nerve-associated structures. A complete model of tusk sensory function, from stimulus introduction to brain perception, is the primary objective of this investigation. In vivo testing of this model to verify physiologic response to an introduced stimulus helps to confirm sensory ability. Demonstration of this model includes: (1) an anatomical description of open channels in a porous cementum layer covering the erupted tusk; (2) further and more complete imaging of a dentinal tubule network in various expressions of the narwhal tooth organ system, including the male erupted tusk, male and female unerupted tusks, and vestigial teeth in both sexes; (3) evidence of sensory-associated cells and tissue in the narwhal pulp and circumpulpal dentin; (4) detection of sensory-associated genes and antigens in the narwhal tusk pulp; (5) published findings of nerve connection at the base of the erupted tusk and continued connection to the maxillary division of the fifth cranial nerve to the brain; and (6) new results associating physiologic heart response in response to sensory stimuli of alternating fresh and high-salt water solutions introduced to the tusk.
MATERIALS AND METHODS
An intact, erupted tusk (“A”) including pulp tissue—taken from a male narwhal in 2003, 48 km from Pond Inlet, Baffin Island, in the eastern portion of Eclipse Sound, 24 km southwest of Cape Graham Moore, Bylot Island [72N 36′ 53.1″, 75W 37′ 56.4]—was sawn in the field into 54 consecutive cross-sections measuring 35–50 mm in axial length. Overall erupted tusk length was 218.4 cm, with diameter increasing from 14.8 mm at the tip to 57.4 mm at the base (Fig. 1A). Segments were fixed in 70% ethanol. Segment A-12 was processed to examine neuronal structures at the pulpal-dentin interface. The midpoint of segment A-12 was located 45.7 cm from the tusk tip, with a 2.78 cm mean outer diameter and 0.87 cm mean pulp diameter. Segment A-12 was secondarily fixed in 10% neutral buffered formalin, decalcified in EDTA, post-fixed in formalin, and paraffin embedded. Ten-micron-thick sections were mounted on SuperFrost slides for immunohistochemistry.
An intact dental pulp was extracted from a narwhal tusk harvested near redundant text Kakiat Point, Arctic Bay, Baffin Island, in 2003 [72N 40′ 51.2″, 86W 41′ 16.9″]. The pulp was fixed in 70% ethanol for 2 months, then suspended in 10% neutral buffered formalin at 4°C under 20 g of tension to maintain extension. Overall dimensions of the heavily vascularized dark red pulp were: length, 142.2 cm; diameters 0.3 cm (tip), 1.5 cm (mid), and 4.8 cm (base/apical tusk). The pulp base was bulbous and calcified. Ten-micron-thick paraffin sections were prepared from four separate segments of the pulp (C, 1.7 cm from pulp tip; F, 15.8 cm; I, 51.4 cm; and L, 90.8 cm). Segment L represented the pulp tissue at the approximate base of the erupted tusk that would be exposed to seawater (Fig. 1B). A fetus was also collected and head dissected into separate maxilla and mandible sections and stored in RNAlater® solution (Life Technologies, Carlsbad, CA) and kept in ice until laboratory arrival. Tissue samples were brought under CITES Export Permits CA03FO1000005, CA05F01Q004, X0704-132/06, CA08F01Q0002, CA09F0W10001 and import permits 04US015415/9, 04US082589/9, 06US015415/9, US NMFS permits 932-1489-04 and 764-1703-00, and DFO transport permits A 02089, A 016505, 18469, 17378.
Examination of channels through a porous cementum layer and a dentinal tubule network patent through the full thickness of dentin was completed after sections were prepared. Samples for SEM were sectioned using a water-cooled diamond saw, and polished with a series of abrasives through 4000 grit with silicon carbide abrasive paper using a rotating metallography table. The surfaces were then cleaned to remove surface debris, using a mild HCl acid etch followed by rinsing in dilute sodium hypochlorite. The cleaned sections were then washed and dried in a vacuum desiccator and gold-sputter coated for SEM observation. Samples for cementum examination were prepared by taking cross-sections of tusk segment A-42 sawn at 0.9-mm thickness with a precision low-speed water-cooled saw. Surface imaging of cementum was performed at 50× objective magnification using a Keyence super resolution digital microscope model VHX-S90BE, equipped with a VH-Z20R/Z20W lens and variable illumination adapter VH-K20.
Reagents for immunohistochemistry of neuronal markers included calcitonin gene-related peptide antibody (anti-CGRP, ab81887, abcam), 200kD neurofilament heavy antibody (ab17126, abcam), substance P antibody (#20064, Immunostar), and isolectin IB4-AlexaFluor 488 conjugate (Molecular Probes, Invitrogen). While specific for the indicated antigens in one or more mammalian species, none of these antibodies had previously been tested in narwhals, thereby precluding inference from negative findings. Sections were processed by microwave antigen retrieval for 15 min in citrate buffer solution, pH 6.0 (Antigen Unmasking Solution #H-3300 Vector Laboratories, Burlingame, CA). Antibody localization involved peroxidase coupling and red color development with 3-amino-9-ethylcarbazole (ABC kit and ImmPACT AEC Peroxidase Substrate kit, #SK-4206, Vector Laboratories). Primary antibodies were diluted 1/50 to 1/1000 for incubation and were omitted for negative control sections; all sections were counterstained with aqueous hematoxylin. Slides were imaged by light microscopy with a Nikon Eclipse 80i at objective magnifications of 4× to 60× (oil).
Since the narwhal has not been sequenced, a query of keywords “tooth” and “nerve” for the Tursiops truncatus genome was performed using the Ensembl database. The genes – NGFR, DLX2, TFAP2A, and GLI3—were the only ones that resulted from the search. Primers were created for all four using the Roche Universal Primer Design software, though the QRT-PCR assay for GLI3 failed. Another approach to identifying additional sensory genes was based on the human genome: FAM134B and WNK1 (formerly HSN2) genes have been shown to be involved in hereditary sensory neuropathy type II, which has a pathology for the reduction or loss of the sensory perception of pain, temperature and touch (Kurth, 2010). These genes were cross-referenced to the Tursiops truncatus genome, and to our surprise, based on the Ensembl database search, there was a sequence similarity for these two genes
The narwhal pulp tissue was stored in DNAgard (Biomatrica, San Diego, CA) and the control samples (muscle tissue around the blow-hole and gingiva from the inner portion of the mandible) were placed in RNAlater solution following the respective company protocol within 15 min of access to a fetus collected at Kakiat Point, then all samples were stored at −20°C until arrival at HC-LITT, where they were stored at −80°C. Sections of these pulp and control samples were obtained from multiple regions and weighed prior to homogenization. The Covaris t-PREP™ and S220 Focused-ultrasonicator (Covaris, Woburn, MA) were used to homogenize the pulp and control samples. Prior to Adaptive Focused Acoustics™ on the S220 instrument, 250 μL of Qiagen Buffer RLT (Qiagen, Valencia, CA) was added to the samples. After processing, 750 μL of TRIzol LS was added to extract the RNA from the samples using the standard three-step phenol/chloroform extraction method according to the manufacturer's instructions and prior to making cDNA. cDNA was synthesized from 1.5 µg of total RNA for each sample using Roche reverse transcriptase. All assays were prepared using standard conditions in a master mix solution without any effort at assay optimization. For each data point, there were three replicates.
The sensory and housekeeping genes were designed based on the Tursiops truncatus and Stenella coeruleoalba genomes, respectively. The Ensembl database (http://uswest.ensembl.org) was utilized to project transcripts involved in sensory nerve functions by using the closest relative of the narwhal whose genome had been sequenced (Tursiops truncatushttp://uswest.ensembl.org/Tursiops_truncatus/Info/Index [Database version: 71.1]) (Hubbard et al., 2002; Flicek et al., 2012). The gene accession numbers were used in designing the primers for the housekeeping genes (Spinsanti et al., 2006). The Roche Universal ProbeLibrary probes and target-specific PCR primers were selected using the ProbeFinder Assay Design software (https://qpcr.probefinder.com/organism.jsp, “Other organism” selection) (Rozen and Skaletsky, 2002; Kuo et al., 2006). Please refer to Table 1 for primer sequences.
The primers were used to confirm relative changes in mRNA levels by QRT-PCR using the Roche 480 LightCycler. Reactions were performed in microliter reaction volumes for six genes, using 1 μL of cDNA under the following conditions: 95°C for 10 min, followed by 45 cycles of 95°C for 10 s and 60°C for 30 s (Kuo et al., 2006). Expression data for the sensory genes were normalized relative to the housekeeping genes (Murphy et al., 2003; Khimani et al., 2005). Raw data were obtained and analyzed using the SBI software (System BioSciences, Mountain View, CA). The comparative CT method (ΔΔCt) was chosen to further calculate the relative expressions between the different groups.
Carbon (δ13C) and nitrogen (δ15N) stable isotope analyses were performed on male (n = 91) and female (n = 91) narwhal skin samples collected by Inuit hunters in the Canadian High Arctic from 1982 to 2009. Narwhal skin subsections of approximately 0.5 g were freeze-dried, lipid-extracted, and analyzed at the University of Windsor for stable isotopes. Burning of fossil fuels emits anthropogenic CO2 that is depleted in 13C, resulting in increasingly depleted C animal tissue signals over the industrial period: this pattern is known as the Suess effect (Sonnerup et al., 1999). Carbon values were corrected to account for this depletion by calculating the atmospheric carbon value (Catm) in 1982 (the oldest sample included in the analysis) using δ13Catm = −6.429 to 0.0060 exp (0.0217[t−1740]), where t is the year (Feng, 1998). Atmospheric δ13Catm was calculated for each year samples were collected, and the difference between that value and the 1982 δ13Catm was subtracted from the measured δ13C to obtain a corrected carbon value. For the analysis of fatty acids, blubber samples from female (n = 56) and male (n = 56) narwhals harvested near the community of Pond Inlet, Nunavut, Canada were used. Lipids were extracted from narwhal blubber samples for fatty acid analysis using the modified Folch procedure (Folch et al., 1957; Budge et al., 2006) and converted to fatty acid methyl esters (FAME). FAME were analyzed using the Agilent Technologies 7890A GC system.
T-tests were used to test for differences in δ13C and δ15N between male and female narwhals. Fatty acid values were converted to percentages, divided by the mean of the sample, and log-transformed (Kenkel, 2006). A principal component analysis, which included 33 dietary fatty acids (Iverson et al., 2004), was followed by a multivariate ANOVA on the first five principal components (cumulatively describing over 70% of the data) to determine if males and females differed significantly in their fatty acid values. All statistical analyses were conducted using JMP 7.0.1.
Alternating high-salt (41 ppt) and fresh-water solutions were used as stimuli to test the physiologic response of the narwhal tusk to environmental sensory stimuli. Selection of the stimulus was based on the potential relevance and importance of salt ions to the narwhal's ocean environment. A tusk jacket was constructed from a 10.2 cm diameter by 45.0 cm length clear acrylic tube with a 0.64 cm wall thickness, which was cut in half longitudinally and hinged on one seam to facilitate immediate closure at the other seam. At each end a 5.7 cm wide by 3.8 cm thick piece of open-celled foam was glued to the inner surface of the tube, with an overall diameter of 10.2 cm and an inner open diameter of 2.8 cm, to form a centering gasket around the tusk at the tube ends. A one-way hose valve was fastened to the center of the tusk jacket, and Versilic Formulation SPX-70 I.B. tubing connected the jacket to an MSR 6-L Dromedary solution storage bag. Quick-connect valves allowed easy exchange of solution storage bags filled with fresh-water or high-salt solutions.
The Midmark Holter IQ electrocardiograph (ECG) with a five-lead montage and eight-second data averaging was used for the six male narwhal samples tested in Tremblay Sound (TS) near the base of the Alpha River outside Pond Inlet [72N 21′ 38″, 81W 05′ 59″], two each in years 2010, 2011, and 2012. Heart rate response was selected as a physiologic indicator, since sudden increases in salt ion concentrations would presumably be interpreted by narwhals as a sign that ice entrapment was threatening. Thus, during the introduction of high-salt solutions narwhal heart rate was expected to rise due to physiologic response to rapid ice formation; conversely, heart rate was expected to fall during the introduction of fresh water, as this stimulus is associated with the waters of summer inlets where ice is melting and there are fresh-water run-offs.
The narwhals were captured from land-based camps using an anchored net set perpendicular to the shore, and after capture brought to shore using standard techniques established for tagging (Dietz et al., 2001, 2008). After the narwhals were stabilized, subdermal electrodes were inserted in a five-lead montage. Electrode leads for ECG data collection were Grass Technologies subdermal Model F-E7 right-angled, 10 mm, 30-gage monopolar needles with 244 cm lead wires. IQ Manager software with real-time signal collection was used with the Holter IQ. Eight-second interval averages were used in the final analysis. Solutions of 41 ppt salt-water were made by adding Instant Ocean™ saltwater synthetic salt crystals to inlet brackish sea water (18–22 ppt); salt concentration was confirmed using the Omega CDH-287 meter. Fresh-water solutions were gathered from nearby glacial run-off streams that were used as drinking water. McCormick red and green food coloring dyes were added (10 drops/gallon) to identify the two different solutions, and to verify the clearing of one solution in the tusk jacket and displacement by the alternating solution. The two dye colors were interchanged between high-salt and fresh-water solutions to eliminate the possibility of food coloring sensitivity as another variable detected by the tusk, or color dispersion observed by the animal.
The ECG was turned on after electrode attachment. After a few minutes of heart monitoring, the tusk jacket was attached at or near the base of the tusk, and alternating solutions of high-salt and fresh water were injected into the tusk jacket. Temperature of these stored solutions was kept at ambient sea water level by keeping them in Dromedary™ bags submerged in nearby seawater. Time periods for the solution injections were varied between animals to account for any effect on changes in heart rate response linked to capture time. Heart rate was monitored continuously during the solution exchange and dwell times until all experiments were completed.
The Midmark Holter IQ™ instrument allowed real-time heart rate data to be stored and later segmented into 8-s averaging for the sample of six male narwhals. The data recording was electronically marked at the beginning of each stimulus period to provide an accurate association of recorded heart rate to stimulus. The five-lead montage provided a manageable configuration of electrodes for field use, and very few interruptions in data were experienced due to displacement of the subdermal electrodes. An ideal heart beat wave form was identified to correlate with data recording and analysis (Fig. 2).
Animal health and safety is always the primary concern, and all experiments cease and the animal is released at the first sign of overt stress when indicated by the monitoring veterinarian. Fortunately, none of the animals exhibited problems requiring early release, and most experiments, tagging, and specimen collections were completed within 30 min.
Experiments and research were conducted under Canadian Department of Fisheries and Oceans' Animal Care Committee permits FWI-ACC-2008–2009-007 and FWI-ACC-2008-2009-008 and Animal Use Protocol permits FWI-ACC-2009-024, FWI-ACC-2010-008, FWI-ACC-2011-016, and FWI-ACC-2012-009. Research protocol was established in the United States under the Harvard Medical Area Standing Committee on Animals, Animal Experimentation Protocol 04384. The Hunters and Trappers Organization (HTO) in Pond Inlet, Nunavut voted to approve these research protocols in TS.
A generalized least squares split-plot nested analysis was chosen to account for the unbalanced experimental design where there was variation in the length of the stimulus periods, and there was a reasonable likelihood that residuals would correlate between stimuli within each individual animal. Statistical analysis on the six animals captured at TS was done with whales and water treatments as the primary fixed factors and split-plots nested within whales and water treatments, to determine parameter estimates and the significance levels of the fixed factors and factor interactions, fitting the model: µ + Wi + Tj + (WT)ij k + εijkl + γ(ij). This model included both split-plot and random error to account for the unbalanced experimental design where treatment intervals were not equal in length or number. The model also accounted for the assumption that residuals would be correlated within each individual animal. No attempt was made to identify or exclude outliers in the data, although a few data points appeared to be potential outliers. Tukey Honest Significant Difference (HSD) simultaneous confidence intervals were generated at 95% to evaluate the sizes of the differences among the whales' mean heart rates. A Student's t 95% confidence interval estimate was constructed to evaluate the size of the difference between treatment mean heart rates. Statistical analyses were conducted using JMP Pro 10.0.0.
Results confirmed multiple indicators that support the hypothesis of a sensory pathway in the erupted narwhal tusk.
Scanning electron micrographs of patent dentinal tubule networks provide evidence of the pathway for sensory conduction in multiple forms of narwhal tooth expression, including both male and female erupted tusks, unerupted tusks, and vestigial teeth (Fig. 3).
Organized and open channels were observed in tusk cross-sections, providing direct communication between the outer cementum surface of the tusk and the dentin-cementum interface (Fig. 4). The average channel diameter is 31 ± 14 (S.D.) μm, and the open areas comprise about 0.5% of the cementum. This sinusoidal channel network appears to surround bundles of Sharpey's fibers, and represents a possible pathway for sea water conduction through the cementum layer to the underlying dentin.
Putative nerve fibers in narwhal pulp and dentin were identified by immunohistochemical detection of neuronal markers. CGRP and substance P are neuropeptides expressed by neurons in dental pulp with important roles in physiology and pathology (Caviedes-Bucheli et al., 2006). CGRP+ fibers were found localized to peripheral pulp (Fig. 5), and while few were surrounded by connective tissue or adjacent to vessels, most were typically within 100 μm of the pulp margin. This outermost pulpal region borders the odontoblast layer. Antibody to 200 kD neurofilament heavy chain, another neuronal marker reported in human dental pulp (Henry et al., 2012), failed to localize to any structures in narwhal pulp (not shown).
Substance P was clearly present in the peripheral pulp (Fig. 5). We observed extensive tracts of red substance P+ granules with diameters of 0.5–2 μm (Fig. 6, panels A, B, D). Substance P+ granules were found in four pulp segments representing the major linear dimensions of the pulp from tip to base of the erupted tusk (C, 1.7 cm from pulp tip; F, 15.8 cm; I, 51.4 cm; and L, 90.8 cm). In all pulp segments (where cross-sectional dimensions ranged from 0.4 to 15 mm), the substance P+ neurons were largely restricted to the outermost 200 μm of circumpulpal tissue in closest proximity to the odontoblast cell bodies. Consistent staining of small discrete loci always at the pulp periphery and in the exact same positions through consecutive serial sections help to confirm the accuracy and precision of the technique. The granule distribution and size is suggestive of parallel, oriented dendrites or peripheral processes of sensory neurons, which are similar to the boutons en passant (beads on a string) morphology of substance P+ neurons in the CNS (Behan et al., 1993). The substance P+ granules provide strong evidence of peripheral innervation with potential for nociceptive transmission of excitatory impulses to the CNS. Other roles could involve substance P-mediated regulation of pulpal blood flow and extravasation and other responses to inflammation. Substance P, a peptide product of the Tachykinin 1/TAC1 gene, acts on its receptor NK1R (Tachykinin receptor 1/TACR1) expressed by neurons and endothelial cells. Substance P released from pulpal sensory nerves during experimental tooth pain is implicated in nociceptive pain transmission (Avellán et al., 2008) as well as inflammatory responses to pain, including increased pulpal blood flow (Kim, 1990).
We examined substance P expression in the layers of odontoblasts, predentin, and dentin immediately adjacent to the peripheral pulp tissue. Odontoblasts were not evident at the pulp surface (Figs. 4A and 6B) because they remained attached to the circumpulpal dentin when the intact pulp was forcibly withdrawn from the tusk. Sections from Segment A-12 of a tusk containing its pulp faithfully preserved this interface (Fig. 6D–F). There is evidence of substance P+ nerve fiber adjacent to odontoblast cell bodies, and further evidence of substance P+ fibers in close proximity to odontoblastic processes in the zone of dentinal tubules (Fig. 6E–F). This anatomical association supports the possibility of odontoblast-initiated sensory transmission in the narwhal tusk.
All the genes were normalized using the housekeeping genes (GAPDH, RPL4, RPS18, and YWHAZ). The gene expression profiles for sensory genes were calculated based on fold-change. DLX2, FAM134B, NGFR, and TFAP2A genes were highly expressed in the pulpal tissue when compared to both muscle and mandibular soft tissues (Fig. 7). When compared with pulpal tissue, the fold-change ranged from 1.64 (for FAM134B) to 21.03 (for DLX2), and three of the genes that had a fold-change greater than 16.5 when compared to muscle were highly expressed. Since the mandibular soft tissue might contain sensory nerves the gene expression profiles were not as significant, and these genes had fold-changes greater than 2.5. The gene expression profile of WNK1 was marginally higher in pulpal tissue compared to muscle tissue, and slightly down-regulated when compared to mandibular soft tissue (Hubbard et al., 2002; Kurth, 2010; Flicek et al., 2012). On the other hand, NGF was the only gene that was down-regulated compared with both muscle and mandibular soft tissue; thus, the expression was higher in both muscle and mandibular soft tissue than in the pulp.
Carbon and nitrogen stable isotope values and dietary fatty acid percentages are both biomarkers that provide an indication of diet. Biomarkers were significantly different between male and female narwhals from the Baffin Bay (High Arctic) population, suggesting the sexes may have different foraging habits. Carbon and nitrogen stable isotope data were normally distributed, as described by normal quantile plots, and variances were homogenous (Levene's test: Carbon, F1, 432 = 0.433, P = 0. 51; Nitrogen, F1, 432 = 0.28, P = 0.60). Male narwhals (n = 278) had significantly higher δ13C (t-test: t = 4.82, P < 0.0001) and δ15N (t-test: t = 3.5, P = 0.0004) values compared to female narwhals (n = 156). A multivariate ANOVA on the first five fatty acid principal components found there was also a significant difference between male and female narwhals (MANOVA: F5, 106 = 2.42, P = 0.041, n = 56). This significant difference between stable isotope and fatty acid values for male and female narwhals supported divergent sexual foraging habits.
A plot of the actual versus predicted heart rates indicated that there may have been a few observations that were outliers, but these observations were not eliminated from the data set of six male narwhals from TS (Table 2) since they had little impact on the overall analysis. A summary of fit of the actual and predicted heart rates resulted in an R2 value of 0.87. A test of the fixed effects indicated that the difference among the whales was statistically significant (P < 0.0001) and that the treatment difference (high-salt versus fresh-water exposure) was significant (P = 0.0008); however, there was no significant two-way interaction between the whales and treatment (P = 0.36). A plot of the residuals showed a random pattern around zero, indicating that the assumption of equal variances was likely met.
The Tukey HSD set of confidence intervals of the differences (Table 3, bottom) indicated that there was no difference in the estimated mean heart rates between whales 1 and 2, whales 1 and 3, and whales 2, 5 and 6 at the 95% confidence level. Whale 4 was significantly different from all others.
The least squares means differences in Student's t test of the salt and fresh-water treatment conditions (Table 3, top) resulted in an overall mean estimated heart rate of 60.42 bpm for the 41 ppt salt-water stimulus and an estimated 52.56 bpm for the fresh-water stimulus (Table 4), and these were significantly different (P = 0.0008). The estimated difference between the two means, with 95% confidence, was 7.86 bpm. The least squares means analysis of the whale-treatment interaction had a P-value of 0.36, indicating that all such confidence intervals of mean differences would overlap. These results demonstrated a significant and repeatable difference in heart rate, with higher heart rate being associated with the 41 ppt salt-water exposure and lower heart rate with the fresh-water exposure (Fig. 8). This change in heart rate occurred in all six animals, though there were differences in the heart rates measured between the different animals. The magnitude of change for the salt stimulus mean was between 3.7 and 12.0 beats per minute.
Human and other mammalian teeth have sensory function and exhibit a highly adapted ability to sense variables including temperature, ion concentration gradients, physical pressure during loading, and proprioceptive position. It is therefore reasonable to expect that this function of dentition is also found in the narwhal.
Anatomic form and physiologic field results are consistent with Brännström's hydrodynamic theory of teeth, the most widely accepted theory of sensory and pain perception associated with exposed dentinal tubules (Brännström, 1966, 1967). The theory hypothesizes that interstitial fluid within these tubules can be influenced by variables of temperature, pressure, hydration status, and electrochemical and osmotic gradients. Solutions introduced at the tusk surface have the effect of penetrating a porous cementum outer layer through a series of open and organized channels that create fluid flow within the dentinal tubules, outward during the exposure to high-salt solutions and inward after exposure to fresh water. Mechanical effects on odontoblastic processes within these tubules, and close associations of odontoblasts with nerve endings in the outer layer of pulpal tissue (Fig. 9), are involved in the initiation of sensory perception. Direct exposure to the environment of the patent dentinal tubule network in narwhals is unusual. In contrast, the beluga (Delphinapterus leucas)—the closest relative to the narwhal—exhibits the more characteristic occluded dentinal tubules within the dentin layer, preventing any fluid conduction from the enamel-covered tooth surface to the ocean environment (Sergeant, 1959 and 1973).
Table 1. QRT-PCR primer sequences
Primer sequences used in QRT-PCR experiments lists the Roche Universal Probe number and corresponding forward, reverse and amplicon sequences for the Universal Probe Library reactions. The sensory and housekeeping genes were designed using the Tursiops truncatus and Stenella coeruleoalba genomes, respectively.
Table 2. Gross anatomic measurements for the capture and released animals at TS
Table 3. Generalized least squares means differences 95%Tukey HSD of heart rates among whales
Least sq mean HR
Mean HR difference
Std Err difference
Lower CL HR
Upper CL HR
p - value
Table 4. Generalized least squares means differences student's t test of whale heart rates between high salt and fresh water stimuli: confidence interval at 95%
Least sq mean HR
Mean HR difference
Std Err difference
Tubules are present in all mammalian dentin and serve to supply the physiologic fluids that maintain the integrity of collagen that forms the collagen fibril network within dentin (Pashley, 1996; Kinney et al., 2003). It is well known that in human root-canal-treated teeth, where the pulpal tissue has been removed and thus the internal supply of physiologic fluid, the dentin becomes more brittle and loses elasticity over time (Jameson et al., 1993; Papa et al., 1994). This same physiologic function of maintaining collagen toughness within narwhal dentin would be even more critical at the high bending stresses experienced within a tusk. Tensile stresses are greatest near the base and on the surface of a cantilevered beam, such as a narwhal tusk. Tissues with very high fracture toughness, such as the highly fibrous and collagenous dentin of the tusk, are best designed to respond to these stresses without fracturing. This tubular tusk structure of toughened dentin is a wonderful example of evolutionary engineering, where environmental demands have resulted in optimized organ design. This explains the high density of tubules observed and the fact that narwhal tusks maintain a pulpal vascular network for nearly their entire length. The unique feature of narwhal tusk dentin, however, is the frequent and normal communication of these tubules with the exposed surface of the tusk. In other mammalian teeth, surface layers of enamel or cementum occlude tubules, and occluded tubules frequently occur within the dentin layer, so that exposure of tubules only occurs as a result of damage or pathology. The narwhal tusk has no surface enamel layer (Ishiyama, 1987), and the cementum observed on the surface is highly porous with a network of open and organized channels surrounding abundant Sharpey's fiber bundles of hypomineralized collagen, thus providing the patent communication needed to sense surface stimuli.
Dissection and imaging of male and female narwhal heads reveal connection of the maxillary division of the fifth cranial nerve from the brain to the tusk. This connection is supported by intracranial dissections, tusk alveolar examination, and visible pathways of nerve conduction observed by dissection and radiographic imaging (Tyler, 2006; Nweeia et al., 2009).
Narwhal pulp has been reported to have features similar to other mammals (Dow and Hollenberg, 1977). Nerve-associated tissue discovered in narwhal pulpal tissue (Figs. 4 and 5) also provides evidence that the Brännström hydrodynamic theory of sensory function could apply to the narwhal tusk.
Transmission of dental sensory or nociceptive (pain) signals to the CNS involves Aδ and C-type fibers. Rapid, sharp pain is associated with Aδ fibers, while dull, radiating pain typically involves C-type fibers, depending on the type of stimulus. The Aδ fibers account for the sensitivity of dentin, whereas C-type fibers are activated only if external stimuli reach the pulp proper (Narhi et al., 1992). Hydrodynamic sensory perception in the narwhal tusk is supported by our findings of extensive innervation and expression of substance P, a known nociceptive neurotransmitter, in the peripheral region of the pulp. The close apposition of odontoblasts and substance P+ fibers is consistent with odontoblast-dependent sensory transmission initiated by fluid flow in dentinal tubules. Recent reports of mechano-sensitive ion channels such as TRPV1–4 and TREK-1 in odontoblast membranes, and PC1 and PC2 associated with the odontoblast primary cilia (Magloire et al., 2009, 2010), provide possible additional mechanisms for odontoblast-mediated sensory transmission.
Significant presence and detection of a set of sensory-related genes in the pulp (including DLX2, FAM134B, NGFR, TFAP2A, and NGF), either highly expressed or less expressed relative to control samples of muscle and mandibular soft tissue, suggest a genetic basis for sensory function.
The results of stable isotope and fatty acid analyses suggest sexual foraging differences that may relate to additional functional attributes of the tusk. Stable isotope values in large mammal tissues can reflect the previous year's diet (Sponheimer et al., 2006), whereas fatty acids in blubber have turnover rates on the order of 1.5–3 months (Nordstrom et al., 2008). The majority of our samples were collected in July and August. Thus, fatty acid signatures would reflect dietary habits during the late spring to early summer period, which may be after the breeding season in either mid-April (Best and Fisher, 1974) or May-June (Heide-Jørgensen and Garde, 2011). Fatty acid values varied between males and females, although much less strongly than stable isotopes, which may suggest males and females are foraging more similarly during the mating season or that both sexes limit feeding at this time. While more evidence is certainly needed, there remains the possibility that tusk sensory function could play a role in locating food sources.
Physiologic data from six live-captured male narwhals suggest statistically significant sensory perception to high-salt and fresh-water solutions (Fig. 7). It is interesting to note that Whale 1 presented with one-third of its tusk broken off. Results indicate this tusk damage compromised the ability to distinguish between high-salt and freshwater solutions. Whale 2 also presented with a broken tusk, but only a minor 2–3 cm loss of length, and thus sensation of salinity was only slightly compromised when compared to Whales 3–6.
An established sensory pathway has thus been discovered from the external ocean environment through porous cementum, penetrating patent tubules conducting fluid shifts to sensory cells in the pulp, and transfer to the maxillary division of the fifth cranial nerve which connects to the base of the brain. This anatomic model (Fig. 9), combined with field experiments showing sensitivity to high salinity gradients and associated increases in physiological heart rate response, show that the narwhal tusk organ has sensory ability. This is an organ with the ability to sense characteristics of its external environment. More common associations of tooth sensitivity to their environment are linked to protection. Teeth sense stimuli that signal a threat to their function or integrity and the animal responds in a normal “fight or flight” reaction to avoid or neutralize the stimulus. Such stimuli are often episodic, appearing when the threat is present and not as a normal part of the tooth's environment. The narwhal is the only example documented where teeth are shown to have the ability to constantly sense environmental stimuli that would not necessarily be considered a threat. Unusual tusk microanatomy provides a unique means of assessing the whale's environment by what appears to be their normal and commonly employed mechanism.
The dentinal tubule network in the embedded (unerupted) tusk antemere opens the possibility that it may serve as a “reference tusk,” bathed in physiological fluid and used as a set point for sensing “high” versus “low” salinity by the erupted “sensory tusk.” Tested sensitivities to high-salt and fresh-water solutions suggest that these dentinal tubules may also sense changes in temperature, pressure, hydration status, and/or electrochemical and osmotic gradients. Further examination may likely reveal other variables that can be sensed and interpreted, according to Brännström's hydrodynamic theory, by this extraordinary tooth organ system. Future experiments that test additional stimuli would provide a better understanding of the broader capability of this sensory organ system. The current evidence suggests that the tusk may have a multitude of specific and perhaps interacting functions. Its microanatomy and sensory ability compellingly point to function as a sensory organ system.
Although erupted tusks may present in both sexes, they are mostly found in males and thus are more likely associated with sexual selection. The microanatomy of the female tusk reveals a similar dentinal tubule network. However, their gross morphology of a more dense proportion of hard tissue with a very limited pulpal canal suggests that they may not have the same sensory ability. Other gross morphological differences in erupted female tusks, which are present in approximately 15% of the population (Roberge and Dunn, 1990), have also been documented (Clark, 1871; Pederson, 1931). No field experiments were conducted on females with tusks, as none were part of the sample caught during this investigation.
The use of teeth in intrasexual selection or “male-to-male rivalry” and competition amongst odontocetes has been documented in the literature, and may have relevance to narwhal tusk function. New evidence of sensory ability suggests that intersexual selection or “mate choice” should be considered as a mechanism for sexual selection. Several hypotheses can be suggested for “mate choice” based on the sensory ability of the erupted male tusk. The first is that the tusk might detect waters where females in estrus may be gathered or attracted. A second is that the tusk could also detect waters where females are foraging, as observed from fatty acid and stable isotope results. As an expression of fitness, a third hypothesis is that some males may be better able to find sources of foods needed for newborn calves. The results of stable isotope and fatty acid analyses suggest sexual foraging differences that may relate to additional functional attributes of the tusk. Tusk sensory ability as a retained function from an evolutionary precursor is a fourth hypothesis; a fifth is that the tusk is a currently evolving sensory organ. Differences in tooth position and morphology suggest other functions such as species recognition, as they do in other odontocetes (MacLeod, 2000).
Since behaviors linked to mating are difficult to collect throughout the year, and a significant time period from late October to March can be forbidding due to weather and ocean conditions including moving pack ice within Baffin Bay in the early spring (Mansfield et al., 1975), observations of narwhal behavior will continue to be difficult to gather. Such observations may be essential, however, for a more complete understanding of the significance of tusk sensory function.
Thanks are extended to the National Science Foundation Offices of Polar Programs and Integrative and Organismal Systems, elders and hunters from the High Arctic communities of Nunavut, Canada and northwestern Greenland, and the Arctic field teams and The National Geographic Society.