Isotope analysis of human dental calculus δ13CO3 2−: Investigating a potential new proxy for sugar consumption

Rationale Dental calculus (mineralised dental plaque) is composed primarily of hydroxyapatite. We hypothesise that the carbonate component of dental calculus will reflect the isotopic composition of ingested simple carbohydrates. Therefore, dental calculus carbonates may be an indicator for sugar consumption, and an alternative to bone carbonate in isotopic palaeodiet studies. Methods We utilised Fourier transform infrared attenuated total reflectance analysis to characterise the composition and crystallisation of bone and dental calculus before isotope analysis of carbonate. Using a Sercon 20‐22 mass spectrometer coupled with a Sercon GSL sample preparation system and an IsoPrime 100 dual inlet mass spectrometer plus Multiprep device to measure carbon, we tested the potential of dental calculus carbonate to identify C4 resources in diet through analysis of δ13C values in paired bone, calculus and teeth mineral samples. Results The modern population shows higher δ13C values in all three tissue carbonates compared to both archaeological populations. Clear differences in dental calculus δ13C values are observed between the modern and archaeological individuals suggesting potential for utilising dental calculus in isotope palaeodiet studies. The offset between dental calculus and either bone or enamel carbonate δ13C values is large and consistent in direction, with no consistent offset between the δ13C values for the three tissues per individual. Conclusions Our results support dental calculus carbonate as a new biomaterial to identify C4 sugar through isotope analysis. Greater carbon fractionation in the mouth is likely due to the complex formation of dental calculus as a mineralized biofilm, which results in consistently high δ13C values compared to bone and enamel.


| INTRODUCTION
Investigations into past diets frequently draw upon direct isotopic measurements of the surviving body tissues of consumers. Bone and teeth most commonly survive in archaeological contexts but soft tissues (e.g. skin), hair and fingernails can also provide dietary information if they are preserved. Dental calculus (mineralised dental plaque) has recently received attention for its potential to reveal aspects of past diets, primarily through the analysis of biomoleculesproteins, DNA 1,2and micro-debris trapped within the mineral matrix. 3 Only a few studies have explored the potential for isotopic analysis of dental calculus but they have mainly analysed bulk carbon (δ 13 C) and nitrogen (δ 15 N) isotope compositions of the organic matrix with mixed results. [4][5][6][7][8] One recent study specifically targeted the inorganic fraction of calculus, suggesting that this component may have more validity as a palaeodiet proxy; however, the study was limited in terms of the methodology as well as the small number of samples analysed. 8 The present study goes further to explicitly explore the potential utility of δ 13 C in dental calculus mineral as a dietary proxy for identifying C 4 sugar (maize/cane). Building on previous studies applying Fourier transform infrared (FTIR) spectroscopy to characterise dental calculus composition, 9,10 we apply this analysis to systematically characterise the composition and crystallisation of dental calculus. This study presents the first analysis and interpretation of dental calculus FTIR results as an assessment of diagenesis in line with what has been previously undertaken for bone and enamel carbonate. It goes on to present the first comparison of dental calculus carbonate with both bone and enamel carbonate to determine whether dental calculus carbonate is a suitable substrate for identifying C 4 resource consumption. In addition, the results obtained here are used to determine if there is any intra-individual variation among tissues. This analysis is notable for analysing material from archaeological populations as well as modern individuals, with the latter rarely incorporated into archaeological studies.
The current understanding of diet in palaeodietary studies utilising stable isotopes (δ 13 C, δ 15 N) is mainly based on the analysis of bone and dentine collagen (δ 13 C, δ 15 N) as well as enamel and bone carbonate (δ 13 C). 11,12 Bone and dentine collagen δ 13 C values represent the protein sources in the diet, but a proportion of collagen can be synthesised from lipids or carbohydrates. Bone and enamel carbonate δ 13 C values, however, reflect whole diets including carbohydrates, lipids and proteins. 13 Palaeodietary reconstruction using carbon isotope values is useful in distinguishing between the consumption of C 3 (low δ 13 C value) and C 4 (high δ 13 C value) terrestrial resources as well as between marine (high δ 13 C value) and terrestrial (low δ 13 C value) foods. Nitrogen isotope values provide information on the main protein sources of diet, for example, differentiating between plant-rich protein and animal-rich protein diets. 14 Previously, C 4 cane sugar consumption has only been detected under specific circumstances, such as when sugarcane is an indigenous crop 15 or inferred where sugarcane may have been used as animal fodder, in which case the C 4 signal is acquired by herbivores. 16 Neither maize nor cane sugar would have been available to the medieval populations considered here. 17 The post-medieval period between the 17th and 19th centuries, however, saw an increase in the consumption of cane sugar in England due to Britain's colonisation of the West Indies in the 17th century. 18,19 Although maize was available in post-medieval England, historical sources indicate that it was considered to be suitable for animal fodder or famine relief food, particularly for the Irish poor. 20,21 However, a recent analysis of dental calculus micro-debris of 36 individuals from the middle-class Cross Street population revealed that two individuals had the presence of maize (not included in this study). 3 Maize is therefore another possible source of C 4 carbohydrate for the postmedieval population; however, sugar will have been more ubiquitous.
Although hypothesised in some bone collagen isotope studies of postmedieval populations in England, [22][23][24][25] the consumption of cane sugar has been hard to detect despite it being a key element of the dietary staple. In this study, it is hypothesised that dental calculus carbonate could offer a novel and more sensitive way of identifying cane sugar/ maize than the traditional isotopic methods focusing on bone.

| Dental calculus and potential for isotopic analysis
Calculus is frequently preserved in archaeological contexts and has been found to survive on the teeth of late Pliocene hominins 26 and Miocene apes dating up to 8 to 12 million years ago. 27 Dental calculus results from the calcification of plaque biofilms that accumulate and mineralise during life; however, the mechanism and rate by which dental calculus forms are still not completely understood. 28 Of the two types of dental calculus, supragingival and subgingival, the present study attempted to target supragingival dental calculus for all analyses.
Plaque formation on the supragingival surface of teeth begins when microorganisms, overwhelmingly bacteria, colonise the pellicle on the tooth surface. These bacteria obtain their nutrients primarily from the amino acids, proteins, glycoproteins and peptides from saliva to grow, confluence and produce a biofilm. 29 During the production of the biofilm, extracellular polymer synthesis occurs resulting in glucans and fructans from sucrose (refined carbohydrate) metabolism becoming part of the plaque matrix. 29,30 Plaque may begin to harden after about ten days to form dental calculus which then builds up over time. Depending on an individual's hygiene, diet and lifestyle, the deposition of plaque begins soon after tooth eruption, and the quantity increases over time, ceasing at death when the production of saliva stops. Mineralisation of dental plaque only occurs in the presence of saliva when an individual is alive. 31,32 Calculus formation is facilitated by alkaline conditions in the mouth which in turn increases the precipitation of minerals from the saliva. 33 Mineralisation also depends on the food being consumed, salivary flow, oral hygiene and the genetics of the individual. 34 The mineral composition of calculus varies according to the concentration of calcium and phosphorus, the presence of calcification promoters such as urea, fluoride and silicon 35 and presumably also by the bicarbonate composition in the salivaitself a function of the rate of carbohydrate metabolism. 30 Dental calculus is composed of about 20% organic and 80% inorganic constituents. 29 Calculus deposits generally contain the inorganic mineral calcium phosphate: crystalline forms of hydroxyapatite, octacalcium phosphate and whitlockite in varying quantities, with hydroxyapatite usually being the most abundant (ca 58%), which has high levels of carbonate. 36 The organic matrix contains trapped proteins, glycoproteins, plant fibres, lipids and carbohydrates. 29 Unlike the carbonate component of bone and enamel, which derives from blood bicarbonate, 37 supragingival calculus derives its carbonate from the precipitated bicarbonate of salivary fluids. 38 Experiments have indicated that salivary fluids derive bicarbonate from two sources in the body: (i) transfer from the blood and (ii) bicarbonate that is produced in the cells of the salivary glands. 39 The salivary gland bicarbonate is from carbon dioxide (CO 2 ) resulting from the secretory activity of the salivary gland cell (based on the equilibrium CO 2 + H 2 O ⇌ HCO 3 À + H + ) during the action of microbes when metabolising carbohydrates in the mouth. The concentration of bicarbonate is greatly increased during food intake and mastication. 30,40 As previously mentioned, dental calculus formation is, in part, Previous isotope analysis has revealed that medieval populations consumed C 3 terrestrial resources. 45 Similarly, diet for post-medieval individuals has been found to have been dominated by C 3 terrestrial resources, although C 4 cane sugar/maize is known to have been part of their diet or available. 25,46 Finally, the modern population represents North American individuals, who almost certainly consumed an abundance of C 4 sugar and maize. Currently, the intake of sugar refined from C 4 plants such as corn and cane syrup makes up to 78% of the sugar consumed in the United States and forms around 16% of the total calories consumed; however, in some cases it exceeds 35%. 47 The modern group, therefore, provides a control for diets rich in C 4 sources.

| FTIR with attenuated total reflectance (ATR)
Sample preparation and analysis of 57 bone and 52 dental calculus samples were executed according to the Kontopoulos et al 48 method.
We carried out FTIR-ATR analysis before acetic acid treatment. The use of ATR during FTIR analysis has the advantage of being generally insensitive to sample thickness. 49 Bones were cleaned prior to analysis using a sterile scalpel blade and dental calculus samples were rinsed with deionised water to remove dirt and contaminating material. Once cleaned, the samples were ground using an agate mortar and pestle. The powdered samples were then sieved through Endecotts woven stainless steel mesh sieves with an aperture size of 20 and 50 μm so that only grains between 20 and 50 μm particle size would be collected. Spectral analyses were performed using OPUS software (Bruker). Spectra were collected in 144 scans, in the 400-4000 cm À1 wavenumber range, with a spectral resolution of 4 cm À1 and zero-filling factor of 4. Each sample was measured in triplicate and ca 2-3 mg of powder was pressed onto the diamond crystal and measured. We cleaned the instrument's crystal and arm's tip with tissue paper soaked in propanol before each measurement. Baseline correction and spectra normalisation were carried out using OPUS software as reported in Kontopoulos et al. 48 We followed the established bone FTIR analysis method for dental calculus samples as there is no agreed standard for the latter yet. We used two quality parameters, the infrared splitting factor (IRSF) and the carbonate-tophosphate ratios (C/P), to assess our samples. IRSF is used to evaluate the crystallinity (structural order) within the mineral component of the bone while the C/P ratio is a measure of diagenesis that reflects the changes to the carbonate in bioapatite crystals relative to the phosphate content ratio in a bone sample. We calculated IRSF indices following the method of Weiner and Bar-Yosef 50 that measures the heights of the double peaks of the phosphate antisymmetric bending frequency between 550 and 650 cm À1 , 51 divided by the trough between them: The C/P ratio was estimated using the method of Wright and Schwarcz 52 by dividing the main v 3 carbonate peak height with the main v 3 phosphate vibrational band: For bone, the peaks were at 565 cm À1 (v 4 PO 4 3À ), 605 cm À1 (v 4 PO 4 3À ), 1035 cm À1 (v 3 PO 4 3À ) and 1035 cm À1 (v 3 CO 3
Therefore, IRSF and C/P ratio were calculated as follows: ), resulting in IRSF and C/P ratios being calculated as follows: We also utilised the mean values of a modern bovine bone as a reference throughout.

| Bone collagen extraction
Lipids were removed from all modern material prior to collagen and carbonate analysis following Colonese et al. 53 The samples were rinsed six times in a 2:1 dichloromethane-methanol solvent solution (3 Â 2 mL), ultrasonicated for 15 min and centrifuged (850g) for 10 min. They were then rinsed with deionised water and dried at room temperature. Collagen extraction from the 57 bone samples followed the Longin 54 method modified by Brown et al. 55 Each bone sample was cleaned using a scalpel to remove contaminants from the outer layer of bone. Following this, bone chunks of ca 300-500 mg were demineralised in 8 mL of 0.6 M hydrochloric acid (HCl), agitated twice daily. The acid was changed every two days until demineralisation was complete. Next, the supernatant was removed, and the samples were rinsed thrice using deionised water and then gelatinised using HCl (pH 3) at 80 C for 48 h. Next, the supernatant liquid containing the collagen was filtered using Ezee™ filters to remove unwanted particulate matter from the collagen solution and was then frozen for a minimum of 12 h at À20 C before being freezedried for 48 h. Collagen yields were estimated by dividing the collagen mass after filtration by the original bone mass after cleaning.

| Bone preparation for carbonate analysis
Bone carbonate analysis followed a procedure adapted from Snoeck and Pellegrini 56

| Enamel preparation for carbonate analysis
Enamel carbonate preparation followed the method described in Miller et al. 58 We separated a section of enamel, approximately 3 mm wide that spanned from the cervical margin to the cusp, from the tooth crown. In order to minimise contamination, we cleaned all tools prior to use and between samples. All surfaces of the enamel samples were lightly drilled with an abrasive drill bit, set at the lowest speed to avoid heating. Any evident cracks, as well as cut surfaces, were lightly drilled.
Each enamel chip was placed in a 2 mL Eppendorf vial with deionised water and sonicated for 3 min to remove any fine powder. In cases where water remained cloudy after sonication, repeat washes were performed. Enamel was then finely ground using an agate mortar and pestle to a particle size of less than 50 μm.

| Analytical measurements
All δ 13 C and δ 15 N ratios are expressed using the delta notation (δ) in parts per thousand (‰) relative to international standards, VPDB for δ 13 C and atmospheric N 2 (AIR) for δ 15 N, using the following equation:

| Suess effect correction
All the modern tissue δ 13 C values were corrected for the Suess effect (Table S4) which is defined as the global decrease of 14 C and 13 C relative to 12 C in atmospheric CO 2 which occurred primarily due to fossil fuel burning since the Industrial Revolution. 65

| Statistical analysis
Statistical analysis was carried out using R statistics, 66

| Infrared splitting factor
The crystallinity (IRSF) values in modern bones ranged from 3.04 to 3.38 (mean = 3.28 ± 0.09; see Table S3; Figure 1A), slightly lower than the value (IRSF = 3.357 ± 0.007) that was obtained for the modern bones in Kontopoulos et al. 71 The crystallinity in all the consider the effect of sample particle size on FTIR measurements as observed previously in bones. 48 There seems to be no clear relationship or a constant offset between each individual's bone and calculus IRSF values ( Figure 1A).

| Carbonate-to-phosphate ratio (C/P)
The modern bone CO 3 /PO 4 absorbance ratios range from 0.21 to 0.36 (mean = 0.28 ± 0.04) and those for the archaeological samples range from 0.12 to 0.22 (medieval mean = 0.17 ± 0.03; post-medieval mean = 0.17 ± 0.02) indicating remarkable similarity between the two archaeological sets of samples (Table S3; Figure 1B). The C/P values for all of the archaeological bones fell below the mean C/P values that were obtained from modern bones analysed in this study as well as those that have been previously obtained in modern unaltered bone (mean C/P = 0.24 ± 0.003) 71 indicating a loss of the carbonate fraction from the bone apatite.
The C/P ratios for modern dental calculus samples ranged from 0.03 to 0.13 (mean = 0.07 ± 0.02) and those for archaeological samples ranged from 0.04 to 0.17 (medieval mean = 0.11 ± 0.03; post-medieval mean = 0.09 ± 0.03) (Table S3). Similarly, there seems to be no relationship between the individual bone and calculus C/P values ( Figure 1B). Hayashizaki et al 9 revealed that the carbonate content in dental calculus was higher when compared to other biological apatites such as bone, enamel and dentine; therefore, the C/P ratios in dental calculus were expected to be higher in this study.
However, the C/P ratios of dental calculus in this study are lower than that of bone, suggesting that there could be other contributing factors. Dental calculus contains non-apatitic calcium phosphates, 9 which are not present in other normal mineralised tissues. 73 Dental calculus may therefore have an inherent higher phosphate content since, in addition to non-apatitic calcium phosphates, it also has hydroxyapatite leading to a potentially lower C/P ratio relative to bone. The mean value of specific phosphorus concentration in human rib bone weights has been found to be 8.42 ± 2.14% of dry bone weight 74 whereas that of dental calculus has been found to be 19%. 75 This strongly suggests that phosphates may be the cause of the lower C/P values in dental calculus when compared to bone.
Moreover, it was also observed that unlike in bone, the most recent calculus samples have the lowest C/P ratio followed by the post-medieval samples and then finally the medieval samples. This pattern seems to indicate that the C/P ratio of dental calculus increases with the age of the sample. On the other hand, Hayashizaki et al 9 revealed that carbonate content in dental calculus depended on its location in the mouth such that the lower anterior teeth have higher carbonate content compared to the upper posterior teeth. All the calculus in this study was collected from posterior teeth, but randomly from either the mandible or maxilla. Therefore, if differences in carbonate content can occur due to the location of where calculus was formed, 9 it is possible that the apparent trend in C/P ratios observed in this study could simply be a function of where each sample was formed ( Figure 1B).

| IRSF and C/P relationship
Overall, the bone IRSF and C/P ratios display a very strong inverse correlation for modern and post-medieval bone samples (Figure 2A).
The weaker correlation in medieval samples may relate to the varied ages of the burials (7th to 16th century) as well as the different environments from which they were recovered. Environmental factors that degrade samples differ from site to site and, even in the same setting, the type of burial can influence how environmental factors interact with archaeological remains. Moreover, the longer the remains are buried, the greater the diagenetic alteration. 76,77 Alteration of the carbonate content in skeletal material has also been shown to be site-specific. 78,79 Both the modern and post-medieval populations were each obtained from a single site while the medieval samples were collected from three sites. There is, however, a strong inverse correlation for dental calculus samples in all periods ( Figure 2B). Since the relationship is negative for both bones and dental calculus (Figure 2), there is a general trend of increasing IRSF values with a reduction in C/P values reflecting the loss of carbonate with increasing crystallinity. This is in keeping with the previously reported work on bones. 80

| Isotopes
Collagen quality was assessed using the established collagen quality criteria. 81,82 All samples in this study produced sufficient collagen for mass spectrometry (Table S1) Figure S1; Table 1). There are statistically significant differences in both δ 13 (Table 1). There was, however, no statistically significant difference between the medieval and the post-medieval populations (Table S5).

| Correlations between tissues
When considered as a collective dataset (the medieval, post-   Table S2. However, there is no consistent offset between the δ 13 C values for the three tissues per individual (Table S2;  In contrast to Price et al, 8 we suggest that a direct comparison between different tissues' isotopic composition is not strictly possible as there may be a difference between the isotopic composition of carbon preserved in bone, teeth and dental calculus. For example, the isotopic composition of carbon in bone carbonate and enamel carbonate has been investigated by Warinner and Tuross, 84 who compared bone and enamel carbonate of pigs raised on controlled diets containing either raw maize or nixtamalized maize for 13 weeks.
The results revealed a significant difference between bone and enamel carbonate in the animals with the δ 13 C values in enamel being higher by 2.2‰ and 2.3‰ in the nixtamalized and raw diets, respectively.
Since simultaneously forming bone and enamel carbonate were used, the offset could not be attributed to 'preferential or differential digestion since the carbonate in both the enamel and bone apatite was deposited from dissolved blood bicarbonate during the same experiment'. 84 Considering that enamel, once formed, cannot be remodelled, whereas bone undergoes constant remodelling, the authors suggested that the enamel carbonate and bone carbonate of adult animals, therefore, represented 'temporally segregated isotopic deposition events'. Therefore, since enamel reflects diet at the time of formation (childhood diet) and bone represents an average from the last few years of life, 85,86 we propose that the inconsistent offsets between enamel and bone are most likely to be due to each individual's different childhood and adult diets.
Additionally, we propose that, unlike bone and enamel, the isotopic fractionation in dental calculus may not be associated simply with metabolism and food sources. In the case of dental calculus, which is formed from a living biofilm, the isotopic fractionation would likely increase with calculus biofilm thickness, similar to what has been observed in other biofilms. 87 However, it is also notable that in modern individuals the extent of fractionation is lower than in medieval and post-medieval populations. Although this study does not have the details on the thickness of the dental calculus biofilm, it is proposed here that greater oral hygiene may lead to less wellestablished calcified biofilms. Therefore, it is suggested that since the mechanisms involved in the formation of dental calculus, fractionation effects and turnover time in this tissue are different from that of  (Figure 4), suggesting that any impact of pre-treatment is likely to be minimal.

| Dietary interpretation from the three types of tissues
Generally, the carbon isotope values for the modern North American population are extremely high for all tissue types and consistently higher than those of the medieval and post-medieval populations.
Unlike the archaeological populations from England, the modern individuals are influenced by the dominant consumption of C 4 foods like fructose, maize and its by-products (e.g. corn syrup). 47 The medieval and post-medieval diets, on the other hand, were dominated by C 3 foods. 17,18 Overall values are expected to fall between À17‰ and À13‰ for a pure C 3based diet. 91 The average δ 13 C enamel values for both the medieval and post-medieval populations are therefore consistent with the typical spectrum of a terrestrial diet in England characterised by high consumption of C 3 resources as suggested in historical sources. 17,18 Although there are some suggestions of C 4  In the case of the medieval and the post-medieval individuals, the enamel and the bone δ 13 C values are the same because not only are the bones and enamel tapping from the same pool of bicarbonates in the blood during formation, but also the children were not exposed to sugary foods, unlike their modern counterparts. Unlike modern individuals, the archaeological children's diet did not differ substantially from their adult diet, at least in terms of consumption of C 4 resources. carbohydrates integrated into the post-medieval diet. Additionally, it has been established that, within individuals, dental calculus δ 13 C values were (mostly) consistently higher than those of enamel or bone; however, the offset (difference, Δ) in δ 13 C values between the three tissues was not consistent. This could be due to differences in diet at the time of tissue formation or carbon fractionation in the mouth versus the body. The calculus δ 13 C values will most certainly be high due to the sugars that are being metabolised in the mouth as well as the effect of more fractionation in this much more poorly biologically controlled environment compared to that of bone and enamel. Using information from previous breath studies, this study has also demonstrated that it is highly unlikely that atmospheric CO 2 is causing higher δ 13 C values as suggested by a previous study 8 as δ 13 C breath values are reflective of the whole diet. We instead argue that calculus carbonate δ 13 C values are highly likely to represent the carbohydrates consumed. In spite of these conclusions, there is an awareness that the formation process and composition of dental calculus are highly variable between individuals, and therefore there are still a number of gaps in knowledge that need to be addressed before its full potential as a viable tool for palaeodietary studies can be realised. For instance, the effect of pre-treatment methods and the potential issue of the presence of particles (inorganic debris) in dental calculus are still outstanding, and more information is required to understand how the micro-debris affects the overall isotope values found in dental calculus. We recommend controlled feeding experiments to clearly observe the effect of different carbohydrates on dental calculus carbon isotope ratios.