Method of micro‐sampling human dentine collagen for stable isotope analysis

Rationale Sampling of dentine for stable carbon (δ13C) and nitrogen (δ15N) isotope ratios in the direction of tooth growth allows the study of temporal changes to the diet and physiological stress of an individual during tooth formation. Current methods of sampling permanent teeth using 1 mm increments provide temporal resolution of 6–9 months at best depending on the tooth chosen. Although this gives sufficient sample sizes for reliable analysis by mass spectrometry, sectioning the dentine across the incremental structures results in a rolling average of the isotope ratios. A novel method of incremental dentine collagen sampling has been developed to decrease the collagen increment size to 0.35 mm along the incremental structures, thus reducing averaging and improving the temporal resolution of short‐term changes within the δ13C and δ15N values. Methods This study presents data for a MicroMill‐assisted sampling method that allows for sampling at 0.35 mm width × 1 mm depth increments following the incremental growth pattern of dentine. A NewWave MicroMill was used to sample the demineralised dentine section of modern donated human third molars from Sudan and compared to data from the same teeth using the 1 mm incremental sectioning method 2 established by Beaumont et al. Results The δ13C and δ15N isotopic data showed an increased temporal resolution, with each increment providing data for 2–4 months of dentine formation. Conclusions The data show the potential of this method for studying dietary reconstruction, nutritional stress, and physiological change with greater temporal resolution potentially to seasonal level and with less attenuation of the δ13C and δ15N values than was previously possible from human dentine.


| INTRODUCTION
Dietary reconstruction using stable carbon and nitrogen isotope ratios (δ 13 C and δ 15 N) from bulk sampling of collagen from both bone and dentine has become a useful tool for understanding the lifeways of past individuals. Bulk dentine provides data from childhood and adolescence averaged over the period of tooth formation, whereas bulk bone gives data which has been averaged over longer periods of life. In adults, differing rates of bone turnover allow us to investigate data over 10 years or more, and in rib bones, from the last 5 years of life. [1][2][3][4][5][6][7] However, the use of incremental dentine sampling has allowed us to study dietary shifts related to events such as breastfeeding, weaning, and physiological changes caused by undernutrition during tooth formation previously invisible in a single bulk sample. [8][9][10] Dentine is an ideal material for incremental sampling research because it forms in predictable temporal increments and does not remodel after formation. 11 δ 13 C and δ 15 N variations in the dentine collagen incremental profiles show clear patterns that have allowed researchers to distinguish between dietary and physiological changes. [12][13][14][15][16][17][18] Different studies have used variations on the early sampling methods to improve temporal resolution and reduce averaging with varying success. 4,5,9,14,19,20 Teeth are often well preserved in archaeological sites, 21 and in the absence of other incrementally forming tissues such as hair and nails, the development of reliable sampling methods has made teeth a preferred resource for research on dietary changes and life events in childhood and adolescence. [22][23][24][25] Methods of incremental sampling were previously limited by the amount of collagen required to obtain reliable data using a continuous-flow isotope ratio mass spectrometer (IRMS). Fuller et al (2003) used deciduous and permanent teeth divided samples into [3][4] horizontal sections to study the breastfeeding and weaning patterns of archaeological remains from Wharram Percy, England. 20 Eerkens et al (2011) used permanent first molars divided into 5-10 horizontal incremental sections to study weaning and childhood diets of six individuals from Marsh Creek banks in central California. 4 Burt and Garvie-Lok (2013) developed a micro-sampling method on modern deciduous teeth from Canadian individuals to study pre-and postneonatal line sections to observe different life stages. 19 The method involved punching disks out of the dentine on both sides of the neonatal line.  developed methods for sampling dentine collagen in 1 mm increments which excluded the filtration stage and thus increased the yield for each section. 9,26 For Beaumont Method 2 the whole root of a multi-rooted tooth, or half of a single rooted tooth from crown to apex, is demineralised under refrigeration in 0.5 M HCl, and then sectioned by hand using a metal ruler and scalpel. The 1 mm increments are then denatured by heating at 70 C in a pH 3 solution of de-ionised water and HCl, then frozen and freeze-dried. 9 This results in sufficient collagen for duplicate 0.5 mg samples to be weighed into tin capsules and measured using IRMS.
This method was used on 19th-century teeth from London to investigate childhood diet and migration to London for survivors of the Great Irish famine. 22 Root dentine is formed in a pattern similar to a series of concentric cones with varying angles of formation within different sections of the tooth. For example, in the crown of a molar, the dentine growth is nearly horizontal while the angle increases as tooth growth proceeds down the root relative to the pulp cavity and also changes as it approaches the root apex. 4 Henderson et al (2014) adjusted Beaumont et al (2013) Method 2 by sectioning the demineralised dentine with five smaller sections in the crown and five larger sections from the remaining root in an attempt to make the time represented in each increment equal. 5 With the exception of Burt and Garvie-Lok (2013) all incremental methods involved cutting sections transversely from the crown to the apex at predetermined intervals resulting in analysis of multiple formation periods. 27 Though proven to be effective in providing temporal resolution which is greatly improved over bulk sampling, there is still an averaging of the isotopic values because of the overlapping growth patterns of the developmental structures (Andresen bands). Andresen bands are patterns formed by diurnal rhythms of dentine formation causing different densities that can be seen under light microscope in both intact and demineralised dentine ( Figure 1). 28 Sampling in the direction of these bands rather than transversely across them would result in reducing the averaging of the isotopic signal between formation ages; however, this is a difficult task because they are visible only under a microscope. 9,29 Method development has been undertaken in Czermak et al the averaging of the isotopic ratios within incremental sampling by eliminating the potential of secondary and tertiary dentine and cementum from contaminating the primary dentine collagen samples. This was done by implementing the use of a 1 mm diameter biopsy punch to remove increments from demineralised dentine collagen.
Our aim in this study was to develop a micro-sampling method that can be reliably used for sampling dentine collagen and will both improve the temporal resolution and reduce the averaging of δ 13 C and δ 15 N. The method we describe here uses a NewWave MicroMill™ to add a level of precision to the incremental sampling process that is not guaranteed when sampling by hand. Though similar in concept to previous sampling methods, this novel method reduces the amount of sample lost while significantly increasing the temporal resolution of the isotope data.

| Standards, chemicals, and instruments
Chemicals used for this research were U.V. Glue (Bondic ® , Cambridge, UK), and hydrochloric acid diluted to 0.5 M using de-ionised H 2 O. The instruments used for sectioning dentine samples were Leica SP1600 Saw Microtome and ESI New Wave Micromill. The instrument used for analysing collagen samples was an EA IsoLink™ IRMS System, using Thermo Delta V Advantage IRMS coupled to a Thermo Flash 1112 Elemental Analyser via a ConFlo III.
Results are reported using delta (δ) notation in parts per thousand (per mil or ‰) relative to international standards. The carbon isotope ratios are expressed relative to Vienna Pee Dee Belemnite (VPDB), and the nitrogen isotope ratios are expressed relative to AIR (AIR N 2 ). 36,37 The analytical error is 0.2‰. International Atomic Energy Agency (IAEA) standards 600, N1, CH3 and laboratory in-house standards fish gel and bovine liver were used when analysing the collagen samples. δ 13 C and δ 15 N values for all standards used are listed in Table 1. Standards were distributed throughout and analysed with the samples in each analytical run.

| Methods
(see Figure 2 for images and flowchart).

| Sample preparation/cutting
Each tooth was sectioned longitudinally, selecting the longest available root and corresponding crown tissue using a Dremel handheld diamond-edged saw. The roots were secured in a cork holder and placed in a Leica SP1600 Saw Microtome (see workflow chart) used to section the samples because of the precision of the saw. A 1.5 mm thick longitudinal section was cut from the root.

| Demineralisation
The longitudinal sections were then demineralised using a modified Longin method. 38 The sections were demineralised in 0.5 M hydrochloric acid (HCl) at 4 C for approximately 4 days with the acid being changed regularly. When demineralisation was complete, the samples were rinsed 12 times with de-ionised water to remove contaminants.

| Freezing and freeze-drying
The collagen sample sections were then placed between two glass microscope slides and wrapped in Parafilm ® (Bemis Company, Inc.,

| Assigning age-at-formation to dentine increments
The assessment of the age-at-formation was approximated based on  Table 2.
Because of the low weight achieved the incremental dentine collagen samples acquired through the novel method were analysed once. The incremental samples acquired using Beaumont method 2 were analysed in duplicate by Jackson (2017) and the mean values.
The δ 13 C values of ET-7889 range from À18.1‰. to À16.9‰.     This profile suggests an individual who is consuming C 4 relief foods with a corresponding catabolic increase in δ 15 N at the start and end of the profile, and a reduction in δ 15 N when access to other foods is available. 42 The development of the micro-milling method went through several stages to determine the ideal sample thickness, channel diameter, and order of steps. For example, to drill a sample using the MicroMill, the sample has to be in a solid state: thus the sample would have to be milled either before the dentine collagen was demineralised or after the collagen had been freeze-dried. During the development of this method, it was determined that there was a large volume of milled sample lost during demineralisation that did not occur if the increments were milled after demineralisation and freeze drying. Cutting the sample channels at an angle, as shown in half months of isotope ratio data. The 2 to 4 months of isotope data is an improvement on the 6 to 9 months that are provided from these teeth using the incremental methods described in . 26 The temporal resolution achievable will vary due to the different growth rates between different tooth types and the length of the root sampled: thus a tooth which takes a long time to develop will have poorer resolution than a fast-growing tooth, and a tooth with a long root such as the permanent maxillary canine will have better resolution than a short-rooted third molar. 39 profiles that do not consistently correspond between the data from the 0.35 mm sampling method and the 1 mm sampling method. The lack of corresponding timing of these data points could be related to methodological differences affecting the accuracy of measuring coforming tissues. The 1 mm sectioning method relies on cutting increments by hand and by eye rather than using measuring tools and microscopic views. In this pilot study the incremental sections were taken from different roots from the same molar tooth, which may have been different lengths. In a prospective study, the samples for both methods would be taken from the same root. Furthermore, the dentine collagen measurements in this study measured a single (rather than duplicate) collagen sample due to the low average weight persists across multiple incremental data points. Although hair keratin and dentine collagen have different chemical makeups, the similarities between the isotopic profiles indicate that the increased temporal resolution in the dentine collagen may be capable of providing information that was not previously observable. The observed isotopic pattern was not seen in the less detailed isotopic data from the Beaumont method.
The level of collagen preservation will be a vital aspect for the use of this novel method on archaeological dentine samples. 14,44 The preservation of samples was not a concern with these modern samples, and further work is being carried out to repeat the comparison between the two methods using well-preserved archaeological teeth. An increased temporal resolution applied to deciduous teeth could also provide more detailed information about maternal health, pregnancy, and weaning practices.

| CONCLUSIONS
The isotopic data generated using longitudinal sectioning and micro-milling in the direction of the incremental structures matches well with previous data produced with Beaumont Method 2 from the same tooth while demonstrating that there is more detailed variation which has been lost in the established method.
The study demonstrates the potential for a method of precision sampling of human dentine collagen along incremental structures which can improve the temporal resolution currently possible with other micro-sampling approaches and thus the ability to investigate shorter-term periods of nutritional or physiological stress, including potential seasonal dietary changes. Further work is in progress to determine the effect of taphonomic and diagenetic modifications and preservation of collagen on the reliability of the method.