In order to understand and characterize diagenetic change in vertebrate remains, a thorough understanding of the nature, characteristics, components and processes involved in the biogenesis of original tissue is required. Without knowledge of the chemistry of original components present in bone or teeth, it is impossible to quantify changes from that state.

The structure, formation and composition of vertebrate mineralized tissues are thoroughly discussed in the literature (e.g. [54] and [169]) and in-depth treatment of this subject need not be repeated here. However, in brief, bones (and other biomineralized elements such as teeth) are a composite material, consisting of an intimate interaction between mineral and organic components. Mineralized tissues are formed by differentiated cells (e.g. osteoblasts, odontoblasts) that secrete an organic matrix consisting primarily (∼90%) of collagen protein, with some other non-collagenous proteins such as osteocalcin, osteonectin, and serum proteins such as albumin [65], [124] and [176]). Understanding the molecular characteristics of extant bone yields a starting point for identification of similar compounds in fossil bone.

The isotopic composition of the mineral incorporated into bone during the life of an organism can reflect the ecology [81] and [151], diet [83] and [106] and possibly, metabolic rates [10], [11] and [12] of the organisms. Isotopes of oxygen and carbon incorporated into bones and teeth during formation and remodeling are affected by the chemistry and temperature of pore waters and diet of the organism. Thus, isotopic analysis holds potential for shedding light on many aspects of the biology of extinct organisms. However, possible modification of these values over time from the living state [18] and [87] requires a detailed understanding of the effect of the burial environment on the original isotopic composition of bioapatite before these data can be accepted as reflecting in vivo values.

Postmortem diagenesis results in a range of chemical and mineral changes to the organic and inorganic constituents of fossil bone and teeth. These processes cause trace-element enrichment, dissolution, precipitation, and recrystallization of the original crystal structure of bioapatite, resulting in modifications to fossil bones and teeth from different burial sites and ages [110], [154] and [174]. The resolution and precision of analytical methods for documenting these primary mechanisms of diagenesis in fossil bone have improved since the 1960s and early 1970s with systematic studies of chemical and physical changes of fossilization [16], [20], [34], [47], [59], [69], [76], [84], [85], [90], [111], [113] and [159]. It is well established that different diagenetic factors act in different burial environments, even if it is difficult to link a specific process to a specific diagenetic change [126]. A well-preserved ‘pristine’ fossil with excellent texture and micron-scale preservation of primary structures does not guarantee the preservation of original isotopic values [87] and [93], nor, for that matter, preservation of original molecular signature [157] and [158].

The widely used proxy for evaluating the robustness of isotopic values derived from fossil bioapatite is more difficult if a modern analog is unknown, or if diet and environmental conditions are not well documented, or have changed through time. This presents a problem when assessing preservation of original carbon isotopic values using the pre-Quaternary fossil record or in deep time before the evolution and global spread in the Miocene (7–10 Ma) of C₄ grasses. To further complicate matters, dietary components that are isotopically distinct may be obscured by secondary fractionation and variable carbon isotope turnover rates in paleoecological studies [152] and [168].

Oxygen isotope variation within apatite carbonate is more difficult to predict than carbon isotopes derived from the same mineral [21]. The δ ¹⁸O of biogenic apatite is strongly correlated with that of body water in extant mammals. A linear correlation between δ ¹⁸O of bone phosphate and δ ¹⁸O of bone carbonate from the same bioapatite mineral would support the preservation of original isotope values in extant mammals in most cases [77a].

Plugging calculated δ ¹⁸O values into the respective temperature equations for the same phase carbonate and phosphate is also used to evaluate isotopic values from fossils [141]. The primary basis for evaluating oxygen stable isotope composition in fossil bones and teeth relies on the comparison of δ ¹⁸O values between bone, tooth enamel, and dentine. Ranges of these values are compared with known extant systems to test whether fossil isotope values are well supported [7]. Wang and Cerling [166] concluded that δ¹⁸O values of structural bone carbonate are vulnerable to diagenesis and unlikely to record original isotopic signature, based on comparison of Tertiary fossil teeth, bone, and sediments from Badlands National Park, South Dakota. Furthermore, efforts to correlate microscopic integrity of dinosaur bone with pristine geochemical integrity [87] were not supported, but instead demonstrated the importance of burial environments in identifying diagenetic alteration. Kolodny et al. [87] calculated δ ¹⁸O from bioapatite in Cretaceous teleost fish, dinosaurs and reptiles from the same localities across a latitudinal gradient (Alaska, Wyoming, and Texas) and found values determined more by locality, than by expected physiological or life history signal, suggestion environmental overprinting of original endogenous values. No systematic isotopic differences between these diverse taxa were preserved. The similarity of δ ¹⁸O phosphate in modern fish, turtle, and crocodilians in their data set did not support the preservation of original isotopic composition of drinking water or environmental waters they lived in, but rather indicated isotopic equilibration occurred between fossils and groundwater from the burial environment. They concluded that differences in dinosaur δ ¹⁸O phosphate composition did not reflect in vivo body temperature variability when the bones in the skeleton were formed, but more likely the chemistry of the burial environment (contra [10], [11], [12], [13], [14] and [55]). Chenery et al. [30] argued that early diagenetic recrystallization resulted in overprinting the δ ¹⁸O signature of hadrosaurid dinosaurs buried in Late Cretaceous marine sediments in Dinosaur Provincial Park, Alberta. And, while Trueman et al. [160] reached the same conclusions from bones buried in terrestrial and marine sediments from Campanian Bearpaw and Dinosaur Park Formations, Thomas and Carlson [151] concluded that, while absolute oxygen isotopic values may be altered, a seasonal signal may be preserved in their ontogenetic series of the hadrosaurid, Edmontosaurus.

Preservation of original isotope values in fossil bioapatite may also be affected by the enzymatic microbial degradation of phosphate. Microorganisms (e.g., bacteria, algae), through enzyme mediated nutrient cycling of organic constituents of skeletons may be responsible for anomalous phosphate oxygen isotope compositions and trace element concentrations in fossil biogenic apatite [16] and [61]. Through biogeochemical cycling of phosphate in aquatic environments, sediments, and soil, isotopic fractionation may occur during early stages of diagenesis. Complete re-equilibration of oxygen isotope values or re-setting phosphate oxygen composition from this secondary organic phosphate source may be responsible for inconsistent oxygen isotope values in fossil bioapatite [16] and [87].

Biogeochemists who use fossil bone and tooth bioapatite as source material for stable isotopes have utilized a variety of methods to determine the effects of burial environments on preservation of in vivo isotope values. This is necessary if these data are used to support hypotheses on climate change, animal and plant migration, evolution, and life histories of extinct animals. For the most part, divergence of fossil isotope values from modern counterparts is attributed to diagenesis, while similarities between the two are evidence of diagenetic resistance [86]. In any case, these techniques alone are insufficient to demonstrate preservation of original isotope values. This has resulted in conflicting conclusions in determining if isotopic values are original or altered in prehistoric and fossil hydroxylapatite.

Even though δ ¹³C values in modern mammals are a result of diet only [84], both δ ¹⁸O and δ ¹³C isotopic compositions in fossils are potentially exposed to postmortem diagenetic overprinting in the burial environment [89]. Deviations from predicted isotopic values may reflect either diagenesis or changes in diet and physiological adaptations [83]. Despite the variety and complexity of factors acting on a bone or tooth when it leaves the biosphere and enters the geosphere [111], many paleobiological and paleoclimatological isotope studies still do not consider the potential modifying effects of diagenesis from the burial environment on the original isotopic composition of their sample source materials when faced with anomalous results [93].

The above concerns dictate that pretreatment of modern and fossil bone and tooth bioapatite is absolutely necessary to remove exogenous organic matter and diagenetic minerals. Standard protocols recommend the pretreatment of powdered samples with a 2–3% reagent grade sodium hypochlorite (NaOCl) to remove organics (e.g., [82]). Approximately 1 ml of solution is required for each 25 mg of apatite (0.04 ml/mg). Samples are centrifuged and rinsed five times in distilled or megapure water, then soaked in 1 N calcium acetate buffered acetic acid overnight. Alternate rinsing and centrifugation cycles are followed by lyophilization, or oven drying prior to stable isotopic analysis by mass spectrometry.

Using fossil apatite to infer aspects of paleoecology or paleophysiology when derived from specimens across a broad geological time frame, without a stable support for the preservation of its original isotope values remains a much debated issue in biogeochemistry [18], [85], [119], [141] and [160], particularly because the mechanisms of alteration of carbon isotope signals derived from fossil bone carbonate have not been fully elucidated [89].

Several methods may be applied to analyze specimens for chemical/molecular content. These will be divided into those that require minimal sample preparation (i.e., bulk sample analyses), and those that require some form of pretreatment.

This method is coupled to TEM systems, and transmits a very high energy electron beam (200 kV) through the sample. The beam is diffracted at angles determined by Bragg's law, applied to a set of reflecting crystal planes separated by the d-spacings of the crystals. In extant, or relatively unaltered fossil bone, the resultant diffraction pattern occurs as partial rings [140]. In replaced or recrystallized bone, the pattern is diffuse or geometric, reflecting the arrangement of crystals when no pattern is imposed from collagen fibrils [177].

Diffraction patterns can allow identification of in situ mineral phases with high accuracy, and can differentiate between bioapatite and other forms by D-ring spacing [140], or between different forms of iron-bearing minerals. It is useful in quantifying degree of diagenetic alteration in fossil samples, or in identifying remnants of original biominerals.

Like other methods of mass spectrometry, PY-GC-MS capitalizes upon the fact that every molecule consists of a unique combination of atoms that can be identified with high resolution by their atomic masses [77b]. This method has been used to identify the remnants of original molecular components, including protein fragments, lignins and chitin, of fossils dating back to 25MY BP [144], [145], [146], [147] and [163], and has been applied to systematic differentiation of plant materials [145].

Py-GC-Mass spectroscopy uses intense heat (∼600 °C) to split and volatize molecular components, which are then separated according to fragment mass by gas chromatography. Each separated peak separated is analyzed by a quadrupole mass spectrometer to identify the number and mass of each volatized fragment, producing a spectrogram. Because this is a bulk sample analysis, it is not possible to localize the source of these molecular components within a heterogeneous sample. However, because this method does not require chemical extraction, it reduces the chances of inducing artifact into the analyses. It is not ideal for identification of proteins and DNA, but is highly sensitive to other biomolecular fragments.

Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) rasters a high-energy (e.g., gallium ions, typical energy: 15 keV) ionizing beam with a pulse width of ∼1 ns and a repetition rate of 10 kHz over a predetermined area to release molecular fragments from a sample surface. Only a small fraction (10⁻⁶ to 10⁻¹) of the total released fragments are measured, because only charged particles (ions) are detectable [165]. Mass of each fragment is then calculated based upon the time it takes for the molecule to move from the source to the detector. This ultra high-resolution detector is capable of differentiating fragments that differ in mass by as little as 0.001 atomic mass unit (amu; 12 amu is, by definition, the mass of a carbon atom). Mass resolution (mass/mass uncertainty, m/Δm) in practice varies from 2000–5000, and is affected by primary pulse width, sample type and surface flatness, and charge compensation. In addition, whether or not detectable secondary ions are generated is highly dependent on characteristics of the sample; therefore, peak intensities (i.e., peak areas) may differ between samples [77b], but mass resolution is not affected. Other mass spectrometry methods use spectral patterns in addition to absolute mass values to identify fragments, but in ToF-SIMS, the spectral patterns will vary because of inherent differences in the surface composition. For example, fresh bone may yield a spectrum with a pattern that differs from fossil bone, even though they may contain some of the same molecules. However, absolute masses (or peak location) will be the same, regardless of surface, if the source fragments are the same.

ToF-SIMS provides at least three avenues of information not available by other mass spectroscopy methods. Because only a small fraction (0.1%) of a small area (<200 μm²) of the outermost surfaces is ionized, this method is less destructive, allowing the same sample to be used for other analyses, once mass data is obtained. This can be important with rare, minute, or fragmentary fossil material. Fig. 4 shows a ToF-SIMS profile obtained on dinosaur vessels recovered from demineralized bone. In addition to several peaks that can be associated with amino acids, we have used this method to detect novel compounds involving organics and iron, supporting our hypothesis of a possible chemical pathway for preservation [139] and [140].

Second, ToF-SIMS measures both positive and negative ion spectra, thereby providing complementary information about fragmentation patterns of sample-derived molecules [2], [3] and [140]. Third, unlike other mass spectrometric methods, ToF-SIMS allows the localization of molecular sources to specific regions of the sample. Therefore, in fragile specimens where it is desirable to limit handling or sample preparation, the study specimen can be left within the encasing matrix. Comparative mass data can be obtained from both fossil tissues and the surrounding matrix, demonstrating that molecular signal originates with the specimen, and not from exogenous environmental contaminants. Like all mass spectrometric methods, it is imperative that ultraclean methods be used in handling the specimens, and particular to ToF-SIMS, because only surface layers are analyzed, it is important that all samples are handled using aseptic techniques. Additionally, freshly fragmented specimens that expose internal surfaces for analysis, minimize chances of detecting exogenous contaminants. The advantages and disadvantages of other mass spectrometry methods that require chemical extraction of samples will be discussed below.

X-ray photoelectron spectroscopy focuses monochromatized X-rays on study specimens under ultra high vacuum (UHV), to produce photoelectrons varying in intensity and kinetic energy. Each element within the sample produces characteristic XPS lines, the intensity and position of which determine the concentration and local chemistry or bonding environments of the atom detected. For example, the sulfur 2p line can show as much as 7 eV chemical shift between sulfur in a disulfide environment (S²⁺) and one in a sulfate (SO₄) environment (S⁶⁺). With sub-eV resolution, such shifts can easily be monitored. This method is an excellent complement to both TOF-SIMS and EDX, confirming the presence of elements identified in EDX, and, since it is slightly more sensitive and discrete, identifying light elements not detectable or quantifiable by EDX, or which are masked by other elements, as is nitrogen. XPS can identify the presence of nitrogen unambiguously, as well as identify its bonding environment. Nitrogen is an essential component of both DNA and proteins, and while its identification in fossil samples does not guarantee the presence of these molecular compounds, it is certain that neither DNA nor proteins will be found in fossil specimens if nitrogen is not identified.

Finally, XPS complements both EDX and mass spectrometry (MS) methods, in that the oxidation state of the elements identified by EDX, or molecular fragments identified by MS, can be linked to local bonding environments by XPS [22]. For example, if TOF-SIMS suggests the presence of sulfur-containing amino acids such as cysteine in a sample, XPS will verify that sulfur is present in the sample, and that it exists in a state consistent with sulfur which has been oxidized either to S²⁺ for disulfide bonds, or S⁴⁺, consistent with cysteic acids [134]. This method differentiates sulfurs in bonding environments consistent with organics, as opposed to those derived from inorganic sedimentary sulfur.

Many analytical methods besides mass spectrometry require fossil material to be chemically extracted, and several methods can be applied, depending on the downstream analytical technique employed. Organic extractions (i.e., using organic reagents such as benzene) are useful for isolating particular compounds, such as kerogens, lipids and other geochemical components (e.g., [38]). However, for isolation of water-soluble molecular components such as protein fragments, other methods are more appropriate and/or more efficient.

Where possible, all surfaces should be cleaned by abrasion to remove external contaminants. Fossil material is then crushed to a fine powder and washed 2–3 times in water or dilute salt solution to remove any loosely bound contaminants. Wash buffers are removed using centrifugation, and an extraction buffer is applied to the bone powder and allowed to incubate for several hours to overnight, depending on individual protocols. Extracting reagents must be removed before further analyses, usually by ion-exchange column chromatography or dialysis. This removes low molecular weight salts and other buffer components, while retaining the larger organic components of fossil tissues. Fossil extracts should then be dialyzed against sterile distilled water or a neutral physiological buffer such as phosphate-buffered saline (PBS), using low molecular weight cut off (MWCO) membrane, assuming any endogenous organic components are fragmented and therefore smaller than comparable molecules derived from fresh tissues. After dialysis, ultrafiltration or lyophilization can be applied to concentrate the sample, thus optimizing chances of identification of endogenous molecules in low concentration.

Various extraction buffers have been used to isolate organic material in fossils. We briefly review several here that have been used with success with fossil and sub-fossil material. The first three protocols below have in common a demineralizing agent to remove the mineral phase of bone, and one or more protein denaturants. Others contain a reducing agent, such as dithiothreitol (DTT), included for breaking intra- or intermolecular disulfide bonds or other cross-linking bonds. (1)

Gurley et al. [62] used a reducing buffer consisting of 0.3 M NaCl, 5% glycerol, 5 mM DTT (dithiothreitol), 2 mM EDTA, 1% CHAPS (3-[3-cholamidopropyl) dimethylammonio]-1-propanesulfonate), 6 M guanadine-HCl, and 100 mM Tris-HCl, to extract dinosaur bone for identification of amino acids.

Once the fossil tissues have been extracted, purified and concentrated, there are many analytical techniques that may be performed. The following are some of the analytical techniques that may be performed on extracts of fossil specimens.

Organic molecules absorb light at specific wavelengths in the electromagnetic spectrum, and may be identified by the specific absorbance pattern they exhibit. Lyophilized extracts are solubilized in PBS or sterile distilled water, placed in a special ultraclean quartz cuvet, and continuous wavelength light (∼190–800 nm) is transmitted through the sample. When light interacts with a molecule, it is absorbed in quantum packets, the characteristics of which are dependent upon the atomic make-up and overall molecular structure of the sample, and the absorbance can be measured quantitatively. Most proteins demonstrate an absorbance maximum at 280 nm that can be attributed to the presence of aromatic amino acid residues (i.e., tryptophan, tyrosine, phenylalanine). Some proteins, such as hemoglobin, have shifted absorbances (∼410 nm, [29]), due to their association with pigment molecules. Maximum absorbance characteristic for nucleic acids is seen at 260 nm. Ratios of absorbance measured at 260 nm: 280 nm is often used to determine the purity of a DNA preparation. For absorbance values of commonly occurring amino acids, see Fasman [52].

Heterogeneous mixtures resulting from bulk extraction of fossil materials present analytical challenges. Chromatographic separation enhances the sensitivity and efficiency of other analytical techniques. In addition, because some components in bulk extractions of fossils may inhibit both polymerases and proteases in amplification of DNA [125] and [159], chromatographic methods can be used to purify samples.

Column chromatography separates individual components from a complex mix using inherent differences in molecular structure of the constituent compounds. Components may be separated by mass, charge, or hydrophobicity, depending upon the column packing chosen. Differences in migration rates that allow for separation are the result of selective retention on the column packing (referred to as the stationary phase). The mobile phase, in which the analyte is suspended, can be gas or liquid. In gas chromatography, the sample is vaporized and the vaporized molecules are carried in an inert gas through a fine capillary tube, lined with a high molecular weight liquid. Separation is monitored by a detector that measures the exit of the molecules from the capillary tube. This, in turn, is determined by the interaction of the vaporized molecules in the mobile phase with the liquid of the stationary phase [77b].

In high-performance liquid chromatography (HPLC, [27]), a heterogeneous sample is dissolved in an inert liquid and applied to a column under high pressure. Components of the mixture will adsorb to the column material according to the characteristics of hydrophobicity or charge, and are then eluted individually from the column as the mobile phase changes across a gradient. In reverse phase HPLC, the water-soluble compounds adhere to the non-polar (hydrophobic) material of the column packing. A hydrophilic or polar buffer is then washed over the column, and will remove the most polar components of the mixture as these are least likely to stick to the hydrophobic column. Less polar substances elute from the column later, as the mobile phase changes from polar to relatively non-polar [100].

Chromatographic separations are achieved by varying the characteristics of the column (stationary phase). For organic compounds more likely to be soluble in lipids than water (such as kerogens) normal phase chromatography is often used, where the stationary phase is polar and the mobile phase is non-polar. Separation by charge is accomplished using a stationary phase containing charged groups (e.g., NO₃ ⁺, or SO₃ ⁻) that interact with charged regions of the molecules carried in the mobile phase. Size exclusion packing is used to separate molecular components by molecular weight. Larger molecules elute earlier, while smaller molecules are retained as they pass through small pores in the packing material.

Ionic exchange is another means of chemical separation, whereby the constituent molecules are separated by charge. This method is excellent for removing buffer salts, detergents, or other remnants of the extraction procedure.

Finally, DNA may be selectively removed from bone extracts using apatite columns. Once separated, individual components can be characterized by mass spectroscopy, UV/VIS, or molecular analytical methods such as PCR amplification or antibody testing (see below).

Raman spectroscopy uses a laser tuned to a single wavelength to elucidate information about molecular structure [153]. When stimulated, a photon is scattered inelastically, transferring energy and momentum to molecules in the sample. Monitoring light wavelengths that are scattered inelastically yields information on how individual bonds stretch or vibrate as energy is absorbed. Intramolecular bonds such as C C or C O can be detected and differentiated, thus yielding information on molecular structure. Resonance Raman (RR) is particularly useful in the study of biomolecules. As the wavelength of the stimulating laser approaches the same frequency as electronic transitions of certain component molecules, Raman scattering is greatly enhanced. Resonance Raman will preferentially enhance certain aspects, at the expense of others, thus identifying unique ligands of known molecules that may arise through chemical cross linking in the fossilization process.

Raman spectroscopy requires only minute samples (e.g., single crystal grains or microliters or less of solubilized material), making it an ideal method for analyzing rare fossil material [60]. In addition, by carefully selecting the excitation wavelength of the stimulating laser, the characteristics of a single molecular class in a complicated mixture [153] can be analyzed. Because organic compounds are often irreversibly crosslinked with other molecules in fossil tissues, purification and/or separation of proteinaceous material is difficult. Raman spectroscopy is particularly well suited for detecting fossil specific molecular modifications. Finally, Raman spectroscopy is a highly accurate and sensitive means of identifying specific molecular compounds such as porphyrins or heme [133] and [149].

Porphyrins are a class of small biomolecules found in many fundamental molecules such as chlorophyll, hemoglobin, myoglobin, and cytochromes. The function, identification and stability of these small molecules depend upon the central atom. In hemes (which form the reactive portion of hemo- and myoglobin and cytochromes, this atom is iron. In chlorophylls, it is magnesium. The metal atom at the center of the ring imparts unique characteristics and vibrational energies to porphyrins, determining function in life and allowing their detection and unambiguous identification. Porphyrin-based molecules are extremely stable, and have been identified, almost unaltered, in organic kerogens from Carboniferous sediments [41]. This stability makes them an ideal target for molecular studies of fossil bone. Cytochrome porphyrins are also found in aerobic microbes, making this molecule a possible indicator of extraterrestrial life as well [149].

In addition to being a valuable tool for the identification and characterization of biogenic material including proteins, Raman spectroscopy is frequently used to study heme and its ligands [4], [17] and [28], such as oxygen, carbon dioxide, or carbon monoxide. Hemoglobin is a prominent component of extant bone, as bone in most taxa is highly vascular. For a listing of Raman characterizations of different groups common to organic compounds, the reader is referred to Grasselli and Bulkin [60].

NMR is routinely performed as a means of determining overall structure of biomolecules. The nuclei of atoms in certain molecules possess an inherent magnetism, or magnetic moment, which is recognized by NMR. Atomic nuclei with an even number of protons and neutrons have no magnetic moment, and are not responsive to NMR analysis [153]. This excludes naturally occurring carbon (6 protons, 6 neutrons, m = 12) and oxygen (m = 16), but isotopes of these elements can be studied by NMR. A pulsed magnetic field is applied to molecules in solution, which in turn interacts with the magnetic moment of the appropriate atomic nuclei. This causes energy splitting, which can be measured by absorbance of an appropriate light (probes and detectors operate in the radio frequency range of the EM spectrum). The chemical bonding environment of the proton can then be positively defined by the variation in positions of absorbance peaks. For example, a proton (H⁺) bound to carbon will be located in a slightly different region of the spectrum than one bound to oxygen. Additionally, NMR can be used to track changes in conformations of biologically active molecules as their chemical environments change [143]. NMR imparts detailed information about the chemical structure of molecules in an extract, but is not highly sensitive and requires relatively high concentrations of material for detection, a potential problem for analyzing fossil material. Even in low concentrations, however, NMR is adequate for detecting certain characteristics of molecules. Proteins or protein fragments that occur as complexes with metal ions (metalloproteins) are particularly sensitive to changes in magnetic field, because their electrons at the lowest energy state are unpaired, making them paramagnetic. Molecules containing paramagnetic atoms produce a large induced magnetic field easily detected by NMR. In addition, a paramagnetic compound can affect the NMR spectra of other molecules within a sample. Because the core heme of hemoglobin and cytochromes is paramagnetic, NMR is particularly useful in the study of hemoglobin and its related muscle protein, myoglobin (e.g., [72]), and it has been used in conjunction with other evidence to identify the occurrence of heme groups in extracts of fossil bone [133]. Like Raman spectroscopy, NMR has been widely used in the study of heme-containing biomolecules ([143] and references therein).

The size and complexity of molecules influence the degree to which their structure can or cannot be elucidated from an NMR spectrum. For example, the complexity of chemical interactions in an average protein consisting of 500 or more carbons would be difficult to structurally characterize via NMR alone. However, the presence of characteristic peaks can be used to identify the presence or absence of biomolecules in bulk solutions or in eluents from chromatographic separations.

Electron Paramagnetic Resonance (also Electron Spin Resonance, or ESR) is an excellent complement to NMR of paramagnetic molecules. Some atoms in molecular complexes contain electrons that exist in a state of unequal spin, detectable by EPR. The range of applications for EPR is more limited than Raman or NMR; however, it is more sensitive than NMR, and provides slightly different information, such as characteristics of electron transitions and molecular interactions [17]. EPR is well suited for verifying the presence of paramagnetic compounds in extracts of fossil bone, and can provide another line of evidence for preservation of molecular fragments [133].

Some mass spectrometry ionization methods require extracted samples in solution, rather than bulk sample as described previously for PY-GC-MS or ToF-SIMS. These are usually more sensitive and can provide more information about source molecules, provided that these fragments are not lost or modified beyond recognition in the process of extraction. These include electrospray ionization mass spectrometry (ESI-MS) and matrix assisted laser desorption ionization mass spectrometry (MALDI-MS) [102] and [107].

Electrospray ionization (ESI) is favored in the mass spectrometry of large biomolecules (i.e. peptides and proteins) because of its ‘softness’ (the ability to charge molecules for analysis without fragmenting them). Electrospray is also capable of generating multiple charges on some analytes via surface charge interactions in the ionization process, allowing more accurate mass measurement and as a result, a better chance in identifying a molecule's chemical makeup [53]. A drawback to ESI-MS is that a larger quantity of analyte may be required, because most of the components of the analyte spray do not reach the spectrometer and hence, are not measured. Larger amounts of sample diluted in solution yields better detection. The development of more advanced mass analyzers, however, makes this problem less prohibitive.

Matrix assisted laser desorption ionization, on the other hand, is a pulsed laser ionization technique where analyte molecules in solution are applied to a solid matrix and are subsequently liberated from a matrix surface for mass analysis via a rapid pulse from a UV or IR laser. This ionization technique yields singly charged, whole analyte molecules, because the process involves rapid neutralization of multiply charged analytes [79].

Although these techniques yield mainly intact charged molecules rather than fragments, it is common to couple these techniques in tandem experiments such as MS/MS where the analyte can be measured intact via ESI or MALDI and then fragmented via collision induced dissociation (CID). The fragmented analyte is then mass-analyzed by another detector. These types of experiments are most commonly performed using Triple Quadrupole Mass Spectrometry or hybrid instruments coupling multiple detectors in series (i.e., LTQ-FT-ICR, LTQ-Orbitrap and 2D-LTQ).

These techniques have in common the ionization, fragmentation and detection of molecular compounds in solution, with the intent of measuring fragment masses. These methods vary in sensitivity; however if sufficient fossil material is available for extractions and concentration, they are very informative. Mass spectrometry methods can also be used to obtain amino acid sequences if samples are otherwise purified [2], [3], [29] and [136], and can also be coupled with separation techniques to yield even more accurate molecular information.

The identification and quantification of proteins from an organism (proteomics), spans a continuum of technologies and techniques. The vast majority of these experiments employ a separation technique such as sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) or liquid chromatography (LC) to separate components from complex mixtures. Each component is then subjected to digestion by proteases that cleave proteins into peptides of manageable length. The use of trypsin, for example, predictably results in peptides with a C-terminal arginine or lysine residue. These peptide mixtures are submitted to mass spectrometric analysis via either direct infusion, using an LC coupled directly to a mass spectrometer for detection (LC-MS), or by in-line tandem mass spectrometry (MS-MS). Inside the mass spectrometer, molecules can be fragmented using collision induced dissociation with an inert gas, resulting in ion spectra from which the amino acid sequence of the parent peptide can be obtained. The sequence information, coupled with the initial mass of the parent peptide, can be used to search proteome databases to identify the original protein.

Recently, a ∼300,000-year-old mammoth skull (MOR 604) and a 68-million-year-old Tyrannosaurus rex (MOR 1125) were chemically extracted and purified via dialysis, and the resulting product was digested with trypsin and sequenced using solid phase extraction [3], [137] and [141]. Mass spectrometer analyses were performed on peaks separated by elution from a reversed phase microcapillary column coupled to a two-dimensional linear ion trap mass spectrometer. Because no genomic sequences or protein databases exist for these organisms, they were identified by searching sequence data against closely related extant species for which a database existed [2], [3] and [105]. The amino acid sequences can then be used to generate phylogenetic hypotheses, or to predict three-dimensional protein structure.

The amino acid is the basic unit of a protein. Although over 60 amino acids have been identified, virtually all naturally occurring proteins from living organisms consist of only 20 of these.

Amino acid analysis is conducted using either specialized column chromatography or mass spectroscopy. Analysis requires solubilized and relatively pure sample extracts which are hydrolyzed with strong acids or enzymes to break apart the bonds holding the constituent amino acids within the peptide chains. The amino acids are identified by a combination of elution/retention time and UV/VIS absorbance of the eluted peaks when compared to known standards.

Estimations of source proteins for identified amino acids may be based upon ratios of particular amino acids. The sequence of collagen, with its highly constrained tertiary structure, requires every third amino acid to be glycine for proper folding. Therefore, amino acid analysis of pure collagen yields approximately 33% glycine. Additionally, collagen incorporates amino acids which are unique to this protein. Proline and lysine are post-translationally modified by the addition of hydroxyl (OH) groups in the finished protein, and presence of these amino acids confirms the identity of collagen.

In addition, most amino acids possess “handedness”, either dextro (d-) or levo (l-) rotatory, because they have 4 unequal groups about the central, or chiral, carbon (Fig. 5; [8] and [130]). While all proteins in living organisms consist of only the l-form of amino acids, post mortem conversion from l- to d- forms (racemization) progresses to equilibrium (i.e., d/l = 1). The degradation of proteins over time as seen in fossil extracts yields both d- and l- amino acids. Using either a special (chiral) column or a post-column derivatizing agent allows separation of amino acids according to chirality. Many factors affect rates of racemization, but general estimates of racemization rates have been calculated for each amino acid [129], based on reaction kinetics of free amino acids in solution. These figures represent a minimum time for each amino acid to reach equilibrium, and under ideal conditions this rate may be extended by one or more orders of magnitude [46]. Chiral amino acid analyses of fossil specimens provide an independent control for identification of contaminating compounds by more recent amino acids. At least some proportion of the d-isomer of more labile amino acids should be detected in specimens older than a few hundred years. Some investigators have proposed a relationship between degree of racemization and the persistence of endogenous DNA from fossil material [116]. While a causal relationship between amino acid racemization and DNA degradation has not been shown, and protein fragments may persist in samples in which DNA has been degraded [115], chiral analyses may be another way of judging the appropriateness of a fossil for DNA analyses.

Extracts of fossil material may be separated into constituent fragments using polyacrylamide gel electrophoresis (PAGE). Samples are mixed with a special buffer that imparts a uniform negative charge to component molecules, then loaded onto a gel to which an electric field is applied. The migration rate for negatively charged fragments is determined by the molecular size of the fragment. The gel may then be stained by a variety of agents to visualize fragments for size estimates relative to known standards loaded concurrently. Extracts of extant tissues, loaded to the gel simultaneously with fossil material, provides a positive control, as well as verification of fragment size seen in fossil lanes.

Enzymes are proteins that can act with varying degrees of specificity to cut or degrade either proteins or DNA. While enzymes such as proteases (which digest proteins) are commonly used to purify DNA in a mixture by degrading protein fragments, DNAses can also be used to identify DNA in a mixture because of their specific action. Likewise, some collagenases are specific enough that degradation by the enzymes is sufficient to prove the presence of collagen in an extract. If collagenase-treated fossil extracts show smaller fragments of material on PAGE separation relative to untreated material, it provides strong evidence that collagen is present in bone extracts. However, crosslinked material is relatively protected from enzyme digestion, and this may be problematic for protein identification by this method alone. In addition, digestion of fossil extracts by various enzymes may actually have the opposite effect than that seen in modern extracts. While crosslinking is a critical factor in the preservation and retention of original organic molecules over time, enzymatic degradation may partially break these links, thus exposing more antibody binding sites in fossil bone after digestion than is seen in undigested material.

Like protein sequence data, antibody–antigen interactions have potential to address evolutionary relationships, though indirectly, and with less resolution. An antigen (ag) is any substance capable of being recognized by an antibody (ab). An antibody is a protein produced by the immune cells of host organisms in response to stimulation by a foreign substance. An immunogen is a substance capable of invoking an antibody response when introduced to a host. An antigen may not always be immunogenic, but an immunogen is almost always antigenic as well. An epitope is that specific region or three-dimensional conformation that an antibody recognizes, and is sometimes used interchangeably with the term antigen. A microbe or a single protein may have hundreds or thousands of epitopes on its surface, each of which is capable of being recognized by a different antibody. Each antibody, in turn, is produced by a different clonal population of cells.

Antibodies are produced by host B-cells in response to exposure to an immunogen. Each B cell recognizes and responds to only one epitope. Upon exposure to an immunogen, the responsive B-cell divides rapidly to produce a population of identical cells secreting identical antibodies, each of which bind to only the one epitope that stimulated its production. Antibodies are secreted into the serum of the host organism, and collected serum will contain all antibodies produced by the different clonal populations, each recognizing a different epitope of the immunogen. This is called a polyclonal serum (Fig. 6).

Monoclonal antibodies are produced by only one B-cell clonal population; therefore, a monoclonal antibody will only recognize and bind to a single epitope of a foreign molecule (Fig. 6). Monoclonal antibodies are highly specific, and if the test antigen is lacking that one epitope, for example one particular 5-amino acid turn or shape, the monoclonal ab may not bind, while the polyclonal serum from which it is derived will show reactivity. A monoclonal antibody specific to a single epitope of a protein used to immunize, but not present on the same protein from the test taxon will not demonstrate reactivity, even though the protein may be present. Therefore, monoclonal antibodies are not ideal for analyses of proteinaceous material from extinct organisms.

Proteins may be extracted from a tissue, then purified to homogeneity, dissolved in a physiological buffer (saline) and used to immunize a host animal, usually a mouse or rabbit. The host immune system responds by producing antibodies; therefore the injected protein is the immunogen. The antibodies produced in response to the collagen remain in the circulating blood and may be collected. Incubation of the antibody containing serum (antiserum) with extracts or tissues derived from either the original source or from a second source may demonstrate varying degrees of reactivity, depending upon the phylogenetic proximity of the animal providing the immunogen with the animals providing the antigen. Antibodies demonstrate greatest reactivity with molecules most similar to those used to immunize, and therefore binding patterns may reveal evolutionary distances. Thus, antibody reactivity provides a means of confirming phylogenetic hypotheses in addition to providing the means to identify the source of protein fragments preserved in fossil tissues. Antibodies can also be raised against proteins that are taxon specific, thus reducing the chances that reactivity is due to contamination.

Antibody–antigen interactions provide a robust identification of protein survival in fossil specimens when a two-fold approach is employed. Fossil extracts that react with antibodies raised against extant proteins can be used to stimulate antibody production in new host animals. If the resultant fossil antiserum binds purified proteins from extant sources, not only does it confirm the presence of protein, but identifies the source proteins in fossil material. However, producing antibodies to fossil material may require the destruction of more fossil material than is practical, as antigenic protein in fossils is extremely low in concentration, and usually degraded, modified or fragmented, compromising immunogenicity of the fossil material. Thus, the proteinaceous material in fossils may be antigenic, but too small to be immunogenic. If fossil tissues contain material that is antigenic but not immunogenic, they can be coupled to a larger carrier molecule called a hapten, making the complex large enough to be recognized by the host immune system.

Antibodies can recognize molecular fragments only 3–5 amino acid residues in length [8], and do not require that a whole molecule be present for recognition, making this a powerful method for studying degraded fossil specimens. Antibodies can be used to distinguish proteins that differ by only by a few amino acids, if those differing amino acids result in changes in three-dimensional structure.

Antibodies can also be used in ancient DNA studies, both to confirm and/or localize DNA to regions of fossil tissues, such as osteocyte lacunae within bone fragments, or they can also be used to purify DNA or proteins from heterogeneous extracts, and the purified molecules can then be used in additional studies such as genome sequencing (e.g., [117]). Antibodies raised against DNA molecules are not dependent on sequence, so no a priori assumptions regarding DNA sequence are required. While antibodies are not specific enough to distinguish DNA for phylogenetic studies, they can visually confirm the presence of the molecule in fossil tissues.

To purify proteinaceous components from fossil extracts for further and more detailed analyses, antibodies showing reactivity to fossil material may be coupled to a column, binding reactive fragments contained in fossil extracts that are passed over it. Most fossil material contains contaminants or degraded complexes of organics in addition to small amounts of endogenous material, some of which inhibit enzymes or confound other methods of analyses, and affinity purification using antibodies avoids these problems [63].

Antibody–antigen interactions are measured using various, usually complementary means. Each method provides slightly different information, and particularly when dealing with degraded fossil material, some may work, whereas others may be less effective. Several of these methods are described below.

These assays are extremely sensitive, capable of detecting nanograms to picograms of antigenic material. Solubilized antigen is applied to wells of specially designed microtiter plates. Antigen binds to the charged plastic after incubation for several hours to overnight (Fig. 7A). After removal of antigen, a blocking buffer containing a high concentration of non-relevant protein is applied to block reactive sites on the well not coated with the antigen, preventing non-specific binding of antibody. Specific antibody in solution (primary antibody) is allowed to incubate with the bound antigen, and any unbound antibodies are removed by multiple washings. A detector antibody is then applied, which recognizes a region on the primary antibody, now bound to the test (fossil) antigen. This secondary antibody is coupled to an enzyme such as alkaline phosphatase or horseradish peroxidase, and after removing excess or unbound secondary antibody a substrate is applied that causes a color change, which is monitored by a special detector that allows the degree of reactivity to be quantified. If the primary antibody does not recognize the antigen, there will be nothing to bind the secondary antibody-enzyme complex, and no color change will be detected (Fig. 7A). ELISA is relatively simple to perform, and allows reactions with non-denatured proteins or protein fragments. However, the ability of the wells to bind fossil antigen may be compromised by the presence of degraded organics such as humic acids, in extracts of the fossil tissues.

Antigen is applied to a polyacrylamide gel and separated by electrophoresis, as previously described. However, instead of applying special stains to the gel to detect the presence of protein, DNA or other organic compounds, the gel is covered with a special membrane (generally nitrocellulose), placed in a chamber filled with electrolysis buffer, and an electric field is again applied. Instead of moving the protein components downward for separation, as in electrophoresis, the field moves the components out of the gel, and onto the membrane, where they bind irreversibly. After the proteinaceous components are membrane-bound, the membrane is incubated with a blocking solution as above, then incubated in a solution of diluted antibodies several hours to overnight, allowing antibodies to bind to the separated components immobilized on the membrane. Unbound antibodies are removed, and secondary antibody, labeled with fluorescent chemical, enzyme, or other detection chemical is incubated with the membrane where, similar to ELISA, it binds the primary antibody. The membrane is then developed, and reactive fragments of fossil material are identified. Assuming that the concentration of antigenic material preserved in fossils is greatly reduced relative both to modern samples and to other co-extracting contaminants, sensitive chemiluminescent techniques are preferred over other detection systems. These are detected using radiographic film, and provide a permanent record of reactivity that will not fade, a problem with other detection systems.

Immunoblots provide information not only regarding overall antibody reactivity, but also molecular size. Antigenic components within extracts of fossil bone can be compared with electrophoresed products of known and/or purified proteins to gain insight into identity of source proteins and patterns of molecular diagenesis. Usually, however, this method relies on denaturation of antigenic material. If antibodies used to detect reactivity in fossil material are made against native (non-denatured) proteins, the probability of binding is reduced, even if protein fragments are present in the fossil extracts.

This method is an invaluable complement to the above described techniques. IHC relies upon in situ localization of antibody binding to antigen, providing a check for contamination and verifying the presence of reactive compounds in fossil matrices. The procedure is similar to that described for light microscopy, in that fossil tissues are embedded in a polymer, but one specifically designed for antibody studies. The tissues are then microsectioned to thicknesses of 1.0 or 0.5 μm, and adhered to coated glass slides. Sections are then blocked with non-relevant proteins as described above to reduce non-specific binding, followed by incubation with primary and secondary antibodies, similar to the above procedures. For IHC, an additional detector coupled to a fluorescent label is applied, greatly amplifying signal from antibody binding (Fig. 7B). Confocal microscopy or fluorescent microscopy can then be used to visualize signal arising from primary antibodies binding to regions of the sectioned tissue. IHC demonstrates that antigenic components reside within the fossil tissues and are not the result of extraction procedures or contamination. This method has been used successfully to identify proteinaceous components in various fossil tissues (e.g., [101], [103], [134], [135] and [140]).

In addition to providing the ability to visualize samples at nanoscale levels, this technique enables scientists to probe molecular ligand-receptor interactions with pico-Newton resolution. Increasing sensitivity and resolution of computer, laser and material sciences makes it possible to detect sub-nanometer movements of a cantilever tip that transverses the surface of a specimen [88]. The ability to sense interactive forces between the tip and the surface with pN sensitivity allows quantitative determination of elastic and adhesive properties of materials at nanoscale levels [78]. This is particularly useful when the probe is coupled to a specific antibody. When the probe is allowed to interact across the surface of the sample, the strength of antibody binding can be quantified [6], [42], [70] and [97]. In addition, because AF microscopes can image at the level of a single molecule, force maps obtained from ab-ag interactions can pinpoint the location of epitopes on the molecule [45], [122] and [148]. The technique is used often to study protein unfolding and its functionality. We have applied this technique to image morphology of and to test the unfolding of collagen obtained from fossils dating to the Late Cretaceous, and compared their morphological and physical properties with those obtained from extant samples [6], [138] and [140]. Fig. 8 shows demineralized Cretaceous bone (A) imaged by AFM, and compared with tendon (B) from an extant bird. In addition to visually observing whole fragments and demineralized tissues, this method may also be used to determine elastic and mechanical properties of components within tissues (Fig. 8C), and to quantify the force of ab-ag binding at nanoscale levels [42], [70] and [97].

In summary, the increasing sensitivity and resolution of chemical and molecular analytical techniques in the last decade or so has increased our understanding of biological and physiological processes in extant animals. The application of these methods to fossil specimens can likewise increase our understanding of the physiology and evolution of extinct vertebrates as well. Complete chemical characterization of cells, tissues and molecules preserved in fossil bone may also provide the means to elucidate the chemical pathways of diagenesis and degradation, as well as preservation, at the molecular level. Finally, by understanding the interactions between biological and geological components of the fossil record, we may gain additional information about the sedimentary and geochemical conditions contributing to preservation. Vertebrate paleontology is, by its very nature, a cross-disciplinary endeavor. We predict that the application of these analytical methodologies in the future will continue this trend, and that an understanding of the methods, data and interpretations will be crucial to vertebrate paleontology in the future.