Bone is a complex biological tissue that characterizes extant and fossil vertebrates, and consists of a mineralized (calcium, phosphorus) and a non-mineralized (collagen and non-collagenous proteins) extracellular matrix, plus water and some lipids (Boskey & Gehron, 2013; Rey et al., 2009). Cells involved in bone tissue are osteoclasts, osteoblasts, and the most abundant of them osteocytes (Bonewald, 2011). Osteocytes are embedded within the hard-mineralized component of bone throughout life (exceptions being when released by fracture or during remodeling) (Robling & Bonewald, 2020), providing them high preservation potential within fossil bones, which has been extensively documented in different clades of vertebrates (e.g., Bailleul, O’Connor & Schweitzer, 2019; Enlow & Brown, 1956; Pawlicki & Nowogrodzka-Zagorska, 1998; Schweitzer, 2011; Schweitzer et al., 2013; Surmik et al., 2019). Similar preservation of osteocytes- and blood vessels-like has also been documented in fossil turtles, showing that their preservation is independent of geologic time, paleoenvironment, lithology, lineages, and latitude (Cadena, 2016; Cadena, Ksepka & Norell, 2013; Cadena & Schweitzer, 2012; Cadena & Schweitzer, 2014).
Something in common to all aforementioned studies are the analytical tools used to study and characterize these fossil bone microstructures, which include principally: (1) ground sections and observation under transmitted and polarized microscopy (Cadena & Schweitzer, 2012; Surmik et al., 2019); (2) bone demineralization using ethylenediaminetetraacetic acid (EDTA) as a chelating agent (0.5 M, pH 8.0), facilitating release the osteocytes-, blood vessels-, and any other cells- or soft-tissue fibers-like from the bone matrix for their posterior study by transmitted and/or polarized light, scanning and/or transmission electron microscopy and any coupled elemental analyzer, Raman spectroscopy, Fourier-transform infrared spectroscopy (FTIR), immunological and antibody studies (e.g., Alfonso-Rojas & Cadena, 2020; Bailleul, O’Connor & Schweitzer, 2019; Bailleul et al., 2020; Cadena, 2016; Saitta et al., 2019; Schweitzer et al., 2013; Surmik et al., 2019; Wiemann et al., 2018).
The preservation of these soft-tissue microstructures (osteocytes and blood vessels) and their potential original constituents (proteins and DNA) has been questioned and considered a consequence of microbial interactions within fossil bone and its microenvironment or even as a result of cross-contamination in the laboratory (Buckley et al., 2017; Kaye, Gaugler & Sawlowicz, 2008; Saitta et al., 2019). The ‘biofilm hypothesis’ as a source for soft-tissue preservation in dinosaur bones has been rigorously tested, which identified fundamental morphological, chemical and textural differences between the resultant biofilm structures and those derived from dinosaur bone, demonstrating that the recovered microstructures in the reports cited above are endogenous in origin and that the ‘biofilm hypothesis’ should therefore be rejected (Schweitzer, Moyer & Zheng, 2016). Issues concerning cross-contamination and replications, timing of sample collections, and reagents have also been addressed by Schweitzer et al. (2019).
Compositionally, the osteocytes- and blood vessels-like from different clades of fossil vertebrates have been shown to commonly be enriched in iron (Cadena, 2016; Schweitzer et al., 2014; Surmik et al., 2019; Ullmann, Pandya & Nellermoe, 2019), an element that has been suggested to play a key role in preserving and even masking identification of proteins in fossil tissues via Fenton reactions (Schweitzer et al., 2014). Other elements typically found in these fossil bone microstructures are carbon, calcium, and silicon (Cadena, 2016; Ullmann, Pandya & Nellermoe, 2019). At present, all these studies of elemental characterization have been conducted using SEM/EDS on isolated (post-demineralization) osteocytes- and blood vessels-like, or from polished ground sections, which implies some degree of manipulation or contact with reagents or preparation tools, potentially raising skepticism on the elemental results.
Here, I explore the in situ (directly on fresh and untreated surfaces) preservation and elemental composition of bone microstructural elements (cells and blood vessels) of fossil turtle bones from three localities which have completely different geological settings (lithological, taphonomic, and fossil diagenesis), including: (1) Gobi Desert, Mongolia, from the Late Cretaceous (late Campanian–early Maastrichtian); (2) Messel Pit, Germany, from the Eocene; and (3) La Venta fauna, Colombia, from the Miocene. Comparison samples include bone from two extant turtles and a domesticated chicken. I discuss herein the results of these analyses and the advantages of using in situ SEM/EDS for understanding preservation of cells/tissues in fossils.
Materials & Methods
Fossil and extant samples
All the fossil and extant samples analyzed here were free of any resin, glue, or stabilizing additives since field collection. Two small pieces donated by Dr. M Norell (American Museum of Natural History, AMNH) from a partially- articulated shell (carapace and plastron) of Mongolemys elegans (IGM-90/42) were used for this study. Specimen IGM-90/42 has been previously figured, including ground sections that show excellent preservation of osteocytes-like under transmitted light microscopy (Cadena, Ksepka & Norell, 2013, figs. 7, 9). This fossil material was collected by the AMNH and the Mongolian Academy of Sciences joint field expeditions at the Bugin Tsav locality, Gobi Desert, Mongolia, from fine-grained sandstones representing ponds deposits within the Nemegt Formation, considered to be late Campanian–early Maastrichtian (∼80 Ma) in age (Jerzykiewicz, 2000, references therein).
Small isolated fragments from the carapace of an Allaeochelys crassesculpta (SMF ME-2449) were donated by Dr. K Smith (Senckenberg Naturmuseum Frankfurt, SMF); these were collected from the well-known locality of Messel Pit, which represents volcanically-influenced lake deposits from the early-middle Eocene (∼48 Ma) (Lenz et al., 2015). Osteocytes-, blood vessels-, and collagen fibers-like from this specimen were previously described and elementally characterized by Cadena (2016, figs. 4–7).
Carapace fragments from a podocnemidid indet. specimen, (UR-CP-0043), as well as the surrounding rock matrix, were collected in 2018 directly from an excavation site (approximately 1.5 m from the surface) using strict aseptic techniques (nitrile gloves, face mask, wrapped in sterilized aluminum foil and kept in glass containers with silica gel for moisture control until analyses were performed). This fossil material was collected from the Repartidora locality, La Victoria Formation, middle Miocene (13.6 ± 0.2 Ma), Tatacoa Desert, Colombia, from what are interpreted as fluvial deposits (Cadena et al., 2020). Permits for collecting and study of the samples were granted by the Colombian Geological Survey (Radicado No. 20193800017321).
For comparisons, two extant turtle carcasses were sampled directly in the field following the same aseptic protocols used for specimen UR-CP-0043. The first corresponds to carapace fragments from an individual of the sea turtle Lepidochelys olivacea (uncatalogued specimen) collected in January 2017 at the Pacific coast, Santa Elena Province, Ecuador, permit granted by Yachay Tech University. The second (uncatalogued specimen) sampled corresponds to a carcass of the side-necked turtle Podocnemis lewyana found in a sand bed of the Magdalena River, close to La Victoria village, Huila Department, Colombia, under a permit granted by the ethics committee of Universidad del Rosario (Resolución DVO005 672-CV1066) and the Colombian Autoridad Nacional de Licencias Ambientales (Technical concept No. 02263, 2019). A third sample corresponds to a femur fragment from a commercial chicken Gallus gallus obtained directly from a local market. Muscle tissue was removed and small bone fragments were cut using a sterilized scalpel and dried out at room temperature for several days.
Scanning electron microscopy and elemental analysis (SEM/EDS)
Each of the fossils, rock matrix and extant bone samples were placed between two disposable sterilized lab-weighing boats and gently hit with a rock hammer to break them into smaller pieces. Using tweezers (sterilized before every mounting process) one of the smaller pieces of broken bone was transferred to an SEM holder with adjustable screws and secured. To prevent any potential particles or dust from entering the SEM chamber, each sample was gently air cleaned before placing it in the SEM carousel. Elemental analysis was performed in combination with high resolution imaging of the bone surfaces, as well as (in some cases) the rock matrix attached to it using a scanning electron microscope coupled with an energy-dispersive X-ray spectroscopy analyzer (Phenom ProX, at the Paleontological Lab of Yachay Tech University (YTU), San Miguel de Urcuqui, Ecuador). Imaging was performed at 5 kV using different magnification settings, and point-and-map analyses of elemental composition of selected regions or features were performed at 15 kV. At least five or more points were explored for each osteocyte- or blood vessel-like, as well as the surrounding bone matrix or rock. Quality of EDS analyses was evaluated considering only those with one million counts or higher. Full raw data is presented in Data S1.
In order to test for the occurrence and preservation of osteocytes- and blood vessels-like in some of the samples, small bone pieces of Mongolemys elegans (IGM-90/42) and the podocnemidid indet. specimen (UR-CP-0043) were demineralized using disodium ethylenediaminetetraacetic acid (EDTA) (0.5 M, pH 8.0 filter-sterilized using a 0.22 μm filter) as previously described (Cadena, 2016; Cleland et al., 2015) for a period of five days to two weeks, or until osteocytes- and blood vessels-like were detected. Photographs of the recovered osteocytes-like were taken using a transmitted light microscope (Olympus BX-63) and a polarized light microscope (Olympus BX-53) at the paleontological lab of YTU. Some of the isolated osteocytes-like from IGM-90/42 were collected with a tip in a 1.5 ml tube, rinsed three times with E-pure water to get rid of EDTA, being centrifuged at 1500 RPM for 2 min between step. A drop of the supernatant was mounted in a stub, dried out at room temperature inside in a sealed small SEM-stub box to avoid any air or dust particles interact with the sample, and analyzed following the same protocol and SEM/EDS machine aforementioned.
Mongolemys elegans Late Cretaceous of Mongolia
The in situ osteocytes-like of Mongolemys elegans (IGM-90/42) under SEM exhibit a distinct contrast with the surrounding bone matrix, which is exclusive of their three-dimensional volume, and it is also different from the empty osteocytes-lacuna, which exhibits the same contrast as the bone matrix (Figs. 1A–1B). Compositionally, they are predominantly composed of iron, calcium, carbon, manganese, and minor amounts of barium and nitrogen (Figs. 1C–1K; 2A; Data S2; and Fig. S1). There is no evidence of any of these elements in empty osteocyte-lacunae walls, which are composed of calcium and phosphorus, like the bone matrix (Figs. 1L–1N). The isolated osteocytes-like show that iron is concentrated on their external surface and the manganese in the internal, this is clearly evident in elemental maps and a cross-line elemental profile (Figs. 1O–1P; Fig. S1). Observation of some of the isolated (post-demineralization) osteocytes-like under transmitted and polarized light revealed excellent morphological preservation, with some of them emitting low-degree birefringence colors under polarized light (Fig. 3).
Allaeochelys crassesculpta, Eocene of Germany
The most abundant bone microstructures preserved in this sample are blood vessels-like and the walls that formed the Haversian-Volkmann-like (H-V) canals; also, in some, there is evidence of very small (2.5 µm diameter) structures with a striated margin which resemble the morphology of osteoblast cells (Figs. 4A–4D; Fig. S2). The blood vessels-like exhibit a width of 1–3 µm, with an average wall thickness of 0.2 µm (Fig. 4D). Compositionally, the blood vessels-like are mainly composed of carbon and nitrogen, with minor amounts of calcium, phosphorus and iron (Fig. 2B; 4E–4G; 4J –4N; Fig. S2; Data S2). The bone matrix surrounding them lacks nitrogen and carbon, and it is exclusively characterized by calcium, phosphorus, and iron (Figs. 4E, 4I). A bone sample with rock matrix attached shows that the bone is composed of calcium, phosphorus, iron, and nitrogen, and, in contrast, the rock matrix is rich in aluminum and silicon (Figs. 2C–2D; Figs. 4O–4Q; Data S2).
Podocnemidid indet, Miocene of Colombia
The sample of the side-necked turtle from La Venta, Colombia, shows on the bone external cortex preservation of walls that formed the H-V-canals, blood vessels- and osteocytes-like tightly embedded in the very homogenous bone matrix (Figs. 5A–5B; 5F–5G). Elementally, the blood vessels-like and H-V-canal walls are rich in carbon, nitrogen, and calcium, with minor amounts of phosphorous and silicon (Figs. 2B; 5C–5D; 5H–5I; Data S2). In contrast, the osteocytes-like are composed of iron, calcium, aluminum, manganese, phosphorus, and minor amounts of silicon (Figs. 2B; 5H–5J; Data S2). The bone matrix lacks carbon and nitrogen, and it is constituted by calcium and phosphorus mainly (Figs. 2C; 5C, 5E 5H). An isolated bone fragment (post-demineralization) shows some of the osteocytes-like still embedded in the matrix, varying in color from orange to black, the darker ones located closer to black, dendritic mats (Figs. 5K–5M).
In situ extant turtle and chicken bone microstructures
The carapace bone fragment of the extant side-necked turtle Podocnemis lewyana shows osteocytes within lacunae (Figs. 6A–6B). Their composition is rich in carbon, nitrogen, calcium, and phosphorus (Fig. 2A; 6C–6D; Data S2). The bone matrix is relatively richer in calcium (Fig. 2C; 6C, 6E). The H-V canals exhibit a distinct wall and a high concentration of blood vessels and red blood cells, which are rich in carbon and nitrogen (Fig. 2B; 6F–6G; Figs. S3; S4; Data S2). Similar spatial patterns and composition are shared by the bone of the extant marine turtle Lepidochelys olivacea (Fig. 2; 6H–6M; Data S2), and the bone of Gallus gallus (chicken) (Fig. 2; 6N–6P; Data S2).
As previously shown (Cadena, 2016; Schweitzer et al., 2014; Surmik et al., 2019; Ullmann, Pandya & Nellermoe, 2019), the in situ analyses presented here, concur with that iron is a very common constituent of fossil osteocytes-like, such as those found in the Late Cretaceous Mongolemys elegans and the Miocene podocnemidid indet. bone samples studied herein (Figs. 1 and 5). However, this composition is not always homogenous and may vary between the external and the internal layer of osteocytes-like, as shown in a broken and folded osteocyte-like from M. elegans, which exhibits richer content of manganese internally and iron externally (Figs. 1O–1P). High levels of manganese were also detected in osteocytes-like from the Miocene side-necked turtle from Colombia, indicating that besides iron as initially suggested by Schweitzer et al. (2014), manganese may also be involved in the preservation of these bone microstructures in deep time. The source for this rich content of manganese seems to be from manganese oxides such as pyrolusite penetrating bone microfractures, which were found herein in some fragments of the Miocene podocnemidid indet. from Colombia (Figs. 5K–5M), and also has been characterized to occur in dinosaur fossil bones from the same Nemegt Formation, from which the M. elegans studied herein was collected (Owocki et al., 2016). The color variation exhibited by the fossil osteocytes-like of M. elegans and podocnemidid indet. seems to be related to enrichment of manganese, higher their manganese content darker their color. In contrast to the osteocytes of the extant turtle and chicken bone, which are rich in carbon and nitrogen (Figs. 2 and 6), these elements only appear in minor amounts in fossil osteocytes-like. However, these cells exhibited a very distinct composition when compared to the surrounding bone matrix and even the wall surfaces of their osteocytes-lacunae, indicating that their mineralized preservation occurred at micro-scale inside the bone, a hypothesis that should be tested by future studies using additional tools (e.g., Raman and FTIR spectroscopy).
The blood vessels- and H-V canal walls-like preserved in the Eocene Allaeochelys crassesculpta from the Messel Pit and the Miocene podocnemidid indet. specimen from the La Venta not only exhibited a similar morphology, but also exhibited the same elemental composition as their corresponding tissues in extant turtle and chicken bone. (Figs. 2, 4 and 5, Figs. S2– Figs. S4; Data S2). In both cases (extant and fossils) being rich in carbon and nitrogen, and differing from the surrounding bone matrix which is richer in calcium and phosphorus, or the rock matrix which is rich in silicon and aluminum (without any traces of carbon, calcium, or nitrogen) which suggests that carbonates or nitrates were absent in the surrounding microenvironment. The in situ measurements performed on some of the preserved blood vessels-like from A. crassesculpta, exhibiting uniform fabric and thin walls of 0.2 µm thickness (Fig. 4D) suggest that they are not consistent with the characteristics of biofilms, which tend to be amorphous and larger in diameter (Schweitzer, Moyer & Zheng, 2016). Blood vessels constitute one of the most promising microstructures preserved in fossil turtles for molecular paleontology studies, and future studies should focus on their molecular in situ characterization using ToF-SIMS mass spectrometry, similarly as it has been used in dinosaurs and other fossil vertebrates (Alfonso-Rojas & Cadena, 2020; Henss et al., 2013; Lindgren et al., 2018; Schweitzer et al., 2019).
For the first time, I herein report the preservation of osteoblasts-like in fossil vertebrates, particularly in the Messel Pit turtle A. crassesculpta. They occur as oval objects with striated margins that are attached to the H-V canals (Figs. 5B–5D; Fig. S2), and thus resemble the morphology and size of osteblasts observed in electron micrographs of human bone (Nakamura, 2007; Schmidt et al., 2002). At the same time, evidence here provided from the Miocene podocnemidid indet. turtle from Colombia (Figs. 5F–5J) shows that, in the same bone specimen, osteocytes- and blood vessels-like that are only 20 µm away from each other are compositionally different. This indicates that each microstructure went through a different preservational pathway. Osteocytes-like seem to be more mineralized than blood vessels-like in these fossil samples, with high amount of iron and manganese, and less organic components than blood vessels-like (Figs. 2 and 3). In the extant bone of turtles and chicken, osteocytes and blood vessels exhibit similar elemental composition under SEM/EDS, both being rich in carbon and nitrogen, which are typically present in abundance within proteins (Torabizadeh, 2011) (Figs. 2 and 6). A similar composition was detected herein in fossil blood vessels from A. crassesculpta from Germany and the podocnemidid indet., from Colombia (Figs. 4 and 5).
Traditionally, it has been suggested that SEM/EDS has to be performed on homogenous or polished surfaces to avoid topographic effects on EDS analyses (Goldstein et al., 2003). However, as I showed here, such effects were negligible for the analyzed samples with composition and signal intensities being very similar in both the fossil and extant samples (Fig. 2). I therefore suggest that a more critical condition for EDS analysis on untreated samples is acquire the highest maximum count rate possible; above 1 million counts is ideal.
This study provided evidence that in situ analyses using a conventional technique, SEM/EDS, on untreated fresh surfaces of fossil and extant bones constitutes a protocol that should be added to the rigorous plethora of proxies and tools (e.g., those recently reviewed and summarized by Schweitzer et al. (2019) to support and demonstrate the preservation of cells, soft-tissues and their original constituents in deep time. Furthermore, in situ analyses of fossil and extant bone samples may also help eliminate any potential skepticism of results obtained by molecular paleontology studies, because, as demonstrate here, it requires minimal sample preparation/manipulation, use of reagents, or contact with lab tools that could cause possible contamination.