Preservation of Bone Collagen from the Late Cretaceous Period ...
Published on: Mar 4, 2016
Transcripts - Preservation of Bone Collagen from the Late Cretaceous Period ...
Preservation of Bone Collagen from the Late Cretaceous
Period Studied by Immunological Techniques and Atomic
R. Avci,*,† M. H. Schweitzer,‡,§ R. D. Boyd,† J. L. Wittmeyer,‡ F. Teran Arce,† and
J. O. Calvo|
Montana State University, Department of Physics, EPS 264, Bozeman, Montana 59717,
North Carolina State University, Department of Marine, Earth and Atmospheric Sciences,
Raleigh, North Carolina 27695, North Carolina Museum of Natural Sciences,
Raleigh, North Carolina 27601, and Museo de Geologia Palentologia, Universitad Nacional
del Comahue, Neuquen, Argentina
Received September 16, 2004. In Final Form: February 1, 2005
Late Cretaceous avian bone tissues from Argentina demonstrate exceptional preservation. Skeletal
elements are preserved in partial articulation and suspended in three dimensions in a medium-grained
sandstone matrix, indicating unusual perimortem taphonomic conditions. Preservation extends to the
microstructural and molecular levels. Bone tissues respond to collagenase digestion and histochemical
stains. In situ immunohistochemistry localizes binding sites for avian collagen antibodies in fossil tissues.
Immunohistochemical studies do not, however, guarantee the preservation of molecular integrity. A protein
may retain sufficient antigenicity for antibody binding even though degradation may render it incapable
of original function. Therefore, we have applied atomic force microscopy to address the integrity and
functionality of retained organic structures. Collagen pull-off measurements not only support immu-
nochemical evidence for collagen preservation for antibody recognition but also imply preservation of the
whole molecular integrity. No appreciable differences in collagen pull-off properties were measured between
fossil and extant bone samples under physiological conditions.
Introduction compared with similarly treated material from related
We have previously demonstrated by immunological extant organisms and can be used to address the pres-
methods that protein epitopes may survive across geo- ervation of molecular function.
logical time in fossil materials that demonstrate mor- Recently, exceptionally well-preserved avian eggs con-
phological and histological integrity,1-3 and we have shown taining embryonic skeletal elements were described6 from
that antibody-antigen interactions can be used to identify exposures of the Bajo de la Carpa Member of the Upper
and characterize organic components of fossils. However, Cretaceous Rio Colorado Formation in the city of Neuquen,
measurable antibody-antigen interactions require only Argentina. The strata containing the eggs have been
a few intact amino acids4 to show reactivity; hence, interpreted as interdune deposits corresponding to an
immunological methods alone do not provide information eolian environment with fluvial influences, in a semiarid
regarding the integrity or mechanical properties of protein climate.7 Cladistic analysis of both the bones and the
fragments in fossil material. Atomic force microscopy eggshell resulted in their assignment to a basal bird, most
(AFM) has the potential to complement and extend other likely an enantiornithine.6 Thin and fragile embryonic
analytical methods applied to fossils by determining the bone tissues surrounded a calcite core that had precipi-
integrity and mechanical properties of organic molecules tated within the original medullary space of the hollow
under physiological conditions, and indeed, AFM in long bones.6 The distinctive taphonomy and preliminary
combination with laser Raman spectroscopy has been evidence of unusual preservation in gross and histological
applied to characterize Precambrian microfossils and shed preparations of these embryonic tissues warranted further
light on the geochemical maturation of ancient organic chemical and molecular analyses.
matter.5 Here, we show that data obtained by AFM, either Because collagen proteins are key components of
alone or in addition to other analytical tools, can be directly virtually all vertebrate tissues and are extremely well
* Corresponding author. E-mail: firstname.lastname@example.org. characterized across a wide range of both tissues and taxa,
Phone: (406) 994-6164. Fax: (406) 994-6040. we have selected this protein as a model system with which
† Montana State University. to demonstrate its morphological and unfolding properties.
‡ North Carolina State University.
We have used a combination of high-resolution imaging
§ North Carolina Museum of Natural Sciences.
| Universitad Nacional del Comahue.
and force-extension measurements and have compared
(1) Schweitzer, M. H.; Watt, J. A.; Avci, R.; Knapp, L.; Chiappe, L.;
the results with those of partially demineralized extant
Norell, M.; Marshall, M. J. Exp. Zool. 1999, 285, 146-157. and fossilized bone tissues and nonmineralized (tendon)
(2) Schweitzer, M. H.; Watt, J. A.; Avci, R.; Forster, C.; Krause, D. tissues.
W.; Knapp, L.; Rogers, R.; Beech, I.; Marshall, M. J. Vert. Paleo. 1999,
(3) Schweitzer, M. H.; Marshall, M.; Carron, K.; Bohle, D. S.; Busse, (6) Schweitzer, M. H.; Jackson, F. D.; Chiappe, L. M.; Schmitt, J. G.;
S.; Arnold, E.; Johnson, C.; Starkey, J. R. Proc. Natl. Acad. Sci. U.S.A. Calvo, J. O.; Rubilar, D. E. J. Vert. Paleo. 2002, 22, 191-195.
1997, 94, 6291-6296. (7) Heredia, S.; Calvo, J. O. In Actas del XV Congreso Geologica
(4) Child, A.; Pollard, M. J. Arch. Sci. 1992, 19, 39-47. Argentino; Cabaleri, N., Cingolani, C. A., Linares, E., Luchi, M. G. L.
(5) Kempe, A.; Schopf, J. W.; Altermann, W.; Kudryavtsev, A. B.; d., Panarello, H. O. O. y., Eds.; Calafate: Santa Cruz, Argentina, 2002;
Heckl, W. M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 9117-9120. article 196197 (CD-ROM).
10.1021/la047682e CCC: $30.25 © xxxx American Chemical Society
Published on Web 00/00/0000 PAGE EST: 6.8
B Langmuir Avci et al.
The vast majority of fossilized remains are those “hard” forming cross-links that stabilize and protect the polymer
tissues resulting from biomineralization of organic ma- backbone. No molecular-level understanding of the origin
trixes during life, typically including bones and teeth. of the sacrificial bonds has been offered at this time;
Interplay between protein and mineral in biomineralized however, experimental evidence for their existence is
vertebrate tissues is complex but invariably contributes irrefutable. The resilience and flexibility of bone is derived
to the potential of preservation in the fossil record. A better in part from these bonds breaking when stressed, thus
understanding of these natural “composite materials” is dissipating stress energy in a nondestructive manner.
crucial to our understanding of a range of processes, from Sacrificial bond breakage is measured by AFM and
diagenetic changes occurring during molecular degrada- reflected in a characteristic saw-toothed pattern18 seen in
tion to preservation of vertebrate remains over geological AFM force-extension curves. These intermolecular bonds
time spans. The diagenetic alteration of fossil tissues at and interactions give collagen fibrils great tensile strength
the molecular level is not well understood; however, it and increase its resistance to heat and chemical degrada-
has been shown that, in biomineralized tissues, proteina- tion,18 characteristics that make collagen a frequent target
ceous matrix components may be stabilized and protected for molecular investigations of fossil specimens.
by association with (extrafibrillar) and incorporation into Preliminary experiments suggested that collagen may
(intrafibrillar) mineral crystals.8-11 Atomic force micros- be present in some elements of these Cretaceous embryonic
copy (AFM) studies demonstrate that, in extant (bovine) bones. A comparison between these fossil and extant
bone, the ratio of extrafibrillar to intrafibrillar crystals is tissues using immunological and AFM techniques has been
about 70-80% to 20-25%.12 undertaken in the present study. Bone tissues from an
In biomineralized tissues such as bones and teeth, extant neornithine hatchling bird (courtesy of J. Horner,
charged amino acid residues within the protein component Museum of Rockies, Bozeman, MT) were used for com-
interact with hydroxylapatite mineral crystals to form a parative analyses, and nonmineralized, paraformalde-
highly oriented, flexible, and strong composite. The organic hyde-fixed tendons from extant chicken were similarly
matrix consists primarily of collagen (∼90%),13 a fibrillar prepared and analyzed.
protein composed of amino acids dominated by glycine,
4(R)-hydroxyproline, and proline residues. An individual Materials and Methods
type I collagen molecule, herein referred to as a collagen Tissue Preparation. Isolated chicken tendons were fixed in
fibril, has a length of ∼300 nm and width of ∼1.4 nm14 4% paraformaldehyde in sodium-cacodylate buffer, rinsed with
and behaves like a flexible nanostring.15 The tertiary cacodylate buffer, and manually separated into individual fiber
structure of the protein incorporated into mineralized bundles. Small fragments of extant avian hatchling bone and
tissues consists of three polypeptide chains in a helical fossil embryo bone were partially decalcified for 2-8 h in 0.5 M
arrangement, and recent AFM studies have provided EDTA. All specimens were rinsed in sterile water, dehydrated
further structural information by showing that, rather in an ethanol series (30, 50, 75, 95, and 100%), and then
than being homogeneous, collagen fibers (a bundle of equilibrated with LR White Hard Grade embedding resin (London
Resin Company Ltd, Berkshire, U.K.) at 4 °C with 10 changes
individual fibrils) are mechanically nonuniform.16 The
of resin. To achieve total infiltration, samples were placed under
“stacking” of helices to form the three-dimensional func- a vacuum for 24 h and then polymerized at 55 °C for 24 h. Sections
tional protein fibers results in a characteristic pattern of (∼250 nm) were taken on a clean diamond knife using a Sorvall
∼67-nm bands, consistent across taxa and verified in ultramicrotome and then heat-fixed at 55 °C to gelatin-coated
ultramicrotomed sections of both tendon and bone collagen glass microscope slides for immunohistochemical analysis or to
by AFM studies.12,13,17-19 8-mm gelatin-coated circular glass or IR transparent Ge disks
The structure of collagen imparts unique properties to for AFM, Fourier transform infrared (FTIR), and time-of-flight
the molecule. One of these properties is for the molecule secondary ion mass spectroscopy (ToFSIMS) analysis. A sche-
matic is given in Figure 1.
to act as a nucleus for mineral precipitation in biomin-
Immunohistochemistry and Collagenase Experiments.
eralized tissues. The ability of bone to resist sudden Bone sections (250 nm) were incubated with 0.5 M EDTA for 15
mechanical impact is attributed to so-called sacrificial min to further decalcify, rinsed with a phosphate-buffered saline
bonds between collagen fibrils across the network of the (PBS, 12 mM PO4- and 150 mM Na+), and etched over three
composite material.18 These (sacrificial) bonds are at- 10-min incubations with 1 mg/mL sodium borohydrate. Sections
tributed to inter- and/or intramolecular noncovalent bonds were rinsed multiple times with 25 mM Tris, 5 mM CaCl2 (enzyme
reaction buffer) and allowed to equilibrate in this buffer for 3 h.
(8) DeNiro, M. J.; Weiner, S. Geochim. Cosmochim. Acta 1988, 52,
Following incubation, all sections were rinsed three times in
2415-2423. enzyme reaction buffer, followed by being rinsed three rinses in
(9) Weiner, S.; Traub, W.; Elster, H.; DeNiro, M. J. Appl. Geochem. PBS. The sections for immunohistochemical analysis were
1989, 4, 231-232. incubated in 4% normal goat serum (NGS) in PBS for 5 h to
(10) Glimcher, M. J.; Cohen-Solal, L.; Kossiva, D.; Ricqles, A. d. reduce nonspecific staining and were then incubated as follows:
Paleobiology 1990, 16, 219-232. (1) test condition of purified polyclonal antiserum (10 µg/mL in
(11) Sykes, G. A.; Collins, M. J.; Walton, D. I. Org. Geochem. 1995, 4% NGS) raised against chicken collagen I (Chemicon, AB752P);
(12) Sasaki, S.; Tagami, A.; Goto, T.; Taniguchi, M.; Nakata, M.; (2) 4% NGS not containing primary antibodies; (3) inhibited
Hikichi, K. J. Mater. Sci.: Mater. Med. 2002, 13, 333-337. antibodies (500 µL of 10 µg/mL primary antibodies inhibited
(13) Mbuyi-Muamba, J.-M.; Dequeker, J.; Gevers, G. Baillere’s Clin. with 6.4 mg of lyophilized chicken collagen, 2.5 h at 21 °C); (4)
Rheumatol. 1988, 2, 63-100. sections were incubated for 3 h in collagenase enzyme (Sigma,
(14) Voet, D.; Voet, J. G. Biochemistry, 2nd ed.; John Wiley and Sons: C-1639) (1 mg/mL in enzyme reaction buffer), and controls were
New York, 1995; p 156. incubated in reaction buffer only. Sections were incubated with
(15) Sun, Y. L.; Luo, Z. P.; Fertala, A.; An, K. N. Biochem. Biophys.
Res. Commun. 2002, 295, 382-386. primary antibodies or controls for a duration of 12 h at 4 °C.
(16) Gutsmann, T.; Fantner, G. E.; Venturoni, M.; Ekani-Nkodo, A.; Sections were then incubated sequentially with biotinylated Goat
Thompson, J. B.; Kindt, J. H.; Morse, D. E.; Fygenson, D. K.; Hansma, x-Rabbit IgG (1:500, Vector) for 1.5 h at 21 °C and Avidin-FITC
P. K. Biophys. J. 2003, 84, 2593-2598. (1:500, Vector) for 1.5 h at 4 °C (in the dark). Each incubation
(17) Ottani, V.; Raspanti, M.; Ruggeri, A. Micron 2001, 32, 251-260. was followed by three rinses in PBS/0.1% Tween 20 and twice
(18) Thompson, J. B.; Kindt, J. H.; Drake, B.; Hansma, H. G.; Morse, with PBS to remove unbound antibodies.
D. E.; Hansma, P. K. Nature 2001, 414, 773-775.
(19) Avci, R.; Schweitzer, M.; Boyd, R.; Wittmeyer, J.; Steele, A.; For humic immunological experiments, a 5 mg/mL stock of
Toporski, J.; Beech, I.; Arce, F. T.; Spangler, B.; Cole, K.; McKay, D. S. humic acid in PBS (Fluka Chemika, 53680) was diluted to 0.5
Langmuir 2004, 20, 11053-11063. and 0.25 mg/mL in pure water. Three 0.5-µL drops per well were
Preservation of Late Cretaceous Bone Collagen Langmuir C
for ∼24 h prior to washing, drying, and imaging. Following
incubation, the sections were rinsed five times in PBS and imaged
in air using tapping mode AFM.
Humic Sample Preparation. In addition to immunological
controls, FTIR and AFM pull-off experiments were conducted on
complex humic substances (Fluka Chemika, 53680). For FTIR
experiments, a solution of humic substances was prepared at an
∼1 mg/mL concentration in pure water, and ∼400-µL drops were
spotted onto IR transparent Ge disks (WJD-U25, Harrick Sci.
Co., Broadway, NY) and dried overnight. Additionally, 250-nm
sections of fossil and extant bone, chicken tendon, and embedding
resin were placed directly on the Ge disks and allowed to dry
thoroughly before being subjected to FTIR experiments. Absor-
bance was determined using an FTIR spectrometer (Nicholet,
model 740) in transmission mode.
Humic complexes were immobilized on amino-functionalized
Si3N4 wafers by the following procedure: ∼400 µL humic solution
was diluted in 0.1 M MES (containing 0.5 M NaCl, pH 6.0) and
mixed with 0.4 mg of EDC and 0.6 mg of NHS. This mixture was
allowed to react for 15 min at room temperature before 1.4 mL
of 2-mercaptoethanol was added. The amino-functionalized
wafers were incubated in this solution for 2 h at room temperature
and then transferred to a beaker containing 1 mL Tris (20 mM,
pH 7.3). The wafers were rinsed four times with deionized water
and air-dried as described above for bone and tendon sections.
AFM Measurements. All AFM images were taken using a
Multimode nanoscope IIIa system, equipped with a vertically
engaged J-scanner. Si cantilevers (TAP3000 HD, NanoDevices,
Santa Barbara, CA) having nominal resonance frequencies of
∼300 kHz and spring constants of ∼40 N/m were used to obtain
air tapping mode images (Figure 1) from a 4 × 4 µm2 area of
tissue embedded and sectioned as described above. After obtain-
ing images in air, the tip was replaced with a silicon-nitride tip
(Veeco Metrology) with a spring constant of 0.01 N/m, and a
contact mode image of the bone section was performed in air
with the soft tip to ensure that the 4 × 4 µm2 area previously
identified by the stiff tip was not lost. In some cases, a Dimension
300 AFM system (Veeco Metrology) was used to obtain AFM
tapping mode images in air to accommodate samples larger than
the Multimode instrument is able to handle (∼1 cm diameter).
Finally, PBS (100 µL) was injected into the liquid cell and
incubated for 1 h to hydrate and equilibrate tissues before AFM
measurements in liquid.
Force-volume measurements were conducted over the 4 × 4
µm2 area divided into 32 × 32 pixels. Each force curve was
obtained with a 1-Hz frequency and had a z-scan range of 500
nm and a maximum deflection set point of 50 nm (corresponding
to a maximum load of 0.5 nN). After force measurements, AFM
images in liquid (either contact or tapping mode) were obtained
to ensure that the bone section under study was intact.
FTIR and ToFSIMS Analysis. For FTIR analysis, fossil bone
sections and similarly prepared sections of chicken tendon, extant
bone, and resin were placed on IR transparent Ge disks and
Figure 1. AFM images obtained in air on 250-nm-thick sections examined in the spectral range from 800 to 1800 cm-1 (amide
of resin-embedded tissues. The schematic representation (center band region).
part) shows a sample embedded in resin and sectioned using To give support to our conclusion that fossil material contains
an ultramicrotome to 250 nm. The AFM images obtained in air organic substances, these sections were analyzed using imaging
are the following: (A) Embedding resin with no specimen. The time-of-flight secondary ion mass spectroscopy (ToFSIMS) in
sectioning marks are clearly visible, but resin is otherwise
order to assess the organic and inorganic material content and
featureless. (B) Partially decalcified extant bone. The fibrillar
distribution of these sections. No special sample preparation was
nature of collagen matrix is apparent but somewhat obscured
by overlying mineral. The 67-nm banding pattern characteristic needed for these studies: the same sample that was used in the
of collagen I is visible (e.g., see arrow) on individual fibers. (C) FTIR studies worked excellently for ToFSIMS analysis. We
Fossil bone, visualized under identical parameters. Scattered employed the PHI-EVANS’s TRIF I ToFSIMS system housed at
fiberlike features are visible among the wavy texture, but the Image and Chemical Analyis Laboratory at Montana State
banding is not discernible. (D) Extant chicken tendon, as above. University (www.physics.montana.edu/ical).
The 67-nm banding pattern is obvious across the tendon. Statistical Analysis. A MatLab program was used to analyze
the force curves obtained by the force-volume technique. Each
heat-fixed to PTFE-coated microscope slides (EMS, 63419-08) pull-off curve was analyzed by first finding the zero force line;
and then treated with antisera or controls as described above. the intersection of this line with the pull-off curve (blue curves
Cover slips were applied using Vectashield mounting media in Figure 3) identifies the tip contact point in the blue curve.
(Vector), and sections were imaged using a compound microscope This point corresponds, for example, to ∼72 nm in extant bone
(Zeiss, Axio Skop2) equipped with a FITC filter. Exposure times (Figure 3B). The MatLab program identifies the unbinding events
ranged from 150 to 700 ms. by searching the local minima.20 For example, the second event
For AFM collagenase experiments, ∼150 µL of collagenase
(with a concentration of ∼2.5 mg/mL in enzyme reaction buffer) (20) Arce, F. T.; Avci, R.; Beech, I.; Cooksey, K.; Wigglesworth-
was pipetted onto the embedded sections and allowed to incubate Cooksey, B. Biophys. J. 2004, 87, 4284-4297.
D Langmuir Avci et al.
Figure 3. Force curves obtained on sections of (A) fossil embryo
bone, (B) extant bird bone, (C) resin, and (D) tendon. The green
lines correspond to tip approach and the blue lines to pull-off
curves. Sharp minima (e.g., arrow in part B) mark the unbinding
events. The more gradual slope of the approach curve seen in
fossil bone after tip-surface contact suggests that fossil bone
is more flexible (“softer”) than other surfaces examined,
consistent with partial degradation of collagen bone matrix. In
some cases, the fossil bone surface shows elastic degradation,
indicated by a separation in the blue and green lines during tip
contact. The red dotted lines correspond to slopes of the force
curves just before unbinding.
Results and Discussion
To maximize the number of results and controls working
with a very small amount of fossil material, we conducted
Figure 2. Immunohistochemical images obtained using optical experiments on sections that varied in thickness from 200
fluorescence microscopy on partially demineralized 250-nm to 300 nm, with tissue sizes ranging from 100 to 200 µm
sections of extant hatchling (A, C, E, and G) and fossil embryonic in diameter. The immunohistochemical work with optical
(B, D, F, and H) avian bone: (A and B) Sections are incubated fluorescence microscopy and AFM studies uses similarly
with primary antibodies against avian collagen I followed by prepared sections. Here, we introduce the schematic of a
secondary antibodies conjugated with a fluorescent label.
Osteocyte lacunae are marked by white ellipses. The scale bars
typical thin section, depicted in the center part of Figure
in each part represent 20 µm. The green color marks the 1. The other parts, labeled as A, B, C, and D in Figure 1,
locations of reactive collagen epitopes, demonstrating that the correspond to the morphological images associated with
immunological response of fossil collagen is similar to that of resin, extant bone, fossil bone, and tendon sections,
extant collagen. (C and D) To demonstrate the specificity of respectively. The discussion of these samples and their
primary antibody binding, anti-collagen antibodies were blocked morphologies is deferred to the AFM Results section below.
by incubating first with collagen and then exposed to bone Immunohistochemical Results. In situ immunohis-
sections. (E and F) To demonstrate that collagen proteins are
the target of the antibodies, sections were digested with the tochemical analyses identified antigenic material within
collagen-specific enzyme collagenase and then incubated with samples. Scanning confocal laser microsocopy (SCLM, not
primary and secondary antibodies as in parts A and B. (G and shown) and optical fluorescence microscopy (Figure 2) were
H) To control for nonspecific binding of secondary antibodies, used to monitor antibody binding on the labeled sections.
sections were incubated with secondary antibodies only, and Antibody binding was observed in both extant and fossil
no primary anti-collagen antibodies were added. All other bone exposed to collagen antiserum (green patches in
parameters were identical to parts A and B, and the imaging Figure 2A,B). The fluorescent signal corresponding to
conditions were identical in all cases.
antibody-antigen interactions was significantly greater
than that seen in the negative controls. When collagen-
from the left in the extant curve (marked by an arrow in Figure specific antibodies were first incubated with excess avian
3B) appears at ∼208 nm at ∼ -0.52 nN. The -0.52 nN collagen to block binding sites on collagen antibodies and
corresponds to ∼52 nm of cantilever deflection. From these data, then incubated with sections as described, the fluorescent
the collagen extension is then calculated by subtracting the
signal was completely blocked, testifying to the specificity
cantilever deflection and the displacement associated with the
contact point from the displacement associated with the particular
of antibody-antigen interactions in these samples (Figure
unbinding event; the example here yields L ) 208 - 72 - 52 ) 2C,D). When sections were subjected to digestion with
84 nm. The unbinding force, calculated as the difference between collagenase and then incubated as described for Figure
the forces corresponding to the minimum (-0.52 nN) and the 2A,B, the immunological signal was significantly reduced
maximum to its right (-0.34 nN), yields an unbinding force value (Figure 2E,F). Finally, sections incubated with no primary
of -0.18 nN. Only those forces greater than 0.025 nN were antibody, but otherwise identical in treatment with the
considered to distinguish unbinding events from instrument test conditions, show that the fluorescent signal is not
noise. due to nonspecific interaction of secondary antibodies with
Preservation of Late Cretaceous Bone Collagen Langmuir E
tissue but dependent upon interactions of the anti-collagen 3A), corresponding to sudden changes in the tip equilib-
antibodies with epitopes preserved within the tissue rium as it returns closer to the zero force line. In the
sections (Figure 2G,H). Osteocyte lacunae, delineated by examples shown in Figure 3, the extant bone sample
white circles on the images of both the extant and fossil (Figure 3B) shows six separate events in the force-
bone sections (Figure 2A,B), indicate that antibody binding extension curve, while five events are obtained from fossil
in both cases is localized to the bone matrix and is not the bone (Figure 3A), suggesting similarities in inherent
result of antibody interactions with embedding resin. molecular pull-off properties. The contact regions of the
However, in some regions of the fossil bone, the fluorescent force curves (upward bend in the green/blue curves in
signal was more uneven than that in extant samples, Figure 3) are sensitive to the microelastic properties of
indicating a differential and patchy preservation of fossil the surface.21 For example, the pull-off (blue) curve derived
bone epitopes consistent with partial degradation/ from fossil bone lags behind relative to the approach
alteration of organics in some regions of the tissue. (green) curve in the contact region (Figure 1A), suggesting
AFM Results. A schematic illustration of section plastic deformation and hence moderate degradation of
preparation and representative images is shown in Figure elastic properties in the Cretaceous sample. However, in
1. Images obtained using AFM in air clearly show a fibrillar some cases, approach and pull-off curves overlapped
pattern with characteristic cross-banding in extant samples perfectly with no degradation in elastic properties.
(Figure 1B,D), although in extant bone the pattern is less Taken together (Figures 1-3), these data support the
distinct than that in tendon fibers, due to uneven masking hypothesis that the material in the fossil bone is organic
of overlying biominerals remaining after partial dem- and consistent with collagen in molecular unfolding
ineralization. Images from fossil embryo bone show properties and antigenic response. The conclusion that
fiberlike features distributed more or less homogeneously the fossil section is organic is further supported by
over a wavy texture (Figure 1C), but cross-banding is ToFSIMS analysis (not shown) and the collagenase
difficult to assess. These features are possibly due to partial digestion experiments presented below (Figure 7). Imaging
degradation/alteration of fossil collagen fibers. Resin ToFSIMS shows an abundance of organic fragments
morphology (Figure 1A) showed no distinctive pattern generated from the fossil section mixed with inorganic
except sectioning artifact; therefore, tissues and embed- components dominated by calcium.
ding resin could be easily differentiated by both optical A full comparison of the statistical distributions of
and AFM imaging in liquid, verifying that force-extension unbinding events in fossil and extant sections requires
data were obtained from the regions of interest. Further- further analysis. Using a MatLab-based analysis,20 1200
more, ToFSIMS analysis of these sections (not shown) unbinding events were identified in the extant bone and
demonstrates clearly that resin material did not diffuse 1711 events were identified in the fossil. It is important
into the fossil matrix and that the organic signature of to note that the number of events varies from sample to
the fossil, in terms of fragmentation patterns, is very sample. In some cases, we observed less than 1000 events.
different from that of the resin. The statistics discussed here pertain to the particular
After images were obtained in air, samples were force-volume data presented here. In this case, 18 out of
immersed into physiological buffer (PBS) in a liquid AFM 1024 force curves obtained from the extant bone and 5
cell and force-extension measurements were obtained from the fossil were identified as corrupt curves and are
on fossil and extant bone sections to determine the pull- therefore not included in the analyses. The force curves
off properties of the protein matrix. Figure 3 represents obtained for a given section are uniformly distributed over
force-extension curves obtained on sections of fossil bone a 4 × 4 µm2 area (force-volume data). The events in each
(Figure 3A), extant bone (Figure 3B), resin (Figure 3C), pull-off curve have been classified as no event, single event,
and extant tendon (Figure 3D). The saw-toothed features, or multiple events based on the profile obtained for each.
herein called unbinding events or just events (an example The distribution of these events depends partly on the
is marked by an arrow in Figure 3B), seen in the force- quality of the section under study and partly on the sample
extension curves are due to one of the three following itself. For example, for the sample studied here, ap-
possibilities: (1) the detachment of the tip from the surface proximately 30% of the curves obtained on extant bone
(a nonspecific short-range event), (2) the detachment of showed no events, 38% showed single events, and 32%
a single fibril (individual collagen molecule) stretched showed multiple events. On the other hand, for the
between the surface and the tip, or (3) the rupture of inter- distribution of curves obtained on the fossil bone, 13%
or intramolecular bonds (relaxation of sacrificial bonds) showed no events, 36% showed single events, and about
associated with a single fibril, causing a sudden reduction 51% showed multiple events.
of load on the atomic force microscope tip. The concept of Figure 4 compares the magnitude of the unbinding force
sacrificial bond rupture was first introduced and used by and the corresponding extension for each event for extant
Hansma et al.,18 and here, we adopt their interpretation. (black points) and fossil (red points) bone. Data from extant
To assess the similarity of force curves, slopes were sections show distributions of greater forces (some larger
calculated for the curves near the tips of the saw-toothed than 0.6 nN) concentrated around 100-nm extensions,
patterns and are represented by red dotted lines in Figure while those from fossil bone show force distributions of as
3. The absolute value of the slope obtained on the fossil much as 0.4 nN with extensions exceeding 150 nm. The
embryonic bone (4.0 ( 0.7 pN/nm) was similar to those variations in magnitude of the unbinding forces suggest
obtained on extant bone (5.1 ( 0.8 pN/nm) and tendon multiple collagen attachments to the atomic force micro-
(4.0 ( 0.8 pN/nm) and clearly distinct from values obtained scope tip, where subsequent detachments of individual
on the resin control (7 ( 1.5 pN/nm), indicating similarities fibrils plus the unbinding of sacrificial bonds give rise to
in mechanical properties among the biological samples. the observed force-extension profiles. We attribute most
Unembedded fossil bone immobilized on a mica surface of the events with unbinding forces <0.1 nN to relaxation
using a commercial epoxy (Devcon, Rivieray Beach, FL) of the sacrificial bonds, as suggested by Hansma et al.18
yielded curves very similar to those in Figure 3A, As mentioned earlier, some of the events are due to the
supporting the evidence that the observed unbinding
events are intrinsic to fossil bone. All force-extension (21) Arce, F. T.; Avci, R.; Beech, I. B.; Cooksey, K.; Cooksey, B. W.
curves show multiple unbinding events (arrow, Figure J. Chem. Phys. 2003, 119, 1671-1681.
F Langmuir Avci et al.
Figure 4. Distribution of unbinding force vs collagen extension,
comparing events encountered in partially demineralized extant
and fossil bone sections. Each data point corresponds to an Figure 6. Negative control experiment: FTIR studies com-
event marking the location of a collagen extension and the paring fossil sections with extant collagen sections (mixtures
unbinding force associated with that extension as explained in of bone and tendon), resin sections, and humic acid. Notice that
the text. fossil and extant collagen show some agreement while the humic
spectrum is totally different from the rest.
around 300 nm for both fossil and extant samples, Figures
4 and 5 indicate that only a small fraction of fibrils are
extended to their full length. As a whole, the force and
length distributions presented in Figures 4 and 5 for fossil
sections suggest the collagen fibers in these sections are
To eliminate the possibility that the antibody reactivity
and/or force curve results were due to interactions between
organic breakdown products, either alone or complexed
with bone matrixes, we conducted further experiments
using FTIR (Figure 6) and collagenase digestion (Figure
7) to confirm the presence of collagen in the fossil sections.
The results were compared with the FTIR spectra of the
solubilized humic substance in the spectral region where
Figure 5. Histograms comparing the events recorded in the amide bands (I and II) are observed, as shown in Figure
extant section (red) with those recorded in the fossil section 6. Because of the small size of the sections, the quality of
(green). Part A shows the normalized number of events vs the these spectra is not ideal, but the humic spectrum is so
collagen extension in increments of 20 nm, while part B shows
the normalized events vs the unbinding forces in increments different from the rest of the data that further data
of 20 pN. refinement was not pursued. The spectra for fossil and
control sections show general agreement, while those
breaking of the bonds between the individual fibers and obtained for humics present a completely different profile.
the atomic force microscope tip. For example, the event Because humic substances are tightly associated with
with the largest extension (the last unbinding event) in most fossil material, and are derived from organic
each pull-off curve is attributed to the latter effect. breakdown products, it could be argued that these
We see a small difference in the force versus extension substances were the source of nonspecific positive re-
distribution between the pull-off lengths of fossil and sponses. To control for this, AFM pull-off experiments
extant bone collagen. The difference suggested by Figure (not shown) were conducted using humics as a substrate.
4 is that, for small extensions (0-150 nm), slightly larger The humics yielded a smaller number of unbinding events
forces are needed to pull off the fibrils in the case of extant that differed in character (mostly short-range and non-
bone collagen when compared to the fossil. This could be specific) from those observed with either extant or fossil-
explained by the degradation of the bone mineral matrix derived collagen. While bicinchoninic acid (BCA, Pierce)
material and perhaps by partially denatured/degraded results showed measurable protein content (∼75 µg/mg)
collagen, resulting in slightly smaller pull-off forces. in commercial humics, immunological cross-reactivity was
However, in other samples, these differences become not observed when humics were tested against avian
insignificant. collagen antibodies using either enzyme-linked immun-
Event distributions associated with the extension of osorbant assay (ELISA) or in situ immunohistochemistry
collagen in extant (red bars) and fossil (green bars) bone (data not shown). Therefore, we are confident that
sections were normalized to account for unused corrupt coeluting/coexisting humic substances are not the source
curves and are shown in Figure 5A. A greater number of of either AFM pull-off curves or immunological reactivity
unbinding events was observed in the fossil bone (green) observed in fossil bone. While we recognize that com-
than in the extant (red) samples, suggesting that fibrils mercially prepared humic substances may not be identical
are easier to pull from the fossil. The force distribution to naturally occurring humics, we feel that the quantitative
data for both samples (Figure 5B) show a majority of events and qualitative differences between collagen and humics,
concentrated around 20-80 pN, suggesting that most of yielded by all assays, make them an adequate control.
the forces are used to break the sacrificial bonds.18 Therefore, a comparison of the immunological, FTIR, and
Considering that the length of type I collagen is ∼300 AFM pull-off results of bone sections and humic acid clearly
nm14,15 and only a handful of collagen extensions are eliminates the possibility that interactions with humic
Preservation of Late Cretaceous Bone Collagen PAGE EST: 6.8 Langmuir G
morphological alteration of fibrils in a fossil section due
to enzymatic digestion. The collagen fibers, initially rather
uniformly distributed in the fossil material, were reduced
to small particles after collagen digestion.
There is great similarity between fossil and extant bone,
with respect to immunological response, enzymatic deg-
radation, and force-extension distribution as measured
by AFM. Together, these data provide strong evidence for
the integrity and functionality of collagen remnants in
fossil bone dating back to the Cretaceous period. Molecular
preservation is supported by immunohistochemical data,
showing localized and specific binding of modern avian
collagen antibodies to fossil tissues. The pattern of
antibody binding visualized by this method indicates that
preservation at the molecular level is somewhat patchy
across the bone surface, although all bone tissues tested
indicate the preservation of antigenic material. Negative
controls support the endogeneity of this material and
strongly indicate a collagen source. Finally, the question
of integrity of collagen proteins is addressed by AFM
pulling curves. When compared with extant bone similarly
treated, pull-off force measurements obtained by AFM on
bone sections indicate that the stretching and unfolding
properties of collagen in Cretaceous embryonic bird bone
are directly comparable to those observed in the extant
bird bone and extant tendon. These data provide strong
evidence that collagen proteins preserve not only antigenic
properties but also molecular integrity and functional
properties. We hypothesize that unusual depositional
conditions and unique taphonomy22 acted in concert to
Figure 7. Digestion of fossil collagen by collagenase: (A and preserve the embryonic remains of this primitive Creta-
C) representative AFM height and amplitude images, respec-
tively, at two magnifications of a partially demineralized fossil ceous bird at the microscopic and molecular levels. Bones
bone section before exposure to collagen-specific enzyme are preserved suspended in three dimensions and ar-
collagenase; (B and D) AFM height and amplitude images, ticulated, indicating that mineralization occurred before
respectively, of the representative areas of the same sizes as soft tissue degradation was complete. The rapidity of this
in parts A and C after the fossil section was incubated with process most likely contributed to molecular preservation.
collagenase for 24 h. The continuous texture of collagen as shown
in parts A and C is totally disrupted as a result of interaction Acknowledgment. Funding from NASA (EPSCOR
with collagenase, leaving behind speckles of ∼50-nm particles
in its place, as shown in parts B and D. grant no. NCC5-579, ONR grant no. N00014-02-1-063)
and the matching funds from Montana State University
are gratefully acknowledged. We would like to thank Dr.
substances were responsible for the collagen pull-off
P. Suci for his help with FTIR analysis and data reduction
results reported here.
and Dr. Z. Suo and Dr. B. Spangler for immobilizing the
More strong evidence that fossil material contains
humic acid on Si3N4 substrate for AFM pull-off experi-
collagen comes from the experiments done with collage-
ments. We further acknowledge the help of Ms. K. M.
nase, an enzyme that specifically cleaves the modified
Thieltges for biological sample handling and of E. Lamm
bonds of collagen while leaving other proteins largely
and M. Fox for fossil bone preparation and histological
intact. The reduction in antibody binding following brief
digestion with collagenase shows that the epitopes rec-
ognized by collagen antibodies are susceptible to colla- LA047682E
genase digestion (Figure 2F). This was further verified by
the AFM imaging presented in Figure 7, which shows (22) Hedges, R. E. M. Archaeometry 2002, 44, 319-328.