Chemistry supports the identification of gender-specific reproductive tissue in Tyrannosaurus rex

Medullary bone (MB), an estrogen-dependent reproductive tissue present in extant gravid birds, is texturally, histologically and compositionally distinct from other bone types. Phylogenetic proximity led to the proposal that MB would be present in non-avian dinosaurs, and recent studies have used microscopic, morphological, and regional homologies to identify this reproductive tissue in both theropod and ornithischian dinosaurs. Here, we capitalize on the unique chemical and histological fingerprint of MB in birds to characterize, at the molecular level, MB in the non-avian theropod Tyrannosaurus rex (MOR 1125), and show that the retention of original molecular components in fossils allows deeper physiological and evolutionary questions to be addressed.

microanatomical, traditional staining, and immunohistochemical staining. The specimen was loaned to SW by Guillermo Zavala (corresponding author of 3 ) for these and additional studies (Werning et al. in preparation).
The osteopetrotic lesion in this tarsometatarsus extends nearly the entire length of the bone, is widest at mid-diaphysis (3-4 times the normal width), and tapers towards the proximal and distal ends. This is the typical distribution of avian osteopetrosis lesions, which tend to give the leg bones a fusiform profile 6 . In cross section, the lesion is visible endosteally as well as periosteally; very little of the original metatarsal medullary cavities remain. In the deep cortex and endosteally, the lesion is compacted, but large circular and radially-oriented oval spaces are visible just under the periosteal surface around most of the circumference of the bone. This microstructural pattern is also typical of avian osteopetrosis ( 3,7,8 )

Ground (petrographic) sections
MB and overlying cortical bone (CB), still attached, from extant ostrich and Tyrannosaurus rex were embedded in Silmar 41 Clear Polyester Casting Resin (USComposites, Cat #SM-S41100). A relatively thick slice taken from Buehler isomet 1000 precision saw was attached to a glass slide with epoxy and ground to desired thickness (0.2-0.1mm)with Buehler Ecomet 4000 Grinder-polisher, and imaged with Zeiss Axioskop 40 microscope equipped with 10x objective CP-ACHROMAT 10x/0.25 (440930) for polarized light microscopy.

Bone sample preparation
MB and overlying cortical bone (CB), still attached, from extant ostrich and chicken femora known to be in active reproduction at time of death 2 were fixed with Neutral Buffered 10% formalin overnight, then demineralized in 500 mM EDTA (pH 8.0) until all mineral was removed, and again subjected to fixation as above. CB and MB from MOR 1125 were analyzed separately. Both bone types were demineralized in 500 mM EDTA. Because of the fragile nature of demineralized fossil bone, CB and MB from MOR 1125, taken from the region at right indicated by the red square, were embedded in 3% agar (Becton Dickinson Cat# 214530) to stabilize the tissues prior to sectioning. Extant CB and MB samples, and agarembedded CB and MB from MOR 1125 were then subjected to routine dehydration (via sequential incubations in 70%, 80%, 90%, 95%, and 100% alcohols for ~ 1 hour each. The tissue blocks were then placed in an additional two 100% ethanol solution to ensure that all water is removed, followed by three, 30 minute incubations in 100% xylene to clear the tissue. Tissues were then transferred to 100% paraffin, three separate, sequential incubations for 30min each to complete infiltration, then embedded in paraffin wax (Paraplast Plus EMS CAt#19216) for sectioning. Sections were taken at 5 um, using a Leica RM 2255.

Staining procedures
Histochemical stains used to differentiate and diagnose MB in living birds were applied to both dinosaur tissues and extant controls. These stains are not conclusive, as other tissue components will react to the stains (e.g acidic polysaccharides in tissues such as cartilage); however in bone samples these can differentiate type and amount of mucopolysaccharides and glycosaminoglycans incorporated into the matrix.
A) HID/Alcian blue stain. Following the high iron diamine (HID) method outlined by Spicer 9 , paraffin embedded sections were deparaffinized with xylene and dehydrated through a graded ethanol series. Demineralized bone sections were oxidized in 1% H 5 IO 6 (periodic acid) for 10min, rinsed under running tap water, and stained with freshly made high iron diamide (HID) solution, prepared as follows: 120mg of meta diamine and 20mg of the para isomer were dissolved in 50ml of water, then poured into a glass container with 1.4ml of N.F. 10% ferric chloride (FeCl 3 ), equivalent to a 62% w/v solution of FeCl 3 .6H 2 O) for 18hr. Sections were then rinsed with water then stained in alcian blue -8GX (1% in 3% acetic acid) for 30min, rinsed again, then dehydrated with a graded ethanol series, followed by several incubations with 100% xylene. Mounting medium (Poly-Mount, PolySciences Cat#08381) and cover glass were applied for visualization.
B) HID Stain only. The same procedure as above was applied to a second set of bone sections, with the omission of the Alcian blue step. This demonstrates that HID components alone reacted with specificity to the differences in bone matrix chemistry.
C) Alcian blue -8GX. Demineralized, paraffin embedded sections of modern and ancient bone tissues were dehydrated as described above. Demineralized bone sections were oxidized in 1% H 5 IO 6 (periodic acid) for 10min, rinsed under running tap water, then exposed to Alcian blue -8GX (1% in 3% acetic acid) for 30min, rinsed again, then dehydrated with a graded ethanol series, followed by several incubations with 100% xylene. Mounting medium and cover glass were applied as above.

D) No stain control
Sections of bone tissues were treated as above; i.e., demineralized fragments of CB and/or MB were embedded in either paraffin, or agar and paraffin. 5 m sections were taken as described, then deparaffinized in xylene and dehydrated in a graded ethanol series. Demineralized bone sections were oxidized in 1% H 5 IO 6 (periodic acid) for 10min, rinsed under running tap water, then dehydrated with a graded ethanol series, followed by several incubations with 100% xylene to clear remaining paraffin. Mounting medium was applied, and sections were imaged without further treatment.

Immunofluorescence
Demineralized T. rex (MOR 1125) MB, T. rex CB and 10% formalin-fixed demineralized ostrich and chicken MB and CB, and chicken osteopetrotic bone, were embedded in LR White resin blocks after partial dehydration in 70% ethanol and infiltration with pure LR White water permeable embedding medium (Ted Pella) as previously described (e.g. 10,11 ). 200 nm sections were cut on a Leica EM UC6 Ultramicrotome, transferred to six-well, Teflon-coated slides, and dried overnight at 45 o C on a warming plate. Sections were etched with Proteinase K (PCR grade, Roche, 25 ug/ml) in 1X phosphate buffered saline (PBS) buffer at 37˚C to expose epitopes, followed by two incubations in 500 mM EDTA (pH 8.0) and two incubations in 1 mg/ml sodium borohydride for 10 minutes each for antigen retrieval (http://www.ihcworld.com/_intro/antigen-retrieval.htm). Sections were then incubated for 2 hours in normal goat serum (NGS) diluted to 4% in PBS to occupy non-specific binding sites and prevent spurious binding. Sections were then incubated with primary antibody consisting of monoclonal mouse anti-KS [Keratan Sulfate (5D4)] (Cosmo Bio Co., LTD Cat# PRPG-BC-M01), diluted 1:20 in primary dilution buffer overnight at 4 o C as recommended by the manufacturer. Sections were washed multiple times to remove unbound antibody, then all sections, including controls (no primary antibodies were applied but all other steps and conditions were kept identical), were then incubated with secondary antibody (biotinylated goat anti-mouse IgG (H+L) (Vector BA-9200), diluted 1:500 in antibody dilution buffer 12 for 2 hours at room temperature. Fluorescein Avidin D (FITC, Vector Laboratories A-2001) diluted 1:1000, was applied to all sections and allowed to bind for 1hr at RT. All incubations were separated by sequential washes (2 times for 10 minutes each) in PBS w/Tween 20 (ACROS Organics) followed by two 10minute rinses in PBS. Finally, sections were mounted with Vectashield Anti-Fade mounting medium (Vector H-1000), and coverslips applied. Sections were examined with a Zeiss Axioskop 2 plus biological microscope and captured using an AxioCam MRc 5(Zeiss) with 10x ocular magnification, and data collected using the Axiovision software package (version 4.7.0.0). Figure S1. Complete ground section of MOR 1125 femur fragment. CB shows secondary osteons interspersed with increasingly large erosion rooms (ER, yellow arrows). Red arrows show distinct boundary between CB and non-lamellar MB. Figure S2. Petrographic sections of ostrich cortical (A) and medullary (B) bone, and T.rex (MOR 1125) cortical (C) and medullary (D) bone viewed in polarized light. In A and C, cortical bone demonstrates secondary osteons and birefringence, reflecting the organized lamellar nature and orientation of collagen fibers. In contrast, MB (B, D) shows non-lamellar bone with non-birefringence and isotropy, consistent with rapidly deposited, random arrangement of fibers.   Figure S5. Controls for immunohistochemical analyses, in which no primary antibody was applied, but all other parameters, including data collection parameters, were kept identical to test conditions, to control for false positives due to spurious binding of secondary antibody. These images were taken from the same samples as presented in the text, at the same parameters; no reactivity to antibodies is