Is there any potential evolutionary explanation for the paradoxical transitions of cell fate described above?

It appears likely that chondrogenesis and osteogenesis developed independently during evolution. Cartilage is a very ancient tissue the development of which is regulated by SoxD (mammalian orthologs Sox 5/6/13), SoxE (mammalian orthologs Sox8/9/10), and ColA (mammalian orthologs Col2a1, Col1a1). This tissue appeared prior to the splitting of the animal kingdom into protostomes and deuterostomes, i.e., 670 million years ago, and cartilaginous structures are present in such diverse creatures as mammals, squids, and horseshow crabs [59,60,61].

The family of calcium-binding phosphoproteins, including enamel and dentin proteins, sialoproteins, osteopontin, and casein, began to evolve from SPARC-like gene duplications approximately 600 million years ago [62, 63]. At the same time, phosphate-based skeletal tissues, i.e., dentin, enamel, and bone, first appeared during evolution as tooth-like structures called odontodes in the skin of jawless eel-like chordates known as conodonts [64]. Subsequently, scale-like dermal skeletal structures resembling these odontodes rapidly evolved into dermal bone plates that form extensive exoskeletons in ostracoderms (jawless fish) and placoderms (jawed fish) [65] approximately 470–430 million years ago. Thus, the cartilaginous endoskeleton and bony dermal exoskeleton are likely to have evolved independently.

In modern tetrapods, the formation of the cranial skeleton is intramembranous, in a manner similar to the corresponding process in armored fish such as ostracoderms and placoderms. At the same time, most of our skeletal structures are formed via endochondral ossification, i.e., based on a cartilage template. So, when, why, and how does this occur and what role may transdifferentiation play in this context?

Ossification of the endoskeleton appeared first in early Gnathostomata (jawed animals) in the form of a perichondral ossification [66]. This process involves the formation of bone on the surface of cartilage produced by the adjacent mesenchymal cells, usually as a tube-like structure around the midsection of the cartilage, which probably provides additional stiffness to the cartilaginous elements. At the same time, periosteal ossification also occurs in many extant animals, including the formation of the jawbone of mammals around Meckel cartilage and in most bones of teleost fish. In this case, bone is formed without the intermediate involvement of chondrocytes, building around a cartilage anlage and thereby armoring the initially cartilaginous endoskeleton. The cartilage anlage can remain inside its bony shield for a long time, gradually degenerating.

Thus, perichondral ossification resembles dermal ossification, except that the former takes place along cartilaginous elements. It seems plausible that the genetic program that evolved for intramembranous ossification was later adopted for perichondral ossification in order to reinforce the endoskeleton with more rigid structures while retaining the growth rate and potential for locomotion.

Endochondral bone formation, in connection with which a cartilage template is replaced by the bone, appeared more recently during evolution than intramembranous and perichondral ossification. Samples of stem sarcopterygians (lobe-finned fish), 380 million years old, exhibit signs of such bone formation [67], and all terrestrial animals that evolved from these fish, including humans, also demonstrate endochondral bone formation. Since certain teleost fish also show endochondral ossification and growth plates, it is possible that endochondral bone formation evolved in vertebrates prior to their split into sarcopterygians and actinopterygians [66] 400 million years ago.

It is generally considered that endochondral bone formation gives rise to both bone collars (i.e., cortical bone) and spongy or trabecular bone. Recent lineage tracing in mice revealed that bone collars are formed directly from the perichondrium, without a cartilage intermediate [68] and, therefore, are formed by perichondral ossification, which appeared prior to endochondral ossification during evolution. In the case of mammals and reptiles, the cortical bone collars grow at the periphery of the growth plates, while the trabecular bone is formed immediately beneath this plate, with both processes being, in general, tightly coupled. However, in amphibians, this coupling is not as tight; for instance, in Urodela (e.g., newts and salamanders), periosteal ossification often lags behind the bone growth, whereas in Anura (e.g., frogs and toads), periosteal ossification is more rapid than endochondral, so that a large proportion of the cartilage remains within bony collars [69,70,71]. Moreover, in connection with limb regeneration in newts and salamanders, these two processes are largely uncoupled, with periosteal ossification taking place after the cartilage template has regrown almost to full size [72]. Interestingly, uncoupling of periosteal and endochondral ossification can also be observed in humans under certain pathological conditions, e.g., metatrophic dysplasia [73].

Thus, both evolutionarily and developmentally, the formation of bone collars and of trabeculae can be viewed as two separate processes. From this perspective, hypertrophic chondrocytes function specifically as a source of cells of the osteolineage for trabecular bone formation. Trabeculae, which improve the strength of bony collars, are particularly beneficial for the weight-bearing demands placed on terrestrial animals.

During endochondral bone growth, longitudinal trabeculae are formed on the calcified cartilage template. An additional source of cells for use in the formation of trabeculae would be clearly advantageous in evolutionary terms. Interestingly, in extant Teleost fish and Urodele amphibians, trabeculation is relatively undeveloped, with the bone marrow cavity being filled predominantly with adipose cells [74,75,76]. Whether these adipocytes are descendants of hypertrophic chondrocytes remains unknown, although there is one report demonstrating such a transition in zebrafish [76]. On the other hand, longitudinally oriented trabeculae were reported in stem lobe-finned fish [67], suggesting that extant Urodele may have lost the ability to form these structures during evolution.

In conclusion, at present, it is impossible to say whether the hypertrophic chondrocytes of teleost and mammals acquired plasticity independently, whether initially hypertrophic cells transitioned into adipose tissue, and/or whether osteopotential was acquired later in the evolution. Nevertheless, chondrocyte hypertrophy and their subsequent transition toward osteolineage coupled with the formation of trabeculae have a clear evolutionary advantage, particularly for weight-bearing terrestrial vertebrates.