Initiating Skeletal Growth

Chapter 30

Initiating Skeletal Growth To the elbow I go; from the knee I flee.

Growth and morphogenesis are multidirectional processes that proceed in space and over time, usually along detectable axes and with a measurable polarity. D’Arcy Wentworth Thompson’s analysis of polarisation and transformation of form of adult organisms (Figure 30.1) is a classic example of this approach. Do not be deceived by Figure 30.1. Thompson modelled change in form as transformation of the adult of one species to the adult of another. However, and it is a critically important ‘however’, new adult forms arise evolutionarily (and developmentally) by changes in the form of parts of embryos, with all the timing and integration that implies. Adults do not transform into adults during evolution. Thompson (1917) analysed form as adaptation to the environment, but chose not to consider the relationship between ontogenetic adaptation and the inheritance of form and pattern. Although it often is difficult to separate control of growth from control of morphogenesis, it is informative to analyse the two independently. The concern in this chapter is with the cellular mechanisms responsible for growth. Morphogenesis, in the sense of the fundamental form of long bones, is discussed in the following chapter. We start with the obvious question.

WHAT IS GROWTH? I am attracted to the definition of growth proposed by Melvin Moss: ‘In the broadest sense, growth may be defined as any temporal change in any parameter that is measurable’ (Moss, 1972b, p. 127). Unfortunately, this definition also applies to morphogenesis, and it is important to distinguish growth from morphogenesis: G G

Growth is permanent increase in size. Morphogenesis is the attainment of form (shape) or change in form.

Much of the discussion in the preceding chapters is relevant to skeletal growth and morphogenesis. In this chapter, I concentrate on the processes that initiate and Bones and Cartilage. DOI: http://dx.doi.org/10.1016/B978-0-12-416678-3.00030-6 © 2015 Elsevier Inc. All rights reserved.

regulate growth as a developmental process, with minimal discussion of how to measure rates or amounts of growth. Growth results from a set of processes controlling increase in cell populations and regulating the physiological activity of those cell populations1. Production, differentiation and maintenance of skeletogenic cells, along with deposition of extracellular matrices (ECMs) (Chapters 2
5), provide major components of skeletal growth. Although the mechanisms controlling matrix accumulation differ from those that control cell proliferation, they are connected through feedback interactions. At the very basis of growth as a developmental process are mechanisms that determine how many cells will be available for an individual skeletal element. This topic begins our discussion of growth.

NUMBERS OF STEM CELLS Proliferation of progenitor cells and their accumulation into condensations represent the earliest growth of the skeleton, growth that is largely determined by the number of stem cells and their rates of proliferation, accumulation and/or loss. While we know that condensation size influences the size of a skeletal element (Chapters 19 and 20), little information is available on the number of stem cells required to produce a condensation or a skeletal element. Exceptions are the studies of Beatrice Mintz, who created ‘tetraparental’ mice (cellular genetic mosaics) by fusing blastocyst cells of two strains (C3H and C57BL/6), which have morphologically distinct vertebral characteristics (Mintz, 1972; Gearhart and Mintz, 1972; Moore and Mintz, 1972). Individual vertebrae in these chimeras also are cellular genetic mosaics
C3H on one side and C57BL/6 on the other
demonstrating that the left and right halves of each vertebra are of independent origin (Chapter 41). The interpretation given to this finding is that vertebrae are clonally derived, each lateral sclerotomite providing one cell line to each vertebra2. Indeed, presence within a single somite of both forms of the dimeric enzyme glucosephosphate isomerase, one dimer

475

476 PART | X Growing Together and Growing Apart

e

d

c (A)

(B)

e d c b a 0 0

1

2

3

4

5

6

1

FIGURE 30.1 D’Arcy Thompson (1917) used deformation of a rectangular coordinate system (a Cartesian grid) to show how the morphology of one species, in this example, the puffer fish Diodon holocanthus (A), could be transformed into the morphology of a second species, in this case the ocean sunfish Mola mola (B). The lines connecting the four landmarks ( ) in each species show that the bulk of the change is in the posterior two-thirds of the body and in the fins. (Free image obtained from http://www.mun.ca/biology/scarr/Thompson_ Transformation.htm)

b

2 3 4 a

5 6

having been contributed from each parent, was used by Gearhart and Mintz (1972) as evidence that individual somites do not each form from a single clone. A further finding was that the paired bones of the skull
notably the palatines, pterygoids and occipital condyles
occasionally show evidence of a mosaic organisation, suggesting that they too originate from at least two cell lines. Resegmentation leads us to the same conclusion (Chapter 16). On this basis, each vertebra would be derived from four clones of cells. Gearhart and Mintz also used their data to conclude that vertebral cell lines are established after 5 days of gestation
when a sufficient number of cells are first present to form all the necessary clones for the vertebral column
and before 7 days, which is when somites first arise. [Criticism of this technique for determining the time of clonal initiation was raised by J. Abbott et al. (1972), J. H. Lewis et al. (1972) and McLaren (1972). O’Higgins et al. (1986) repeated this study using Fourier analysis, concluding that the clonal theory is too simplistic.] Growth of the components of each vertebra could be regulated by, or vertebral defects could arise from: i. elimination of a cell lineage or portion of a cell lineage, for example, by apoptosis, resulting in a smaller than normal mesenchymal condensation; ii. excess proliferation or duplication of a cell lineage or part thereof, resulting in a larger than normal mesenchymal condensation; or

iii. disruption of a whole body or skeletal element gradient as seen, for example, when temperature perturbs vertebral development and changes vertebral number (Chapter 43)3. Laird (1966) and Saxe´n (1976) developed similar celllineage based concepts using comparative growth and teratogenesis, while Atchley and Hall (*1991) developed an integrated developmental and quantitative genetics model for mandibular development based on the properties of modular cell lineages.

CELL MOVEMENT AND CELL VIABILITY Maintaining that cell movement within limb buds is an important component of limb outgrowth, Donald Ede’s group in Glasgow provided ultrastructural evidence of the attachment of mesenchymal cells to, and movement toward, the limb-bud epithelium. Ede’s studies on cell adhesion in the talpid3 chick mutant are important in this regard and are discussed in Chapter 39. talpid mutant chick embryos are discussed more fully in Chapter 39. Limb buds from H.H. 24 and 26 wild-type and talpid3 embryos were dissociated into single cells and patterns of reaggregation in rotation culture observed. Rate and duration of cell movement in talpid3 cells increase the probability of cell-to-cell contact in vitro. Viscometric analysis confirmed that talpid3 cells are more adhesive than cells from wild-type embryos. Because they are both

Chapter | 30 Initiating Skeletal Growth

more adhesive and less mobile, talpid3 cells aggregate more rapidly and form smaller aggregates than do wildtype cells. In vivo, this behaviour is expressed in the failure of condensations to separate from one another, fusion of adjacent skeletal elements, and defective growth4. Culturing limb buds with vitamin A or applying vitamin A to pregnant mice decreases cell motility and so decreases limb size. Vitamin A also initiates apoptosis via Caspase 3. Similarly, and as discussed in Box 27.1, vitamin A blocks migration and proliferation of neural-crest cells and so inhibits the development of the upper and lower jaws. Any reduction in cell motility or decrease in surface adhesiveness will tend to increase the generation time for cell division and be seen as a lower labelling index after 3H-thymidine5.

Epithelia, Fgf/FgfR2 and Mesenchymal Cell Proliferation From the data discussed in Parts VI and VII, we know that proliferation and viability of skeletogenic mesenchymal cells depend on interactions with epithelia. Within the developing head, facial epithelia are required for mesenchymal viability. This requirement is stage dependent and cell-to-cell contact dependent. Consequently, growth of facial mesenchyme is promoted by influences from facial epithelia. While epithelia from different facial processes in chick embryos are interchangeable in this role, interestingly, frontonasal epithelium promotes the outgrowth of limb buds but mandibular and maxillary epithelia do not. As can be demonstrated using micromass culture, differential growth of facial primordia is regulated by growth factors. Fgf2 stimulates growth of embryonic chick frontonasal but not maxillary or mandibular processes and enhances chondrogenesis from frontonasal but not from the other processes. Fgf can replace epithelium to promote outgrowth of facial processes6. Epithelia and mesenchyme signal through FgfR1 and FgfR2; epithelial FgfR1 maintains FgfR2 expression in chick facial mesenchyme. In the absence of the epithelium, FgfR2 and Col2a1 both are down-regulated (Matovinovic and Richman, 1997). Although both frontonasal and maxillary processes contain high levels of Fgf2, FgfR2 is high in the frontonasal and lower in the maxillary process, correlating with differential growth of the two facial processes. Similarly in limb buds, the epithelial b variant of FgfR2 is active in limb epithelium and in the apical epithelial ridge (AER), where it plays a role in limb outgrowth. Chimeric mouse embryos containing a high proportion of embryonic stem cells with a lack-offunction mutation of FgfR2 fail to form limbs

477

(Gorivodsky and Lonai, 2003). Cells with the loss-offunction mutation are predominantly in limb epithelium (but not in the AER) and in the subjacent mesenchyme where Shh and Msx1 are down-regulated. Epithelial expression of Engrailed 1 and Wnt7a is discontinuous, reduced and ectopic, implying that both are under the control of FgfR2.

METABOLIC REGULATION We know that use/disuse, amount and quality of diet, nutrition and metabolic rate all have important influences on skeletal growth in neonates, juveniles and adults. Metabolic regulation is important for initiation and initial growth of skeletal elements. The microenvironment immediately surrounding skeletal cells
local shifts in pH, temperature, pO2 and pCO2 tensions, osmolarity and metabolic inhibitors
modify the rates of physiological processes and so alter growth rates. Little is known about this microenvironment in vivo, so it is difficult to determine accurately the relative roles played by each in vivo. Alterations in pericellular ECMs are more amenable to analysis and have yielded valuable information on the control of cartilage growth by positive feedback from matrix products to chondroblastic and chondrocytic biosynthetic activity. As I discussed when evaluating achondroplasia in Chapter 27, growth is retarded by any disruption in feedback or interactions between components of the ECM. To turn to mutants again, the creeper (cp) mutant in domestic fowl provides a nice example of how metabolic inhibition can severely retard growth and morphogenesis.

Creeper (cp) Fowl Creeper (cp), a dominant mutation on the C chromosome of the common fowl, is homozygous lethal. Homozygous embryos, which die at the end of the third day, have under-developed heads, limbs and extraembryonic vasculature. Creeper is genetically linked with rose comb, a useful linkage, as creeper embryos are often difficult to distinguish from wild-type embryos at early ages. Heterozygotes are viable but phocomelic; proximal limb elements are reduced so that the digits are located much closer to the body than is normal. Thalidomide taken during sensitive phases of pregnancy can result in a phocomelic (‘seal limb’) phenotype of the type discussed in Chapter 387. I revisit creeper after discussing development of the tibia and fibula in wild type chicken embryos.

Tibia/Fibula Long bones, especially those of the hind limbs, and especially the tibiae, are severely shortened. The tibiae are

478 PART | X Growing Together and Growing Apart

severely bent, although the fibulae may be larger than normal. Dissimilarity in the action of cp on tibia and fibula reflects a faster growth rate of the tibia between 7 and 9 days of incubation in heterozygotes and that cp retards growth rates8. Although in chick embryos the prechondrogenic condensations for tibia and fibula are initially nearly the same size, differential growth of the tibia subsequently outstrips that of the fibula. This is so even when the two skeletal elements are maintained in vitro, a result consistent with an intrinsic pattern of growth that is disrupted in the mutant. Hinchliffe and Johnson (1983) discuss and illustrate this study of the organ culture of tibiae and fibulae from 6-day-old embryos in various combinations (pp. 276
280 and Figures 7
10). Interestingly, maintaining tissues from cp embryos in vitro revealed that only the heart showed decreased growth, implying that growth retardation in other tissues/organs is secondary.

Competing for Mesenchyme For a fibula (or any skeletal element) to develop at all, a critical volume of mesenchyme must be present in the early limb bud. The fibula is especially affected when mesenchyme is excised from limb buds, demonstrating that development of the fibula is differentially sensitive to reduction and illustrating the importance of elementspecific cell number and condensation size for initiation of chondrogenesis. Rotating fibular mesenchyme through 90 and reimplanting it into a limb bud changed the direction of fibular growth, allowing it to grow to the same length as the tibia (Wolff and Hampe´, 1954; Hampe´, *1960). Hampe´ did other experiments, discussed in Box 30.1, that produced what were interpreted as atavistic skeletal and muscular changes reminiscent of the ancestral condition seen in Archaeopteryx. Studies published in the 1950s and 1960s were interpreted as indicating ‘competition’ for mesenchyme within

BOX 30.1 Armand Hampe´’s Experiments Recreating Ancestral Musculoskeletal Patterns by Manipulating Fibula Mesenchyme in Embryonic Chick Limb Buds Armand Hampe´ published some classic experiments in which he manipulated the developing limb buds of chick embryos and obtained what were interpreted as reversals to skeletal patterns of Archaeopteryx or ancient reptiles. In reptiles, the tibia and fibula are equal in length and articulate with a series of small tarsal bones in the ankle, which articulate distally with a series of metatarsal bones as shown in Figure 30.2. In Archaeopteryx the tibia and fibulae retain the reptilian condition of being equal in length but tarsal elements are reduced to two and metatarsals to three

separate elements, again shown in Figure 30.2. In modern birds, the fibula is reduced to a splint with only a proximal epiphysis and the tibia fuses with the tarsals bones and articulates with the metatarsals, which fuse laterally into a single bone in adult birds (Figure 30.2). Given how little we know of other species of birds, this may or may not be the typical avian condition; Pourlis and Antonopoulos (2013) provide details of the development and ossification of the leg bones in Japanese quail Coturnix coturnix japonica, in which ossification of metatarsal I starts 4 days after ossification of the other metatarsals.

Reptile Modern bird (chick)

Archaeopteryx Tibia

Modern bird (adult)

Fibula

Tarsals

Ankle

Metatarsals

Experimental chick FIGURE 30.2 The tibia and fibula are fully developed in reptiles and articulate with a set of tarsal bones that in turn articulate with several metatarsals. In Archaeopteryx, the numbers of tarsals and metatarsals are reduced but the tibia and fibula retain the reptilian conditions. In birds, shown on the right, the fibula is reduced to a splint and the tarsals have fused with the fused metatarsals. In the experimental manipulations discussed in the text and shown as ‘experimental chick’ in the figure, the skeletal relationships show a remarkable similarity to those seen in Archaeopteryx but with a tarsal incorporated into the distal end of the fibula. Modified from Frazzetta, 1975.

(Continued )

Chapter | 30 Initiating Skeletal Growth

a. See Hampe´ (*1960) for the early studies, and G. B. Mu¨ller (1986, *1989, *1991) and Streicher and Mu¨ller (*1992) for the foil barrier and extirpation studies. In response to the modified proportions of the tibia/fibula, the associated muscles changed to patterns typical of reptilian muscle, especially M. flexor perforans, M. popliteus and M. fibularis brevis (Mu¨ller, *1989).

a single condensation between presumptive tibia and presumptive fibula. Madeleine Kieny obtained experimental data that she interpreted as evidence for a competitive interaction for cells between fibular and tibial primordia, the ‘stronger’ tibial rudiment taking cells from the ‘weaker’ fibular rudiment (Kieny, 1967). During embryonic development, the fibula loses its distal epiphysis by H.H. 27 or 28, a normal process in skeletal development in most birds resulting in a splint like fibula attached

Modern bird

Experimental Reptile bird (crocodylus)

Traditional

Archaeopteryx

(B) Embryonic patterns

Hampe´ (1960) grafted additional mesenchyme into the limb bud, rotated the fibula mesenchyme to alter the directions of growth of the fibula with respect to the tibia and inserted a mica plate within the single condensation for the tibia/fibula. The fibula developed a distal epiphysis, grew to the same extent as the tibia and failed to fuse with the tarsal elements, which remained separate; the metatarsals remained as three individual bones, shown as the experimental chick image in Figure 30.2, essentially mimicking the ancestral condition seen in Archaeopteryx. This ability to create atavisms experimentally is intriguing. Gerd Mu¨ller repeated Hampe´’s third experiment by implanting a foil barrier between tibial and fibular rudiments in the hindlimb bud between H.H. 22 and 24. Figure 30.3 shows the traditional depiction of the adult morphology (A) (also shown in Figure 30.2) and embryonic (B) and adult morphology (C) in birds, reptiles and in experimental chick embryos as described by Mu¨ller (*1989), who interpreted the changes in the experimental group as a consequence of greater growth of the fibula and patterning changes of the tarsals and metatarsals. Interestingly, a phylogenetic analysis of nonavian theropods and basal birds by Dececchi and Larsson (2013), discussed further in Chapter 44, supports the conclusion that the relative elongation of the forelimbs in avian ancestors was driven by reduction in body size, not elongation of forelimb length. Rather than evoking atavisms as if that term described a mechanism (which it does not), Mu¨ller invoked a process-oriented interpretation, heterochrony, discussed in Chapter 44a. Mu¨ller and Streicher (1989) used a similar experimental approach in their examination of the ontogeny of the syndesmosis tibiofibularis in hind limbs of domestic fowl. This element is initiated as a cartilage
essentially a sesamoid
that is later incorporated into the tibiotarsus as a bony crest. Theropod dinosaurs, but not other tetrapods, have a similar crest. Because the cartilaginous precursor is mechanically induced, it fails to form in paralyzed embryos. One can readily imagine that altered patterns of muscle-induced stress evoked the cartilage in theropod dinosaurs (Mu¨ller, 2003).

(A)

(C)

Final patterns

BOX 30.1 (Continued)

479

FIGURE 30.3 The traditional interpretation (A) of the tibia and fibula in chick embryos in which a barrier is placed between developing tibia and fibula (top row) is that the absolute length of the fibula and the patterns of the tarsals and metatarsals all change. The embryonic (B) and adult (C) patterns found by Mu¨ller (*1989) differ from those shown in Figure 30.2. Image kindly provided by Gerd Mu¨ller.

proximally to the tibia (Figure 30.2). Archer et al. (1983) proposed that loss of the epiphysis rather than competition between condensations is a more likely explanation for the slowed growth of the fibula and therefore for differences between fibula and tibia
differential growth, not competition
although, as they also showed, the diameter of the tibia is already greater than that of the fibula at the onset of chondrogenesis, which could result from competition.

480 PART | X Growing Together and Growing Apart

Growth Retardation

A Growth Inhibitor

All who have studied creeper embryos attribute the slowed long-bone growth to the failure of tibial chondrocytes to hypertrophy on schedule, which in turn slows subperiosteal osteogenesis. See Landauer and Dunn (1930), Lerner (1936) and Cock (1966) for growth retardation, and see Chapters 1, 2 and 22 for subperiosteal osteogenesis. The theory that growth retardation can result in specific skeletal defects was tested 75 years ago by Fell and Landauer (1935) by culturing limb buds from chick embryos of 3.5
5 days of incubation in a medium that restricted their growth. The medium was modified from the one they had developed for organ culture of skeletal elements by diluting the embryo extract. Limb-bud growth was retarded, as today we would expect, but those were the pioneering days of organ culture, immune to revisionist interpretations. Of greater interest, although the smaller limb buds produced chondrocytes, they did not hypertrophy. Consequently, the perichondrium failed to transform to a periosteum and subperiosteal osteogenesis was not initiated. This result provides one basis for the concept
discussed in Chapter 22
that chondrocyte hypertrophy plays a role in transforming the adjacent perichondrium to a periosteum and initiating subperiosteal osteogenesis. Mandibular membrane bones cultured on the same growth-restricting medium ossified normally. Fell and Landauer extrapolated these results from organ culture to propose that retardation of chondrocyte maturation is sufficient to explain defects seen in creeper mutants in ovo9. Fell and Landauer’s study stimulated research on the cp mutant. Indeed, more studies using in vitro cultivation or transplantation may have been undertaken to investigate the mode of action of cp than for any other skeletal mutant. Despite this, the action(s) of cp, like the thalidomide syndrome discussed in Chapter 38, is not fully understood; both are fundamentally defects in growth, and little is known about how the many processes involved in the control of growth are regulated and coordinated. Morphogenesis of cp limb buds remains abnormal after they are transplanted into wildtype hosts, indicating that cp acts directly on the skeleton. Hamburger and Waugh (1940), however, showed that limb buds from cp embryos were less retarded than limb buds from wild-type embryos when grown as transplants Accordingly, they cautioned against using transplantation as the means to analyse growth and its control. I raised a similar caution in the context of chorioallantoic grafting; both techniques impose space limitations that impede full growth potential (Hall, *1978d).

A systemic effect of the creeper gene on the whole embryo potentiates the action of other agents that influence skeletal development and growth. As two examples: creeper embryos show earlier onset and more pronounced symptoms of rickets in response to vitamin D than do wild-type embryos; and retardation of tibial growth in response to sex hormones (testosterone, estradiol, dehydroepiandrosterone) is more pronounced in creeper than in wild-type embryos (Wolff and Kieny, 1963; Kieny and Abbott, 1962; Elmer, 1968). Creeper embryos produce a soluble molecule(s) that inhibits growth but not morphogenesis of tibiae from wildtype embryos cultured in its presence. In addition to providing presumptive evidence for a growth inhibitor, this is a nice illustration of how morphogenesis and growth can be controlled independently. Tibial growth is retarded by some 12% when cultured in a medium containing embryo extract from creeper embryos rather than embryo extract from wild-type embryos. Total protein is reduced because of deficient leucine metabolism. Coculture of wild-type and creeper tibiae retards growth of wild-type tibiae in comparison to their growth when cocultured with a second wild-type tibia. The tibiae of creeper mutants, therefore, appear to produce the growth inhibitor, releasing it in vitro, and perhaps synthesising it as well. Tibiae from wild-type embryos accelerate accumulation of protein and growth rate between 8 and 9 days of incubation. Both parameters are slowed in tibiae from creeper embryos. Protein/µg DNA and hydroxyproline/µg protein decrease at the beginning of the eighth day in creeper tibiae relative to wild-type embryos. Levels of hydroxyproline/µg DNA are within wild-type levels, suggesting that noncollagenous protein is reduced in creeper10.

MECHANICAL STIMULATION AND CHONDROBLAST DIFFERENTIATION AND GROWTH Dennis Carter has promoted the role of intermittent stress in skeletal differentiation, assigning a role to mechanical forces early in ontogeny (D. R. Carter et al., *1996, Carter and Wong, *1988; Carter and Beaupre´, 2001). Carter also argues that if mechanical loading is important in bone development, it must play an important role in bone evolution, an issue taken up by Khayyeri and Prendergast (2013) with a model for the origination of mechanosensitive genes and the evolution of endochondral ossification. Different types of mechanical stress, as listed below (and see D. R. Carter et al., 1998a), affect skeletal tissues and initiation of skeletal growth differently.

Chapter | 30 Initiating Skeletal Growth

G G

G

G

G

G

Cyclic motion and associated shear promote mitosis. High tensile stress promotes differentiation of fibrous tissue. Tensile strain accompanied by hydrostatic compressive stress promotes fibrocartilage. Hydrostatic compressive stress promotes chondrogenesis (M. O. Wright et al., 1992). Intermittent loads of low stress or low tensile strain promote intramembranous ossification. Intermittent hydrostatic compressive stress inhibits endochondral ossification.

Three examples of chondrogenesis and intermittent load are provided. In the first, chondrogenic cells from limb buds of H.H. 23 or 24 chick embryos were cultured in agarose and exposed to static compressive loading (a constant 4.5-kPa stress) or cyclic compressive loading (9 kPa peak at 0.3 Hz). Cyclic loading doubled the number of cartilage nodules that formed (doubling the number of chondrogenic cells) and increased incorporation of 35S. Static loading had little effect on either parameter (Table 30.1). In the second, chick epiphyseal chondrocytes cultured under high cell density produce more proteoglycan and more aggregated proteoglycan when exposed to intermittent compression than when not. On the other hand, cultured embryonic chick tibiae respond to 20 min of continuous exposure to 0.4 Hz by maintaining levels of alkaline phosphates (levels decline in controls) and with enhanced synthesis of type I collagen11. In the third, periosteal cells isolated from human tibiae were allowed to proliferate in monolayer culture and then seeded into agarose gel constructs. RT-PCR was used to quantify changes in levels of mRNA expression after cells were exposed to intermittent dynamic

TABLE 30.1 Average Number of Cartilage Nodules Formed after 2 h of Static or Cyclic Loading of Chondrogenic Cultures from Wing and Limb Buds of Chick Embryosa Area (µm2 3 102)

0
2

Nodules/10 mm2

Control

Static Loading (Constant 4.5 kPa)

Cyclic Loading (0.25
9 kPa at 0.3 Hz)

7.5

9

19

2
4

7

6

22

4
6

7

6

16.5

6
8

7

8

13

8
10

3

3

6

101

11

5.5

8

a

Based on data from Elder et al. (2000).

481

compression of 1 Hz and 15% strain for 4 days with or without addition of Tgfß3 at 10 ng/mL. Both osteogenic and chondrogenic genes (Runx2, Sox9, ALP, Col2a1, ColXa1) were up-regulated but the response to Tgfß3 and compression was seen only in cells from one of the donors, a donor specificity both unexpected and raising a substantial caution when considering extrapolating results from individual to individual (Bonzani et al., 2012)12. Osteoblastic and chondrogenic cells respond to different types of mechanical stimulation. Fluid shear, especially intermittent or pulsatile fluid flow, rather than hydrostatic compression and/or stretching promotes osteogenesis in vitro (Basso and Heersche, 2002; Datta et al., 2006). As discussed in Box 29.1, for progenitor cells to respond to mechanical stimuli by initiating chondrogenesis rather than osteogenesis, the mechanical stimulus must be intermittent. Several classes of independent evidence support this statement: G

G

G

G

The number of chondrogenic cells entering the S phase of mitosis is regulated by intermittent compressive forces. Progenitor cells within mesenchyme take up lipid and modulate to fat cells (adipocytes) in the absence of mechanical stress, but synthesise sulphated GAGs and differentiate into chondroblasts in response to high rates of change in compression. Sensitivity to intermittent compression induces cardiac cartilage formation, discussed further in Box 17.2. Bipotential progenitor cells on avian membrane bones initiate secondary chondrogenesis in response to intermittent mechanical forces. I explored this class of evidence in Chapter 1213.

MECHANICAL STIMULI AND METABOLIC ACTIVITY If progenitor cells differentiate into chondroblasts instead of osteoblasts or fibroblasts in response to intermittent mechanical stimulation, we might expect particular cellular metabolic activities to be especially sensitive to changes in the mechanical environment surrounding progenitor cells. Several types of response are known, eight of which are listed below. i. Hyaluronan, which plays a role in condensation (for which see Box 4.1 and Chapter 20), is extremely sensitive to mechanical stresses. Indeed, seeking any correlation between cellular constituents and biomechanical data from the same cells reveals that GAGs are often the only component whose levels correlate with the mechanical data. ii. Synthesis of hyaluronan, collagen, chondroitin-6sulphate and DNA all are enhanced when smooth

482 PART | X Growing Together and Growing Apart

iii.

iv. v.

vi. vii.

viii.

muscle cells are stretched on membranes, particularly when stretching is cyclical. Rat calvarial osteoblasts respond to intermittent or constant stretching with enhanced cell division and enhanced synthesis of noncollagenous proteins. Osteoblast proliferation, behaviour and collagen synthesis all respond to mechanical stresses, Avian calvarial osteoblasts respond to cyclic tension in vitro by increasing DNA synthesis and enhancing mitosis. Osteoblasts exhibit behavioural changes by orienting at 90 to the imposed strain. Subjecting cranial bones from embryonic chicks to high levels of compression in vitro slows conversion of procollagen into collagen. Filling root canals stimulates collagen synthesis in the alveolar bone that anchors the teeth to the jaw, which has been interpreted as mediated by the second messenger cAMP (Box 30.2)14.

Transduction of Mechanical Signals Obviously, levels of major ECM components of cartilage and bone vary in accordance with the mechanical environment, reflecting the responses of cells to alterations in that environment. How do skeletal cells sense those changes? Bioelectrical changes are one way by which cells could transduce mechanical stimuli into signals they can recognise. Other possibilities include indirect sensing by changes in blood vessels or nerves resulting from the liberation of metabolites, or direct sensing by membrane components of skeletal cells
membrane receptors act as binding sites for ions, hormones and macromolecules.

The periostea of mammalian long bones are innervated by postganglionic sympathetic neurons; coculture of cells from an osteocyte-like cell line (MLO-Y4) and neurons from dorsal root ganglia was used by Boggs et al. (2011) to show that neuronal connections to the MLO-Y4 cells enhances synthesis and expression of genes associated with the osteogenic phenotype. Adrenergic (norepinephrinebased) innervation in neonatal rat calvariae is restricted to the bone; the sutures are not innervated. Adrenergic neurons arise from the neural crest and their differentiation is enhanced in culture media conditioned by neural tubes, but not in media conditioned by notochord or somites. Given that the sutures are neural crest in origin (Chapter 17) one wonders whether sutures might inhibit adrenergic neuronal differentiation15. With respect to periosteal
nerve interactions, Asmus et al. (2000) examined the ability of periostea to change the transmitter properties of the sympathetic neurons that innervate the periosteum. Working from knowledge that sweat glands modulate neurons from noradrenergic to cholinergic (acetylcholine-mediated) and peptidergic, Asmus and colleagues found that rat sternal periostea could modify neurons from cholinergic
their state when they reach the periosteum
to acetylcholine-secreting neurons after the cells contact the periosteum. As a final proof of this role, periostea transplanted to the skin induce skin neurons to switch from noradrenergic to cholinergic and peptidergic. Sensory and autonomic innervation augments osteoblastic activity; reduced incorporation of 3H-proline into hydroxyproline for collagen synthesis is seen in mandibular and femoral diaphyseal osteoblasts after resecting the inferior alveolar nerve or after chemically induced

BOX 30.2 Teeth, Alveolar Bone and Cyclic AMP Davidovitch and Shanfeld (1975) assayed levels of cAMP in tibiae from cats and chicks, which range from 0.2 to 0.4 pmol cAMP/mg wet weight, in comparison with alveolar and mandibular bone of cats, in which cAMP levels are somewhat lower at 0.1
0.3 pmol/mg wet weight. Orthodontic tipping of cat canines, by applying an initial force of 100 g by elastic and maintaining the elastic in position for 7 or 15 days, increases levels of cAMP by 50
130%. When 80 g of initial force is applied with a coiled spring, an initial (24-h) decrease in cAMP levels is followed by a prolonged period (28 days) when cAMP levels are some 50% above the levels in untipped canines. When cAMP is localised intracellularly using an immunohistochemical reaction, however, only a few cells have elevated levelsa. So, we know that cAMP is stimulated at pressure sites when cat canines are tipped with orthodontic force. We do not know

(i) the precise cellular response that causes this elevation, (ii) which cells produce the response, (iii) how the mechanical stimulus is perceived by these cells or (iv) whether alteration in cAMP levels is causally related to the remodelling that is initiated after orthodontic tipping. One possibility is that stress applied to alveolar bone alters the electrical environment, which in turn regulates the differentiation of osteogenic progenitor cells. Certainly, application of electric current as low as 15 6 2 µA to feline alveolar bone increases cAMP and cGMP levels and the numbers of cAMP- and cGMP-positive cells (De Angelis, 1970; Davidovitch et al., 1980). a. See Davidovitch and Shanfeld (1975), Davidovitch et al. (*1978) and Shanfield et al. (1975) for levels of cAMP in bone, Davidovitch (1973) and Davidovitch and Shanfield (1975) for the approaches using orthodontic appliances, and Davidovitch et al. (*1978) and Gustafson et al. (1977) for the response as restricted to a few cells.

Chapter | 30 Initiating Skeletal Growth

sympathectomy. As is so often the case, alveolar bone responds differently; surgical sympathectomy in rats induces resorption at the base of the incisors within a day. Although axon-derived proteins of the semaphorin family are expressed in osteoblasts and osteoclasts (Gomez et al., 2005), Semaphorin3A regulates bone mass by modulating the development of sensory nerves, not by acting directly on osteoblasts (Fukuda et al., 2013). Do skeletal cell also have receptors/mechanisms to allow them to respond directly to changes in the mechanical characteristics of the environment?16 It has been proposed that integrins, which play an important cell-signalling role in condensation (Figure 5.1), also act as mechanochemical transducers and transducers of mechanical signals. As assessed using wing-bud cells, an association between focal adhesion kinase and fibronectin is required for precartilage condensation; condensation is enhanced via ß-integrin-mediated interaction with fibronectin via tyrosine phosphorylation of focal adhesion kinase (O. S. Bang et al., 2000).

Cellular Tensegrity The proposal for integrins and mechanical transduction arises, in part, from the theory of tensional integrity versus compressional continuity used in architecture. Application of the principle of tensional integrity goes back to the Romans; the Roman arch is held in place by gravity pulling the arch tensionally downward. R. Buckminster Fuller pulled exactly the opposite way when he argued that a structure is pulled outward (and thus supported) by tensional forces inherent in and restrained by the structure, rather than structural weight providing compressive continuity. Ingber (*1993) extended this theory to the cytoskeleton in his theory of cellular tensegrity. Although little empirical information is available, it is known that ßeta 1 integrin promotes formation of focal adhesions in mediating changes in the ECM to the cell surface and then to cytoskeletal mechanotransduction. Integrin subtypes change as osteoblasts differentiate. Whether such changes drive or react to changing mechanotransduction was unclear until it was demonstrated that both ßeta 1 integrin and focal adhesion kinase are up-regulated in experimentally expanded (stretched) rat midpalatal sutures (I. Takahashi et al., 2003)17.

Membrane Potential Depending on the amount of deformation to which their membranes are exposed, the electrical potential of fibroblast cell membranes varies from 28 to 217 mV (Bard and Wright, 1974). From information on transmembrane potential we can conclude that osteoblasts contain metabolic pumps. The

483

resting potential of osteoblasts cell membranes is similar in different cell populations in vitro: 220.3 6 3.8 mV, with a range of 211 to 230 mV for isolated murine calvarial osteoblasts in vitro, which contrasts with 216.9 6 0.64 mV for osteoblasts from cortical endosteal rabbit long bones. The resting potential of osteoblasts from rabbit parietal bones in situ is much lower at 23.93 mV18. Surface osteoblasts sense strain via electric coupling between adjacent cells; membrane polarisation is responsive to hormones, PTH eliciting depolarisation and calcitonin-induced hyperpolarisation. If mechanoreceptors exist
and there is evidence that they do, perhaps localised on the cilia, as described in Chapter 1
then there must be a mechanism(s) for transducing the mechanical stimulus into a biochemical signal(s) to which cells can respond. As introduced in Box 30.2, transduction via cAMP
activation of adenylate cyclase and alteration in intracellular levels of cyclic AMP
is an attractive possibility.

SKELETAL RESPONSES MEDIATED BY cAMP A molecule with a ubiquitous distribution, adenosine 30 , 50 -monophosphate or cyclic AMP (cAMP) mediates external signals received via a cell-surface mechanism, thereby regulating intracellular metabolic activity in a great diversity of cell types and over a wide range of metabolic activities. These include but are by no means limited to the following: G

G

G G

G

G

induction of flagellar protein and sugar-digesting enzymes in bacteria, with starvation as the external signal; functioning as the attractant (acrasin) in the aggregation of slime moulds, again with starvation as the stimulus; aggregation of polymorphonuclear leucocytes; Catalysing phosphorylation of glycogen to glucose 1-phosphate; mediating synaptic transmission, activation of melanophores, the immune response, neural induction, cell division, assembly of microtubules and microfilaments and release and activation of steroid hormones; responding to adenosine receptors to participate in the commitment of osteoprogenitor cells as chondroblasts or osteoblasts (S. H. Carroll and Ravid, 2013).

A stimulus perceived by a cell activates synthesis of membrane-bound adenylate cyclase and catalyses the transformation of ATP into cAMP, which activates inactive enzymes and protein kinases to stimulate metabolic processes. Through the action of phosphodiesterase, cAMP is transformed into the inactive 50 -AMP (adenylic acid). Therefore, levels of cAMP increase via activation of adenylate cyclase or inhibition of phosphodiesterase,

484 PART | X Growing Together and Growing Apart

and decrease by inhibition of adenylate cyclase or activation of phosphodiesterase. Levels of cAMPphosphodiesterase are elevated in limb mesenchyme in the mouse mutant hemimelia-extra toes (hx), an elevation that could enhance mitosis and initiate abnormal outgrowth (Knudsen et al., 1985). Hemimelia-extra toes results from a cis-regulatory element that controls the express of Shh (Sagai et al., 2004), an element that is conserved between tetrapods and teleosts (Knudsen and Kochhar, 2010).

Matrix Synthesis and Condensation Alterations in cAMP levels can affect synthesis of ECM products by skeletal cells. Dibutyryl 30 , 50 -cAMP (i) modulates transformation of hamster ovarian cells into fibroblasts and induces synthesis of collagen within those transformed cells; (ii) increases rates of synthesis and secretion of GAGs by transformed fibroblasts; and (iii) blocks hyaluronan-induced inhibition of chondrogenesis by somitic cells maintained in vitro19. cAMP increases cell-to-cell communication as cells condense, stimulating chondrogenesis of limb mesenchymal cells in a density-dependent mechanism, provided that the cells are cultured at high densities (high is 2.5
5.0 3 104 cells/10 µL; low is 1
2 3 105 cells/10 µL). Importantly, limb mesenchymal cells can chondrify without condensation if cAMP is added to the medium (Hattori and Ide, 1985; Rodgers et al., 1989). cAMP can regulate collagen synthesis. In the presence of dibutyryl cAMP or 1.8 mM CaCl2, rabbit articular chondrocytes in suspension culture can be induced to down-regulate synthesis of type II and up-regulate synthesis of type I collagen. Direct application of dibutyryl cAMP to clonal myogenic cells depresses myogenesis and stimulates the synthesis of collagen and GAGs20. However, chondrocytes from nanomelic chicks
which lack almost all cartilage-specific GAG (Chapter 23)
respond to pressure by altering cAMP in the same way as do chondrocytes from wild-type individuals, which led Bourret et al. (1979) to conclude that cartilage-specific GAGs play no role in the chondrocytic response to mechanical stress.

Hormones Hormones such as thyrocalcitonin, thyroxine and growth and parathyroid hormones elevate intracellular levels of cAMP in the entire skeleton and in isolated populations of osteoblasts, although the mechanisms vary. Johanne Heersche and his coworkers showed that elevated levels of cAMP in rat calvariae exposed to calcitonin result from activation of adenylate cyclase. cAMP is elevated after exposing calvariae to dibutyryl 30 , 50 -cAMP because

of inhibition of phosphodiesterase. As outlined in Box 30.2, cAMP mediates the action of orthodontic forces in moving teeth through the jaws by modulating the formation of alveolar bone21. I have not discussed the possible role of cyclic GMP (cGMP), which is antagonistic to cAMP in hormone induction. Levels of cGMP in skeletal tissues follow the pattern of cAMP. Facial bones respond to PTH by a rapid elevation of the levels of both cyclic nucleotides; over a 1-h period, cGMP increases by 750%, cAMP by 150%. In mechanically compressed tibiae, cGMP decreases in the hypertrophic chondrocyte zone, but increases in the proliferative and resting chondroblast zones, parallelling changes in cAMP22.

cAMP AND PRECHONDROBLAST PROLIFERATION Prechondroblasts can translate pressure changes into altered levels of cAMP. Three situations for which information is available are long-bone development in chick embryos, amphibian limb regeneration and mammalian condylar cartilage. Levels of cAMP are high in blastemata during the first 10 days of regeneration of amputated amphibian limbs. The proposal is that cAMP aids in the accumulation of immature blastemal cells, perhaps by enhancing their proliferation. cGMP also has been implicated in regulating cell proliferation during amphibian limb regeneration (Liversage et al., 1977) and cAMP in proliferation of chondrogenic cells. Taban and Cathieni (1989) demonstrated greater increase of cAMP than cGMP in blastemal cells and the reverse pattern in stump tissues, concluding that these second messengers were involved in cell cycling in the blastema and in cell movement and differentiation in the stump23.

Long Bones in Chick Embryos Gideon Rodan and colleagues adapted a tuberculin syringe to deliver a known compressive force to embryonic tibiae in vitro and used it to evaluate the effects of compression on glucose metabolism and 3H-thymidine incorporation into DNA. Intact tibiae, slices of tibial epiphyses and isolated epiphyseal cells from 16-day-old chick embryos all responded to a pressure of 60 g/cm2 by accumulating or losing cAMP and cGMP. A preferential decrease in cAMP in cells of the proliferative zone matched the decrease in whole tibiae, meaning that the change in prechondroblasts was sufficient to explain the decrease in the entire skeletal element24. Prechondroblasts from the proliferative zones and hypertrophic chondrocytes from embryonic chicks tibiae were then exposed to the same pressure for 15 min.

Chapter | 30 Initiating Skeletal Growth

cAMP levels were 20% lower in the pressure-treated proliferating cells than in non-pressure-treated, although levels in hypertrophic chondrocytes were unaffected (Table 30.2). The response of proliferating chondroblasts to short exposure to pressure was mimicked by the calcium ionophore A23187 in the absence of pressure, while the effects of pressure on cAMP levels were abolished by the chelating agent EGTA (Table 30.2). Bourett and Rodan (1976a,b) concluded that an increase in Ca11 concentration mediates a decrease in cAMP levels within proliferating cells. However, pressure increased calcium levels within hypertrophic chondrocytes without altering cAMP. Adenylate cyclase in these cells is not sensitive to changing calcium levels; calcium inhibits adenylate cyclase in plasma membrane preparations derived from proliferating chondroblasts, but not in membrane preparations from hypertrophic chondrocytes (Rodan et al., 1977). Therefore, G

G

G

short, 15-min exposure to physiological pressure (60 g/cm2) enhances Ca11 uptake into chondroblasts and chondrocytes; proliferating chondroblasts respond to increased Ca11 levels by a decrease in adenylate cyclase and lowering cAMP levels; and the effects of pressure are localised on immature cells by loss of the ability to respond to heightened Ca11 as chondroblasts differentiate25.

An intriguing result may implicate Indian hedgehog (Ihh) in the transduction of those mechanical stimuli that influence chondroblast proliferation. Ihh plays a role in regulating cell proliferation, a role that is summarised in Figure 15.2. Q.-q. Wu et al. (2001) showed that cyclic mechanical stress induces expression TABLE 30.2 A Summary of the Levels of cAMP in Proliferating Prechondroblasts from Tibiae of 16-DayOld Chick Embryos after Exposure to Pressure (60 g/cm), the Calcium Ionophore A23187 and/or the Chelating Agent EGTAa,b c

Treatment

cAMP

Krebs
Ringer
glucose (control)

4.08 6 0.37

Per cent of Control

Pressure

3.21 6 0.36

221

A23187

3.47 6 0.47

215

Pressure 1 A23187

3.48 6 0.42

215

EGTA

6.93 6 0.31

170

Pressure 1 EGTA

6.79 6 0.11

166

a

EGTA, ethylenebis(oxyethylenenitrilo)tetraacetic acid. b Data adapted from Bourret and Rodan (1976a,b). c pmol cAMP/106 cells, mean 1 SEM, based on 6
12 replicates.

485

of Ihh by chondroblasts and that expression is abolished by gadolinium, an inhibitor of stretch-activated channels. Block Ihh during mechanical loading of chick sternal chondrocytes and the stimulatory effect of loading on cell proliferation is abolished. Bmp2 and Bmp4 are also upregulated in response to mechanical loading, an upregulation that is mediated by Ihh.

Mammalian Condylar Cartilage When occlusal changes are created in rat mandibular condyles, cells in the intermediate layer but not those in the hypertrophic layer reduce levels of cAMP and enhance levels of cGMP, the latter associated with cell proliferation. Continuous compression applied to condylar cartilage isolated from 4-day-old rats and maintained in vitro reduces the levels of both acid and alkaline phosphatase, increases cell proliferation and decreases 35S incorporation into acid mucopolysaccharides. Intermittent compression increases alkaline phosphatase in the hypertrophic zone, decreases proliferation and increases 35 S incorporation, inhibiting condylar cartilage growth beyond that seen with continuous compression (Table 30.3). Proliferation is inhibited with intermittent compression. At least some of these changes are mediated by cyclic nucleotides: G

G

an increase in intracellular cAMP is associated with reduced chondrocyte proliferation and enhanced hypertrophy, that is, with enhanced differentiation, while an increase in intracellular cGMP is associated with increased proliferation without affecting hypertrophy, that is, cell proliferation is enhanced without any action on differentiation26.

TABLE 30.3 Effects of Continuous and Intermittent Compression on the Behaviour of Mandibular Condylar Cartilage Isolated from Four-Day-Old Rats and Maintained in vitroa Continuous Compression

Intermittent Compression

Alkaline phosphatase activity

Decreased in hypertrophic and erosive zones

Increased in hypertrophic zones

Acid phosphatase activity

Decreased

Unaltered

Prechondroblast proliferation

Increased

Decreased

Incorporation of 35 S-sulphate

Decreased

Increased

Parameter Measured

a

See Copray and Jansen (*1985).

486 PART | X Growing Together and Growing Apart

In conclusion, alteration in intracellular cAMP is an attractive mechanism allowing proliferating chondroprogenitor cells to respond to mechanical changes in their environment and transform to chondroblasts to initiate cartilage growth.

12.

NOTES

13.

1. See Bryant and Simpson (1984), Atchley and Hall (*1991) and Hall (1991c) for control of skeletal growth. P. D. F. Murray and Selby (1930), Moss (1972b,c), the volume edited by Moffett (1972) and Bryant and Simpson (1984), Hall (*1985a), Atchley and Hall (*1991), Storey et al. (*1992) and Herring (1993, 1994) provide discussions of the relative contributions of intrinsic and extrinsic factors to the growth of cartilage and bone and facial development. For the importance of this epigenetic approach to understanding skeletal growth and morphogenesis, see Hall (1983g, 1984a
e, 1987a, 1990b,c, 1991d, *1995a, *1998a) and Hallgrı´msson and Hall, 2011). 2. Cartilage and bone also develop when blastocysts are cultivated in vitro or ova are grafted into the testes (Stevens, 1968; Hogan and Tilly, 1977). 3. See Gru¨neberg (1954), Gru¨neberg and des Wickramaratne (1974) and Cooke (1975) for the three possible mechanisms, investigation of any of which requires looking for control of growth processes during early embryonic development, not during the ‘growth period’. 4. See Ede and Agerback (1968), Ede and Law (1969), Ede and Flint (1972, 1975), Ede et al. (1974) and Wilby and Ede (1975) for adhesiveness of a movement of cells in limb buds from talpid3 embryos. 5. See Ali-Khan and Hales (2003) for apoptosis, Kwasigroch and Kochhar (1975) for the mouse studies, and Riley (1974) for the link with generation time. 6. See Minkoff and Kuntz (1978), Minkoff (1984) and Minkoff and Martin (1984) for the growth of craniofacial processes, Richman and Tickle (1992) for the role of facial epithelia, Richman and Lee (*2003) for Fgf2, and Hall (1980a), Coffin-Collins and Hall (1989), Saber et al. (1989) and Hall and Coffin-Collins (1990) for epithelia and mesenchymal viability. 7. See Landauer (*1934), I. M. Lerner (1936), Hamburger (1941), Rudnick (1945b), Kieny and Abbott (1962), Elmer (1968) and Shibuya and Kuroda (1973) for the Creeper phenotype. 8. My thanks to Gerd Mu¨ller (University of Vienna) for his comments on this section of text and on Box 30.1. 9. See Van Limborgh (*1982) for discussion of the differing controls operating in the growth of cartilage and bone. The cp strain that arose in Japan is said to show precocious differentiation of chondrocytes coupled with decreased rates of proliferation of precursor cells (Shibuya and Kuroda, *1973). However, I am unable to completely follow their evidence. 10. See Elmer (1968) and for these studies, and Loewenthal (1957) for some histochemical data on young embryos. 11. See Elder et al. (2000) and Van Kampen et al. (1985) for the two studies with intermittent load, and Zaman et al. (1992) for the study with chick tibiae. Burger et al. (1991) provides an insightful review of mechanical stimulation modulating osteogenesis from fetal bones

14.

15.

16.

17. 18.

19. 20.

21.

22.

23.

24. 25.

26.

in vitro, especially onset of movement and its relation to mineralisation in vivo and in vitro. ES cells that form embryoid bodies can redifferentiate as chondrocytes that hypertrophy, mineralise and are then associated with an osteoblast-like phenotype, a redifferentiation that is blocked by Tgfß3. Clonal cultures established from these chondrocytes express an adipogenic potential (Hegert et al., 2002). See Veldhuijzen et al. (1979) for proliferation and Rodbard (1970) for modulation to chondroblasts. ˚ keson et al. (1974) and Woo et al. (1975) for data on levels See A of GAGs and mechanical stimulation, Leung et al. (1976, 1977) for culture of smooth muscle cells, Hasegawa et al. (1985) for the response of calvarial osteoblasts, Buckley et al. (1988) for the data from avian osteoblasts, Ehrlich and Bornstein (1972) for procollagen, and Espie (1975) for responses to root canals. See Cowin et al. (1992a,b) for excellent analyses of mechanosensory system in bone. See Hohman et al. (1986) and Alberius and Skagerberg (1990) for long bones and calvariae, Howard and Bronner-Fraser (1985) for the culture study, and Herskovits et al. (1991) for innervation of bone. See I. J. Singh et al. (*1982) for the 3H-proline study, Sandhu et al. (1990) for the incisor study, Herskovits et al. (1991) for an overview of innervation, and Roth (1973), Edelman (1976), Salomon and Pratt (1976), Hogg et al. (1980), S. J. Jones et al. (1981), Maclean and Hall (1987), W. Knudson and Knudson (1991), Matzner et al. (1992) and Killian et al. (1993) for membrane receptors. See N. Wang et al. (1993) for promotion of focal adhesions, and J. H. Bennett et al. (2001) for changing subtypes. See Schusterman et al. (1974) and Chow et al. (1984) for the resting potential data, Harrigan and Hamilton (1993) for strain detection, and see Jeansonne et al. (1978) for polarisation. See Hsie et al. (1971) and Goggins et al. (1972) for the fibroblast data, and Toole (1973) for somatic chondrogenesis. See Deshmukh and Sawyer (1977) and Schubert and Lacorbiere (1976) for the studies with articular chondrocytes and myogenic cells. See Toole (1973), S. B. Rodan and Rodan (1974) and Davidovitch et al. (1977) for hormonal elevation of cAMP, and Heersche et al. (*1974) for the mechanism. See Davidovitch et al. (1977) for the cGMP in facial bones, and Rodan et al. (1975b) for the data on tibiae. As discussed in Chapter 15, cGMP mediates the action of nitric oxide in enhancing fracture repair. See Jabaily et al. (1975) and Liversage et al. (1977) for regeneration, and Burger et al. (1972) and Rodan et al. (1975b, 1978) for cAMP and proliferation. See Rodan et al. (1975a,b) for the method and application to tibiae, and Bourret and Rodan (1976a,b) for the subsequent studies. See McMahon (1974), Rebhun (1977) and Rasmussen and Goodman (1977) for overviews of the interplay between Ca11 and cAMP in regulating cell proliferation and differentiation. See Ehrlich et al. (1980) for occlusal changes, and Copray and Jansen (*1985), and Kantomaa and Ro¨nning (1992) for continuous and intermittent compression.