Endochondral Ossification Is Accelerated in Cholinesterase-Deficient Mice and in Avian Mesenchymal Micromass Cultures

Most components of the cholinergic system are detected in skeletogenic cell types in vitro, yet the function of this system in skeletogenesis remains unclear. Here, we analyzed endochondral ossification in mutant murine fetuses, in which genes of the rate-limiting cholinergic enzymes acetyl- (AChE), or butyrylcholinesterase (BChE), or both were deleted (called here A-B+, A+B-, A-B-, respectively). In all mutant embryos bone growth and cartilage remodeling into mineralizing bone were accelerated, as revealed by Alcian blue (A-blu) and Alizarin red (A-red) staining. In A+B- and A-B- onset of mineralization was observed before E13.5, about 2 days earlier than in wild type and A-B+ mice. In all mutants between E18.5 to birth A-blu staining disappeared from epiphyses prematurely. Instead, A-blu+ cells were dislocated into diaphyses, most pronounced so in A-B- mutants, indicating additive effects of both missing ChEs in A-B- mutant mice. The remodeling effects were supported by in situ hybridization (ISH) experiments performed on cryosections from A-B- mice, in which Ihh, Runx2, MMP-13, ALP, Col-II and Col-X were considerably decreased, or had disappeared between E18.5 and P0. With a second approach, we applied an improved in vitro micromass model from chicken limb buds that allowed histological distinction between areas of cartilage, apoptosis and mineralization. When treated with the AChE inhibitor BW284c51, or with nicotine, there was decrease in cartilage and accelerated mineralization, suggesting that these effects were mediated through nicotinic receptors (α7-nAChR). We conclude that due to absence of either one or both cholinesterases in KO mice, or inhibition of AChE in chicken micromass cultures, there is increase in cholinergic signalling, which leads to increased chondroblast production and premature mineralization, at the expense of incomplete chondrogenic differentiation. This emphasizes the importance of cholinergic signalling in cartilage and bone formation.


Introduction
Endochondral ossification in vertebrates, that is formation of long bones, presents an excellent model to analyze cellular and molecular functioning of so-called non-neuronal cholinergic a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 the effects of deletion or inhibition of ChEs affects on bone formation in embryonic mouse and chick models. We analyzed three ChE mutant mice, in which either one, or both ChEs were deleted. In all cholinesterase mutants from E13.5 until birth, growth dynamics and endochondral ossification were much affected, along with disruption in expression of skeletogenic master genes and a remarkable loss of proteoglycans. Notably, all ChE mutants presented distinct phenotypic differences, indicating that both ChEs contribute differently to the entire process of skeletogenesis. Independently, an improved in vitro micromass culture system of chicken limb bud cells showed that cholinergic stimulation, or AChE inhibition accelerated chondrogenic differentiation and mineralization. In summary, this study demonstrates leading roles of AChE and BChE in bone development, predominantly but not exclusively by terminating ACh-mediated activation of nAChR. The biomedical implications of these findings are briefly discussed.

Animals
Formalin-fixed samples of WT, AChE -/-/BChE +/+ , AChE +/+ /BChE -/-, and AChE -/-/BChE -/knockout mouse embryos were kindly provided by Dr. O. Lockridge (University of Nebraska Medical Center, Omaha, USA). The AChE -/mice were generated in the 129SV strain and characterized as described previously [42]. After birth, double knockout mice (AChE -/-/BChE -/-) did not breathe and hence died immediately after birth. AChE -/-KO mice survive to adulthood when maintained on a liquid diet [43], however, these mice do not breed. Therefore, the AChE knockout colony was maintained by breeding heterozygotes. BChE -/knockout mice were produced by gene targeting and bred to mice in strain 129S1/SvImJ [44]. BChE -/mice are healthy and indistinguishable from wild-type mice under normal conditions. The AChE -/-/BChE -/phenotype was analyzed and the mice were genotyped by PCR. Samples were immediately taken for genotyping and then mice were fixed in 4% PFA/PBS, pH 7.4.
Ethics statement: 1. For studies on mice: all tissue materials from mice as used in this study were provided as fixed samples from Dr. O. Lockridge (Omaha, USA; for ethics declarations see [36][37][38]), except for in situ hybridisation results of Runx2 and Ihh, which were derived from mutant KO mice from the same lines, raised in our own colony at TU Darmstadt (permission, Regierungspräsidium Giessen, AZ IV 44-53r30.03TUD10.01). 2. For studies on chick embryos: according to German animal welfare regulations ("Deutsches Tierschutzgesetz"), chicken embryos before hatching are not assigned the legal status of "animals"; therefore approval from ethics committee was not required for this study.
loosened ectoderm was then removed manually with fine preparation forceps. Purity of the mesenchymal tissue (absence of epithelial cells) was established microscopically. The mesenchymal tissue was mechanically dissociated in culture medium, and the cell number was adjusted to 2x10 7 cells/ml. A 10μl drop, containing 2x10 5 cells, was carefully placed in the center of each cavity of a 24-well plate. The cell drops were incubated for 1 h at 37˚C and 5% CO 2 ; subsequently 500μl aliquot of culture medium was added to each cavity. After two days of incubation, the culture medium was changed and replaced with 1 ml of fresh differentiation medium and afterwards changed every two days. To study the cholinergic system in vitro, the micromass cultures were treated with different concentrations (conc., see in Results and Figure legends) and combinations of either BW284c51 (specific AChE inhibitor), nicotine (an ACh agonist) or MLA (an antagonist to α7-nAChR), all 3 purchased from Sigma.

Histological stainings of micromass cultures
The cultures were washed with PBS and fixed with 4% PFA for 2 h. After removal of PFA, the cultures were stained by Alcian Blue (A-blu), Alizarin Red (A-red), or Alkaline phosphatase (ALP). For Alcian Blue staining, the cultures were pre-incubated with 3% acetic acid for 10 min, then incubated in Alcian Blue solution (0.1mg/ml) for 1hr at room temperature. The cultures were then washed with PBS and immersed in glycerol. Alizarin Red staining was performed by pre-incubating the cultures in 0.5% KOH for 10 min, and then incubated in 0.1% A-red staining solution for 1 hr at room temperature. The coloring reaction was stopped by washing with PBS, and then cultures were immersed in glycerol. Alkaline phosphatase, which catalyzes the mineralization of bone, was studied in vitro. The cultures were pre-incubated twice in freshly prepared alkaline phosphatase buffer (100mM Tris, 50mM MgCl 2 , 100mM NaCl, 0.1% Tween 20 in distilled water), and then treated with staining solution (3.5 μl BCIP + 4.5 μl NBT/ml alkaline phosphatase buffer) for 1 hour in dark. The reaction was stopped by washing the cultures in PBS, and finally immersed in glycerol.

Skeletal preparation and histology
For morphological analysis embryos were eviscerated and the skin was removed. After overnight fixation in 4% PFA at 4˚C, embryos were stained in Alcian blue solution (0.1mg Alcian blue in 2% acetic acid/EtOH) overnight. After several hours in 95%, 70%, 40% and 15% ethanol, they were transferred to a 1% trypsin solution for 2 hours to digest and clear the tissue. After staining in Alizarin red S solution for 16 h (40mg/l Alizarin red in 0.5% KOH), skeletons were cleared in a series of 0.5% KOH/25% glycerol, 0.5% KOH/50% glycerol and stored in 0.5% KOH/70% glycerol at 4˚C.
For histochemistry, limbs (whole front, or hind limbs) from newborn mice were fixed in 4% PFA in phosphate-buffered saline at 4˚C overnight, and were dehydrated in 30% sucrose solution for 2 days. Sagittal 12 μm cryosections were prepared, stained with Alcian blue solution (see above) and counterstained with 0.1% solution of nuclear fast red to assess histology of limb growth plates.

Cholinesterase stainings
Acetylcholinesterase staining was performed using the Karnovsky and Roots technique [47]. In brief, frozen cryosections were dried on a heating plate, incubated twice for 30 min in Tris-Maleate buffer (0.1M, pH 6.0) and treated with the reaction buffer (0.1M Tris-Maleate buffer, 0.1M sodium citrate, 30 mM copper sulfate and 5mM potassium hexacyanoferrate). The reaction was conducted for 1 hour at 37˚C until appropriate staining was reached, with 6 mM acetylthiocholine iodide (0.72mg/ml) as substrate and 0.1mM iso-OMPA to inhibit BChE, which led to the production of a brown insoluble precipitate (Hatchett's brown) at the site of enzymatic AChE activity, which was inspected under the light microscope. For visualizing BChE activity, butyrylthiocholine iodide (1mg/ml) was used as substrate, and 0.5mM BW284c51 to inhibit AChE, respectively.

Statistics
For descriptions of A-blu and A-red-stained whole mounts of mice embryos, 20 AChE -/-/ BChE -/double knockout mice and 10 wild type mice of stages E13.5 to P0 were used (all of them were shipped as fixed samples from the Lockridge lab). For histochemical analyses (IHC, ISH), limbs from 20 AChE -/-/BChE -/double knockout mice and 14 wild type mice of stages E13.5 to P0 were used (all of them were shipped from Omaha (as above), except for in situ hybridisation results of Runx2 and Ihh, which were derived from mutant KO mice, raised in our own colony at TU Darmstadt.
Quantification of hybridization signals and histological stainings with imageJ. Briefly, images were captured using bright field microscopy by ZEISS Axiophot microscope. Histological assessments of Alcian blue, Alizarin Red and Alkaline Phosphatase staining and IHC markers for Col-X, Ihh and MMP-13 were performed. For quantifying in situ hybridization signals, we used a purple-blue color from NBT formazan, which is a generated product of alkaline phosphatase reaction using BCIP and NTB substrate. In order not to interfere with the mRNA signal, we renounced on the counterstain. Before using ImageJ (freely available on http://rsb. info.nih.gov/ij/download.html.), the area, minimum and maximum grey area, and mean grey value can be selected in Set Measurements under Analyze. We used the tool "free hand selections" surrounding areas for measurements. All measurements were read clicking on Measure under Analyze. All areas measured for a hybridization signal were corrected for background reading. To define background, 5 areas without hybridization signal were selected and averaged, then subtracted from the signal value. All obtained numbers are percentages of stained area of the same correlating bone sections (n = 5 sections from 3 mice per group, and total 15 units were measured in each group), or of the entire micromass culture (n = 5 with separate replicas for each treatment and total 20 units were measured in each experimental setup). For Alcian blue, Alizarin Red and Alkaline Phosphatase staining, the minimum and maximum grey area and also the mean grey value were adapted to the corresponding colour.

Microscopy and photography
Wholemount wild type and mutant mice, and also their upper and lower extremities stained with A-blu and/or A-red were analyzed and photographed using a ZEISS stereomicroscope using digital imaging with a TZ7 Panasonic camera. Microscopic images of immunostained cryosections were generally taken with bright field microscopy using a ZEISS Axiophot microscope. Using a ZEISS stereomicroscope, the three-dimensional micromass cultures were imaged and merged into a 2D stack.

Cholinesterase deletion mutants present strong phenotypes
To elucidate the role of cholinesterases during early skeletogenesis in vivo, we characterized the phenotypes of AChE -/-/BChE +/+ , AChE +/+ /BChE -/and AChE -/-/BChE -/-KO mice from E13.5 until birth (for simplicity, these are furtheron called A -B + , A + Band A -B -, respectively). Compared to wild type mice (wt), all three KO mice grew faster during their first two weeks of gestation, but slowed down in the last gestational week, as demonstrated by representative images, body weight and body length for A -Bmutant (S1 and S2 Figs; [15]). While at E13.5 the weight of the A -Bmutant was more than 3-fold more than that of wild type, by E16.5 its weight was 15-20% less than that of wt (S2 Fig). Furthermore, the mutant mice exhibited a strongly bended shape, a thickened skin, which was leathery rough (S1  mice was severely affected. For example, in A -Bmutant mice, A-red staining of their dorsal ribs was somewhat weaker, but extended farther ventrally (cf. Fig 1C and 1D). At the same time, mutant vertebrae and ventral ribs remained unstained by A-blu (cf. Fig 1A, 1B, 1E and 1F). The caudal vertebral columns in the mutants, A-blu staining was significantly diminished compared to wild type (wt). Several distinct differences of stainings in the skull were also noted. For example, head plates were stained more strongly by AB (however note: the skull will not be further discussed, since it follows the process of direct ossification).
Growth of long bones is accelerated in ChE mutants. To analyze cartilage and bone formation in one representative body part of all three mutants in detail, Fig 2 presents developmental series of A-blu and A-red double-stained in the front limbs from wild type and three mutant mice (Fig 2; with a temporal delay, hind limbs showed similar results). At E13.5, all three mutants had reached a much larger size as compared to WT. Notably, mutant limb development was approximately two days ahead of wild type (see also S1 and S2 Figs; cf. [15]). At this time, in mutants with deleted BChE (A + B -, A -B -) mineralization had clearly commenced, while in wt and the A -B + mutant A-red staining was still absent. As development proceeded, mineralization appeared rather normal. By P0 in A -B + , and more conspicuously so in A -Bmutant ( Fig 2I and 2Q), A-blu staining was absent in epiphyses, but now was found at each end of diaphyses, admixed with A-red staining. In summary, absence of either ChE had accelerated early limb growth, absence of BChE preferentially accelerated mineralization, while absence of AChE affected more strongly cartilage remodeling.
Growth plate and cartilage remodeling is disturbed in newborn ChE mutant mice. The epiphysis (growth plate, GP; Fig 3A; see also Comments on terminology) in a long bone of a newborn wild type mouse consists of the following areas: resting (rc), proliferating (pc), pre-  Fig 3A). In wt bones at P0, AChE or BChE activities were restricted to MZ and specific places of perichondria and periostea (Fig 4A and 4B); both ChE activities were completely absent in A -Bmutant sections (not shown; see also [42,44]). Notably, a sharp border restricted their expressions at the dia-/epiphysis border (arrows in Fig 4A and 4B). Consistently in all mutants the structure of growth plates, in particular of the degeneration zone (DZ) appeared widened, often even destructed (cf. Fig 4C-4F). DAPI staining of cell nuclei revealed many brightly stained, mostly apoptotic cells, spread over most of the epiphysis and reaching even into the mutant mineralizing zone (MZ, details in Fig 4F; further see Discussion); these were absent in wt mouse (Fig 4C  and 4E).
Since cartilage remodeling in mutants appeared severely impaired, most prominently so between E18.5 and P0, we compared structural and molecular changes in femural cryosections of P0 mutants with wt bones (Fig 3). Sections were stained with Alcian blue (A-blu) at pH 1.0, revealing sulfated glucosaminoglycans (GAGs; Fig 3, upper), while A-blu staining at pH 2.5 visualized carboxylated and sulfated proteoglycans (PGs; Fig 3, lower), In wt mice, epiphyses were strongly stained at both pHs (Fig 3A and 3E), while in their diaphyses only some remnant A-blu staining was detected, which appeared somewhat stronger at pH 1.0 ( Fig 3A). In all three ChE mutants, A-blu staining at both pH values was completely different from wt mouse. At pH 2.5, it was nearly absent in all three mutants (Fig 3F-3H and 3J-3L). However at pH 1.0, A-blu staining presented a differential picture. In A -B + mutant, cells in epiphysis showed an increase in A-blu staining from resting (rc) to hypertrophic zones (hc), reflecting a decreasing loss of A-blu from rc to hc (Fig 3B). In A + Bmutant A-blu was restricted to a narrow band of hypertrophic cells (hc), while staining in the mineralizing zone (MZ) and in parts of perichondrium was significant (Fig 3C). Finally in double A -Bmutant, A-blu staining was nearly abolished in epiphysis, but now was most prominent within MZ (Fig 3D). Thus, absence of ChEs led to a premature (or, accelerated) degradation of cartilage matrix in epiphyses, whereby carboxylated and sulfated proteoglycans (PGs) were affected more significantly, earlier than sulfated glucosaminoglycans (GAGs). Most importantly, effects of both ChEs appeared additive in the A -Bmutant. In summary, absence of either ChE selectively affected the ECM composition in epiphyses, documenting their strong accelerating effects on remodeling of cartilage.
Skeletal master genes are disturbed in growth plates of A -Bmice. Collagen-II is a reliable marker of pre-hypertrophic chondrocytes, which is exchanged for collagen-X, as these cells become hypertrophic [19,30]. As shown by ISH in wt mouse at P0, Coll-2 was found almost exclusively in the epiphysis, with increasing staining towards the hypertrophic zone ( Fig 4G). In this zone of wt, Coll-10 was strongly expressed in a wide and a separate narrow band, from which few stained cells reached into the diaphysis (Fig 4H). MMP13 is a major protease involved in cartilage degradation, which was found in the mineralizing zone of wt mice (below DZ, arrow, Fig 4I). The distributions of all three markers were severely affected in A -Bmutant (Fig 4J-4L; for simplicity, only A -Bmutant is presented; results for the two others were comparable). At P0 Coll-2 was decreased in the epiphysis (Fig 4J; see further Discussion); Coll-10 was weakly expressed (Fig 4K). MMP13 was almost completely lost over the entire diaphysis; only some small band was left below DZ ( Fig 4L). Together, these experiments revealed an extensive derangement of major cartilage constituents in the A -Bmutant, and show that cartilage matrix degradation in the mutant diaphysis was disturbed (cf., Figs 2 and 3, S4 Fig).
ALP, which marks hypertrophic cells before mineralization onset, is essential for normal mineralization [49][50][51]. The expression of this enzyme is lost in the mutant animals at P0 but comparable to wt by E18.5 (cf. Fig 5A and 5B with 5G and 5H; S4 Fig). The expression patterns for Ihh and Runx2, as analyzed by in situ hybridization (ISH) on long bones of E18.5 and P0 wild type mice, provided explanations for these changes (Fig 5C-5F and 5I-5L). Ihh, a marker of pre-hypertrophic chondrocytes and a regulator of Runx2, was strongly expressed in wt mice at E18.5 in a limited band of the epiphyseal HC zone. This pattern had almost completely shifted into the distal mineralizing zone by P0 (Fig 5C and 5D). Thus, Ihh shifted to positions where Runx2, a marker for hypertrophic chondrocytes, was expressed in wt mice at both timepoints (Fig 5E and 5F), showing that in wild type between E18.5 and birth Ihh became colocalized with Runx2. In contrast in the mutant (Fig 5I-5L), both genes were expressed at a very low level at E18.5, whereby Runx2 was weakly expressed in the mineralizing zone ( Fig  5K). Taken together, in P0 mutant legs the expression of decisive genes and proteins that direct hypertrophic differentiation and transition into mineralization had become suppressed. Loss of cholinesterase activities results in dramatic defects in chondrocyte maturation and matrix remodeling in the hypertrophic and in mineralizing zones, including increased levels of apoptotic cells in the degenerating zone (DZ; Fig 4C-4F, S4 Fig).

Cholinergic regulation of skeletogenesis in mesenchymal chick limb micromass cultures
Based on the above findings with ChE mutant mice, and also on previous studies by Kawakita et al. [25], we hypothesized that the defects on epiphyseal growth plate chondrocytes were mediated through neuronal nicotinic acetylcholine receptors (nACHRs). To test our hypothesis in vitro, we utilized micromass cultures of primary embryonic limb bud mesenchyme [46]. In this regard, mesenchymal progenitor cells undergo condensation into compact aggregates (so-called nodules), prior to differentiation into chondrocytes and synthesis of cartilage-specific ECM components, such as collagens and proteoglycans (review, [52]). However, using the conventional technique it was not possible to distinguish the spatial arrangement of resting, proliferative and hypertrophic chondrocytes (cf. Fig 6A, left,-CS). Based on our culturing experiences with retinal spheroid cultures [15,53], we investigated whether supplementation with 2% chicken serum (CS) could improve the spatial organization in mesenchymal micromass cultures. Under conventional culture conditions, we observed typical cellular condensation, differentiation and extracellular matrix secretion (Fig 6A, left). In contrast, with supplementation with 2% chicken serum (Fig 6A, right) a distinct separation of cartilage matrix (stained by A-blu), calcified matrix (A-red) and differentiated chondrocytes (ALP staining) was observed, whereby the differentiation process spread from inside to the periphery (Fig 6A and 6C; +2% CS). To analyze the effect of chicken serum on cell viability, we used FDA and EtBr stainings (Fig 6B). In absence of CS, a low number of dying cells was distributed evenly in micromass cultures, whereas the addition of CS resulted in a circular arrangement

Fig 5. Changes of major gene expressions in A -Bmutant tibias between E18.5 and P0. (a, b, g, h) ALP as expressed in hc and diaphysis of WT (a, b), disappeared completely by P0 in mutant (g, h). (c-f, i-l) Expressions of Ihh and Runx2 disappear in mutant tibias.
In WT, ISH staining for Ihh expression shifted from a small zone in epiphysis into the diaphysis (c, d), while Runx2 was expressed at both stages in diaphyses (e, f). Note that in mutant at P0 both markers were not detectable anymore (j, l); however, in the E18.5 mutant, Ihh was nearly absent (i), while Runx2 was weakly expressed in diaphysis (k). n = 3 per each sample. of areas with high cell death (EtBr, orange/red in Fig 6B). This ring corresponded with a ring of ALP activity (cf. Fig 6A and 6C), the latter marking terminal hypertrophic chondrocytes. These results indicate that chick limb micromass cultures supplemented with 2% chicken serum differentiate circularly from inside out, thereby following distinct differentiation steps of chondrogenesis (Fig 6C).
Using this improved culture procedure, we investigated cholinergic regulation of chondrogenesis and cartilage matrix secretion (Fig 7). At div 9, a time when only mineralization sets in, the AChE inhibitor BW284c51 increased calcification and ALP activity in a concentrationdependent manner (Fig 7C, AB staining is not affected by BW284c51 at this stage, not shown). By div 13, BW284c51 led to a concentration-dependent decrease of A-blu staining, regressing from periphery to the center (Fig 7A). In parallel, calcification expanded towards inside, and ALP activity disappeared (Fig 7A, A-red and ALP). To analyse an involvement of nAChRs, micromass cultures were treated with nicotine, an agonist of ACh (Fig 7B). At a low concentration of 5μM nicotine, A-blu staining was increased, indicating an advanced initiation of prechondrogenic condensation and fusion of nodules of chondrocytes (Fig 7B). At higher nicotine concentrations A-blu staining (i.e., proteoglycan contents) decreased in a concentrationdependent manner, reaching very low levels at 20μM of nicotine. Similarly, ALP activities also decreased, being nearly absent at the highest nicotine level. In contrast, calcification increased simultaneously and reached into the inner cell mass of the micromass cultures at div13 (cf. quantification in S5 Fig). Thus, not only inhibition of AChE, but also cholinergic stimulation of nAChRs promoted cartilage-characteristic matrix secretion and "terminal" hypertrophic differentiation of chondrocytes. Then, we applied 20μM of nicotine together with increasing concentrations of MLA, a specific inhibitor of the α7-nAChR (Fig 7D). In absence of MLA (control), both A-blu and ALP stainings were very weak (cf., Fig 7B and 7D; S5C Fig), while with increasing concentrations of MLA both A-blu and ALP stainings were completely restored in a concentration-dependent manner (Fig 7D). Thus, terminal hypertrophic chondrocyte differentiation was mediated in vitro at least in part through cholinergic stimulation via the α7-nAChR.
As further support for a stimulatory role of inhibited AChE in promoting terminal chondrogenic differentiation in vitro, we used a combined FDA/EtBr staining (live cells are green, dead cells are red), which allowed an easy distinction of in vitro zones, corresponding in vivo with chondrogenic (GP), degenerating (DZ) and mineralizing zones (MZ; cf. Figs 6 and 8 with 3). Application of BW284c51 showed that cell death in the inner core of div13 micromass cultures strongly increased in a concentration-dependent manner (red, Fig 8; GP, below stippled lines), while it decreased in their outer compartments. Concomitantly, cell viability (FDA, green) decreased in chondrogenic, and increased in mineralizing zones. These findings again demonstrate that inhibition of AChE activity accelerated hypertrophic differentiation and their death, while proliferation (by viability test) increased, indicating a decisive involvement of AChE on chondrocytes in epiphyses (growth plates).

Discussion
Here we have analyzed three mutant embryonic mice, in which AChE, BChE or both genes were deleted [42][43][44]. All three mutant mice presented strong, yet distinct skeletal phenotypes. Our in vivo findings in mice were supported by an improved in vitro 3D micromass culture model, using mesenchymal stem cells from chicken embryonic limbs [46] in which not only AChE inhibition, but also nicotine promoted in vitro bone development. The present study extends a recent in vivo study on living chicken embryos, in which beads soaked with cholinergic components were implanted into one limb bud ( [13], see below).

Acetylcholine functions in epiphyses of forming bones
The main synthesizing enzyme for ACh is ChAT, with a minor possible contribution by carnithin acetyltransferase (carAT, [5]). Stem cells in the resting zone of growth plates produced ACh [5,32], as do cultured osteoblasts [4,27]. In vivo, ChAT has been identified in early mesenchyme and chondroblasts of the chick limb bud [8,25,26]. Specific types of AChRs have been detected in resting, proliferating and pre-hypertrophic chondrocytes (incl. α7-nAChR) of the murine epiphysis (growth plate, GP), as well as in osteoblasts, spongy bone and periosteum, but not in hypertrophic chondrocytes ([4, 25, 26]; see schematic, Fig 9A). Muscarinic and nicotinic types of AChRs were found in osteoblasts in vitro, and have been shown to be involved in their proliferation and differentiation [27]. In vivo, a major fraction of osteoblasts originate from the epiphysis, migrating into the diaphysis (cf., string of highly Coll-2 + cells; also called degeneration zone, DZ). Note that osteoblastic precursors (obp, orange circles), a second source of proliferative cells besides pc, will migrate from GP into MZ [31]. (a, right) represents spatial distributions of ACh, of α7-nAChR and of AChE. Note that in regions of ChE activity in MZ, ACh concentration and thus, cholinergic signaling will be low (cf. Fig 4A and 4B).  Fig 4J; [32,[54][55][56][57][58]). Therefore, ACh and the responding elements of cholinergic stimulation are preferentially distributed in epiphyses of forming vertebrate long bones, there increasing proliferation of both chondroblasts and osteoblasts (schematic, Fig 9, see further below).

Cholinesterases are rare in epiphyses of forming bones
In considering how absence of ChEs could affect bone development, their distinct in vivo localization needs attention. Besides their expression along perichondrium and periosteum, both AChE and BChE were found in diaphyses of the wild type long bones (Fig 4A and 4B). Of note, AChE is expressed in cultured osteoblasts [11,59]. ChE activities were absent inside the epiphysis, except a narrow band of AChE + cells in/near the degeneration zone (DZ, see below). Although ChEs are probably secreted by osteoblasts [11,12,60], significant amounts of both ChEs [61] are most likely delivered to the diaphysis via the blood supply. This process itself is regulated by cholinergic mechanisms [62][63][64], a topic which is not considered here. Immediately after vascularization, mineralization commences. This suggests that areas of ACh action and of its degradation are locally separated in epiphysis and diaphysis, respectively (cf. Fig 9). Hence, cholinesterases in diaphyses could act as sinks for ACh to effectively degrade it, since ACh as a very small molecule can easily diffuse from epi-into diaphysis (schematic, Fig  9A). Therefore it appears likely that in mutant mice in which either one, or both ChEs were missing, levels of ACh in epiphyses of forming bones were elevated.

Lack of cholinesterases disrupts orderly matrix remodeling
Regulators of endochondral ossification [21][22][23][24][65][66][67][68][69], which were disrupted in ChE mutant mice, act predominantly in the epiphysis. The Runx2 protein, a master regulator of chondrogenesis [19-22, 30, 65], has a binding site on the AChE promotor [18], which could be relevant for AChE expression near the degeneration zone. In ChE mutants the period between E18.5 and birth showed an extremely rapid remodeling of the cartilage matrix, that is epiphyses became void of A-blu and ALP stainings (see . As a consequence of absence of Ihh and Runx2 expressions in the newborn A -Bmutant, also MMP-13, Col-X and ALP were also nearly absent (cf., S4 Fig). In micromass cultures, the effects of AChE inhibition on ALP were perplexing at a first glimpse (Fig 7). At div 9, when remodeling was still ongoing, increasing AChE inhibition drove the system towards mineralization (note in Fig 7C: ALP also in inner core of micromass). However by div 13, mineralization became restricted to the micromass periphery, indicating its near completion (now, inner core was free of ALP; Fig 7A). Hence at div 13, ALP expression decayed, which was again promoted by increasing AChE inhibition.

Does elevated ACh in epiphysis stimulate chondroblast proliferation?
The observations presented here, using completely different approaches, are consistent with the notion that bone development is accelerated by an underlying nicotinic stimulation. However, it has been reported that nicotinic stimulation could extend chondrogenesis [5]. This is further strengthened by the observation that maternal nicotine exposure resulted in somewhat smaller sizes of α7-nAChR +/+ fetuses by E15.5, but not in α7-nAChR -/fetuses [25]. The apparently conflicting results compared to our study may be due to the fact that effects of nicotine in different systems are difficult to compare. For instance, nicotine affected osteosarcoma cells dose-dependently. That is, their proliferation was stimulated at lower concentrations, but was inhibited at higher doses [33]. More specifically, the growth curves of ChE mutant embryos document (S2 Fig), that their sizes and weights were much advanced by E13.5, and were accompanied by acceleration of both chondrogenesis and mineralization (Fig 2). But after E13.5, their growth rates slowed down (S1 and S2 Figs), so that from E16.5 onwards ChE mutant mice remained smaller until birth. Thus, a larger population of chondroblasts needed longer time to go through all steps of chondrogenesis, such that an extension brought with it a delay, which compromised an orderly matrix remodeling (Fig 9B, left pathway). For instance, cells with matrix features of pre-hypertrophic chondrocytes (e.g., A-blu + staining) found themselves in the mineralizing zone (Figs 2 and 3; see further below), showing that a) these cells had not completed matrix remodeling before their arrival in the diaphysis, and that b) the process of ossification in ChE mutants was initiated prematurely and independently of completion of chondrogenesis. Taken together, our observations are consistent with the notion that nicotinic stimulation primarily regulates the size of the chondrocyte population [25].

Each cholinesterase exerts distinct skeletogenic effects
Since all three phenotypes presented distinct differences, further specific contributions from either cholinesterase have to be considered. BChE has been associated with sustaining cell proliferation in development and cancers [7,10,14]. In most cases, including bone development, BChE may exert its effects by degrading ACh. Its availability in bones depends-at least partially-on blood supply (see above). Noticeably, supplementation of micromass cultures with chicken serum (CS) provided them with higher levels of BChE [70]. Indeed, BChE in CS could have promoted their spatial restructuring (Fig 5), since BChE had improved a laminar structure in retinal spheroids [53]. The accelerating effect of BChE absence was most obvious in the epiphysis of the A + Bmutant, where the process of secondary ossification had clearly commenced at birth (Fig 3G, star). In wild type mouse, this does not happen before P6/7 [71], indicating that development of long bones in this mutant mouse was advanced by almost one week! A-blu staining at pH1 (indicating sulfated GAGs) also demonstrated a massive acceleration of bone differentiation in absence of BChE (Fig 3C and 3D). In addition to elevating ACh levels, BChE could act indirectly by its complexing in blood with other proteins, e.g. transferrin [70], and thereby affect ALP and mineralization [72,73].
Also for AChE additional actions must be envisaged, since in A -B + embryos onset of mineralization was delayed after E13.5 (Fig 2), contrasting the other two mutant mice. This assumption is supported by bead implantations into chicken limb buds, whereby AChE inhibitors did not stimulate, but rather decelerated skeletogenesis [13]. In fact, AChE has been associated with several non-conventional activities, independent of its ACh-degrading capacity [7,10,[15][16][17]. These include its actions on cell differentiation, on neurite outgrowth, in stress responses and on apoptosis ( [74][75][76][77]; see Introduction). Cell adhesive mechanisms in epiphyses will organize cells during cartilage remodeling. Thus, specific surface features of AChE protein, such as their CHED domain [78], its peripheral anionic binding site (PAS, [14]), or its abilty to complex with other proteins, such as collagens, could be involved in separating cartilage matrix (A-blu) and ALP from MZ (A-red, Figs 2-4 and 6B). A tetrameric ColQ-form of AChE possesses a collagen tail [7,14], which in turn can bind to ALP [79], the latter disappearing at onset of calcification [72] (cf., Fig 6A and 6C). Furthermore, an enzymatic side activity of AChE could also affect bone formation ( [13,80,81]). In particular, AChE´s possible role in apoptotic processes is being much disputed [15,[74][75][76][77]. In growth plates of wild type mice, apoptosis occurs in a narrow stripe of cells near the epi-/diaphysis border, where a slight elevation of AChE activity was noted (not shown). Interestingly, in a zone of micromass cultures that corresponds precisely with this border (DZ; Fig 6B), cell death was strongly increased as AChE was inhibited (Fig 8E-8H). The precise molecular mechanism of how AChE affects apoptosis remains elusive, but could be due to AChE affecting nuclear condensation and caspase-9 activation [82].

Conclusion
In conclusion, in cholinesterase knockout mice skeletogenesis was much disturbed, due to accelerated chondrocyte proliferation and premature onset of mineralization, leading to a premature expression of skeletogenic master genes and disturbed cartilage matrix remodeling. A primary effect of inhibition (or absence) of cholinesterase(s) appears to be an increase of chondroblasts, due to an elevated level of ACh. This conclusion is supported by in vitro micromass cultures from chicken limb buds. Thus, cholinesterases are major players to define local (graded) concentrations of ACh in the forming bone, which regulates periods of proliferation and differentiation of chondrocytes and of osteoblasts. Additionally, distinct roles of both ChEs most likely also contribute to skeletogenesis. These findings are biomedically highly relevant (see, Introduction and [33][34][35][36][37][38][39][40]58]) and deserve much more investigations.
Comments on terminology: In the literature, different terms are used for the same components of growing bones. Therefore, our definitions follow here (cf., Fig 3A): the bone can be subdivided into epiphysis and diaphysis (mineralizing zone, MZ). In the developing bone, epiphysis is coincident with the growth plate (GP; also called physis); GP consists of resting chondrocytes (rc), proliferating chondrocytes (pc), pre-hypertrophic chondrocytes (pc), and hypertrophic chondrocytes (hc), followed by the ossification fissure (also called degeneration zone, DZ, or, chondro-osseous junction, COJ), and then MZ. However, at a more mature stage, a secondary ossification center will develop within the epiphysis (cf. Fig 3G); from now onwards epiphysis and GP are not congruent any longer.