How thyroid hormones and their inhibitors affect cartilage growth and shape in the frog Xenopus laevis

Abstract

Understanding how skeleton changes shape in ontogeny is fundamental to understanding how its shape diversifies in phylogeny. Amphibians pose a special case because their jaw and throat skeleton consists of cartilages that are dramatically reshaped midway through life to support new feeding and breathing styles. Although amphibian metamorphosis is commonly studied by immersing larvae in thyroid hormones (TH), how individual cartilages respond to TH is poorly understood. This study documents the effects of larval stage and TH type (T4 vs. T3), dose and deprivation on the size, shape and morphogenesis of the lower jaw and ceratohyal cartilages in the frog Xenopus laevis. It uses thyroid inhibitors to isolate the effects of each hormone at specific concentrations. It also deconstructs the TH responses into the effects on individual dimensions, and uses measures of percent change to eliminate the effects of body size and growth rate variation. As stage increases, T4 and T3 responses become increasingly similar to each other and to natural remodeling; the differences at low and intermediate stages result largely from abnormal responses to T3. Most notably, the beak‐like lower jaw commonly observed at the lowest stage in other studies results largely from arrested growth of cartilage. TH responses are superimposed upon the growth typical for each stage so that cartilages can attain postmetamorphic shapes through dimensional changes that exceed those of natural metamorphosis. Using thyroid inhibitors alters the outcome of TH‐induced remodeling, and T4 has almost the same capacity to induce metamorphic shape changes as T3. The results have implications for understanding how the starting shapes of larval elements affect morphogenesis, how chondrocytes behave to change cartilage shape, and how intracellular processing of TH might contribute to interspecific differences in shape change. Also, the data on animal mortality and which stages and doses most closely replicate natural remodeling have practical value for researchers who treat Xenopus tadpoles with TH.

Introduction

Amphibian metamorphosis appeals to developmental and evolutionary biologists because it presents an organism‐wide suite of postembryonic genetic, cellular and morphological changes that are all mediated by the same primary agent, thyroid hormone (TH). This means that these changes can be induced by immersing animals in specific concentrations of exogenous TH, and they can be blocked with numerous chemical, surgical and transgenic means of inhibiting TH production or TH activity. Developmental biologists have used these techniques for over 100 years to investigate the cellular and biochemical processes, and genetic and morphogenetic pathways involved in tissue responses in almost all amphibian organ systems (Gudernatsch, 1912; Allen, 1916, 1918; Hughes & Astwood, 1944; Dodd & Dodd, 1976; Fox, 1983; Tata, 1993; Gilbert et al. 1996; Shi, 2000; Buchholz et al. 2006; Brown & Cai, 2007). They have also used them to infer the plasma‐ and tissue‐level profiles of TH levels required to regulate metamorphic development (Etkin, 1935, 1968; Becker et al. 1997; Page et al. 2008). Evolutionary biologists meanwhile have pursued a different perspective, using these techniques to test whether specific changes in TH production and activity might account for the evolution of interspecific differences in amphibian development (Swingle, 1922; Noble, 1924; Noble & Farris, 1929; Noble & Richards, 1930; Lynn, 1948; Lynn & Peadon, 1955; Rose, 1995a,1995b, 1996, 1999; Callery & Elinson, 2000a,2000b). Whereas one research agenda assumes or hopes for similarities between induced and natural remodeling, the other looks for differences in induced remodeling that mimic evolutionary changes and might thus help explain the latter's developmental basis.

There are many reasons why induced remodeling might not resemble natural remodeling or be informative of evolutionary processes. Larval amphibian tissues acquire their competency to respond to TH largely through the expression of TR receptors and the activity of deiodinating enzymes, processes that occur at different stages in different tissues and are regulated by different mechanisms (Buscaglia et al. 1985; Yaoita & Brown, 1990; Becker et al. 1997; Huang et al. 2001; Brown, 2005). Such processes appear to account for why limb bud formation but not tail resorption can be induced in small Xenopus tadpoles. Amphibian tissues also usually undergo a phase of growth before metamorphosis, and skeletal tissues at least appear to undergo significant changes in shape and histology as they differentiate into mature larval elements (Rose, 2009; Rose et al. 2015). It is hence plausible that the patterning and morphogenesis required to transform certain larval elements into their adult counterparts cannot occur normally in tissues that are still actively growing and have not reached the appropriate starting size, shape or histology for metamorphic remodeling. This prediction is supported by observations of abnormal remodeling in precociously induced salamander cranial cartilages (Rose, 1995a,1995b). Also, regardless of whether it is done precociously or at a metamorphic stage and age, induced remodeling requires that exogenous TH be applied at a constant concentration of either tri‐iodothyronine (T3) or thyroxine (T4). The data available from radioimmunoassay measurements of plasma TH levels at larval and metamorphosing stages generally indicate that T4 and T3 are both present, and that they increase and then decrease over these stages (Rose, 2003b). The few studies with techniques sensitive enough to measure TH levels in individuals also suggest that T3 and T4 profiles vary significantly among individuals (Rose, 2003b). Since T4 became regarded as a prohormone, or inactive precursor, to the more potent T3, most frog researchers have used T3 to induce remodeling. This is despite T4 being the primary form of TH secreted by the thyroid gland and T3 inductions often being lethal and producing body forms not observed in natural development. Following the classical tradition of using T4, salamander researchers found that T4 results in lower mortality and stronger resemblance to natural development (Rose, 1995a,1995b, 2003a; Page & Voss, 2009).

To date, however, there has been no systematic attempt to work out the effects of TH type, dose and larval stage on induced remodeling for any tissue type in either of the lab animals commonly used to study amphibian metamorphosis, the aquatic frog Xenopus laevis and the paedomorphic salamander Ambystoma mexicanum. Evolutionary morphologists are generally interested in how skeletons diversify and developmental biologists are specifically interested in why the lower jaw of Xenopus transforms into an abnormal beak shape when induced in small tadpoles (Schreiber et al. 2001; Buchholz et al. 2003). We have carried out a systemic study of induced remodeling on the feeding skeleton, specifically the lower jaw and hyoid arch cartilages, of X. laevis. Tadpoles were treated at early, mid‐ and late larval stages with either T4 or T3 at doses that encompass the plasma concentrations that have been measured in naturally developing Xenopus species (Leloup & Buscaglia, 1977; Buscaglia et al. 1985). The four doses used here (1, 5, 10 and 50 nm) are commonly used to induce tissue responses in Xenopus and other amphibian species in the literature (Buckbinder & Brown, 1993; Shi & Brown, 1993; Rose, 1995b; Brown, 1997; Schreiber et al. 2001; Das et al. 2006; Page & Voss, 2009). Our TH‐treated and control animals were additionally treated with chemical inhibitors of TH production and deiodination, methimazole (Cooper et al. 1984) and iopanoic acid (Buscaglia et al. 1985; Galton, 1989), respectively. These agents ensure that tissues respond only to the applied TH and not to TH produced by the animal's thyroid gland or to T3 produced by the deiodination of T4; iopanoic acid also ensures that the applied TH is not deiodinated into nonactive forms. Both reagents were applied at concentrations routinely used to block TH production and TH deiodination in Xenopus tadpoles (Buckbinder & Brown, 1993; Denver et al. 1997; Huang et al. 2001; Cai & Brown, 2004; Brown et al. 2005; Zhang et al. 2006; Mukhi et al. 2009; Choi et al. 2017).

Additionally, whereas previous researchers focused their analyses on comparing the shapes of skeletal elements between treated and control animals, we focused on the individual dimensions that collectively define those shapes and might later prove useful in understanding the cellular basis of the underlying shape changes. Also, rather than comparing final dimensions among individuals, we compared percent changes in dimensions which allowed us to eliminate the confounding effects of variation in body size and growth rate. This required keeping all animals separate in each experiment, and using allometric equations from a previous study (Rose et al. 2015) to estimate their starting cartilage dimensions based on their starting body dimensions. This approach allows one to interpret differences in cartilage shape in terms of changes in specific cartilage dimensions and to distinguish TH‐induced remodeling from TH‐independent larval growth, which is expected to be strongly stage‐dependent and could also vary considerably among individuals. This study follows the morphometric analyses and landmarks descriptions used by Rose et al. (2015) to quantify the larval growth and metamorphic shape change of the same cartilages in untreated X. laevis. This study also describes occurrences of abnormal cartilage morphogenesis in treated and untreated specimens and provides mortality data as a reference for future studies that use T4 and T3 to induce remodeling in this species.

Material and methods

Xenopus tadpoles were produced from hormone‐induced spawnings, raised in the lab using published techniques (Rose et al. 2015) and staged using the Nieuwoop and Faber (NF) staging system for X. laevis (Nieuwkoop & Faber, 1956).

Twelve experiments were done, 11 with the TH inhibitors methimazole and iopanoic acid, and one without. For the 11 experiments with inhibitors, early larval (NF 46) tadpoles were treated with 1, 5 and 10 nm TH, mid‐larval (NF 53/4) and late larval (NF 57/8) tadpoles were treated with 5, 10 and 50 nm TH, and early metamorphic (NF 59) tadpoles were treated with 10 and 50 nm TH. The NF 46 experiment at 10 nm was replicated without the TH inhibitors to simulate the treatment commonly cited in the literature as producing beak‐like lower jaws.

Each experiment except the NF 59, 50 nm one, had six controls and six T4‐ and six T3‐treated animals, all from the same egg clutch. NF 46 experiments were started on day 7 postfertilization after 2 days of feeding. For the others, 30–40 animals were raised to two NF stages before the required start stage and then pretreated for 7–10 days in 1 mm methimazole to arrest their development; the 18 animals that arrested closest to the required start stage were selected for each experiment. Due to difficulties arresting animals at NF 59, the NF 59, 50 nm experiment had to be done with five controls, five T4‐treated animals and four T3‐treated animals. Before starting each experiment, specimens were photographed in ventral view after being anesthetized to the point of being nonresponsive to touch by immersion in 0.2% benzocaine hydrochloride (Sigma) in 0.1× Marc's Modified Ringers (MMR). Experiments were carried out in 0.1× MMR with 1 mm methimazole (Sigma) and 10 μm iopanoic acid (Sigma) for NF 46 and 53/4, and with 1 mm methimazole and 4 μm iopanoic acid for NF 57/8 and 59. The switch to 4 μm iopanoic acid for the higher two stages was based on preliminary experiments at 10 μm that resulted in limb spasticity and locking of knee, ankle and toe joints in extended positions, followed soon after by death. The TH and iopanic acid were prepared from thawed aliquots of frozen 1‐ and 10‐mm stocks, respectively; the methimazole was diluted from a stock of 100 mm kept protected from light at 4 °C.

NF 46 animals were raised singly in 75 mL in small glass stender dishes, and higher stage animals were raised singly in filled 1‐L polystyrene boxes (Tri‐State Plastic). All containers were placed in natural light, under a black felt cover to eliminate direct light, and at room temperature (22 °C). Animals were checked on a daily basis, fed every 1–2 days with a blended suspension of Frog brittle for Tadpoles (NASCO), and had their solutions changed twice a week. Feeding was terminated if animals were no longer swimming, and TH treatments were terminated when the longest surviving specimen had either ceased undergoing external morphological change or died. In experiments where T4‐treated animals were anticipated to long outlive T3‐treated animals, one to two control and T4 animals were sampled along with the last surviving T3 animals to allow for rate‐of‐change comparisons. Specimens that were found dead were fixed immediately, and all others were euthanized by a 5‐min immersion in 0.2% MS222 (Sigma). Specimens were fixed for at least 24 h in 10% neutral buffered formalin, photographed in ventral view, stained with Alcian blue for cartilage and Alizarin red for bone, and cleared in glycerol; lower jaw, hyoid arch and branchial arch cartilages were dissected out for all stages except NF 46 and photographed in ventral view.

All photographs were taken using the same Zeiss Stemi SV 11 dissecting scope and Leica camera, and at the same magnification for each experiment. The whole‐animal photographs taken at the start of experiments were landmarked in photoshop and digitized in imagej to produce measurements of starting snout‐belly length. The cartilage photographs were landmarked and digitized to produce final cartilage dimensions as in Rose et al. (2015). For NF 53/4, 57/8 and 59 experiments, starting cartilage dimensions were estimated from starting snout‐belly lengths using allometric equations (Supporting Information Table S1) generated from 26 untreated specimens spanning NF 47–59 (Rose et al. 2015). To maximize the accuracy of starting cartilage dimension estimates for the smallest tadpoles, new allometric equations (Table S1) were generated for each NF 46 experiment using six start controls that were sampled at the start of each experiment, photographed as whole animals, stained and cleared, and rephotographed for cartilage. Percent changes in cartilage dimensions were calculated as 100 × (final dimension − starting dimension)/starting dimension, and left and right percent changes and dimension ratios were averaged. Of the 212 specimens in this study, 18 of 72 NF 46 specimens and three of 54 NF 53/4 specimens had to be excluded from the analysis, as mold growth after death prevented cartilage staining. This research received IACUC approval and conforms to NIH guidelines.

Results

Effects of NF stage and TH type and dose on tadpole survivorship

Control mortality was limited to five of 71 specimens: four NF 46 specimens in the 1‐, 5‐ and 10‐nm experiments and one NF 53/4 specimen in the 10‐nm experiment (Tables 1 and 2). That four NF 46 control tadpoles survived and grew for 84 days suggests that long‐term exposure to the TH inhibitors at the concentrations used here is not lethal and does not inhibit growth when applied at NF 46.

Table 1.

Durations for experiments

TH type	NF stage	TH dose
		1 nm	5 nm	10 nm	10 nm a	50 nm
Controls	46	4–84+	4–20+	7+	8+
	53/4		27+	2–7+		7+
	57/8		28+	28+		10+
	59			28+		10+
T4	46	5–84+	20+	7+	8+
	53/4		5–27+	2–7+		6–7
	57/8		28+	14–28+		8–10+
	59			28+		10+
T3	46	8	2–3	6–7	8+
	53/4		3–4	2–4		3
	57/8		4–5	2–9+		4–10+
	59			4–10+		8–10+

Range indicates the day of earliest death and the day at which the last surviving animal died or was euthanized. Plus indicates that one or more animals had survived to that day at which point the experiment was terminated by euthanization.

^aNF 46 experiment replicated without methimazole and iopanoic acid.

Table 2.

Mortality for experiments

TH type	NF stage	TH dose
		1 nm	5 nm	10 nm	10 nm a	50 nm	All doses
Controls	46	2/6	1/6	1/6	0/6		4/24
	53/4		0/6	1/6		0/6	1/18
	57/8		0/6	0/6		0/6	0/18
	59			0/6		0/5	0/11
	All stages						5/71
T4	46	2/6	0/6	0/6	0/6		2/24
	53/4		1/6	4/6		6/6	11/18
	57/8		0/6	1/6		3/6	4/18
	59			0/6		0/5	0/11
	All stages						17/71
T3	46	6/6	5/6	5/6	0/6		16/24
	53/4		6/6	6/6		6/6	18/18
	57/8		6/6	5/6		5/6	16/18
	59			4/6		2/4	6/10
	All stages						56/70

^aNF 46 experiment replicated without methimazole and iopanoic acid.

T3 caused high mortality at all stage and dose combinations (Tables 1 and 2). Though NF 46 specimens treated at 10 nm without TH inhibitors had not died by 8 days, their skeletal deformation, abdomen shrinking, and cessation of feeding and swimming behaviors were consistent with changes soon followed by death in other treatments. In follow‐up tests to identify a non‐lethal T3 concentration, 18‐day‐old tadpoles immersed in 1 nm T3 died after 7 days and 25‐day‐old tadpoles immersed in 0.5 nm T3 died after 17 days. Both groups died soon after exhibiting the same changes, the onset of which was increasingly delayed with decreasing concentration.

T4 caused high mortality at NF 53/4, 50 nm and moderate mortality at NF 53/4, 10 nm and NF 57/8, 50 nm (Tables 1 and 2). The NF 59, 10 and 50 nm treatments had no mortality and tadpoles completed what appeared to be almost normal metamorphic changes in 4 weeks and 10 days respectively. NF 46 tadpoles treated at 1 nm T4 appeared to grow and develop slowly through larval stages and arrest in a healthy state and at a stage‐appropriate size near the start of metamorphosis.

Background for investigating TH effects on cartilage size and shape

Though previously described in the literature (Rose et al. 2015), the natural size and shape changes exhibited by the lower jaw and ceratohyal cartilages of Xenopus are reviewed here to provide background for analyzing the induced changes. Figure 1 shows the shapes of the lower jaw and ceratohyal cartilages at early and late tadpole and early postmetamorphic stages, as well as the four dimensions used in Rose et al. (2015) and the current study to measure their size and shape. Figure 2a shows the percent changes in these dimensions across tadpole growth and metamorphosis. In tadpole growth, the lower jaw length, gape width and ceratohyal length increase much more than the lower jaw width. In metamorphosis, the lower jaw length increases slightly, and the other three dimensions decrease slightly. Figure 2b shows how these changes in dimensions affect the shapes of the gape and ceratohyal as measured by the gape depth‐to‐width and ceratohyal width‐to‐length ratios (ceratohyal width is included in this study only for calculating the latter ratio). The gape depth‐to‐width ratio increases slightly in tadpole growth and dramatically in metamorphosis, and the ceratohyal width‐to‐depth ratio does the opposite. In other words, the gape becomes slightly deeper or more pointed as the tadpole grows and much more pointed as the tadpole transforms to a froglet. The CH becomes a little less blocky in larval growth and shrinks into a shorter, much narrower cylindrical shape in metamorphosis. Under optimal growth conditions, these changes occur gradually over 5–6 weeks and 6–8 days for larval growth and metamorphosis respectively (Rose, 2014; Rose et al. 2015).

Figure 1.

Outlines showing cartilage dimensions and shapes at the start and end of tadpole growth (NF 46–59) and at the end of metamorphosis (NF 66) for untreated specimens. The NF 46 and NF 59 outlines (light and intermediate shading) are scaled to have the same anterior‐posterior (vertical) dimension, and the NF 59 and 66 (dark shading) outlines are at the same scale. IR, infrostral; MC, Meckel's cartilage. From Rose et al. (2015).

Figure 2.

Graphs showing the percent changes in the four cartilage dimensions for untreated specimens in tadpole growth and metamorphosis (a) and the cartilage shapes at the start and end of tadpole growth (NF 46–59) and at the end of metamorphosis (NF 66) (b). From Rose et al. (2015).

Some guidelines are also required to distinguish percent changes in cartilage dimensions that are TH‐dependent from ones that are TH‐independent, and, when possible, to recognize TH‐dependent changes that resemble natural metamorphic responses. Control animals are expected to exhibit percent changes in all four dimensions that are greater than or equal to zero, which would signify some or no TH‐independent growth. Since almost all control animals showed some TH‐independent growth, each of the four cartilage dimensions in a TH‐treated animal is expected to exhibit one of four responses to TH: (i) a percent increase greater than in controls, which would mean that TH stimulates growth in that dimension, (ii) a percent increase within the range of controls, which could mean that TH has no effect on that dimension, (iii) a percent increase less than in controls, but still greater than zero, which would mean that TH inhibits growth, and (iv) a percent increase less than zero, which would mean that TH definitely stimulates a reduction, or shrinkage, in that dimension. Since all four dimensions increase in larval growth, the fourth response is the only result that can be interpreted unambiguously as resembling a metamorphic response, and only for the three dimensions that normally decrease in metamorphosis, lower jaw width, gape width and ceratohyal length.

It is also important to consider that larval growth and developmental rates in lab‐raised Xenopus can be extremely variable (Rose, 2014) and that the durations of TH treatments in this study were determined by two factors beyond the experimenter's ability to control or predict, individual mortality and when the last surviving TH‐treated animal would developmentally arrest. Only in an arrested, but otherwise healthy animal can one assume that a tissue response has proceeded as far as the applied hormone can induce it at the given dose and stage. TH responses that overlap with control responses are of little use since one cannot distinguish among no response to TH, a slow response and, in the case of short‐lived specimens, a fast response curtailed by early death. A short‐lived animal might alternatively show a bigger response than a similarly treated, longer surviving animal that responded more slowly to the same treatment. The most informative TH response for each treatment is thus the one showing the maximal deviation from controls, i.e. the one whose percent change lies furthest outside the range of control values. This value, which is hereafter referred to as the TH value for that treatment (and marked by colored circles in Fig. 3), indicates both the direction and magnitude of the response. Treatments with TH values both above and below the controls (black ellipses in Fig. 3) could indicate TH‐independent growth that happens to be more variable than in controls or the beginning of a TH response or individuals responding in different directions to the same treatment. Being able to distinguish among these possibilities would depend on the distribution of responses relative to controls and their treatment times.

Figure 3.

Graphs showing the percent changes in the four cartilage dimensions for all TH‐treated and control specimens; individuals are separated by stage between graphs and by hormone type and dose within graphs. Numbers beside dots indicate days of treatment for each specimen. Legends are shown once per column; NF 57/8 has the same legend as NF 53/4.

Effects of NF stage, TH deprivation and TH type and dose on cartilage dimensions

Almost all controls in all experiments show percent increases for all four dimensions, and the increases in lower jaw length, gape width, and ceratohyal length are generally much larger than in lower jaw width as in larval growth (Figs 2 and 3). The controls for NF 46, 1 nm (Fig. 3a–d) were the only ones allowed to progress long enough (84 days) to approach the amount of growth exhibited in a normal larval period. Of these controls, the most advanced individual reached NF 55 in leg development and NF 57 in ossification (Trueb & Hanken, 1992). Based on external morphological changes, two of the five controls for NF 59, 50 nm (Fig. 3o) failed to arrest and progressed to midmetamorphic stages, resulting in percent decreases in lower jaw width, gape width and ceratohyal length, and higher percent increases in lower jaw length than in the three non‐arrested controls and the controls for NF 59, 10 nm (Fig. 3).

Of the 92 pairwise comparisons of percent changes in dimensions between TH‐treated and TH‐deprived (control) specimens, 59 have TH values that lie below control values (violet circles in Fig. 3), 9 have TH values that lie above controls (orange circles), 9 have TH values that lie both above and below controls (black ellipses), and 15 have TH values that overlap with controls. The overall distribution of TH values that lie either above or below controls shows that the three dimensions that increase in larval growth and decrease in metamorphosis, lower jaw width, gape width and ceratohyal length, generally respond to TH with growth inhibition or shrinkage. However, the one dimension that increases in both larval growth and metamorphosis, lower jaw length, transitions from growth inhibition to growth stimulation between NF 53/4 and 57/8, although the 10 nm T4 treatments at these stages produce both kinds of responses. The other three dimensions show conspicuous increases in growth stimulation between these stages, and all four dimensions exhibit percent changes at NF 57/8 and 59 that exceed what happens in natural metamorphosis.

Another conspicuous stage effect is that many TH values and controls are considerably higher or more positive at NF 46 (Fig. 3a–d) than at higher stages (Fig. 3e–p). This is true for both kinds of TH and is most evident for lower jaw length. Comparing TH values within stages shows that T3 more frequently produces a stronger response that T4, through the differences are not conspicuous, and neither hormone shows an obvious dose effect at any stage.

Effects of NF stage, TH deprivation and TH type and dose on cartilage shape and morphogenesis

Gape depth‐to‐width and ceratohyal width‐to‐length ratios of treated specimens are compared with control values (Fig. 4) and with the values for untreated cartilages at the ends of larval growth and metamorphosis (Fig. 2). Both ratios generally respond to TH in the directions observed in natural metamorphosis, the magnitudes of the responses increase with stage, and neither ratio is obviously dose‐dependent at any stage. T3 produces the greatest gape depth‐to‐width ratios at NF 46, T4 does so at NF 53/4, and both hormones induce similarly high ratios at NF 57/8 and 59, including ones that exceed postmetamorphic values (Fig. 4a–d). Some controls at all stages have ratios that exceed the range for larval growth, and only two of these outliers, ones for NF 59, 50 nm, can be attributed to failure to arrest animals at larval stages. Ceratohyal width‐to‐length ratios of treated specimens are generally more variable than gape depth‐to‐width ratios, (Fig. 4e–h), with a few individuals treated at NF 46, 53/4 and 57/8 having values above 0.4, which is not seen at any stage in natural development. A few treated and control specimens at NF 46 and 53/4 have values that approximate the start of natural metamorphosis, and T3‐treated specimens at NF 57/8 and NF 59 reach values approaching the end of metamorphosis.

Figure 4.

Graphs showing the final gape and ceratohyal shapes for all TH‐treated and control specimens; individuals are separated by stage between graphs and by hormone type and dose within graphs. Numbers beside dots indicate days of treatment for each specimen. Legends are shown once per row; NF 57/8 has the same legend as NF 53/4.

Visual examination of lower jaws revealed three cases of abnormal, stage‐ and hormone‐specific morphogenesis (Fig. 5, 6, 7). Again, some background on natural morphogenesis helps to interpret this variation (Fig. 1 and Rose et al. 2015). In small larvae, the lower jaw overall is bow‐shaped, and each side is comprised of a small median infrarostral and larger Meckel's cartilage that are separate from and not aligned with each other. The left and right infrarostrals meet medially in a variably pronounced chevron or inverted V shape. Over larval growth, the bow shape becomes slightly deeper, Meckel's cartilages become relatively thinner, and infrarostrals become relatively smaller and more aligned with Meckel's cartilages. During metamorphosis, the lower jaw overall transforms from a bow to U shape, as each infrarostral and Meckel's cartilage fuse to form single lower jaw cartilages that also become thinner and longer. Also, the median region of the lower jaw is thickened by the addition of a transverse bar of cartilage along its posterior edge.

Figure 5.

Ventral view photos of lower jaws and ceratohyals for control (left) and T4‐treated (center) and T3‐treated (right) animals for experiments using 10 nm TH at NF 46 (a–c) without inhibitors, NF 46 with inhibitors (d–f; inserts are of specimens sampled at the start of treatment), NF 53/4 (g–i), NF 57/8 (j–l) and NF 59 (m–o). Controls were selected to have the largest size and time of exposure, and treated specimens were selected to show the greatest departure from controls while matching their treatment times as closely as possible. Treatment times were 8 days (a–c), 7 days (d–f), 7, 4 and 4 days (g–i), 28, 28 and 9 days (j–l) and 27, 27 and 10 days (m–o). The smaller scale bar (for a–f) is 1 mm, and the larger is 5 mm. See Rose et al. (2015) for untreated specimens at similar stages.

Figure 6.

Ventral view photos of lower jaws and ceratohyals showing the maximum amounts of change inducible at NF 46 with 1 nm T4 (a), and at NF 59 with 50 nm T4 (b) and T3 (c). Treatment times were 84, 10 and 10 days, respectively. Scale bar: 5 mm.

Figure 7.

Ventral view photos of lower jaws and ceratohyals at NF 57/8 showing abnormal bends in U‐shaped lower jaws induced by 5 nm T3 (a,b) and an abnormal bridge of cartilage in the median lower jaw induced by 10 nm T4 (c). Treatment times were 5, 5 and 7 days, respectively. Specimens for (a) and (b) were photographed prior to dissection to show that the bends are not artifacts of clumsy dissection. Scale bars: 5 and 1 mm.

Experimental lower jaws were scored as having bow (larval), U (metamorphic or postmetamorphic), or beak shapes (Table 3). Beak shapes, which are never seen in normal development, result from straight or slightly convex or concave lower jaw cartilages meeting medially in a point that often lies anterior of the upper jaw cartilage (Fig. 5c,f,i). The infrarostrals, which are usually nonaligned and separate from Meckel's cartilages in bow shapes and aligned with and fused to Meckel's cartilages in U shapes, are variable in these respects in beak shapes. Bow shapes occur in all controls (Fig. 5a,d,g,j,m) except the two non‐arrested NF 59s, and U shapes become increasingly frequent with higher doses and stages of TH treatment (Fig. 5k,l,n,o; Table 3). Beaks occur at all stages, at the three higher doses (5, 10 and 50 nm), and with both hormones, though they are more common with T3 (Table 3). They were not induced by either hormone at NF 57/8, 10 nm and NF 59, 50 nm.

Table 3.

Occurrence of bow‐, U‐ and beak‐shaped lower jaws

Treatment	Stage	1 nm	5 nm	10 nm	10 nm a	50 nm
T4	46	4 bow	6 bow	4 bow	5 bow, 1 beak
	53/4		6 bow	5 bow		4 U, 2 beak
	57/8		5 bow/U, 1 beak	3 bow, 3 U		5 U, 1 beak
	59			6 U		5 U
T3	46	na	5 bow, 1 beak	1 beak	1 U, 5 beak
	53/4		3 bow, 3 beak	2 bow, 3 beak		1 bow, 5 beak
	57/8		5 U, 1 beak	1 bow, 5 U		1 U, 5 beak
	59			5 U, 1 beak		4 U

Na, no specimens produced visible staining due to mold.

^aNF 46 experiment replicated without methimazole and iopanoic acid.

Six NF 57/8 specimens induced by T3 exhibit U‐shaped lower jaws with outwardly directed bends midway along their lengths (Fig. 7a,b). The bends are paired in two specimens treated at 5 nm and unpaired in three others, as well as in one specimen treated at 10 nm. One specimen has a second unpaired bend just posterior of where the infrarostral and Meckel's cartilage had fused (Fig. 7a). All six specimens have a dermal bone (medial angulosplenial) along the inside edge of the lower jaw, and the specimen treated at 10 nm also has a much smaller bone (lateral angulosplenial) along the outside edge. The appearance of the bends appears consistent with lengthening of a central, flexible portion of the lower jaw cartilage, the ends of which are locked in place by adjacent tissues.

Lastly, some specimens exhibit a small bridge of cartilage along the posterior median edge of the lower jaw, which in combination with the infrarostrals, creates a ring of cartilage around a small foramen. These are found in NF 57/8 and NF 59 controls (Fig. 5j,m) and in NF 57/8 specimens that were treated with 5 and 10 nm T4 and retained bow‐shaped lower jaws (Fig. 7c).

Discussion

Although there have been many studies on how exogenous TH affects bone development in amphibians (Terry, 1918; Fox & Irving, 1950; Kemp & Hoyt, 1969; Yamaguchi & Yasumasu, 1977; Hanken & Hall, 1988; Smirnov & Vassilieva, 2014), few have described the effects on cartilage (Hanken & Summers, 1988; Thomson, 1989; Rose, 1995a,1995b). To our knowledge, this is the first study to try out a method for documenting how TH effects on cartilage shape result from changes in individual dimensions, and how the latter are affected by variation in body size and growth rate.

Interpreting the effects of TH deprivation and TH treatment on individual dimensions and cartilage shapes

That almost all dimensions increased in controls is significant in two ways. One is that most reductions in TH‐treated animals are in fact responses to TH and not artifacts of measurement error or unanticipated responses to the TH inhibitors or rearing conditions. Whereas the reductions shown by two NF 59 controls can be attributed to failed developmental arrest, smaller (1–3%) reductions in other controls are likely due to measurement and/or estimation error in slow‐growing individuals.

The second is that cartilages, like most larval amphibian tissues, will grow in the absence of TH. This is not unexpected given past successes in growing tadpoles for months in methimazole and other goitrogens (White & Nicoll, 1981; Brown, 2005; Choi et al. 2017) and the rare occurrence of exceptionally large thyroid‐deficient tadpoles in nature and commercial colonies (Rot‐Nikcevic & Wassersug, 2004; Kerney et al. 2010). This study, however, is the first to show that methimazole and iopanoic acid together can support growth, though why iopanoic acid appears toxic at 10 μm in mature tadpoles is unclear. Its effect on leg joints suggests an impact on muscle contraction and possibly the muscle growth and differentiation that starts at early limb bud stages and continues until tail loss. However, since the effect does not happen with methimazole alone (C. S. Rose pers. obs.) and is not alleviated by exogenous T4 or T3, it would appear to be a TH‐independent effect of iopanoic acid.

The general distribution of TH values in this study supports previous findings that TH applied to tadpoles induces metamorphic‐type responses and not larval growth (White & Nicoll, 1981; Gilbert et al. 1996). However, the TH responses can deviate from natural responses in stage‐, hormone‐ and dose‐specific ways. Most generally, TH responses at the lower three stages appear to be coupled with some amount of growth that happens before or during the TH response, and biases the percent changes at low doses and stages towards higher, or more positive, values than in natural metamorphosis. This is particularly evident at NF 46, at which stage controls can almost double in size in 1 week, and the three dimensions that normally undergo shrinkage at metamorphosis respond to TH with less shrinkage than in metamorphosis or with growth, albeit less than in controls (Fig. 3).

The fourth dimension, lower jaw length, stands out because it switches from an abnormal TH response, growth inhibition, at NF 46 and 53/4 to a normal one, growth stimulation, at NF 57/8 and 59. Though many T3‐treated specimens at low stages died before their accompanying controls were euthanized, that T3 treatment inhibits jaw growth is clearly demonstrated by comparing T3‐induced specimens with controls that were terminated on the same day. These include all NF 46 specimens treated with 10 nm without inhibitors and the single NF 46 specimen treated with 10 nm with inhibitors (Fig. 3b). Whether the inhibition is by T3 retarding or arresting jaw growth can be assessed by comparing specimens that died on different days. The one specimen treated with 10 nm for 7 days exhibited the same growth as the maximum growth exhibited by specimens treated with 5 nm for 3 days. This would support T3 arresting lower jaw growth. The same argument can be made for T3 arresting growth of the gape width at NF 46 (Fig. 3c). The apparent growth arrests of jaw length and gape width are coupled with decreases in jaw widths in some T3‐treated specimens, and the effects on all three dimensions are visible in beak‐shaped cartilages (Fig. 5c,f,i). Whether the small percent increases that happen in lower jaw length and gape width prior to their arrest result entirely from larval growth or from larval growth followed by a TH response that prevents further growth awaits future investigation at a cellular level.

Finding discrepancies between natural and induced remodeling raises obvious caveats for interpreting the shapes of induced cartilages. Whereas induced ceratohyals attain midmetamorphic proportions through dimensional changes that are consistent with natural remodeling, the beak‐shaped gapes induced by T3 at NF 46 attain midmetamorphic proportions via changes that have no resemblance to natural remodeling, and the high gape ratios induced at NF 57/8 and 59 result from percent changes that surpass what occurs in natural remodeling. For example, the two highest gape ratios induced by T3 at NF 59 (1.56 and 1.62) result from increases in lower jaw length of 61 and 65%, and the 1.46 ratio induced by T4 at NF 57/8 results from a 47% increase; the range for this increase in natural metamorphosis is 17–33%. Resolving whether the exceptional lengthening is a result of a normal TH response combined with larval growth or an excessive TH response requires investigation of the cell behaviors involved in untreated and induced development.

The U vs. beak dichotomy of lower jaw shape is interesting because it reveals how the differences in cartilage shape among different larval stages can affect the outcome of induced remodeling. U shapes indicate that cartilage histology and shape, hormone signal strength and cell responses to the hormone are sufficient to support a natural metamorphic shape change. We interpret beaks to signify remodeling that has been misdirected and possibly stalled by one or more of these factors being excessive, inadequate or otherwise inappropriate. Beaks induced at NF 46 are very distinct from U‐shaped lower jaws, and their median tips can extend beyond the upper jaw (Fig. 5c,f). This aspect of their appearance is misleading, as this study shows that beaks result from the arrested growth of cartilages, rather than their elongation. At NF 46, the chevron shape and relatively large size of the infrarostrals (Fig. 1) means that left and right lower jaws at this stage have thick medial ends that meet at an angle. This starting configuration and the arrested growth of cartilage between the jaw joints might be expected to predispose the lower jaw cartilages to retain this angle in induced remodeling. By NF 57/8 and 59, the infrarostrals have become relatively smaller and more aligned with Meckel's cartilages and with each other (Fig. 1). This starting configuration is closer to a U shape, and beaks at NF 57/8 are often more similar to U shapes. They appear to result from less lower jaw lengthening and thinning than in U shapes, coupled with similar reductions in gape width.

The occurrence of beaks and abnormal bends in U‐shaped lower jaws also reveals which stage and hormone combinations can be used to induce a natural shape change and which cannot. That beaks are not induced by either hormone at NF 57/8, 10 nm and NF 59, 50 nm suggests that these are the best treatments for inducing a bow‐to‐U shape transformation. The former treatment with T3, however, can produce bends in U shapes. That beaks predominate at NF 46 and 53/4 suggests that a natural shape change is generally not inducible by either hormone at these stages.

We interpret the bends in U‐shaped lower jaws as the result of an induced shape change that mimics natural development coupled with mechanical or space constraints that are imposed by adjacent tissues at this stage, but not in natural metamorphosis. The most plausible explanation is that as the cylindrical lower jaw cartilage elongates, its more terminal portions remain fixed in place and thus force one or more middle portions to bulge outwards. The terminal portions might be fixed in place by their attachment to adjacent dermal bones, and by failure of the jaw suspension to shift the jaw joint posteriorly and accommodate metamorphic lengthening of the lower jaw (Wassersug & Hoff, 1982). The absence of bends at NF 46 and 53/4 correlates with the absence of dermal bones and the induction of beaks rather than U shapes at these stages.

Comparing the roles of T4 and T3 in cartilage remodeling

The rationale for suppressing endogenous TH production and TH processing by deiodinase enzymes in this study is to compare the potential of T4 and T3 as direct inducers of tissue responses. T4 is generally considered the primary hormone produced and released by the thyroid gland (White & Nicoll, 1981). Radioimmunoassay studies on multiple species show that both hormones are typically present in the plasma throughout amphibian metamorphosis, and that T4 often reaches a higher plasma concentration than T3 does (Rose, 2003b). Nonetheless, most hormone induction experiments on amphibian larvae are now done with T3, presumably because T4 is viewed as not only less potent (Robinson et al. 1977) but less active in tissue induction. This is because most T4 is thought to be converted to T3 by a cytoplasmic deiodinating enzyme (D2) before it enters cell nuclei to bind to nuclear receptors and alter gene expression (Becker et al. 1997; Huang et al. 2001; Cai & Brown, 2004; Brown, 2005). T4 is generally considered to activate tissue responses before being supplanted by the rise in plasma T3 in mid‐ and late larval stages (Shi, 2000).

This study shows that TH inductions done without TH inhibitors produce different results than inductions done with them. NF 46 lower jaws treated with T4 with inhibitors (Fig. 5e) bear a closer resemblance to control specimens (Fig. 5d) than to ones treated with T4 without inhibitors (Fig. 5b). The latter resemble T3‐treated lower jaws (Fig. 5c) in their alignment of infrarostrals and Meckel's cartilages, which is consistent with portions of their cartilages being able to convert T4 to T3 and then respond to the T3. That lower jaws treated with T3 with and without inhibitors respond similarly (Fig. 5c,f) suggests that lower jaws at NF 46 do not express the deiodinating enzyme that inactivates T3 (D3). This agrees with Berry et al.'s (1998) finding that D3 is inactive in Meckel's cartilage until midmetamorphosis.

This study also shows that T4 is capable of directly inducing changes in cartilage dimensions and shapes that closely resemble natural metamorphosis, and with less detriment to overall health and growth than T3 (Fig. 5). Though the irregular sampling times in this study precluded a meaningful comparison of response rates, T4‐treated animals consistently appeared to respond later and more slowly than T3‐treated animals, and in a timeframe more consistent with natural development (C. S. Rose, pers. obs.). The subtle differences in lower jaw curvature and extent of CH resorption between T3‐ and T4‐treated specimens at 10 and 50 nm (Figs 5k,l,n,o and 6b,c) indicates that although T4 is capable of inducing the majority of the shape transformation, T3 is needed to complete it. The same would appear true of tail resorption (data not shown).

Whether T4 participates directly in metamorphic remodeling by entering cell nuclei and binding to receptors would depend partly on whether chondrocytes express the cytoplasmic enzymes that convert T4 to T3 or inactivate both (D2 and D3). Cai & Brown (2004) report that gills in metamorphosing Xenopus (which we assume to mean branchial cartilages since Xenopus tadpoles lack gill filaments) do not express D2 genes. Berry et al. (1998) report that D3 is up‐regulated in midmetamorphosis throughout Meckel's cartilage and in central portions of the ceratohyal. Pursuing this question further would require a more complete developmental analysis of D2 and D3 expression and activity in cranial cartilages, as well as investigating how hormones enter cartilage, which is nonvascular and sealed by a layer of undifferentiated perichondrial cells. Cartilage presumably must receive TH from extracellular fluids that bathe the tissue, but whether hormone simply diffuses across the perichondrium or is actively transported, and how cartilage size and extracellular matrix might affect its delivery remain to be investigated.

The developmental and evolutionary significance of abnormal morphogenesis

The loop‐like configuration of lower jaw cartilage that is induced at NF 57/8 by inhibitors alone or with 5 or 10 nm T4 appears to result from a small bridge of cartilage condensing between the posterior edges of the infrarostrals, coupled with retention of the infrarostrals in a midlarval configuration, i.e. not yet aligned with Meckel's cartilage. To our knowledge, a loop‐like lower jaw cartilage does not occur naturally at any stage in any amphibian species. This result thus provides another indication that changes in hormone profile can alter the timing and pattern of cartilage condensation events to produce experimental artifacts (Rose, 1995a,1995b, 1996). One possible explanation is that the new cartilage represents a precocious appearance of the transverse bar of cartilage that normally arises at midmetamorphosis, at a stage with high plasma T3 and T4 levels, and fuses to the posterior edge of the median lower jaw (Rose et al. 2015). If the cartilage does in fact represent this bar, its appearance in otherwise untransformed lower jaws in the absence of both T3 and T4, and in the presence of only T4 is enigmatic.

The abnormal bends arising in U‐shaped lower jaws help to inform our understanding of how cell behaviors contribute to this shape change in natural metamorphosis. The appearance of a bend in the middle of an otherwise normal U‐shaped cylindrical cartilage as it transforms from a shorter, blocky, bow‐shaped cartilage is strong evidence that the natural elongation of this cartilage occurs through cell division, matrix secretion and other cell behaviors inside the cartilage, as suggested by Rose (2009), and not by the addition of new cells at each of its ends, as suggested by Kerney et al. (2012). The occurrence of two bends on one side of one lower jaw further argues for cell behaviors contributing to elongation throughout the cartilage, and not just in the region that most commonly bends.

Lastly, that high TH doses can induce magnitudes of cartilage elongation and thinning in excess of what occurs naturally in this species may help explain how other amphibian species have evolved their more dramatic shape changes at metamorphosis. For example, all metamorphosing frogs rotate their jaw suspension backwards to accommodate lengthening of their lower jaws, and the angle of rotation, which ranges from 36° for a microhylid species to 69° for Xenopus and to 111° for a leptodactylid species, corresponds to the magnitude of lower jaw lengthening (Wassersug & Hoff, 1982). By this measure, many frogs undergo much greater jaw lengthening than Xenopus, and the African bullfrog, Pyxicephalus adspersus, in fact appears to triple the length of its lower jaw (Haas, 1999). Though the peak T4 and T3 plasma concentrations measured in metamorphosing Xenopus, 9.65 and 7.94 nm, respectively (Leloup & Buscaglia, 1977; Buscaglia et al. 1985), are high compared with the few other frog species for which such data are available, they are not high compared with the concentrations used here or with the peak plasma levels measured in larval salamanders, fish and lampreys (Rose, 2003b). Our results demonstrate that interspecific differences in lower jaw lengthening might arise from how T3 and T4 levels are modulated by deiodinating enzymes in jaw chondrocytes. Greater conversion of T4 to T3 could result in more cell behaviors that contribute to lengthening of this particular cartilage. More generally, the processing of TH within chondrocytes could contribute to the patterning of generative and degenerative cell behaviors that underlies the development and evolution of cartilage shape in amphibians (Rose, 2009). Also, the developmental liability of lower jaw cartilage further illustrates the primacy of this skeletal tissue over bone in accounting for the shape changes in feeding and breathing skeleton that are key to the evolution of amphibian metamorphosis (Rose, 2014).

Summary and directions for future research

Besides providing background data on mortality for researchers who treat Xenopus tadpoles with TH, this study demonstrates a new methodology for quantifying changes in the size and shape of amphibian larval cartilages in response to TH treatment and for interpreting results amidst variation in body size, growth rate and mortality. TH inhibitors are used to isolate the direct effects of T3 and T4 at specific concentrations, and percent changes in linear dimensions are used to distinguish TH responses from TH‐independent growth. The general findings include a stage‐dependent switch from growth inhibition to growth stimulation in lower jaw length, the absence of conspicuous dose‐dependent relationships at all stages, that T4 induces almost the same amount of remodeling as T3 at high stages, and that remodeling induced with TH inhibitors can differ from remodeling induced without them. This approach helps clarify the exact nature of individual tissue responses to each kind of hormone and their contributions to specific shape changes, e.g. growth arrest in beak‐shaped lower jaws and excessive elongation in some U‐shaped lower jaws and bends in others. The stage‐ and hormone‐specific occurrences of abnormal morphogenesis also help illuminate the hormonal, cellular and morphological factors that contribute to normal cartilage morphogenesis and potentially to the evolution of cartilage shape. Though a strictly morphological description of TH responses is of limited value on its own, this study provides a framework for investigating the cell behaviors and molecular mechanisms underlying cartilage shape change in amphibian larval growth and metamorphic remodeling.

Author contributions

C.S.R. raised the animals, performed the experiments, prepared specimens and wrote the paper. J.C. photographed, landmarked and digitized specimens, and did all calculations. Both authors contributed equally to the data analysis.

Conflicts of interest

The authors have no conflicts of interest.

Supporting information

Table S1. Allometric equations for estimating starting cartilage dimensions.