Abstract

Transgenic Xenopus laevis tadpoles that express a dominant negative form of the thyroid hormone receptor (TRDN) controlled by the cardiac actin muscle promoter (pCar) develop with very little limb muscle. Under the control of the tetracycline system the transgene can be induced at will by adding doxycycline to the rearing water. Pre-existing limb muscle fibers begins to disintegrate within 2 days after up-regulation of the TRDN transgene. The muscle cells do not die even after weeks of transgene exposure when the myofibrils have degenerated completely and the tadpole is nearing death. A microarray analysis after 2 weeks of exposure to the transgene identified 24 muscle genes whose expression was altered in such a way that they might cause the muscle phenotype. These candidate genes are normally activated in growing limb muscle but they are repressed by the TRDN transgene. Several of these genes have been implicated in mammalian myopathies. However, the expression of only one of these genes, calsequestrin, is down-regulated in 1 day and therefore might initiate the degeneration. Calsequestrin is one of several affected genes that encode proteins involved in calcium sequestration, transport and utilization in muscle suggesting that uncontrolled calcium influx into the growing limb muscle fibers causes rhabdomyolysis. Many of the same genes that are down-regulated in the tail at the peak of metamorphic climax just before it is resorbed are suppressed in the transgenic limb muscle in effect turning the limb growth program into a tail resorption program.

Keywords: Xenopus laevis; Metamorphosis; Thyroid hormone; Gene expression; Muscle development; Transgenesis

Article Outline

Introduction

Materials and methods

The plasmids used for transgenesis

Raising tadpoles for microarray limb samples

RNA purification, probe synthesis, and microarray hybridization

Histology, immunostaining and in situ hybridization

Results

Induction of the pCar/TRDN phenotype

Gene expression profile of the developing transgenic limb

Candidate genes

The kinetics of muscle degradation and muscle gene down-regulation

Muscle, glycolysis and mitochondrial energy related genes are influenced by the transgene

Other programs in the limb are not affected by the TRDN transgene

Discussion

The development of the muscle wasting phenotype

Candidate TH-regulated muscle genes

The role of calcium in muscle degeneration

Other primary candidate genes are implicated in myopathies

Many of the same genes that are repressed by the transgene in the limb are down-regulated in tail when it resorbs

Acknowledgements

Appendix A. Supplementary data

References

Introduction

A series of transgenic experiments, in which different cell type specific promoters control the expression of a dominant negative thyroid hormone receptor alpha (TRDN) revealed that the multiple programs of limb development are controlled independently by TH (Brown et al., 2005). A ubiquitously expressed promoter driving the TRDN reporter represses TH-induced DNA synthesis in the early limb bud (Schreiber et al., 2001) while cell-type specific promoters inhibit cell autonomously the differentiation of muscle, innervation of the limb muscle, and the growth of the limb skeleton (Brown et al., 2005). The death of muscle in the tail at metamorphic climax is also a cell autonomous TH-induced program (Das et al., 2002, Nakajima and Yaoita, 2003 and Yaoita and Nakajima, 1997). Overexpression of the same TRDN transgene that causes muscle degeneration in limbs protects tail muscle from TH-induced resorption (Das et al., 2002). A transgenic tadpole with the X. laevis cardiac actin promoter (pCar) driving the TRDN transgene dies at the climax of metamorphosis. The animal's limbs never function because they have almost no muscle (Das et al., 2002).

We began a series of microarray experiments designed to compare the TH-induced gene expression profiles in the growing limb and brain with the tail (Das et al., 2006). Not only do the growth and resorption programs differ greatly but also in some cases the same genes had opposite responses to the hormone in tail compared to limb. In this paper we have investigated the limb muscle wasting induced by the TRDN transgene. We prepared transgenic animals in which the pCar promoter drives the expression of the TRDN reporter under the control of the tetracycline inducible system. Induction of the transgene causes the degeneration of limb muscle specifically without affecting the other cell types that differentiate in the limb (Brown et al., 2005). A microarray analysis of the gene expression profile reveals a small set of muscle genes whose expression pattern is inhibited from the up-regulation that occurs normally at metamorphosis in the muscle growth program. These same genes are down-regulated in the muscle death program that occurs normally during tail resorption. Several of these candidate genes have been implicated in human myopathies.

Materials and methods

The plasmids used for transgenesis

Control of the expression of a transgene by the tetracycline system (Urlinger et al., 2000) has been applied to X. laevis development (Das and Brown, 2004). One plasmid called pCS2+(tetO)TRDN/GFP has the tet operator adjacent to a dominant negative form of X. laevis thyroid receptor alpha. This mutant receptor lacks 12 amino acids from its C-terminus, and it is fused to GFP at its carboxy terminus by a small flexible linker. The TRDN part of the resulting protein has a nuclear localization signal so that the fluorescent protein can be visualized in the nucleus. The second plasmid has the X. laevis cardiac actin promoter (pCar) driving a modified tetracycline repressor (pCar/rTA2S-M2). This system is under positive control so that addition of the tetracycline derivative, doxycycline (Dox), induces expression of the TRDN/GFP transgene. The transgenesis procedure (Kroll and Amaya, 1996) used a mixture of the 2 plasmids (Das and Brown, 2004). A doubly transformed male parent was raised to sexual maturity and bred with a wild type female. The transgene can be induced in half of the progeny. Mixtures of DNAs are often integrated together presumably in tandem arrays (Marsh-Armstrong et al., 2004).

Raising tadpoles for microarray limb samples

The F1 progeny were raised to NF 55 without the inducer. Batches of 20 tadpoles were grown in 4 l of 0.1 MMR containing 50 μg/ml doxycycline hyclate (Dox; Sigma) (Das and Brown, 2004). The tadpoles were fed daily and the rearing water was changed twice each week. After 2 weeks and 6 weeks of Dox treatment the tadpoles had advanced to NF 56 and 61, respectively. The latter developmental stage (NF61) is metamorphic climax when the endogenous TH is highest. By NF62 the control tadpoles are swimming with their legs, and all of the transgenic animals at the same stage have the characteristic paralyzed phenotype. Control and transgenic tadpoles were sorted by their GFP fluorescence and their hind limbs were removed for RNA extraction. If allowed to continue development all of the paralyzed transgenic tadpoles will soon die. The fourth RNA sample was isolated from the hind limbs of control frogs (NF 66) that were raised in 50 μg/ml of Dox for 2 weeks after completing metamorphosis. This concentration of the inducer has no effect on the development or growth of normal tadpoles or frogs. The tadpoles and frogs were fed throughout the experiment.

Three separate samples of hind limbs were collected from each of the 4 groups of animals. For the control and transgenic 2-week tadpoles, each sample consisted of hind limbs from 12 tadpoles. We used limbs from 6 tadpoles per replicate for the 6-week Dox treated climax transgenic tadpoles and limbs from 4 control frogs for each sample.

RNA purification, probe synthesis, and microarray hybridization

After amputation the limbs were homogenized immediately in Trizol (Invitrogen) to purify total RNA according to the manufacturer's protocol. Probes for the microarray were synthesized with the Agilent Low RNA Input Fluorescent Linear Amplification Kit (Agilent Kit #5184-3523) that incorporates the fluorescent nucleotide Cy3 CTP (Perkin Elmer). The universal standard labeled with Cy5 CTP was derived from all stages of whole tadpoles and has been described (Das et al., 2006). These cRNA probes were hybridized with a second version of Agilent X. laevis microarray slides (AMADID #013214). Each slide contains 21,654 sense oriented 60-mer oligonucleotides from the complete X. laevis Unigene list that was issued in February 2004 (Build 48). Duplicate entries and other details of these arrays have been described (Das et al., 2006). Since each of the four experimental points were performed in triplicate the statistical significance of expression values could be evaluated by the False Discovery Rate (FDR) method (Benjamini and Hochberg, 1995 and Sharov et al., 2005). We only scored gene expression changes that differed from the control with an FDR value of less than 0.05. The microarray data are presented as the ratio of the log intensity of triplicate hybridization values (2-week and 6-week transgenic limbs and control frog limbs) relative to the control tadpole limb. The tail expression data is derived from our previous array analysis (Das et al., 2006). The NF62 climax tail values are compared to control tail at NF54. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE5249. It can also be accessed at our web site (http://www.ciwemb.edu/brownlab/index.html).

Histology, immunostaining and in situ hybridization

The methods for hematoxylin and eosin staining, immunocytology and in situ hybridization have been described (Cai and Brown, 2004). The antibodies used in this study were directed against skeletal muscle myosin (fast) clone My32 (Sigma, product No. M4276), Alexa-568/488 secondary antibody from Molecular Probes and anti-GFP polyclonal antibody (Torrey Pines Biolab). Probes for in situ hybridization were digoxygenin-labeled antisense RNA. We purchased cDNAs from ATCC to make the probes for the in situ hybridizations.

For electron microscopy frontal and cross sections of limbs were fixed with glutaraldehyde and paraformaldehyde followed by osmium tetroxide and uranyl acetate, dehydrated in ethanol and embedded in epoxy. Sections were cut at about 90 nm thickness, stained with lead citrate and imaged on an FEI Techai 12 using a Gatan camera and software.

Results

Induction of the pCar/TRDN phenotype

The pCar/TRDN transgenic tadpoles form limbs that have almost no muscle (Das et al., 2002). In order to study this phenotype in greater detail we prepared transgenic X laevis with the pCar/TRDN–GFP transgene controlled by the positive tetracycline inducible system (Urlinger et al., 2000). In the presence of the inducer, doxycycline (Dox), the fused transgenic protein (TRDN–GFP) can be detected in limb muscle nuclei when specialized muscle proteins are first synthesized (NF54) (Fig. 1). The co-localization of GFP and myosin confirm that the pCar promoter drives expression specifically in differentiated muscle fibers. When the inducer was added at NF54 the muscle phenotype developed as the tadpoles advance to the climax of metamorphosis. The tadpoles die at NF62 after gill resorption but before tail resorption (Brown et al., 2005). When the TRDN transgene is induced with Dox as late as NF56 the pre-existing differentiated limb muscle fibers degenerate.

The pCar/TRDN-paralyzed limb muscle was analyzed by light and electron microscopy. Muscle fibers have disintegrated, and the muscle nuclei clump together within the bundles that previously contained the myofibrils (Fig. 2). These nuclei are convoluted with prominent nucleoli. Residual patches of parallel filaments (Fig. 2B) are often visible. Mitochondria are clumped as are the sarcoplasmic reticulum. In contrast to tail muscle cells (Das et al., 2002) that are induced by TH to resorb these degraded limb muscle cells were negative for both active caspase-3 and the TUNEL reaction for apoptosis. Even at the latest stages of the phenotype the limb muscle cells are not dead. Satellite muscle cells can be identified with an antibody to Pax 7. These cells continue to incorporate Brdu (Fig. S1). As will be shown many muscle genes including the muscle cell specific transcription factors are expressed even at the late stages of the phenotype. In support of the idea that the muscle nuclei do not die we removed the inducer Dox and reversed the phenotype of one transgenic NF58 tadpole that had developed a visible limb phenotype. The animal recovered and switched to leg swimming at metamorphic climax (data not shown). Since development of these transgenic tadpoles continues inexorably to climax the time for recovery at NF58 is too narrow for efficient recovery of most tadpoles with the limb phenotype.

We have induced metamorphosed transgenic small frogs with Dox for as long as 4 months. These animals grow more slowly than controls and develop abnormal but functional limbs. They never become paralyzed but left in the inducer the frogs stop eating and die. Histology of their limbs revealed normal but smaller muscle bundles than control frogs (data not shown). The inducer was removed from the rearing water of two of these frogs, and they were raised to sexual maturity. They retained a deformed leg structure but otherwise recovered. Therefore the up-regulation of the transgene after metamorphosis has a demonstrable but much less drastic affect on the frog limb muscle than it does on tadpoles.

Gene expression profile of the developing transgenic limb

In order to probe the molecular basis of this wasting phenotype, we carried out a microarray analysis on the hind limbs of normal and Dox induced transgenic tadpoles and control frogs (Fig. 3). The gene expression profiles of the control and 2-week samples from sibling tadpoles that had been raised in the same container with Dox are remarkably similar. Out of the 21,654 entries on the microarray only 42 and 67 genes are significantly up- and down-regulated, respectively, in the transgenic 2-week limb compared to the control limbs (Table 1). Many more genes are differentially expressed 4 weeks later (6-week time point), when the transgenic tadpoles had progressed to the climax of metamorphosis, and in the control frog limb. The final column of Table 1 includes the number of genes that are up- or down-regulated in the tail at climax (NF62) compared to a growing tail (NF54). These data were obtained in our previous microarray study (Das et al., 2006). Note that there are more than twice as many up-regulated genes as down-regulated genes in the tail at climax (NF62) even though the tail has been activated to die and resorb. Most of the up-regulated genes in the tail are expressed in fibroblasts while the majority of down- regulated genes are expressed in muscle.

The developmental changes in gene expression fall into classes (Fig. 4) that reflect the differentiation of the several cell types in the limb, the influence of TH, and the effect of the transgene. Gene expression values of transgenic limbs at the 2-week and 6-week time points as well as the control frog limbs are always compared with those of the GFP negative control tadpole limbs at NF56 (Fig. 3). Although we will concentrate on muscle-specific genes, the array experiments document the dramatic TH-induced transformation of the epidermis and the formation of cartilage, bone, and bone marrow that are not affected by expression of the transgene in muscle. At the 6-week time point, which is the climax of metamorphosis (NF61), these non-muscle cell type specific genes invariably change their expression in the same direction as they do in the frog limb. We call this pattern of expression for TH-controlled genes that are unaffected by the transgene ns-up-up (Fig. 4G) or ns-down-down (Fig. 4I) referring to the three successive developmental stages compared to the control tadpole i.e. unchanged (ns) in the 2-week transgenic, elevated (up) in both the 6-week transgenic and the frog limb. This class of genes (ns-up-up) includes the muscle specific transcription factors that are expressed in myoblasts before the pCar promoter is activated (Table S1). There are 8 tadpole-specific muscle genes that are down-regulated at metamorphosis and follow the kinetics of “ns-down-down” (Fig. 4I, Table S1). Tadpole-specific skin genes also exhibit this pattern (for example, gene 19 (U41861) (Furlow et al., 1997). The majority of genes (13,329) on the array do not change their expression during limb development at these three time points (ns-ns-ns) (Fig. 4F). There are 49 muscle genes on the array with an unchanged expression profile (Table S1). The in situ results of two of these genes, titin and desmin, are shown in Fig. S2.

Candidate genes

Several considerations narrow the search for candidate genes that might be involved in the muscle phenotype. Gene(s) responsible for the phenotype should differ in their expression levels between the control and the transgenic limbs after 2 weeks of Dox induction, because the limb muscle phenotype is detectable by that time. We have established a more precise time for the onset of the muscle destruction phenotype (see below). A candidate muscle gene that normally is down-regulated at metamorphosis should be up-regulated in the transgenic limb. We refer to this pattern as an “up-up-down” profile (Fig. 4E). Only 3 of the 42 genes that are up-regulated in the transgenic limbs after 2 weeks of Dox induction have the profile of continued up-regulation at 6 weeks and finally down-regulation in the frog limb. Their expression changes are less than two-fold. These genes (uroplakin 1A, transportin 1, and an unidentified EST) are not known to be involved in muscle development or maintenance.

Genes having the opposite expression profile (down-down-up and down-ns-up) (Figs. 4A and B) could initiate the phenotype. These genes are down-regulated in the transgenic limb after 2 weeks of Dox, and either down-regulated or unchanged in the 6-week transgenic but always up-regulated in the control frog limb. Only 67 of the 21,654 entries on the microarray are reduced in expression after 2 weeks of Dox treatment compared to the sibling control values. 32 of these 67 genes have the relevant gene expression profile (Table 2). The majority (24) of these genes are known to be expressed in muscle. The muscle localization and the change in expression caused by the transgene of two candidate genes (LIM and actin) is shown in Fig. 5. These two genes and most of the other candidate genes are up-regulated in the control limb by thyroid hormone and down-regulated in the control tail at the climax of metamorphosis. Candidate genes whose expression profiles have been confirmed by in situ hybridization are noted in Table 2. The remainder of the genes in Table 2 has not been studied further so they too might be expressed in muscle. 25 of these 32 genes are down regulated in the tadpole tail at climax (Table 2). The other 7 are unchanged. Many muscle genes continue being expressed in the limb even at the latest stage just before death of the transgenic tadpoles (Table S1; Fig. S2).

The kinetics of muscle degradation and muscle gene down-regulation

The candidate genes that are down-regulated in the microarray after 2 weeks exposure to the inducer could include genes that respond directly or indirectly to the TRDN transgene. We analyzed the kinetics of muscle fiber degradation at varying times after induction of the transgene by whole mount immunocytology with an antibody to myosin (Fig. 6). Remarkably, patches of muscle fiber disruption were seen as early as 2 days after addition of the inducer Dox. The destruction increased at 4 days and was extensive after 1 week of exposure to the transgene (Fig. 6). We screened 7 of the most dramatically regulated candidate genes by in situ hybridization for the earliest time that down-regulation could be detected presuming that down-regulation of the gene(s) causing the phenotype must precede visible degradation of muscle fibers (Table 3). Only calsequestrin was down-regulated at day 1 (Fig. 7). The other 5 genes were unchanged in expression at day 1 but down-regulated partly or extensively by 4 days (Table 3; Fig. S3). The slightly later down-regulation of most of these candidate genes compared to calsequestrin does not rule out their role in the degradation phenotype since the destruction begins in small patches (Fig. 6) and does not involve all of the muscle until 1 week of Dox treatment. In addition there are another 17 muscle genes that have not been tested at these early time points.