Proc. Natl. Acad. Sci.

USA
Vol. 93, pp. 1924–1929, March 1996
Developmental Biology

The thyroid hormone-induced tail resorption program during

Xenopus laevis metamorphosis
DONALD D. BROWN, ZHOU WANG*, J. DAVID FURLOW, AKIRA KANAMORI†, ROBERT A. SCHWARTZMAN,
BENJAMIN F. REMO, AND A LLISON PINDER
Department of Embryology, Carnegie Institution of Washington, 115 West University Parkway, Baltimore, MD 21210

Contributed by Donald D. Brown, December 7, 1995

ABSTRACT Genes that are up- and down-regulated by Xenopus tadpole tails (7) at about stage 51 (8). When TH is
thyroid hormone in the tail resorption program of Xenopus added to the swimming water of a competent tadpole the first
laevis have been isolated by a gene expression screen, se- visible changes are detected 2 days later. If protein synthesis is
quenced, and identified in the GenBank data base. The entire inhibited at that time, tail resorption will continue (7). Other
program is estimated to consist of fewer than 35 up-regulated TH-induced developmental programs such as limb formation
and fewer than 10 down-regulated genes; 17 and 4 of them, unfold over many days (9). TH must be present longer than 24
respectively, have been isolated and characterized. Up- hr for tail resorption to begin at 48 hr, suggesting that the direct
regulated genes whose function can be predicted on the basis response genes must stay activated until the morphological
of their sequence include four transcription factors (including changes begin, but the delayed genes are required to execute
one of the thyroid hormone receptors), an extracellular matrix the program. With these features in mind we carried out a gene
component (fibronectin) and membrane receptor (integrin), expression screen on tail tissue at 2, 24, and 48 hr after the
four proteinases, a deiodinase that degrades thyroid hormone, addition of TH to the water of competent tadpoles (7). Our
and a protein that binds the hypothalamic corticotropin- goal was to identify as many of the regulated genes in the tail
releasing factor, which has been implicated in controlling resorption program as possible. It was clear from the absence
thyroid hormone synthesis in Xenopus tadpoles. All four of up-regulated genes at 2 hr that genes with ‘‘immediate early’’
down-regulated genes encode extracellular proteins that are kinetics that have been described in other programs (10) do not
expressed in tadpole epidermis. This survey of the program play a role in tail resorption. Expression of the up-regulated
provides insights into the biology of metamorphosis. genes falls into two kinetic profiles. The direct response genes
have a lag of from 2 to 4 hr before their mRNA levels begin
Amphibian metamorphosis is controlled by thyroid hormone to rise, they reach a peak at 8–12 hr, and they remain elevated
(TH) (for review see ref. 1). The thyroid gland appears shortly for the full 48 hr. The mRNA transcribed from the delayed
after the tadpole begins to feed. As the tadpole grows in size genes starts rising 12–24 hr after TH induction and reaches a
it synthesizes increasing amounts of TH (2), which induces a maximum during the second day after induction. Several genes
succession of coordinated morphological and physiological have biphasic profiles. Their mRNAs begin to rise early, but
changes in tadpole tissues and organs. A component of most the majority of the response occurs with delayed kinetics.
of these changes is TH-induced tissue remodeling that includes Regulation of direct response genes is resistant to inhibition of
cell death (3). Tail resorption is the simplest of these programs protein synthesis by cycloheximide.
because there is no concomitant growth and differentiation, The subtractive hybridization procedure (6) that we devel-
merely death and resorption. Since TH exerts its effects by oped to isolate TH-induced genes is called a ‘‘gene expression
interacting with nuclear TH receptors (TRs) that have been screen’’ because of its analogy with a genetic screen, in which
shown to be transcription factors (4, 5), we have examined tail one isolates all mutants expressing a particular phenotype until
resorption with the assumption that it is the result of the the same genes are found repeatedly. The frequency of finding
activation or repression of sets of genes. the same gene provides a probability estimate of the total
Amphibian tail resorption and the other varied organ number of genes in the program. Double-stranded cDNAs
changes that are induced by TH in tadpoles are examples of prepared from the two mRNA populations that are to be
complex developmental changes that cannot be explored by compared are cleaved with two four-base-pair-recognition
traditional genetics. A three-phase strategy that uses molecular restriction enzymes, and these cDNA fragments are ligated to
biological methods in lieu of genetics can analyze such complex linkers for PCR amplification. After the subtractive procedure
developmental events. First, the up- and down-regulated genes the enriched cDNA fragments are cloned and then mapped to
are isolated by a ‘‘gene expression screen’’ which gives an a high molecular weight cDNA library to find multiple frag-
estimation of the program’s complexity (6). Second, the genes ments that are derived from the same mRNA. Table 1 sum-
are identified by sequencing and noting similarity to genes in marizes the mapping of 34 different up-regulated cDNA
the published data base. Third, functional assays must be fragments isolated from poly(A)1 RNA 48 hr after TH
devised to test the relevance of regulated genes to the program. induction (7). These cDNAs were mapped to 18 different
In this report, we summarize the first two phases of this project, genes. When these data are subjected to a probability analysis,
namely the identity‡ and characteristics of the TH-induced
genes, for insights into the tail resorption program in particular
and metamorphosis in general. Abbreviations: TH, thyroid hormone; TR, TH receptor; FAPa,
fibroblast activation protein a; CRF, corticotropin-releasing factor;
GPI, glycosyl-phosphatidylinositol.
RESULTS *Present address: Department of Urology, Northwestern University
Medical School, Tarry Building 11-715, 303 East Chicago Avenue,
Complexity of the Tail Resorption Program. The ability to Chicago, IL 60611-3009.
†Present address: Cell Research Engineering Section, National Insti-
respond morphologically to TH (competence) is acquired in
tute of Aquaculture, 224-1 Hiruta, Tamaki, Mie 519-04, Japan.
‡The new sequences reported in this paper have been deposited in the
The publication costs of this article were defrayed in part by page charge GenBank data base (accession nos. U35408 –U35410, U44025,
payment. This article must therefore be hereby marked ‘‘advertisement’’ in U37373–U37377, L49412, U41824, U41839, U41854 –U41861,
accordance with 18 U.S.C. §1734 solely to indicate this fact. U46575, U46576).

1924
 Developmental Biology: Brown et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1925

Table 1. Assignment of cDNA fragments to up-regulated genes in the open reading frame starting with the cDNA fragments that
the tail program had been isolated from the gene expression screen. Two
No. of independent cDNA fragments found per gene directionally oriented cDNA libraries were prepared from
mRNA that had been purified from the tails of stage 54
1 for each of 11 genes tadpole treated for 48 hr with 100 nM 3,39,59-triiodothyronine
2 for each of 3 genes (T3), a dose that induces tail resorption. One library was
3 for each of 2 genes primed with oligo(dT), the other was random primed. The
4 for 1 gene mRNAs that were 2.5 kb and smaller were isolated as full-
7 for 1 gene length cDNAs from the dT-primed library, while larger gene
The size range of the mRNAs is 1.6 to .10 kb; the abundance of the products were cloned by walks along the random-primed
mRNAs is 10 to 260 copies of mRNA per cell equivalent; and the genes library. Various tadpole and adult organs and total embryo
are up-regulated 6- to .20-fold (7). homogenates have been tested by RNA blotting (Fig. 1; see ref.
7 for details).
we estimate that there are about 35 total genes that are Genes 1 and 3 (GenBank accession nos. U35408–U35410)
up-regulated in the tail 48 hr after TH treatment. These encode a 292-amino acid transcription factor that was identi-
include the genes that are induced in the first 24 hr after TH fied originally in rat as a member of the Sp1 family of
induction. At the peak of their induction none of these genes transcription factors called BTEB (14). The DNA-binding
are expressed at very high levels, and some are as low as 10 region near the carboxyl terminus includes three Cys2His2
copies per cell (up-regulated level). This number might be an zinc fingers and is identical in amino acid sequence between
average of cells with widely different ranges of expression (11), Xenopus and rat BTEB. A complex set of large mRNA
since the entire tail was extracted for mRNA. One confirma- transcripts are expressed in all tadpole tissues studied to date
tion of the method’s validity is the identification of two as a direct response to TH.
independent cDNA fragments from TRb, a gene already Gene 2 is a direct response gene that is similar to gene 1 in
known to be up-regulated by TH (12, 13) to about 50 molecules its expression pattern, the abundance of its mRNA, and the
per cell. very large size of its multiple mRNAs. Attempts to ‘‘walk’’ to
The gene expression screen will not isolate all of the genes its open reading frame have been unsuccessful.
in the program. It will not identify genes that are critical to the Gene 4 (U44025) is the Xenopus homologue of integrin a-1
program but whose mRNA abundance is unaltered by TH. An (15), a member of an extended family of heterodimeric recep-
example might be the activation of lysosomal enzymes during tors for extracellular matrix proteins. Due to the large size of
tail resorption (3). The screen will miss regulated genes whose its mRNA it required three successive ‘‘walks’’ through a
cDNAs are not fragmented to the size range of 200–800 bp by random-primed cDNA library to reach the 59 end of the
the two restriction enzymes used to cleave the cDNA. In our mRNA. The gene was identified as integrin a-1 by sequencing
experience PCR does not preferentially amplify DNA by size the ends of the cDNAs obtained in the walk. The a-1 isoform
in the range of 200–800 bp. About 15% of published X. laevis is not expressed in embryos, although a number of a and b
mRNAs lack the recognition sites for the two restriction integrins have been cloned from embryos (16). Integrin a-1
enzymes (6). Any mRNA that has no poly(A) tail will be lost mRNA first appears in tadpole tissues and rises as metamor-
because oligo(dT) was used to prime cDNA synthesis. phosis proceeds (Fig. 1).
The pseudotetraploidy of the X. laevis genome complicates Gene 5 (U37373) encodes a protein of 996 amino acids with
measurements of the complexity of the program. In three no homologues in the data base. The protein is rich in proline
different instances, cDNA fragments that had been thought to (10.6%), basic amino acids (13.6%), and serine and threonine
originate from different genes were found to be derived from residues (17.1%). It has diverged tandem repeats of 5–10
highly diverged 39 untranslated regions of the two copies of the amino acids between residues 660 and 750. About half of the
same gene (genes 1 and 3, 8 and 9, and 11 and A), both of which amino acids are threonine and proline. This is followed by a
are up-regulated. For the complexity measurement cDNA very basic region of about 100 amino acid region at the
fragments from duplicated genes are treated as coming from carboxyl terminus. There is a transmembrane region near the
a single gene even though at least in one case (genes 8 and 9) amino terminus and three sites for N-glycosylation. Gene 5 is
there is an indication that the two genes might have evolved the only gene in the tail resorption program that is expressed
different functions. Cross-hybridizing cDNA fragments, in- exclusively in tails at metamorphosis.
cluding those from the coding regions of two genomic copies, Gene 6 (M35359–M35362) is TRb (17), which had been
would not have been recognized as different. Thus, 35 total shown previously to be a direct response gene in the TH-
genes in the tail resorption program may be an overestimation. induced program (12, 13). TRb mRNA (12) and TRb protein
A key assumption that must be made for the validity of the (18) rise as metamorphosis proceeds, paralleling the increase
complexity estimation is that each gene has the same proba- in endogenous TH. By the climax of metamorphosis TRb
bility of generating cDNA restriction fragments that can be protein exceeds TRa by severalfold in the tail.
amplified by PCR, a procedure that has sequence bias. In other Gene 7 (U37374) is Xenopus Fra-2 ( fos-related). The 320-
words, a cDNA fragment isolated by the gene expression amino acid bZip protein has 258 of its amino acids identical to
screen has the same probability of being derived from any of mammalian Fra-2 (19). The Xenopus gene is biphasic in its
the response genes. If this is true then the genes that we have response to TH, with most of the up-regulation of its mRNA
isolated are a random sample of all of the regulated genes, delayed. Two prominent sizes of mRNA are expressed. The
including those expressed as low as 10 copies of mRNA per higher molecular weight form is detectable at low levels in
cell. embryos (Fig. 1) and then drops in young tadpoles. Both sizes
Identity of the Up-Regulated Genes. In the discussion that of mRNA then rise in amount as the tadpole progresses
follows we refer to genes by the numbers used in previous through metamorphosis, reaching their highest levels at cli-
publications (6, 7). The original cDNA fragments isolated in max.
the screen were obtained by priming mRNA with oligo(dT) Genes 8 (U37375) and 9 (U41859) are two genomic copies
and were therefore more likely to be located within or near the of a bZip gene. The closest gene in the data base is human
39 end of their respective mRNA. Since many of the up- E4BP4 (20). Although the similarity is greatest in the DNA-
regulated mRNAs are very large, their cDNAs are not full- binding and leucine zipper regions (70% identity for gene 8
length in a dT-primed library. To identify these up-regulated and 75% identity for gene 9 over 53 amino acids), the
genes it was necessary to ‘‘walk’’ by multiple cloning steps to C-terminal regions of the gene 8 and 9 products also are
1926 Developmental Biology: Brown et al. Proc. Natl. Acad. Sci. USA 93 (1996)

FIG. 1. Developmental RNA blots for 16 of the 17 up-regulated genes and one of the 4 down-regulated genes (gene 18). The first three lanes
are blots of total embryo; the remaining lanes were from tail extracts (7). Stage numbers were assigned according to Nieuwkoop and Faber (8).
ORF, open reading frame.

homologous with the C-terminal region of human E4BP4 loproteinase family. Xenopus collagenase 3 has 277 amino
(43% identity over 30 amino acids for both). However, there acids identical with human collagenase 3 out of a total of 472
is a proline-rich region between the bZip and C-terminal amino acids (23). Genes 11 and A encode a different collage-
regions that is conserved completely between genes 8 and 9 yet nase from that reported recently from bullfrog tadpoles and
absent from E4BP4. There are 331 and 335 amino acids, implicated in epithelial changes during metamorphosis (24).
respectively, in the open reading frames of genes 8 and 9; 45 Although genes 11 and A are both up-regulated in tail as part
of them differ, raising the possibility that the two genes may of the delayed program, gene 11 mRNA rises to about 10 times
have evolved separate functions. One significant difference is the levels of gene A in the tail.
in their DNA-binding domains. Gene 9 has alanines at residues Gene 12 (U41854 and U41855) is a direct response gene with
101 and 102 in the basic region which are highly conserved in
nucleic acid homology to a murine (and human) gene that is
most bZip transcription factors (21). The equivalent residues
in gene 8 at residues 97 and 98 are valine and glutamic acid. expressed abundantly in fetal brain (25). The authors who
The crystal structure of the yeast bZip transcription factor originally cloned the mouse gene pointed out a small open
GCN4 shows that these two alanines make hydrogen bonds reading frame encoding 67 amino acids near the 59 end which
with bases in the DNA-binding site (22), suggesting that genes they found to be conserved between humans and mice. They
8 and 9 have evolved to recognize different DNA elements. also noted extensive nucleotide sequence homology between
The kinetics of their up-regulation and tissue distribution of the two mammalian cDNAs in regions with no open reading
their mRNAs cannot be distinguished. frame. We have sequenced the cDNAs from both of the
Genes 11 (L49412) and A (U41824) are the two genomic Xenopus genomic copies of gene 12. The two transcripts of
copies of collagenase 3, a member of the matrix metal- 2200 and 2410 bp are greater than 99% identical for the first
 Developmental Biology: Brown et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1927

1700 nucleotides, whereafter they diverge completely for the (36). Gene D is, to our knowledge, the first eukaryotic member
remainder of their length. Neither transcript has an open of this novel family of proteinases. The gene D protein has no
reading frame longer than 80 amino acids. The homologous signal sequence, transmembrane domain, or N-glycosylation
stretch between the Xenopus and mouse sequences begins sites, which suggests that it is located in the cytoplasm like its
about 600 bp from the 59 ends of the cDNAs, which is bacterial homologue.
downstream from the putative open reading frame in the Gene E (U41858) is corticotropin-releasing factor (CRF)-
mammalian sequence and extends for about 700 bp. There are binding protein, which was identified originally in mammals
additional short homologous sequences periodically all the way (37). The rat (322 amino acids) and Xenopus (321 amino acids)
to the end of the two cDNAs. The extensive homology that binding proteins are 69% identical along their lengths, whereas
exists between the Xenopus and mammalian genes does not the CRF hormones have 38 of 41 identical amino acids (38).
include an open reading frame. The identification of the up-regulated genes is summarized
Gene 13 (U41856) encodes the Xenopus homologue of in Table 2.
fibroblast activation protein a (FAPa) (26), an integral mem- Identification of the Down-Regulated Genes. An estimation
brane member of a family of heterodimeric serine dipeptidyl of the complexity of genes in the tail that are down-regulated
peptidases (27) that was cloned from stromal fibroblasts that genes in the 48 hr after TH treatment gives a maximum of 10
adjoin metastatic epithelial cancer. Xenopus FAPa at 756 genes (7). Four have been isolated and sequenced.
amino acids is slightly shorter at its C terminus than its human Genes 17 (U41860) and 18 (U41839) are related to a set of
homologue (759 amino acids) and has 50% amino acid identity mammalian proteins that are bound to the exterior of cells
along its entire length. Gene 13 is a delayed gene in the tail through a glycosyl-phosphatidylinositol (GPI) anchor. Two
program whose expression is absent from Xenopus embryos. known proteins in this family are uromodulin (39), an abun-
Gene 14 (Z27093) encodes stromelysin 3, another member dant protein in urine that is expressed only in kidney, and GP-2
of the matrix metalloproteinase family (28). In metamorphosis (40), found exclusively in the pancreas. The function of these
it is expressed widely (7, 29), but it is especially up-regulated mammalian proteins is unknown. The overall identity both
in the tadpole intestine as it remodels in response to TH (30). between genes 17 (357 amino acids) and 18 (429 amino acids)
The only adult (frog) Xenopus tissue found to date that with each other and their mammalian relatives is less than
expresses this gene (constitutively) is the intestine, which 30%. The most striking similarities between the Xenopus and
continues to remodel throughout the life of the organism. mammalian proteins are that they share 13 (gene 17) and 17
Gene 14 is the only direct response gene that encodes a (gene 18) precisely aligned cysteine residues. They both have
proteinase. an N-terminal signal sequence, a C-terminal GPI anchor
Gene 15 (L28111) encodes a type III iodothyronine deio- sequence, and the potential for 7 or 8 N-linked glycosylation
dinase (31), an enzyme that cleaves iodine from the inner ring sites. Despite their similarity, genes 17 and 18 are distinctly
of TH and inactivates the hormone. There is no expression of different genes.
gene 15 in the earliest reacting tissue, the limb, until late in its Gene 19 (U41861) is a member of the mucin family of
morphogenesis. In contrast, the latest reacting tissue, the tail, O-linked glycosylated proteins (41). It contains amino acid
has a substantial baseline expression of gene 15 throughout repeats that are high in proline (10%) and threonineyserine
tadpole development. As endogenous TH increases the level (45%). Only about 2.2 kb of the 8-kb mRNA has been cloned
of gene 15 mRNA in the tail rises to peak before climax, earlier and sequenced. The sequence at the 59 end of this clone begins
than any other mRNA in the tail program. At climax when the in the open reading frame in the middle of a proline-, serine-,
other TH-response genes in the tail are most active, the gene and threonine-rich tandem mucin repeat; the succeeding three
15 mRNA level has dropped (see Fig. 1). The levels of type III mucin repeats are each 77 amino acids in length. They are
deiodinase enzymatic activity follow its mRNA (31). The only followed by a 153-amino acid C terminus which does not
previously cloned deiodinase gene (32), from rat, encodes a contain a transmembrane domain or GPI anchor sequence.
type 1 enzyme which synthesizes triiodothyronine from thy-
roxine and therefore activates the hormone. Amongst the Table 2. Summary of up-regulated genes
similarities between the type I and III proteins is the presence Gene Product
of a selenocysteine codon (TGA) in the middle of the gene’s
open reading frame that usually encodes a termination signal. Transcription factors
There is also a conserved structure in the 39 untranslated 6 TRb*
regions of the mRNAs that directs the insertion of selenocys- 8, 9 bZip (E4BP4)†
teine into the protein. 7 bZip (Fra-2)†
Gene 16 (U41857) encodes an as-yet-unidentified 397- 1, 3 Zinc finger (BTEB)*
amino acid protein that contains repeats of the WD-40 se- Proteinases
quence motif that has been found in a variety of proteins, 11, A Collagenase 3‡
including the b subunits of G proteins (33). Gene 16 contains 14 Stromelysin 3*‡
three recognizable WD-40 repeats. The protein structure 13 FAPa‡
suggests that it is localized intracellularly. D N-Aspartyl dipeptidase
Gene B (U37376) encodes a protein of 688 amino acids with Extracellular matrix and receptors
a hydrophobic signal sequence. It has a region identified in C Fibronectin‡
proteins that are anchored to cell membranes called an MAM 4 Integrin a-1‡
domain (34). This domain is thought to adhere to other Other
proteins by way of conserved cysteines. The function of gene 15 Type III iodothyronine 5-deiodinase*
B cannot be determined conclusively from its sequence. E CRF-binding protein‡
Gene C (M77820) is Xenopus fibronectin, originally cloned 16 WD-40 domains*
from embryos and studied in early embryogenesis (35). Of all 12 No open reading frame*
the genes in the tail resorption program this one has the highest B MAM domains‡
expression in embryos. It is up-regulated by TH greater than 5 Tail-specific‡
15-fold over a baseline of expression in tadpoles. 2 Unknown*
Gene D (U37377) encodes a 243-amino acid protein that has *Direct response genes.
51 residues identical to a 229-amino acid cytoplasmic bacterial †Biphasic response.
peptidase (pepE) which cleaves N-terminal aspartyl peptides ‡Membrane bound or extracellular protein.
1928 Developmental Biology: Brown et al. Proc. Natl. Acad. Sci. USA 93 (1996)

Gene 20 (U46575 and U46576) is a novel gene whose spontaneous changes in the limb TRb is not up-regulated (7).
predicted protein of 997 amino acids contains a signal peptide In contrast, TRb becomes the predominant receptor form in
and six N-linked glycosylation sites. Like its down-regulated the tail at the climax of metamorphosis (18). This suggests that
counterparts, it is predicted to be a secreted or membrane- limb changes are controlled by TRa while tail changes are
bound glycoprotein. controlled by TRb. It also provides a possible explanation for
These four genes have identical kinetics of down-regulation the biphasic kinetics of the bZip genes, namely that the low
(7). Most of their mRNA has disappeared 12 hr after TH level first (direct) response is induced by TRa, while the
treatment. These genes are expressed only in tadpoles and delayed response is induced by TRb.
their tissue and organ distribution are the same. Recent in situ Roles for Genes in Metamorphosis. The chronology of
hybridization experiments have localized all four mRNAs to metamorphosis refers to the sequential response by tissues to
the tadpole epidermis (J. D. Furlow, D. L. Berry, and D.D.B., the rising concentration of endogenous TH during prometa-
unpublished data). morphosis. The expression of the type III deiodinase gene (31)
correlates with a role in the chronology of metamorphosis by
regulating the intracellular level of TH. The earliest respond-
DISCUSSION ing organ is the limb bud, which requires a very low level of TH
Gene Expression Changes in the Tail Resorption Program. to grow and develop; the latest is tail resorption, which requires
There are two kinetic waves of TH-induced up-regulated gene a very high endogenous level of TH. Limbs have no detectable
expression and one of down-regulated that precede the first deiodinase expression until late in their development. In
visible signs of tail resorption at 48 hr (7). Other morphological contrast, the tail has a basal level of deiodinase mRNA
changes in metamorphosis also can be seen about 48 hr after throughout prometamorphosis that rises to a peak before
TH induction. Whereas the tail resorption program is ex- climax (Fig. 1; ref. 7). It then falls before any other mRNA in
pressed by 48 hr, the programs that involve growth and the program has reached its peak.
remodeling such as the limb require additional periods of gene Gene E, the CRF-binding protein, is a candidate to account
expression that continue for many days (9). for the development of the thyroid–pituitary negative feed-
The expectation for any complex program is that transcrip- back loop that differentiates late in metamorphosis. Denver
tion factors are among the direct response genes; they in turn (43) has shown that mammalian CRF but not thyroid-releasing
activate delayed genes that are responsible for executing the hormone (TRH) induces the tadpole pituitary to secrete
program. In the tail resorption program TRb and Xenopus thyroid-stimulating hormone (TSH). According to this hy-
BTEB are unequivocally direct response genes, while the bZip pothesis, CRF-binding protein is induced late in metamorpho-
genes (8, 9, and Fra-2) are biphasic in their response. Included sis and inactivates CRF, thereby shutting off the pituitary’s
amongst the delayed genes that are predicted to play a role in production of TSH. In the absence of TSH the thyroid gland
tail resorption are three of four of the proteinases. The stops synthesizing TH.
exception to this rule is stromelysin 3, which is a direct response The tail resorption program includes the extracellular matrix
gene, suggesting that this enzyme might activate other pro- protein fibronectin, whose membrane receptor is reported to
teinases. be an integrin. The member of the integrin family that is
Up-regulation dominates the changes in gene expression in up-regulated as part of the tail resorption program is integrin
a diverse array of TH-induced programs of metamorphosis (7, a-1, which has been implicated as a receptor for laminin and
9, 29, 42). All four down-regulated genes in the tail program collagen (44). A possible role for these proteins in tail resorp-
encode extracellular proteins. Since no new genes are down- tion might be their participation in the recognition by macro-
regulated in the second 24 hr of the program, cell death cannot phages of cells or matrix fragments that they will engulf (45).
be attributed to gene repression. This conclusion is reinforced Macrophages are reported to play a major role in cleaning up
by the observation that no gene for a transcription factor has cellular debris during tail resorption (46).
been found to be down-regulated in any of the programs Role of Proteinases: Remodeling, Metastasis, and Apopto-
investigated to date. sis. For many years it has been proposed that proteolytic
Tissue Distribution of the Resorption Program. The ex- enzymes play a part in tail resorption and organ remodeling (3,
pression of the delayed genes is generally more restricted in 47). Collagenase enzymatic activity was demonstrated 30 years
tissue distribution than that of the direct response genes. ago to be greatly increased during tail resorption (47). The
However, only gene 5 is entirely tail-specific (7). Some genes activation of lysosomal enzymes has been implicated as well
up-regulated in tail by TH are also up-regulated in organs that (3). Four proteolytic enzymes have been identified as part of
are remodeled during metamorphosis, such as the intestine the tail program. Collagenase 3 (23) and stromelysin 3 (28) are
(29). The majority of the up-regulated genes are not expressed members of a family of matrix metalloproteinases, while the
in embryos (7). Most of the direct and delayed response genes third enzyme, FAPa, is an integral membrane protein that is
are expressed constitutively at low levels in frog tissues (data a member of a family of serine proteinases (27). The expression
not shown). This basal level may itself be the result of of stromelysin 3 has also been correlated with remodeling and
endogenous TH. The expression of all four down-regulated apoptosis (48). All three of these proteinases have been
genes is tadpole-specific. implicated in the stromal response to cancer metastases. The
The growth (limb) and resorption (tail) programs are dra- invasion of the collagen basement lamella by mesenchymal
matically different. None of the genes found to be up-regulated cells (49) during tail resorption is reminiscent of metastasis.
by TH in the limb (9) can be induced by TH in the tail, nor are The fourth proteinase (encoded by gene D) appears to be
they up-regulated during spontaneous metamorphosis of the the first eukaryotic example of a class of proteinases related to
tail. The identity of the limb genes that are up-regulated in the a bacterial cytoplasmic peptidase (pepE). If pepE is cytoplas-
first 24 hr after TH induction (9) suggests that they are related mic as its structure suggests, then it resembles the interleukin-
to growth, which is the limb’s first morphological response. converting enzyme (ICE) that has been shown to be involved
Considering the known complexity of genes that are up- in apoptosis (50) and also cleaves at aspartate residues (51).
regulated as a response to growth (10), it is not surprising that There are no other obvious similarities between the two
TH induces the expression of a large number of genes in the proteins.
limb in the first 24 hr. Although many of the up-regulated tail The prominence of proteinases whose genes are up-regu-
genes can be induced in the limb bud (pre-metamorphosis) by lated in the resorption program raises the question of whether
exogenous TH, these genes are not up-regulated during spon- the digestion of the extracellular matrix by these enzymes is
taneous metamorphosis in the limb (7). During the early responsible for the death of tail cells as well as their resorption.
 Developmental Biology: Brown et al. Proc. Natl. Acad. Sci. USA 93 (1996) 1929

Isolated tails in organ culture (3) and tail cells in primary 18. Eliceiri, B. & Brown, D. D. (1994) J. Biol. Chem. 269, 24459–
culture (52) have been shown to respond to TH by resorption 24465.
and growth arrest, respectively, but there is also evidence that 19. Nishina, Y., Sato, H., Suzuki, T., Sato, M. & Iba, H. (1990) Proc.
epidermis is required for tail resorption (53). Therefore the Natl. Acad. Sci. USA 87, 3619–3623.
20. Cowell, I. G., Skinner, A. & Hurst, H. C. (1992) Mol. Cell. Biol.
extent that tail resorption is cell autonomous remains a 12, 3070–3077.
question. The large number of delayed genes that encode 21. Hurst, H. C. (1994) Protein Profile 1, 125–133.
secreted or membrane-bound products emphasizes the impor- 22. Ellenberger, T. E., Brandl, C. J., Struhl, K. & Harrison, S. C.
tance of events outside the cell or at its surface in the tail (1992) Cell 71, 1223–1237.
resorption program. 23. Freije, J. M. P., Dı́ez-Itza, Balbı́n, M., Sánchez, L. M., Blasco, R.,
Viewing the Program. This three-step strategy for analyzing Tolivia, J. & López-Otı́n-López, C. (1994) J. Biol. Chem. 269,
the tail resorption program during amphibian metamorphosis 16766–16773.
24. Oofusa, K., Yomori, S. & Yoshizato, K. (1994) Int. J. Dev. Biol.
is a systematic molecular biological approach that can be
38, 345–350.
applied to many complex developmental programs. Once the 25. Studler, J.-M., Glowinski, J. & Levi-Strauss, M. (1993) Eur. J.
genes are identified, it is logical to view the program for clues Neurosci. 5, 614–623.
before proceeding to functional assays. A total of 17 of fewer 26. Scanlan, M. J., Raj, B. K. M., Calvo, B., Garin-Chesa, P., Sanz-
than 35 predicted up-regulated genes have been isolated. Moncasi, M. P., Healey, J. H., Old, L. J. & Rettig, W. J. (1994)
Eleven of them are the Xenopus homologues of previously Proc. Natl. Acad. Sci. USA 91, 5657–5661.
discovered genes. Several others encode a recognizable do- 27. Ogata, S., Misumi, Y. & Ikehara, Y. (1989) J. Biol. Chem. 264,
main that suggests a function or a location for the protein 3596–3601.
28. Basset, P., Wolf, C. & Chambon, P. (1993) Breast Cancer Res.
product within or outside of the cell. The final descriptive Treat. 24, 185–193.
information needed before embarking on functional assays is 29. Shi, Y.-B. & Brown, D. D. (1993) J. Biol. Chem. 268, 20312–
the detailed tissue and cell expression pattern of each gene 20317.
product; these patterns can be learned only by in situ hybrid- 30. Patterton, D., Hayes, W. P. & Shi, Y.-B. (1995) Dev. Biol. 167,
ization. 252–262.
31. St. Germain, D. L., Schwartzman, R. A., Croteau, W., Kanamori,
The authors are grateful to their colleagues for helpful comments. A., Wang, Z., Brown, D. D. & Galton, V. A. (1994) Proc. Natl.
Eddie Jordan and Kenneth Graf provided expert technical assistance Acad. Sci. USA 91, 7767–7771.
during the early stages of this project. J.D.F. is supported by a National 32. Berry, M. J., Banu, L. & Larsen, P. R. (1991) Nature (London)
Research Service Award; R.A.S. is an American Cancer Society fellow. 349, 438–440.
This research has been supported in part by grants to D.D.B. from the 33. Neer, E. J., Schmidt, C. J., Nambudripad, R. & Smith, T. F.
National Institutes of Health and from the G. Harold and Leila Y. (1994) Nature (London) 371, 297–300.
Mathers Foundation. 34. Beckmann, G. & Bork, P. (1993) Trends Biochem. Sci. 18, 40–41.
35. Desimone, D. W., Norton, P. A. & Hynes, R. O. (1992) Dev. Biol.
194, 357–369.
1. Dodd, M. H. I. & Dodd, J. M. (1976) in Physiology of the Am-
36. Conlin, C. A., Hakensson, K., Liljas, A. & Miller, C. G. (1994) J.
phibia, ed. Lofts, B. (Academic, New York), Vol. 3, pp. 467–599.
Bacteriol. 176, 166–172.
2. Leloup, J. & Buscaglia, M. (1977) C.R. Acad. Sci. 284D, 2261–
37. Potter, E., Behan, D. P., Fischer, W. H., Linton, E. A., Lowry,
2263.
P. J. & Vale, W. W. (1991) Nature (London) 349, 423–426.
3. Weber, R. (1969) in Lysosomes, eds. Dingle, J. T. & Fell, H. B.
38. Stenzyl-Poore, M. P., Heldwein, K. A., Stenzel, P., Lee, S. &
(North–Holland, Amsterdam), pp. 437–461. Vale, W. W. (1992) Mol. Endocrinol. 6, 1716–1724.
4. Weinberger, C., Thompson, C. C., Ong, E. S., Lebo, R., Gruol, 39. Pennica, D., Kohr, W. J., Kuang, W.-J., Glaister, D., Aggarwal,
D. J. & Evans, R. M. (1986) Nature (London) 324, 641–646. B. B., Chen, E. Y. & Goeddel, D. V. (1987) Science 236, 83–88.
5. Sap, J., Munoz, A., Damm, K., Goldberg, Y., Ghysdael, J., Leutz, 40. Hoops, T. C. & Rindler, M. J. (1991) J. Biol. Chem. 266, 4257–
A., Beug, H. & Vennstrom, B. (1986) Nature (London) 324, 4263.
635–640. 41. Devine, P. L. & McKenzie, I. F. C. (1992) BioEssays 14, 619–625.
6. Wang, Z. & Brown, D. D. (1992) Proc. Natl. Acad. Sci. USA 88, 42. Kanamori, A. & Brown, D. D. (1993) Proc. Natl. Acad. Sci. USA
11505–11509. 90, 6013–6017.
7. Wang, Z. & Brown, D. D. (1993) J. Biol. Chem. 268, 20312–20317. 43. Denver, R. J. (1993) Endocrinology 91, 38–51.
8. Nieuwkoop, P. D. & Faber, J. (1956) Normal Table of Xenopus 44. Hall, D. E., Reichardt, L. F., Crowley, E., Holley, B., Moezzi, H.,
laevis (North–Holland, Amsterdam). Sonnenberg, A. & Damsky, C. H. (1990) J. Cell Biol. 110,
9. Buckbinder, L. & Brown, D. D. (1992) J. Biol. Chem. 267, 2175–2184.
25786–25791. 45. Fadok, V. A., Savill, J. S., Haslett, C., Bratton, D. L., Doherty,
10. Lau, L. F. & Nathans, D. (1991) in Hormonal Control: Regulation D. E., Campbell, P. A. & Henson, P. M. (1992) J. Immunol. 149,
of Gene Transcription, eds. Cohen, P. & Faulks, J. G. (Elsevier, 4029–4035.
London), pp. 257–293. 46. Kinoshita, T., Sasaki, F. & Watanabe, K. (1985) J. Morphol. 185,
11. Kawahara, A., Baker, B. S. & Tata, J. R. (1991) Development 269–295.
(Cambridge, U.K.) 112, 933–943. 47. Gross, J. & Nagai, Y. (1965) Proc. Natl. Acad. Sci. USA 54,
12. Yaoita, Y. & Brown, D. D. (1990) Genes Dev. 4, 1917–1924. 1197–1204.
13. Kanamori, A. & Brown, D. D. (1992) J. Biol. Chem. 267, 739–745. 48. Lefebvre, O. Wolf, C., Limacher, J.-M., Hutin, P., Wendling, C.,
14. Imataka, H., Sogawa, K., Yasumoto, K., Kikuchi, Y., Sasano, K., Le Meur, M., Basset, P. & Rio, M.-C. (1992) J. Cell Biol. 119,
Kobayashi, A., Hayama, M. & Fujii-Kuriyama, Y. (1992) EMBO 997–1002.
J. 11, 3663–3671. 49. Gross, J. (1964) Medicine 43, 291–303.
15. Ignatius, M. J., Large, T. H., Houde, M., Tawil, J. W., Barton, A., 50. Miura, M., Zhu, H., Rotello, R., Hartwieg, E. A. & Yuan, J.
Esch, F., Carbonetto, S. & Reichardt, L. F. (1990) J. Cell Biol. (1993) Cell 75, 653–660.
111, 709–720. 51. Kumar, S. (1995) Trends Biochem. Sci. 20, 198–202.
16. Whittaker, C. A. & DeSimone, D. W. (1993) Development (Cam- 52. Izutsu, Y., Kaiho, M. & Yoshizato, K. (1993) J. Exp. Zool. 267,
bridge, U.K.) 117, 1239–1249. 605–615.
17. Yaoita, Y., Shi, Y.-B. & Brown, D. D. (1990) Proc. Natl. Acad. 53. Niki, K., Namiki, H., Kikuyama, S. & Yoshizato, K. (1982) Dev.
Sci. USA 87, 7090–7094. Biol. 94, 116–120.