DEVELOPMENTAL DYNAMICS 240:1826–1840, 2011

PATTERNS & PHENOTYPES

Gene Expression Profile of the Regeneration

Epithelium During Axolotl Limb Regeneration
Leah J. Campbell,1 Edna C. Suárez-Castillo,1 Humberto Ortiz-Zuazaga,2,3 Dunja Knapp,4
Elly M. Tanaka,4 and Craig M. Crews1,5,6*

Urodele amphibians are unique among adult vertebrates in their ability to regenerate missing limbs. The
process of limb regeneration requires several key tissues including a regeneration-competent wound epi-
dermis called the regeneration epithelium (RE). We used microarray analysis to profile gene expression of
the RE in the axolotl, a Mexican salamander. A list of 125 genes and expressed sequence tags (ESTs)
showed a
1.5-fold expression in the RE than in a wound epidermis covering a lateral cuff wound. A sub-
set of the RE ESTs and genes were further characterized for expression level changes over the time-
course of regeneration. This study provides the first large scale identification of specific gene expression
in the RE. Developmental Dynamics 240:1826–1840, 2011. V 2011 Wiley-Liss, Inc.
C
Developmental Dynamics

Key words: limb regeneration; urodele amphibian; gene expression microarray; regeneration epithelium

Accepted 3 May 2011

INTRODUCTION and replenish their respective miss- regeneration in denervated newt

ing structures (Kragl et al., 2009) limbs (Kumar et al., 2007). nAG is
Limb regeneration is a unique ability through a recapitulation of signaling expressed in Schwann cells of the
of urodele amphibians, which are the molecules and pathways used in de- nerve sheath as the severed axon
only vertebrates able to replace such velopment (Gardiner et al., 1999). regrows and it has been shown to pro-
a complex structure throughout their Nevertheless, it is unclear how blaste- mote proliferation of blastemal cells
adult life. The process of regenerating mal cells are recruited to the amputa- in culture. Later in regeneration,
an amputated limb proceeds from the tion plane and how urodele amphib- nAG appears in the gland cells of the
early phase of wound healing to digit ians, such as newts and axolotls, are regeneration epithelium (RE). The
development through formation of a able to proceed past wound healing to RE is an epithelial structure covering
mass of cells called the blastema. replace missing limbs. the distal part of the regenerate and
Until recently it was believed that the The nerve has been extensively is also required for successful regen-
blastema was comprised of dedifferen- studied as the source of regeneration eration (Stocum, 2004). The RE is
tiated mesenchymal cells with pluri- signaling molecules since it was first unique to the amputation wound and
potent properties (Lo et al., 1993). On described as required for regeneration forms as a result of epidermal migra-
the contrary, the blastema appears to (Singer, 1952). Recently, the newt AG tion over the wound from around the
be a heterogeneous mass of cells that protein (nAG) was identified as a circumference of the amputation
retain memory of their tissue origin secreted nerve factor that rescues plane (Repesh and Oberpriller, 1978).

Additional Supporting Information may be found in the online version of this article.
1
Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, Connecticut
2
High Performance Computing facility Rı́o Piedras Campus, University of Puerto Rico, San Juan, Puerto Rico
3
Department of Computer Science, Rı́o Piedras Campus, University of Puerto Rico, San Juan, Puerto Rico
4
Center for Regenerative Therapies, Dresden, Germany
5
Department of Chemistry, Yale University, New Haven, Connecticut
6
Department of Pharmacology, Yale University, New Haven, Connecticut
Grant sponsor: NIH; Grant number: GM094944; Grant sponsor: ARO; Grant number: Army US #W911NF-07-1-0252.
*Correspondence to: Craig M. Crews, Departments of Molecular, Cellular & Developmental Biology, Chemistry, and
Pharmacology, Yale University, New Haven, CT 06511. E-mail: craig.crews@yale.edu
DOI 10.1002/dvdy.22669
Published online 3 June 2011 in Wiley Online Library (wileyonlinelibrary.com).

V
C 2011 Wiley-Liss, Inc.
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1827

In addition to nAG, genes such as Sp9

(Satoh et al., 2008) and Dlx-3 (Mullen
et al., 1996) display nerve-dependent
expression patterns in the RE. Histor-
ically, the RE has been referred to as
the wound epidermis (WE) and the
apical epithelial cap (AEC); however,
it has recently been suggested that
the structure be termed regeneration
epithelium (RE) (Satoh et al., 2008) to
distinguish it as a specialized struc-
ture that communicates with the
nerve to promote regeneration. It has
been demonstrated that the RE is
required for successful regeneration
because removal of the structure
delays the process (Thornton, 1957)
and preventing formation of the
structure inhibits regeneration
(Mescher, 1976). Markers for the RE
are limited (Campbell and Crews,
2008), but include transcription fac-
tors such as Msx-2 (Carlson et al.,
Developmental Dynamics

1998), Dlx-3 (Mullen et al., 1996), and

Sp9 (Satoh et al., 2008), FGF signal-
ing molecules (Han et al., 2001; Chris-
tensen et al., 2002), and matrix metal-
loproteinases (Yang et al., 1999; Kato
et al., 2003). Of these, Msx-2 is
expressed the earliest, within hours
after amputation, but is not RE-spe-
cific because it is also expressed during Fig. 1. Schematic depicting tissue dissection and RNA isolation for microarray hybridization.
Axolotl limbs were amputated or wounded with a lateral cuff wound. After 7 days of healing, the
healing of a lateral wound (Carlson regeneration epithelium (RE) and the radial lateral cuff wound epidermis (LE) were dissected and
et al., 1998). Sp9, another early RNA was isolated. mRNA was labeled with Cy3 and Cy5 dyes and hybridized to custom micro-
marker, is RE-specific and expressed array slides designed from expressed sequence tag (EST) assemblies for A. mexicanum, A. tigri-
within 24 hr after amputation (Satoh num, and other salamander species.
et al., 2008). As more investigations
focus on exploring the molecular path-
ways involved in regeneration such as tion on a broader level (Monaghan and with respect to previously pub-
with the Accessory Limb Model (Endo et al., 2007, 2009). These studies have lished regeneration studies.
et al., 2004) or with in vitro work (Fer- introduced many new candidate
ris et al., 2010), a larger set of RE-spe- genes that will enable future regener-
cific markers will be needed. ation studies. RESULTS AND DISCUSSION
The discovery and characterization In the present study we used the Identification of RE-Specific
of molecules and signaling pathways publicly available collections of
involved in limb regeneration has
Gene Expression
Ambystoma genes and ESTs to com-
been improved by the development of pare, by microarray analysis, the It is well documented that the RE is
genomic tools for salamanders. Signif- expression profiles of the RE and the necessary for successful regeneration
icant effort has been put toward wound epidermis covering a lateral in urodele amphibians (Thornton,
sequencing and organizing expressed cuff wound (LE). From these results, 1957; Mescher, 1976), however, there
sequence tags (ESTs) from Ambys- we focused on a list of Ambystoma are few markers for RE-specific gene
toma mexicanum and Ambystoma ESTs and genes that are significantly expression. To identify RE-specific
tigrinum (Habermann et al., 2004; more highly expressed in the RE. A genes, we collected both the LE and
Putta et al., 2004). The Sal-Site at subset of this list was further charac- the RE after 7 days of healing,
http://www.ambystoma.org (Smith terized using quantitative polymerase which represented an early stage of
et al., 2005) provides an Ambystoma chain reaction (qPCR) to show the regeneration between wound healing
gene collection and EST database temporal expression pattern during and early bud formation (Fig. 1).
that has allowed for approaches such the time-course of regeneration. We Messenger RNAs (mRNAs) from the
as microarray analysis and high- discuss the characterized ESTs and two populations were labeled and
throughput 454 cDNA sequencing to genes in terms of best hit candidates hybridized to oligonucleotide arrays
investigate aspects of limb regenera- discovered through BLAST searches containing 42,553 elements that
 1828 CAMPBELL ET AL.

TABLE 1. List of Interesting Hits Up-regulated in Regeneration Epithelium as Compared to Radial Lateral Wound
Epidermis

Fold change Sal ID/ Best hit

(RE/LE) accession no. Gene name BLASTX (E) accession no.
7.79 M003080 Similar to methyltransferase 24 (Gallus gallus) 3E-66 XM_421871
6.47 D82576 Msx2 (Ambystoma mexicanum)
5.73 M011831 N/A N/A No hit
5.30 M064466 N/A N/A No hit
5.18 M002949 Sodefrin precursor-like factor 3E-26 DQ097063
(Desmognathus monticola)
5.11 M006889 hyaluronan and proteoglycan link protein 3 8E-42 NM_001086162
(Xenopus laevis)
5.02 M065735 N/A N/A No hit
4.53 M061758 Dynein, cytoplasmic 1, intermediate chain 1 2E-39 NM_004411
(Homo sapiens)
4.08 Z14047 Wnt-5a (Ambystoma mexicanum)
3.53 M065526 Similar to desmoglein 4 preproprotein (Gallus gallus) 2E-05 XM_426082
3.38 M003433 Apolipoprotein C-I (Hemibarbus mylodon) 6E-06 FJ170109
3.26 M002254 Uromodulin-like (Xenopus tropicalis) 1E-16 XM_002934636
2.86 M008800 laminin, beta 1 (Homo sapiens) 9E-67 NM_002291
2.66 M003964 Prostate stem cell antigen (Homo sapiens) 8E-16 NM_005672
2.25 M004510 Mal, T-cell differentiation protein-like (Homo sapiens) 2E-27 NM_005434
2.13 M062282 EGF-like-domain, multiple 6 (Homo sapiens) 7E-26 NM_015507
Developmental Dynamics

2.06 AY326272 BMP2/4-like (Ambystoma mexicanum)

1.88 U59480 Distal-less (Dlx-3) (Ambystoma mexicanum)
1.68 M061881 LY6/PLAUR domain containing 2 (Homo sapiens) 6E-17 NM_205545
1.58 M062365 Krüppel-like factor 2 (Homo sapiens) 3E-19 NM_016270

TABLE 2. Summary of Fold Level Changes in Expression Between Tissuesa

RE/LE RE/LE Blood/RE RE/NE

Sal ID/gene name (best hit gene name) (microarray) (qPCR) (qPCR) (qPCR)
M003080 (methyltransferase) 7.79 10.64 0.0037 7.44
Msx2 6.47 3.11 0.3995 6.17
M011831 5.73 34.50 0.0118 60.33
M064466 5.30 32.37 0.0468 54.15
M002949 (sodefrin precursor-like factor) 5.18 19.91 0.0083 199.06
M006889 (hyaluronan and proteoglycan link protein 3) 5.11 2.43 0.7441 14.23
M065735 5.02 2.58 0.0441 20.97
M061758 (dynein) 4.53 3.07 0.0015 28.15
Wnt-5a 4.08 2.28 0.2899 22.65
M065526 (similar to desmoglein 4 preprotein) 3.53 3.29 0.0119 23.07
M003433 (apolipoprotein C-I) 3.38 1.68 0.0474 28.49
M002254 (uromodulin-like) 3.26 1.89 0.0278 33.23
M008800 (laminin, beta 1) 2.86 2.68 0.0012 40.71
M003964 (prostate stem cell antigen) 2.66 1.84 0.0190 4.32
M004510 (mal, T-cell differentiation protein-like) 2.25 1.23 0.0089 2.66
M062282 (EGF-like domain, multiple 6) 2.13 1.38 1.1405 1.08
BMP2/4-like 2.06 1.83 0.3872 3.03
Dlx-3 1.88 1.55 0.0974 1.06
M061881 (LY6/PLAUR domain containing 2) 1.68 2.88 0.0756 18.05
M062365 (Kruppel-like factor 2) 1.58 2.31 1.2014 4.05
M001136 (hemoglobin, delta) 2.09 N/A 5225 N/A
M003674 (myosin, heavy chain 3) 0.05 0.09 N/A N/A
M032377 0.04 0.15 N/A N/A

a
RE, regeneration epithelium; LE, lateral cuff wound epidermis; NE, normal epidermis; qPCR, quantitative polymerase chain
reaction (qPCR).
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1829

represented 16,257 sequences from

A. mexicanum and A. tigrinum and
396 sequences from other salaman-
der species.
A total of 698 probe sets demon-
strated significantly different (p 
0.0016) hybridization intensities
between the RE and the LE (266 in
the RE; 423 in the LE; Supp. File S1,
which is available online). We focused
on the significant probe sets that
exhibited at least a 1.5-fold intensity
increase in the RE over the LE (n ¼
195; Supp. File S2). The Sal-Site
(http://www.ambystoma.org; Smith
et al., 2005) was used to BLAST the
probe sets and identify sequences tar-
geted by the probes. Best hits were
identified with NCBI BLAST (Table 1;
Supp. File S2).
Best hit identification revealed that
blood-specific genes, such as globins,
were over-represented in the RE list.
Developmental Dynamics

Due to the extent of injury during

amputation it is not surprising to see
this up-regulation in the RE with
respect to the LE. It is expected that
there was carryover of blood cells in
the dissection and RNA isolation of
the RE samples. Additionally, the best
hit identification revealed that genes
in the list are targeted by multiple
probe sets. To reduce the list of RE
genes we averaged the multiple hits
and achieved a list of 125 genes that
are up-regulated in the RE (Supp.
File S3).

Validation of RE Gene
Overexpression by qPCR
For the purposes of validation and
further investigation we focused on
the top ten genes as well as ten addi-
tional genes of interest (Table 1).
Because we observed several blood-
specific genes in the RE list it was
necessary to confirm that these
selected genes are specific to RE and
not to blood. Total RNA was isolated
from RE at 7 days after amputation,
as well as from blood cells, and qPCR
was used to examine expression levels
of the genes of interest. The geometric
mean of four endogenous genes, glyc-
eraldehyde-3-phosphate dehydrogen-
ase (GAPDH), b-Actin, EF1a, and
Fig. 2. Quantitative polymerase chain reaction (qPCR) validation of fold change gene expression of
regeneration epithelium (RE) up-regulated hits. A: Fold change expressions in blood over RE. B: Fold L27, was used for data normalization.
change expressions in RE over lateral cuff wound epidermis (LE). C: Fold change expressions in RE A delta globin gene with Sal ID
over normal epidermis (NE). Genes are represented by Sal ID or gene name as in Table 1. Data rep- M001136 (Table 2) was used as an in-
resent the mean of three biological replicates and standard error of the mean. *p  0.05; **p  0.01. ternal control. The results confirmed
 1830 CAMPBELL ET AL.

in RE and LE at 7 days after wound-

ing and also in normal unwounded ep-
idermis. As with the tissue collection
for the microarray data set, the RE
tissues were collected at a time-point
corresponding to a phase between
wound healing and early bud. RNA
was isolated only from the epidermal
tissue of the regenerate. Again, the
analysis was performed using the
geometric mean of four endogenous
controls, GAPDH, bActin, EF1a, and
L27, as reference. Two genes,
M003674 and M032377, which were
identified in the microarray as more
highly expressed in LE, were used as
internal controls (Table 2). The qPCR
analysis confirmed overexpression of
all 20 genes of interest in the RE with
respect to LE (Fig. 2B). The two LE
genes were validated with higher fold
levels in the lateral cuff wound than
the amputation wound. M011831,
Developmental Dynamics

M064466, and M002949 demon-

strated much higher fold level differ-
ences in RE than in LE by qPCR anal-
ysis than by microarray (Table 2),
while the remaining genes displayed
relatively similar fold differences by
both methods. These data suggest
that the five genes with the greatest
fold difference between RE and LE
are M011831, M064466, M002949,
M003080, and M065526, respectively.
Normal epidermis (NE) was also
compared with RE using qPCR. The
skin removed from the limb to make
the lateral cuff wound was soaked in
dispase I, which promoted the separa-
tion of epidermis from the underlying
dermis (Kitano and Okada, 1983). As
before, analysis was performed using
Fig. 3. Whole-mount in situ hybridization analysis of M002949 (sodefrin precursor factor) the geometric mean of four endoge-
expression in 7-day postamputation regenerating axolotl limbs. All panels are forelimbs with dis-
tal at the top. A: M002949 expression covering distal tip of regenerating limb indicated in blue
nous controls, GAPDH, bActin, EF1a,
as determined with anti-sense probe. B: Negative control using M002949 sense probe. C,D: Sis- and L27. QPCR analysis confirmed
ter sections of tissue from panel A. Dashed line represents level of amputation. Tissue sectioned that all but two genes were signifi-
at 14 mm. C: Expression in the regeneration epithelium indicated in blue. D: Hematoxylin and cantly overexpressed in RE over NE
eosin. E,F: Higher magnification of the distal tip of the regenerating limb shows that the
(Fig. 2C). M002949 demonstrated the
M002949 expression is localized to the top layers of the regeneration epidermis.
greatest fold difference at nearly 200-
fold (Table 2). The two genes that did
a >5,000-fold level of expression for higher, respectively, in blood over not show significant difference were
M001136 in blood over RE, while 16 of RE, suggesting that M062282 and M062282 and Dlx-3, with fold level dif-
the 20 RE genes were significantly M062365 may be expressed by both ferences of 1.08 and 1.06, respectively.
more highly expressed in RE than in blood and RE. Some of the fold differ- Given the technical difficulty of
blood (Fig. 2A). Two genes, M006889 ence in expression level of M062282 isolating epidermal tissue from the
and Wnt-5a, tended toward higher and M062365 in RE over LE may be underlying stump and blastemal tis-
expression levels in RE than in blood due to blood cell contamination. sue, we cannot eliminate the possibil-
but did not have significant p-values. QPCR was used as an independent ity that gene expression differences
The qPCR analysis with primers verification method of RE gene between RE and LE or RE and NE
for M062282 and M062365 showed expression levels as compared to LE. may be partially due to blastemal cell
expression levels of 1.1- and 1.2-fold The 20 genes of interest were tested contamination. Whole-mount in situ
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1831
Developmental Dynamics

Fig. 4. Expression level time-courses of genes previously reported to play a role in limb regeneration. The y-axis represents normalized RNA level
and the x-axis represents days post amputation. Genes are represented by Sal ID and/or gene name as in Table 1. Each dark circle represents
the mean of three samples 6 standard error of the mean. *p  0.05; **p  0.01.

hybridization analysis shows that stump tissue (Fig. 3C,D). The expres- the mesenchymal tissue of the regen-
M002949, one of the genes with the sion of M002949 in the top layers of erate; however, RE-specific expression
greatest difference between RE and the RE (Fig. 3E,F) confirms that the of M002949 demonstrates that mes-
LE, as well as RE and NE, is identification of this gene by microar- enchymal cell contamination did not
expressed over the distal tip of the ray analysis was due to tissue-specific contribute to the identification of one
regenerating limb at 7 days after expression in the RE as opposed to of the most highly expressed RE
amputation (Fig. 3A). The sense con- contamination of blastemal or under- genes as compared to LE. The micro-
trol showed no stain (Fig. 3B). Sec- lying stump tissues. It is important to array results in combination with the
tioning of the tissue showed that note that this does not exclude the qPCR data demonstrate that the RE
expression is limited to the RE, with possibility that some microarray-iden- is unique in terms of gene expression
no expression in the underlying tified RE genes may be expressed by as compared to the LE or NE. These
 1832 CAMPBELL ET AL.
Developmental Dynamics

Fig. 5. Expression level time-courses of three finger protein family members. The y-axis represents normalized RNA level and the x-axis repre-
sents days post amputation. Genes are represented by Sal ID and gene name as in Table 1. Each dark circle represents the mean of three sam-
ples 6 standard error of the mean. *p  0.05; **p  0.01.

results provide a set of regeneration lected and assayed, expression level that has been described to be expressed
markers for the identification of the changes in the time-course experi- in the RE and distal mesenchyme of
RE in regeneration studies. ments, unlike the microarray and the regenerating limb (Carlson et al.,
qPCR validation experiments, account 1998; Koshiba et al., 1998). The time
RE-Specific Genes Display for gene expression levels in the RE as course data show that the expression
well as possible expression in the blas- level gradually increases over time
Changes in Expression Levels
tema tissue. Therefore it should be from the wound healing stage through
Throughout the Regeneration noted that expression level changes late bud (Fig. 4A). This result concurs
Process could account for gene expression in dif- with the previous report that Msx2 is
The temporal expression patterns of ferent or multiple tissues of the regen- re-expressed after amputation and is
the RE genes of interest were charac- erate during the various phases of limb highly expressed in the late bud stage
terized using qPCR. Regenerating regeneration. In the following sections, (Carlson et al., 1998).
limbs representing six stages of we present the temporal expression Wnt-5a, a member of the Wnt fam-
regeneration were processed for RNA patterns of all 20 genes of interest and ily of secreted proteins, has been pre-
isolation and analyzed for each gene discuss their identity as determined by viously described to play an impor-
with two endogenous controls for nor- BLAST search or previous characteriza- tant role in the early stages of limb
malization: GAPDH and EF1a. Regen- tion in limb regeneration. regeneration during dedifferentiation
erates were collected from slightly (Ghosh et al., 2008). The temporal
smaller animals than those used for Temporal expression patterns of expression pattern described by the
the microarray and for the qPCR vali- qPCR time course agrees with this
genes previously described in
dation data in Table 2. Therefore, the previous report that expression was
7 day post amputation time-point used
limb regeneration. detected from very early stages and
in the microarray study corresponds Five genes identified in the microar- throughout regeneration (Fig. 4B).
with the 2–4 day time-points used for ray have been previously described to BMP2 has been described as an im-
the time-course expression profiling. play a role during limb regeneration. portant factor in the onset of conden-
Because the entire regenerate was col- Msx2 encodes a transcription factor sation during limb regeneration
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1833
Developmental Dynamics

Fig. 6. Alignment of axolotl (A. mexicanum) M002949 amino acid sequence with sodefrin precursor-like factor from four species of salamander:
seal salamander (D. monticola; accession no. AAZ06333), clouded salamander (A. ferreus; accession no. AAZ06336), Blue Ridge two-lined sala-
mander (E. wilderae; accession no. AAZ06337), and Siskiyou Mountain salamander (P. stormi, accession no. AAZ06325). Conserved residues are
indicated by blue and nonconserved residues are indicated by red. Gaps are indicated by a dash (-).

(Guimond et al., 2010). The qPCR indicates that Dlx-3 significantly Temporal expression patterns of
time-course data show an increase in increases early in expression level as three-finger protein family
expression from the stump through compared to stump tissue and tends
members.
medium bud stages and a significant to be expressed at a constant level
decrease from medium bud to digit throughout regeneration with no sig- Three genes (M002949, M003964, and
stage (Fig. 4C). This concurs with the nificant increase or decrease over M061881) are structurally similar to
in situ hybridization patterns that time (Fig. 4D). Prod1 (Garza-Garcia et al., 2009).
show strong staining in the medium M002254 shows a significant Prod1 is a three-finger protein (TFP)
to late bud and then a distinct local- increase in expression upon wound family member that is involved in
ization to the interdigital regions healing and decreases over time from positional identity of the proximal-
(Guimond et al., 2010). the wound healing stage to palette distal axis in the newt limb (da Silva
Dlx-3, or distal-less 3, has been pre- (Fig. 4E). This gene shows similarity et al., 2002) and binds to nAG, the
viously described to play a nerve-de- to Xenopus tropicalis uromodulin-like. nerve factor that rescues regeneration
pendent role in regeneration with an It has been linked to nerve-dependent in denervated newt limbs (Kumar
expression pattern that is very low at blastema outgrowth in axolotl (Mona- et al., 2007). The structural similarity
early stages and peaks at the late bud ghan et al., 2009) and has been shown suggests that these factors may be
stage of regeneration (Mullen et al., to be down-regulated in response to secreted or anchored to the cell mem-
1996). Contrary to the previous thyroid hormone-induced metamor- brane by glycosylphosphatidylinositol
report, qPCR time course analysis phosis in Xenopus (Brown et al., 1996). (GPI) linkage.
 1834 CAMPBELL ET AL.
Developmental Dynamics

Fig. 7. Expression level time-courses of genes involved in cell adhesion and organization. The y-axis represents normalized RNA level and the x-
axis represents days post amputation. Genes are represented by Sal ID and gene name as in Table 1. Each dark circle represents the mean of
three samples 6 standard error of the mean. *p  0.05; **p  0.01.

M002949 shows high expression Temporal expression patterns of M062365 encodes a gene with
level at the wound healing stage and genes with roles in cell adhesion strong similarity to the Krüppel-like
takes a drop in expression by the zinc-finger transcription factor 2. This
and organization.
early bud stage (Fig. 5A). The tendency gene is also described as lung KLF
for lower expression during the Four genes have similarities with cell (LKLF). Knockout mouse studies
remaining stages of limb regeneration adhesion factors. The M065526 pat- have shown that loss of LKLF results
suggests that it may have a very early tern shows higher expression in in a change in smooth muscle cell
role in regeneration. M002949 shows regenerating tissue as compared to morphology and loss of organization
similarity to sodefrin precursor-like stump tissue with fluctuating expres- in the blood vessel wall (Kuo et al.,
factor from several salamander species, sion levels during regeneration (Fig. 1997). The temporal expression pat-
particularly at the N-terminal end and 7A). This EST shows similarity to des- tern shows a significant increase
among the cysteine residues (Fig. 6). moglein 4 preprotein, a cadherin fam- upon wound healing and a tendency
M003964 shows similarity to pros- ily member that is involved in the toward relatively stable expression
tate stem cell antigen (PSCA), a GPI- formation of desmosomes. More spe- level during regeneration (Fig. 7C).
anchored cell membrane protein. The cifically it has been shown to localize The M006889 gene shows similarity
expression pattern of this PSCA-like to desmosomes in the human hair fol- to a hyaluronan and proteoglycan
gene takes a significant increase imme- licle (Bazzi et al., 2006), a regenerat- link protein family member. Align-
diately upon wound healing and then a ing dermal appendage. ment with Xenopus and human
tendency toward low expression levels M008800 shows similarity to lami- sequences shows high conservation in
throughout regeneration (Fig. 5B). nin, beta 1, which has been localized the second hyaluronan and proteogly-
Similarly, the M061881 gene shows a to epithelial basement membranes can binding link domain (Fig. 8). This
significant increase upon wound heal- (Virtanen et al., 2003). During limb family of proteins functions in cell ad-
ing and then a significant drop in regeneration its expression increases hesion and migration by binding hya-
expression level at the early bud stage upon wound healing and signifi- luronan, a glycosaminoglycan, with
(Fig. 5C). This gene shows similarity to cantly decreases by the palette stage proteoglycans to modify the extracel-
the LY6/PLAUR domain containing 2. (Fig. 7B). lular matrix or cell surfaces (Fraser
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1835
Developmental Dynamics

Fig. 8. Alignment of axolotl (A. mexicanum) M006889 amino acid sequence with hyaluronan and proteoglycan link protein 3 from X. laevis (acces-
sion no. NP_001079631) and human (H. sapiens; accession no. NP_839946). Conserved residues are indicated in blue and nonconserved residues
are indicated by red. Gaps are indicated by a dash (-). The shaded boxes identify the hyaluronan and proteoglycan link domains.

et al., 1997). The temporal expression (Fig. 9A). Dynein has been described believed to be involved in lipid trans-
pattern demonstrates an increase in to play an important role in axonal port. M004510 shows similarity to
expression level at the medium bud regeneration by signaling to the cell mal, T-cell differentiation protein-
stage as compared to stump tissue body that an axon has been injured like, which is a protein that appears
and then a decrease in expression at (Tuck and Cavalli, 2010). The qPCR to localize in lipid rafts. Both of these
the late bud stage (Fig. 7D). The data for temporal limb regeneration genes show a significant increase in
expression pattern suggests an inter- expression shows that expression for expression by the early bud stage.
esting initial decrease in expression this gene peaks at the wound healing It has also been suggested that an
level upon wound healing. stage. epidermal growth factor-like molecule
M003433 (Fig. 9B) and M004510 and lipid rafts are involved in cutane-
(Fig. 9C) are two lipid-associated pro- ous wound healing (Mathay et al.,
Temporal expression patterns of teins. Lipid rafts have been described 2007). The M062282 sequence shows
genes with lipid-associated and as crucial for the signaling that drives similarity to an epidermal growth fac-
neurite outgrowth and regeneration tor repeat superfamily member,
neurite regeneration roles.
(Zhao et al., 2009). M003433 shows EGFL6. Genes of this type have been
M061758 shows similarity to dynein, similarity to apolipoprotein C-I, shown to be expressed during early
a minus-end directed motor protein which is a lipid-binding protein development (Buchner et al., 2000b)
 1836 CAMPBELL ET AL.
Developmental Dynamics

Fig. 9. Expression level time-courses of genes with lipid-associated and neurite regeneration roles. The y-axis represents normalized RNA level
and the x-axis represents days post amputation. Genes are represented by Sal ID and gene name as in Table 1. Each dark circle represents the
mean of three samples 6 standard error of the mean. *p  0.05; **p  0.01.

and in the anterior part of the dermal increase in expression at the wound the digit stage (Fig. 10D). The gene
placode, which induces hair follicle healing stage and then a tendency for shows similarity to a putative S-aden-
formation (Buchner et al., 2000a). higher expression through medium osylmethionine-dependent methyl-
The qPCR temporal expression pat- bud and lower expression from late transferase (AdoMet-MTase), class I
tern for EGFL6 shows a very low bud through digit stage (Fig. 10A). (Fig. 11). The class I family of AdoMet-
expression level after amputation and The expression pattern of M064466 MTases is the largest and most diverse
then a significant increase at each is highest at the wound healing stage including members with substrate
time point thereafter, with the high- and then significantly decreases by specificity to small molecules, lipids,
est expression at the digit stage (Fig. the late bud and digit stages (Fig. proteins, and nucleic acids (Martin
9D). Taken in concert with the com- 10B), suggesting that the gene prod- and McMillan, 2002; Schubert et al.,
parison data that suggested nearly uct plays a role in wound healing and 2003). Further investigation is needed
equal expression levels between RE early blastema formation. to identify the substrate of this poten-
and NE (Fig. 2C), M062282 may play M065735, another sequence with tial axolotl AdoMet-MTase and its role
a role in the maturation of the epider- no identifiable similarities, shows a in limb regeneration.
mis covering the regenerate. significant increase upon wound heal-
ing followed by a decrease at early
bud and then an increase at medium CONCLUSIONS
Temporal expression patterns of
bud stage, suggesting a dynamic This study provides the first expres-
sequences with unknown or
expression pattern (Fig. 10C). sion profiling of the RE in urodele am-
hypothetical identity. M003080, which presented in the phibian limb regeneration. The anal-
The M011831 sequence does not show microarray with the highest fold dif- ysis identified 125 genes that
strong similarity to any known genes ference in RE over LE, demonstrated demonstrate higher expression in
but demonstrates a high fold differ- an increase in expression upon wound the regenerative epithelium than in
ence of expression level in RE over LE healing followed by a tendency toward wound epidermis covering a lateral
(Table 2; Fig. 2B). The temporal higher expression during medium and cuff wound, suggesting that the
expression pattern shows a significant late bud stages and then a decrease to expression is specific to the
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1837
Developmental Dynamics

Fig. 10. Expression level time-courses of genes with unknown or hypothetical identity. The y-axis represents normalized RNA level and the x-
axis represents days post amputation. Genes are represented by Sal ID and gene name as in Table 1. Each dark circle represents the mean of
three samples 6 standard error of the mean. *p  0.05; **p  0.01.

regeneration response of an amputa- obtained from the Ambystoma Genetic and LE were collected at 7 days after
tion wound as opposed to general Stock Center at the University of Ken- amputation/wounding. NE was col-
wound healing. The qPCR data for a tucky. Amputations and tissue collec- lected by soaking full-thickness skin
subset of the genes support the tions were performed on animals from the radial lateral wounding in a
microarray findings and show that measuring 8–15 cm from snout to tip 1% solution of dispase I (Sigma-
they are significantly more highly of tail. All animals were anesthetized Aldrich) in 0.8 phosphate buffered sa-
expressed in RE than in NE. Addi- in 0.1% MS222 solution (Ethyl 3-ami- line (PBS) for 5 hr at room tempera-
tional qPCR data show interesting nobenzoate methanesulfonate salt, ture, washing with 0.8 PBS, and then
expression changes for the genes Sigma-Aldrich, St. Louis, MO). Radial gently peeling the epidermis from the
during the time-course of regenera- lateral wounds were created by cutting dermis layer. Dispase treatment was
tion. These markers will provide an through full thickness skin around the only performed on the NE tissue sam-
important tool for studying the early circumference of the limb with spring ple. Blood was collected from the radial
events in limb regeneration. Further scissors and peeling away the full lateral and amputation wounds at the
study into the function of the indi- thickness skin from the underlying time of RE and LE collection. Limb
vidual genes will illuminate the role stump tissue. Animal care and use regenerates for qPCR were collected at
of the RE for successful limb protocols were approved by the Yale 0, 2, 4, 7, 10, 14, and 21 days after
regeneration. University Institutional Animal Care amputation. Tissues were soaked in
and Use Committee. RNAlater (Ambion, Foster City, CA)
before RNA isolation. For microarray
EXPERIMENTAL analysis, three pools of seven animals
Tissue Collection each were used. For qPCR validation
PROCEDURES
Amputations and lateral wounds studies, three pools of four animals
Animal Procedures were made in the zeugopod region of each were used. For regeneration time-
Axolotls (Ambystoma mexicanum) the limbs (between the wrist/ankle course qPCR studies, three pools of
were spawned at Yale University or joint and the elbow/knee joint). RE three animals each were used.
 1838 CAMPBELL ET AL.
Developmental Dynamics

Fig. 11. Alignment of axolotl (A. mexicanum) M003080 amino acid sequence with putative methyltransferase sequences from chicken (G. gallus;
accession no. XP_001232694) and X. laevis (accession no. NP_001136263). Conserved residues are indicated by blue and residues that are not
conserved are indicated by red. Gaps are indicated by a dash (-). The shaded boxes identify the S-adenosylmethionine-dependent methyltransfer-
ase, class I regions in chicken and Xenopus.

RNA Isolation and lection were performed by the White- tine was used to moderate the t-statis-
head Institute Genome Technology tic (Smyth, 2004). Genes were
Microarray Analysis Core (Cambridge, MA). RE samples selected as differentially expressed af-
RNA was isolated using Trizol Rea- were labeled with Cy5; LE samples ter adjusting p-values for repeated
gent (Invitrogen, Carlsbad, CA). Fol- were labeled with Cy3. Each of the tests by controlling the false discovery
lowing isolation, RNA was purified three pools of tissues collected for RE rate (Benjamini and Hochberg, 1995).
and DNase treated using RNeasy and LE were hybridized to three
minicolumns (Qiagen, Valencia, CA). arrays as biological replicates, while a
RNA quality was assessed by spectro- fourth array was used as a technical
Quantitative PCR
photometry using a NanoDrop ND- replicate of one of the biological sam- Reverse transcription was performed
1000 (NanoDrop, Wilmington, D). ples. Analysis of the arrays was done with iScript Reverse Transcription
RNA samples were also analyzed on a using the limma (Smyth, 2005) pack- Supermix (Bio-Rad, Hercules, CA).
Bioanalyzer 2100 (Agilent Technolo- age in Bioconductor (Gentleman Quantitative PCR assays were run
gies, Santa Clara, CA). Microarray et al., 2004). Multiple probes for the using Power SYBR Green PCR Mas-
analysis was performed on custom same clone were averaged. Arrays ter Mix (Applied Biosystems, Carls-
eArrays (Agilent Technologies, Santa were normalized with the loess rou- bad, CA) on a C1000 Thermal Cycler
Clara, CA) using 60-mer probes tine (Yang et al., 2001, 2002; Smyth (Bio-Rad) and analyzed with the
designed against genes and ESTs and Speed, 2003) without background CFX96 Real-Time System. Primer
from Ambystoma mexicanum, Ambys- correction (Zahurak et al., 2007). A sequences and annealing tempera-
toma tigrinum and other salamander linear model was fit to the expression tures for the genes assayed are listed
species. Hybridization and data col- values, and an empirical Bayes rou- in Supp. File S4.
 GENE EXPRESSION DURING AXOLOTL LIMB REGENERATION 1839

Whole-Mount In Situ Christiano AM. 2006. Desmoglein 4 is superfamily implicated in salamander

expressed in highly differentiated kera- limb regeneration. PloS One 4:e7123.
Hybridization and Histology tinocytes and trichocytes in human epi- Gentleman R, Carey V, Bates D, Bolstad
Whole-mount in situ hybridization dermis and hair follicle. Differentiation B, Dettling M, Dudoit S, Ellis B, Gaut-
74:129–140. ier L, Ge Y, Gentry J, Hornik K, Hot-
was performed as previously described Benjamini Y, Hochberg Y. 1995. Control- horn T, Huber W, Iacus S, Irizarry R,
(Gardiner et al., 1995) with modifica- ling the false discovery rate: a practical Leisch F, Li C, Maechler M, Rossini A,
tions. Tissues were fixed overnight at and powerful approach to multiple test- Sawitzki G, Smith C, Smyth G, Tierney
4 C with gentle rocking in freshly- ing. J R Stat Soc Series B Stat Meth- L, Yang J, Zhang J. 2004. Bioconductor:
odol 57:289–300. open software development for compu-
made MEMFA (0.1 M MOPS, pH 7.4, 2
Brown DD, Wang Z, Furlow JD, Kana- tational biology and bioinformatics. Ge-
mM EGTA, 1 mM MgSO4, 3.7% form- mori A, Schwartzman RA, Remo BF, nome Biol 5:R80.
aldehyde) and then dehydrated and Pinder A. 1996. The thyroid hormone- Ghosh S, Roy S, Seguin C, Bryant SV,
stored at 20 C in 100% methanol. induced tail resorption program during Gardiner DM. 2008. Analysis of the
Proteinase K (New England BioLabs, Xenopus laevis metamorphosis. Proc expression and function of Wnt-5a and
Natl Acad Sci U S A 93:1924–1929. Wnt-5b in developing and regenerating
Ipswich, MA) treatment was per- Buchner G, Broccoli V, Bulfone A, Orfa- axolotl (Ambystoma mexicanum) limbs.
formed at 10 mg/ml for 30 min at 4 C nelli U, Gattuso C, Ballabio A, Franco Dev Growth Differ 50:289–297.
followed by 15 min at 37 C to permeab- B. 2000a. MAEG, an EGF-repeat con- Guimond J-C, Levesque M, Michaud P-L,
ilize tissue. Prehybridization was per- taining gene, is a new marker associ- Berdugo J, Finnson K, Philip A, Roy S.
formed at 60 C overnight. Probes were ated with dermatome specification and 2010. BMP-2 functions independently of
morphogenesis of its derivatives. Mech SHH signaling and triggers cell conden-
prepared from pBluescript SK- vector Dev 98:179–182. sation and apoptosis in regenerating ax-
containing the 1,062 bp M002949 frag- Buchner G, Orfanelli U, Quaderi N, Bassi olotl limbs. BMC Dev Biol 10:15.
ment. The vector was linearized with MT, Andolfi G, Ballabio A, Franco B. Habermann B, Bebin A-G, Herklotz S,
PstI and anti-sense probe was tran- 2000b. Identification of a new EGF- Volkmer M, Eckelt K, Pehlke K, Epper-
repeat-containing gene from human lein H, Schackert H, Wiebe G, Tanaka
scribed with T7 RNA polymerase (New
Xp22: a candidate for developmental E. 2004. An Ambystoma mexicanum
Developmental Dynamics

England BioLabs) and digoxigenin disorders. Genomics 65:16–23. EST sequencing project: analysis of
RNA labeling mix (Roche Applied Sci- Campbell LJ, Crews CM. 2008. Wound 17,352 expressed sequence tags from
ence, Indianapolis, IN). Sense probe epidermis formation and function in embryonic and regenerating blastema
was transcribed with T3 RNA poly- urodele amphibian limb regeneration. cDNA libraries. Genome Biol 5:R67.
Cell Mol Life Sci 65:73–79. Han MJ, An JY, Kim WS. 2001. Expres-
merase (Roche Applied Science) and Carlson MR, Bryant SV, Gardiner DM. sion patterns of Fgf-8 during develop-
digoxigenin RNA labeling mix after 1998. Expression of Msx-2 during devel- ment and limb regeneration of the
linearization with XhoI. Hybridization opment, regeneration, and wound heal- axolotl. Dev Dyn 220:40–48.
was performed at 60 C for 72 hr. Alka- ing in axolotl limbs. J Exp Zool 282: Kato T, Miyazaki K, Shimizu-Nishikawa
line-phosphatase (AP) conjugated 715–723. K, Koshiba K, Obara M, Mishima HK,
Christensen RN, Weinstein M, Tassava Yoshizato K. 2003. Unique expression
anti-digoxigenin antibody was RA. 2002. Expression of fibroblast patterns of matrix metalloproteinases
obtained from Roche Applied Science. growth factors 4, 8, and 10 in limbs, in regenerating newt limbs. Dev Dyn
The colorimetric alkaline-phosphatase flanks, and blastemas of Ambystoma. 226:366–376.
reaction was developed with BM Pur- Dev Dyn 223:193–203. Kitano Y, Okada N. 1983. Separation of
da Silva SM, Gates PB, Brockes JP. 2002. the epidermal sheet by dispase. Br J
ple (Roche Applied Science). Tissues
The newt ortholog of CD59 is impli- Dermatol 108:555–560.
from whole-mount in situ hybridiza- cated in proximodistal identity during Koshiba K, Kuroiwa A, Yamamoto H,
tion were cryoembedded and sectioned amphibian limb regeneration. Dev Cell Tamura K, Ide H. 1998. Expression of
at 14 mm. Hematoxylin and eosin 3:547–555. Msx genes in regenerating and develop-
stains (Sigma-Aldrich) were used to Endo T, Bryant SV, Gardiner DM. 2004. ing limbs of axolotl. J Exp Zool 282:
A stepwise model system for limb 703–714.
stain nuclei blue and cytoplasm and regeneration. Dev Biol 270:135–145. Kragl M, Knapp D, Nacu E, Khattak S,
extracellular proteins shades of pink. Ferris DR, Satoh A, Mandefro B, Cum- Maden M, Epperlein HH, Tanaka EM.
mings GM, Gardiner DM, Rugg EL. 2009. Cells keep a memory of their tis-
ACKNOWLEDGMENTS 2010. Ex vivo generation of a functional sue origin during axolotl limb regenera-
We thank the Crews laboratory for and regenerative wound epithelium tion. Nature 460:60–65.
from axolotl (Ambystoma mexicanum) Kumar A, Godwin J, Gates P, Garza-Garcia
helpful discussion. We thank Randall skin. Dev Growth Differ 52:715–724. A, Brockes J. 2007. Molecular basis for the
Voss for providing the vector contain- Fraser JRE, Laurent TC, Laurent UBG. nerve dependence of limb regeneration in
ing M002949. H.O.Z. is supported, in 1997. Hyaluronan: its nature, distribu- an adult vertebrate. Science 318:772.
part, by P20RR016470. We acknowl- tion, functions and turnover. J Intern Kuo CT, Veselits ML, Barton KP, Lu MM,
Med 242:27–33. Clendenin C, Leiden JM. 1997. The
edge the services of the Ambystoma
Gardiner DM, Blumberg B, Komine Y, LKLF transcription factor is required
Genetic Stock Center, which is sup- Bryant SV. 1995. Regulation of HoxA for normal tunica media formation and
ported by NSF-DBI-0443496, and the expression in developing and regenerat- blood vessel stabilization during murine
resources available at the Sal-Site, ing axolotl limbs. Development 121: embryogenesis. Genes Dev 11:
which are supported by NIH-NCRR- 1731–1741. 2996–3006.
Gardiner DM, Carlson MR, Roy S. 1999. Lo DC, Allen F, Brockes JP. 1993. Rever-
R24RR16344. Towards a functional analysis of limb sal of muscle differentiation during uro-
regeneration. Semin Cell Dev Biol 10: dele limb regeneration. Proc Natl Acad
385–393. Sci U S A 90:7230–7234.
REFERENCES Garza-Garcia A, Harris R, Esposito D, Martin JL, McMillan FM. 2002. SAM (de-
Gates PB, Driscoll PC. 2009. Solution pendent) I AM: the S-adenosylmethio-
Bazzi H, Getz A, Mahoney MG, Ishida- structure and phylogenetics of Prod1, a nine-dependent methyltransferase fold.
Yamamoto A, Langbein L, Wahl Iii JK, member of the three-finger protein Curr Opin Struct Biol 12:783–793.
 1840 CAMPBELL ET AL.

Mathay C, Giltaire S, Minner F, Bera E, Repesh LA, Oberpriller JC. 1978. Scan- Stocum DL. 2004. Amphibian regenera-
Herin M, Poumay Y. 2007. Heparin- ning electron microscopy of epidermal tion and stem cells. Curr Top Microbiol
binding EGF-like growth factor is cell migration in wound healing during Immunol 280:1–70.
induced by disruption of lipid rafts and limb regeneration in the adult newt, Thornton CS. 1957. The effect of apical
oxidative stress in keratinocytes and Notophthalmus viridescens. Am J Anat cap removal on limb regeneration in
participates in the epidermal response 151:539–555. Amblystoma larvae. J Exp Zool 134:
to cutaneous wounds. J Invest Derma- Satoh A, Graham GMC, Bryant SV, 357–381.
tol 128:717–727. Gardiner DM. 2008. Neurotrophic regu- Tuck E, Cavalli V. 2010. Roles of membrane
Mescher AL. 1976. Effects on adult newt lation of epidermal dedifferentiation trafficking in nerve repair and regenera-
limb regeneration of partial and com- during wound healing and limb regen- tion. Commun Integr Biol 3:209–214.
plete skin flaps over the amputation eration in the axolotl (Ambystoma mex- Virtanen I, Korhonen M, Petajaniemi N,
surface. J Exp Zool 195:117–128. icanum). Dev Biol 319:321–335. Karhunen T, Thornell L-E, Sorokin LM,
Monaghan JR, Walker JA, Page RB, Schubert HL, Blumenthal RM, Cheng X. Konttinen YT. 2003. Laminin isoforms
Putta S, Beachy CK, Voss SR. 2007. 2003. Many paths to methyltransfer: a in fetal and adult human adrenal cortex.
Early gene expression during natural chronicle of convergence. Trends Bio- J Clin Endocrinol Metab 88:4960–4966.
spinal cord regeneration in the sala- chem Sci 28:329–335. Yang EV, Gardiner DM, Carlson MR,
mander Ambystoma mexicanum. J Neu- Singer M. 1952. The influence of the Nugas CA, Bryant SV. 1999. Expression
rochem 101:27–40. nerve in regeneration of the amphibian of Mmp-9 and related matrix metallo-
Monaghan JR, Epp LG, Putta S, Page extremity. Q Rev Biol 27:169–200. proteinase genes during axolotl limb
RB, Walker JA, Beachy CK, Zhu W, Pao Smith J, Putta S, Walker J, Kump D, regeneration. Dev Dyn 216:2–9.
GM, Verma IM, Hunter T, Bryant SV, Samuels A, Monaghan J, Weisrock D, Yang YH, Dudoit S, Luu P, Speed TP. 2001.
Gardiner DM, Harkins TT, Voss SR. Staben C, Voss S. 2005. Sal-Site: inte- Normalization for cDNA microarray
2009. Microarray and cDNA sequence grating new and existing ambystomatid data. In: Bittner ML, Chen Y, Dorsel AN,
analysis of transcription during nerve- salamander research and informational Dougherty ER, editors. Microarrays: op-
dependent limb regeneration. BMC Biol resources. BMC Genomics 6:181. tical technologies and informatics. Proc
7:1. Smyth GK. 2004. Linear models and em- SPIE 4266:141–152.
Mullen LM, Bryant SV, Torok MA, Blum- pirical bayes methods for assessing dif- Yang YH, Dudoit S, Luu P, Lin DM, Peng
berg B, Gardiner DM. 1996. Nerve de- ferential expression in microarray V, Ngai J, Speed TP. 2002. Normaliza-
Developmental Dynamics

pendency of regeneration: the role of experiments. Stat Appl Genet Mol Biol tion for cDNA microarray data: a robust
Distal-less and FGF signaling in am- 3:Article 3. composite method addressing single
phibian limb regeneration. Develop- Smyth GK. 2005. Limma: linear models and multiple slide systematic variation.
ment 122:3487–3497. for microarray data. In: Gentleman R, Nucleic Acids Res 30:e15.
Putta S, Smith J, Walker J, Rondet M, Carey V, Dudoit S, Irizarry R, Huber W, Zahurak M, Parmigiani G, Yu W, Scharpf R,
Weisrock D, Monaghan J, Samuels A, editors. Bioinformatics and computa- Berman D, Schaeffer E, Shabbeer S, Cope
Kump K, King D, Maness N, Haber- tional biology solutions using R L. 2007. Pre-processing Agilent microarray
mann B, Tanaka E, Bryant S, Gardiner and Bioconductor. New York: Springer. data. BMC Bioinformatics 8:142.
D, Parichy D, Voss SR. 2004. From bio- p 397–420. Zhao H, Cao X, Wu G, Loh HH, Law PY.
medicine to natural history research: Smyth GK, Speed T. 2003. Normalization 2009. Neurite outgrowth is dependent
EST resources for ambystomatid sala- of cDNA microarray data. Methods 31: on the association of c-Src and lipid
manders. BMC Genomics 5:54. 265–273. rafts. Neurochem Res 34:2197–2205.