Developmental and Comparative Immunology 28 (2004) 9–28

www.elsevier.com/locate/devcompimm

Development and maturation of the immune system in zebrafish,

Danio rerio: a gene expression profiling, in situ hybridization
and immunological study
S.H. Lama, H.L. Chuaa, Z. Gonga, T.J. Lama, Y.M. Sina,b,*
a
Department of Biological Sciences, The National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260
b
Singapore Fish Breeding and Immunization Centre, Teo Way Yong and Sons Pte. Ltd, Singapore
Received 12 December 2002; revised 12 May 2003; accepted 13 May 2003

Abstract
The development and maturation of the immune system in zebrafish was investigated using immune-related gene expression
profiling by quantitative real-time polymerase chain reaction, in situ hybridization (ISH), immunoglobulin (Ig) detection by
immuno-affinity purification and Western blotting as well as immersion immunization experiments. Ikaros expression was first
detected at 1 day post-fertilization (dpf) and thereafter increased gradually to more than two-fold between 28 and 42 dpf before
decreasing to less than the initial 1 dpf expression level in adult fish (aged 105 dpf). Recombination activating gene-1 (Rag-1)
expression levels increased rapidly (by 10-fold) between 3 and 17 dpf, reaching a maximum between 21 and 28 dpf before
decreasing gradually. However, in adult fish aged 105 dpf, the expression level of Rag-1 had dropped markedly, and was
equivalent to the expression level at 3 dpf. T-cell receptor alpha constant region and immunoglobulin light chain constant
region (IgLC) isotype-1, 2 and 3 mRNAs were detected at low levels by 3 dpf and their expression levels increased steadily to
the adult range between 4 and 6 weeks post-fertilization (wpf). Using tissue-section ISH, Rag-1 expression was detected in head
kidney by 2 wpf while IgLC-1, 2 and 3 were detected in the head kidney and the thymus by 3 wpf onwards. Secreted Ig was only
detectable using immuno-affinity purification and Western blotting by 4 wpf. Humoral response to T-independent antigen
(formalin-killed Aeromonas hydrophila) and T-dependent antigen (human gamma globulin) was observed in zebrafish
immunized at 4 and 6 wpf, respectively, indicating that immunocompetence was achieved. The findings reveal that the
zebrafish immune system is morphologically and functionally mature by 4 –6 wpf.
q 2003 Published by Elsevier Ltd.
Keywords: Zebrafish; Immune system maturation; Ikaros; Recombination activating gene-1; T-cell receptor alpha constant region;
Immunoglobulin light chain constant region isotypes; Humoral immunity

Abbreviations: dpf, day(s) post-fertilization; HGG, human 1. Introduction

gamma globulin; Ig, immunoglobulin; IgLC1,2,3,
immunoglobulin light chain constant region isotype 1,2,3; ISH, in
situ hybridization; PCR, polymerase chain reaction; Rag-1, The zebrafish, Danio rerio, offers an attractive
recombination activating gene-1; TCRAC, T-cell receptor alpha model for the study of ontogenetic development of
constant region; wpf, week(s) post-fertilization. the immune system [1 – 7]. The relative ease with
* Corresponding author. Address: Department of Biological
which zebrafish embryos and larvae can be studied,
Sciences, The National University of Singapore, 10 Kent Ridge
Crescent, Singapore 119260. Tel.: þ 65-64691619; fax: þ65-
the powerful genetics which can be applied for the
67792486. generation of mutants and transgenic animals, the vast
E-mail address: info@teowayyong.com.sg (Y.M. Sin). genomic resources including the availability of many
0145-305X/04/$ - see front matter q 2003 Published by Elsevier Ltd.
doi:10.1016/S0145-305X(03)00103-4
10 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
molecular markers and soon to be completed genomic
sequence, make zebrafish an extremely versatile model
for immune system developmental studies [6,8]. While
there have been cloning and characterization studies on
some immune-related genes [9 – 14] and studies on the
ontogeny of the lymphoid system during the embryo
period of the zebrafish, mainly to understand the
hierarchical steps and genetic control involved in
lymphopoeisis [1,2,5 – 7], little is known about the
maturation of its immune system, with regard to form
and function, which occurs later in development.
In fish, at the time of hatching, the immune system is
still developing and not all of the structures and
functions present in the adults are in the larvae [15–17].
Lymphoid-organogenesis, involving the development
of the thymic and head kidney primordia, is initiated in
the middle to late embryo period, but remains in its
rudimentary form throughout the early larval stages.
Although these organ primordia are colonized by
lymphoid precursors relatively early, the lymphoid
cells do not become immunocompetent until the
stromal non-lymphoid components of the organs,
mainly epithelial, dendritic and fibroblastic cells,
mature [3,16]. The maturation of the network of
stromal cells provides a supportive microenvironment
for the maturation of lymphocytes which involves
differentiation, selection and expansion of immuno-
competent cells. Correspondingly, the specific immune
system, both cell-mediated and humoral immunity, is
non-functional during the early larval stages and
becomes fully competent after both lymphoid organs
and cells mature several weeks after hatching.
Recently, we have reported the morphological
transformation of the developing zebrafish thymus
from 1 to 15 wpf [18]. We have observed that the
overall morphology of the zebra fish thymus changed
from a small pouch-like shape at 1 wpf to a conical
shape between 2 and 3 wpf before acquiring a more
complex shape from 4 wpf onwards. Rapid growth
along the lateral axis at the region near the pharyngeal
epithelium occurred between 1 and 2 wpf while rapid
growth along the dorsal – ventral axis occurred
between 3 and 6 wpf. Expansion of the thymocyte
population, beginning 1 wpf, became more evident by
2 –3 wpf as indicated by the apparent increase of
lymphocytes of different sizes, Rag-1 and T-cell
receptor alpha constant region (TCRAC)-positive
cells. Tissue-section in situ hybridization (ISH)
analysis with a Rag-1 probe revealed that cortex-
medullary regionalization begins between 1 and 2 wpf
as Rag-1 expression clearly demarcates the cortex
while the medulla is Rag-1 negative. The presence of
TCRAC-positive cells in the medulla by 2– 3 wpf,
suggests that the thymic selection processes have
begun. The zebrafish thymus is morphologically
mature by 3 wpf. Early signs of thymic involution
were observed in zebrafish aged 15 wpf.
In this study, we have extended our investigation
on the maturation of the zebrafish immune system.
The expression of six immune-related genes of the
zebrafish were monitored: Ikaros, which encodes a
transcription factor which is used as an early
lymphoid marker [7]; Rag-1, which encodes a protein
involved in genomic rearrangement (termed V(D)J
recombination) of the TCR and immunoglobulin (Ig)
loci, and is a suitable marker for maturing lympho-
cytes [1]; TCRAC and three immunoglobulin light
chain constant region (IgLC) isotypes genes which
encode portions of the antigen receptors of mature T
and B lymphocytes, respectively. The aim was to
correlate the immune-related gene expression profiles,
the in situ detection of cells expressing these genes in
lymphoid tissues, and the detection of Ig and the onset
of humoral immunity, and enable a better under-
standing of the different maturation state of the
immune system during zebrafish development.
2. Materials and methods
2.1. Animals
Newly hatched wild type zebrafish, Danio rerio,
were maintained in well-aerated aquarium conditions in
dechlorinated filtered water at 27 ^ 1 8C. The fish were
fed twice per day with Paramecium and rotifers until
about 10 dpf followed by Artemia salina nauplii until
8 wpf and thereafter with commercial feed (Tetra).
2.2. Detection of immune-related gene expression
levels by two-step conventional reverse transcription
polymerase chain reaction (RT-PCR) and quantitative
real-time PCR
Total RNA of whole fish aged 1, 3, 7, 10, 14, 17,
21, 24, 28, 35, 42 and 105 dpf was isolated in
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28 11

triplicate for each age group using TRIzole Reagent
(Life Technologies) at a concentration of not more
than 100 mg tissue/ml Trizol reagent as described in
the manufacturer’s protocol. Due to vast differences in
size, different numbers of embryos/fish were used for
each total RNA extraction per age group (ranging
from about 200 developing embryos for 1 and 3 dpf to
5 –6 adult fish for aged 105 dpf). The RNA concen-
trations were determined by OD260 measurements and
2 mg of total RNA was treated with RNase-free
DNase 1 (Life Technologies) prior to reverse
transcription with the First-strand cDNA Synthesis
Kit (Amersham Pharmacia) with Oligo(dT) as the
primer following manufacturers’ instructions. After
reverse transcription, the first-strand cDNA samples
in a final volume of 33 ml were stored at 2 80 8C until
used for PCR.
The conventional PCR was performed in a
thermocycler (Bio-Rad) with the Advantagew 2
PCR Enzyme System (Clontech Laboratories) in
Table 1
Targeted transcripts with GenBank Accession Number, primers correlating to nucleotide position in GenBank sequence, amplicon size and
cycling conditions used for conventional PCR [in parentheses] and quantitative real-time PCR. Amplifications performed using genomic DNA
as a template indicate that the primers are within an exon
a volume of 50 ml with final concentration of 1 £
Advantage 2 Polymerase Mix, 1 £ Advantage 2 PCR
buffer, 0.2 mM dNTPs, 0.4 mM of each primer and
2 ml of first strand cDNA sample. The quantitative
real-time PCR was performed using the Lightcyclere
system (Roche Applied Science) with the Lightcycler-
FastStart DNA Master SYBR Green 1 (Roche
Applied Science) in a volume of 20 ml with final
concentration of 1 £ Lightcycler-FastStart DNA
Master SYBR Green 1, 3 mM MgCl2, 0.5 mM of
each primer and 2 ml of first strand cDNA mixture.
The primers and conditions used for both the
conventional PCR and quantitative real-time PCR
are listed in Table 1. Amplification by conventional
PCR was carried out for 30 – 40 cycles while
quantitative real-time PCR was carried out for 40
cycles and fluorescence readings were acquired at the
end of each amplification cycle at 72 8C. Melting
curve analysis was performed with continuous
fluorescence acquisition from 65 to 95 8C at

Targeted GenBank Accession Number Amplicon size Pre-amplification Amplification

transcripts (primers correlate to nucleotide (bp) (8C/min)
position in GenBank sequence)
Denaturation Annealing Extension
(8C/s) (8C/s) (8C/s)

Ikaros AF092175 (upstream primer: 156 95/10 [95/1] 95/10 [95/30] 57/5 [57/20] 72/12 [72/30]
1167-1186; downstream primer:
1322-1303)
Rag-1 DRU71093 (upstream primer: 144 95/10 [95/1] 95/10 [95/30] 55/5 [55/20] 72/12 [72/30]
3712-3731; downstream primer:
3855-3836)
TCRAC AF246178 (upstream primer: 155 95/10 [95/1] 95/10 [95/30] 56/5 [57/20] 72/12 [72/30]
511-530; downstream primer:
665-646)
IgLC-1 AF246185 (upstream primer: 135 95/10 [95/1] 95/10 [95/30] 60/5 [58/20] 72/12 [72/30]
518-537; downstream primer:
652-633)
IgLC-2 AF246162 (upstream primer: 160 95/10 [95/1] 95/10 [95/30] 58/5 [58/20] 72/12 [72/30]
1852-1871; downstream primer:
2011-1992)
IgLC-3 AF246193 (upstream primer: 131 95/10 [95/1] 95/10 [95/30] 57/5 [57/20] 72/12 [72/30]
567-586; downstream primer:
697-678)
b-Actin AF057040 (upstream primer: 201 95/10 [95/1] 95/10 [95/30] 58/5 [58/20] 72/14 [72/30]
680-700; downstream primer:
880-860)
12 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
a temperature transition rate of 0.1 8C/s to determine
the amplification specificity. The PCR products for
both conventional and quantitative real-time PCR
were detected by electrophoresis using a 2.0– 2.5%
(w/v) agarose gel stained with ethidium bromide. In
all cases, the amplifications were specific and no
amplification was observed in negative controls
(water blanks and total RNA without reverse
transcription).
To determine the copy number of the targeted
transcripts, cloned cDNA plasmids carrying a frag-
ment of the respective targeted genes, including the
amplified region, were used to generate calibration
curves. The pIkaros [GenBank Accession Number:
AF092175; nucleotide (nt) 1026-1700] and pRag-1
(DRU71093; nt 2449-3967), containing a 675 and
1519 bp fragment of the respective coding regions in
pBluescript KS vector (Stratagene), were kindly
provided by Dr Z. Wen of the Institute of Molecular
and Cell Biology, National University of Singapore.
The pTCRAC (AF246178; nt 408-732), pIgLC-1
(AF246185; nt 425-752), pIgLC-2 (AF246162; nt
1679-2030) and pIgLC-3 (AF246193; nt 421-720),
containing 325, 328, 352 and 300 bp fragments of the
respective coding regions in pBluescript SK vector
were kindly provided by Drs G.W. Litman and R.N.
Haire of the University of South Florida, USA. The
zebrafish b-actin (AF057040; nt 49-1183) cDNA
plasmid containing 1135 bp of the coding region in
pBluescript SK vector was cloned in our lab. The
plasmids were purified using Wizard Plus SV
miniprep kit (Promega) according to manufacturer’s
instruction and the concentration was determined by
OD260 measurements. The copy numbers of the
plasmid DNA templates were calculated according
to the molecular weight of the plasmid (660/bp
average value) and then converted into the copy
numbers based upon Avogadro’s number
(1 mol ¼ 6.022 £ 1023 molecules). The plasmid
cDNA was serially diluted in log10 steps from 107
copies down to 101 copies. A calibration curve was
generated by plotting the threshold cycle (ct) versus
the known copy number for each plasmid template in
the dilutions. The copy numbers for all unknown
samples were determined by LightCycler software
version 5.32, according to the calibration curve. To
minimize variability due to differences in the RT
efficiency and RNA quality between samples, data
were normalized by dividing copy number of the
target cDNA by the copy number of b-actin.
Quantitative results are presented as copies of target
gene per 105 copies of b-actin. Each sample was run
in duplicate, together with known dilutions of
respective plasmid cDNA ranging from 106 to 102
copies and the appropriate non-template controls for
each Lightcycler run.
2.3. In situ hybridization detection of immune-related
genes mRNA
Fish aged 1, 2, 3, 4 and 6 wpf were fixed in 4%
(w/v) paraformaldehyde in phosphate-buffered saline
(PBS) overnight at 4 8C. After fixation, samples were
washed twice in PBS and dehydrated in ascending
series of ethanol to 100% ethanol before clearing
overnight in Histocleare. The samples were
embedded in paraffin wax at 60 8C and were sectioned
at 5 – 7 mm on a microtome (Reichert-Jung). The
sections were spread on 3-aminopropyl-triethoxysila-
ne(TESPA)-treated slides and dried completely at
37 8C overnight. The sections were dewaxed in two
changes of fresh Histocleare before rehydrating in a
descending ethanol series from 100% ethanol to
diethylpyrocarbonate(DEPC)-treated water and were
used for ISH.
Digoxigenin (DIG)-labeled RNA sense and anti-
sense probes for zebrafish Ikaros, Rag-1, TCRAC,
IgLC-1, IgLC-2 and IgLC-3 were synthesized with
the DIG RNA labeling Kit (Roche Applied Science)
using their respective cDNA plasmid as template. The
Ikaros RNA antisense probe was transcribed with T7
polymerase using EcoRI-digested cDNA plasmid
pIkaros as template while the sense probe was
transcribed with T3 polymerase using BamHI-
digested pIkaros as template. The Rag-1 RNA
antisense probe was transcribed with T7 polymerase,
using BamHI-digested plasmid pRag1 as template
while the sense probe was transcribed with T3
polymerase using XbaI-digested pRag1 as template.
The TCRAC RNA antisense probe was transcribed
with T3 polymerase using HindIII-digested pTCRAC
as template while the sense probe was transcribed
with T7 polymerase using BamHI-digested pTCRAC
as template. The IgLC-1 and IgLC-2 RNA antisense
probes were transcribed with T3 polymerase using
HindIII-digested plasmids pIgLC-1 and pIgLC-2 as
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
respective templates while corresponding sense
probes were transcribed with T7 polymerase using
BamHI-digested pIgLC-1 and pIgLC-2 as templates,
respectively. The IgLC-3 RNA antisense probe was
transcribed with T7 polymerase using BamHI-
digested pIgLC-3 as template while the sense probe
was transcribed with T3 polymerase using HindIII-
digested pIgLC-3 as template.
ISH of tissue-sections with sense and antisense
RNA probes was performed essentially as described
by Jowett [19] with a few modifications. Rehydrated
tissue-sections were fixed in 4% (w/v) paraformalde-
hyde and subsequently washed three times in PBS.
The sections were then treated in 0.2 M HCl for
10 min, after which they were rinsed three times in
TBS before incubating in two changes of 0.5% (v/v)
acetic anhydride for 5 min each change. Sub-
sequently, the sections were washed three times in
TBS for 3 min each and treated with proteinase-K
(Sigma) at a concentration of 10 mg/ml proteinase-K
and 2 mM CaCl2 in TBS at 37 8C for 10 min. The
proteinase-K treatment was terminated by two rinses
in PBS and incubation in PBS at 4 8C for 10 min. The
sections were prehybridized in a fresh hybridization
buffer (50% (v/v) formamide, 5 £ SSC, 500 mg/ml
Yeast tRNA, 0.5% (v/v) Tween-20, 50 mg/ml heparin,
10 mM citric acid, 10% (w/v) dextran sulphate) at pH
6.0 for 2 h at room temperature. Hybridizations with
respective RNA probes (1 mg/ml) in hybridization
buffer were at 60 8C overnight. Following hybridiz-
ation, the sections were washed twice in 2 £ SSC and
twice in 0.2 £ SSC for 30 min each at 60 8C.
In order to increase the sensitivity of detection, the
Fluorescent Antibody Enhancer Set for DIG Detec-
tion kit (Roche Applied Science) was used. Both the
first (monoclonal anti-DIG) and second (anti-mouse
Ig labeled with DIG) antibodies were from the Kit,
however we have replaced the third antibody (anti-
Dig labeled with Fluorescein) from the Kit with an
anti-DIG labeled with Alkaline Phosphatase (Roche
Applied Science). The antibody system was first
tested by detecting Rag-1-positive cells in the thymus
and compared with using a direct single anti-DIG
labeled with Alkaline Phosphatase, as well as with
negative controls using sense probe, without probe or
without the third antibody. It was found that the
antibody system worked well and by controlling the
substrate incubation/color development time, no
13
increasing background was observed (data not
shown). Prior to antibody incubation, the hybridized
sections were blocked with the blocking solution
provided by the kit in PBST for 1 h at room
temperature. After blocking, the sections were
incubated in a 1:50 dilution of a mouse monoclonal
anti-DIG Ig for 1 h at room temperature before
washing three times in PBST for 5 min each wash.
Subsequently, the sections were incubated with 1:50
dilution of an anti-mouse Ig labeled with DIG for 1 h
at room temperature followed by three washes in
PBST for 5 min each wash. Anti-digoxigenin Ig
conjugated with alkaline phosphatase was diluted
1:500 in PBST and incubated with the sections for 1 h
at room temperature. After incubation, the sections
were treated as above with four washes in PBST
followed by another four washes in NTMT buffer
(100 mM NaCl, 50 mM MgCl2·6H20; 100 mM Tris;
0.1% (v/v) Tween-20; pH 9.5) for 5 min each wash.
Subsequently, the sections were incubated in NBT-
BCIP substrate in NTMT buffer with 10 mM
levamisole, at room temperature in the dark until
optimal color development. The reaction was stopped
by washing three times in PBST and refixed in 4%
(w/v) paraformaldehyde before being counterstained
with water-soluble eosin and mounted in Aqua-
mounte (BDH), or dehydrated for alcoholic eosin
counterstaining and DPXe (BDH) mounting. A total
of four fish per age group were examined.
2.4. Detection of zebrafish immunoglobulin in serum
and whole fish extracts by affinity purification, SDS-
gel electrophoresis and Western blot analysis
Blood was collected from the dorsal aorta of adult
wild-type zebrafish with heparinized capillary tubes
(Oxford) after cutting the tail completely off at a
location midway between the anal fin and the base of
the caudal fin with a sharp scalpel. The blood was
pooled into 1.5 ml centrifuge tubes and allowed to clot
while incubating in ice. As the developing young fish
were too small to be bled, whole fish aged 2, 3, 4 and
6 wpf were macerated and homogenized in ice-cold
PBS at a concentration of approximately 0.1 g
tissue/100 ml. Both the clotted blood and tissue
homogenates were centrifuged at 500g for 5 min at
4 8C to remove tissue material and at 14,000g for
5 min at 4 8C to remove particulate material before
14 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
collecting the serum and supernatant (whole fish
extract) to be stored at 2 80 8C until used. A total of
9 ml of serum pooled from about 3000 adult fish and
three batches of 1 –1.5 ml whole fish extract for each
age group were collected and used in this study.
Zebrafish Ig was purified by affinity chromatography
using protein-A Sepharosew (Amersham Pharmacia) as
described by Scapigliati et al. [20]. Briefly, 1 volume of
pooled serum or whole fish extract was diluted with 2
volume of PBS, 0.2 mm-filtered (Millipore), and
incubated in a PBS-equilibrated 1.2 ml protein A
Sepharosew column for 2 h at 4 8C, before washing
the column with 30 column volumes of PBS. Bound
proteins were recovered by eluting with 10 column
volumes of glycine–HCl (0.1 M, pH 2.5), whereby
1 ml fractions were collected and mixed with 50 ml of
Tris pH 10.5 to obtain a final pH of 8.0. The absorbance
of each fraction was determined at 280 nm wavelength.
The fractions from the fish serum and whole fish extract
samples were analyzed by sodium dodecyl sulfate-
polyacrylamide gel electrophoresis (SDS-PAGE) with
12.5% resolving gel and 4% stacking gel, under
reducing condition and protein bands were detected
by silver staining using standard protocol or Western
blotting using a polyclonal rabbit anti-zebrafish Ig
heavy chain antiserum. This rabbit antiserum was
produced from sepharose-Aw purified zebrafish serum
Ig, whereby the heavy chain band had been excised
from polyacrylamide gels, homogenized and eluted in
PBS before mixing with equal volume of complete
Freund’s adjuvant, Titermax (Sigma) and injecting
three times intramuscularly, with one month interval
between each injection, into New Zealand white rabbits
as described in detail by Crosbie and Nowak [21].
Western blots were performed by transblotting the
separated proteins from the polyacrylamide gel onto
a nitrocellulose membrane with the Mini Protean II
set (Bio-Rad). After transblotting, the membrane was
blocked with 1% (w/v) casein in PBST at 37 8C for
1 h before incubating in 1:2000 dilution of the rabbit
anti-zebrafish Ig heavy chain antiserum for another
1 h at room temperature. After four washes in PBST,
the membrane was incubated in 1:10,000 dilution of
goat anti-rabbit Ig conjugated with alkaline phos-
phatase (Sigma) for 1 h at room temperature and
washed four times in PBST before detecting with
substrate BCIP/NBT (Roche Applied Science) in
Tris/Ca2þ/Mg2þ buffer (100 mM Tris-base; 100 mM
NaCl and 5 mM MgCl2) at pH 9.5. Reaction was
stopped by washing the membrane in distilled water
when color development was optimum.
2.5. Experimental design for determining the onset
of humoral immunity, sampling procedure and tests
for antibody production
The experimental design was as described by Tatner
[22] with few modifications. Starting at 2 wpf, fish were
either immunized by immersion for 1 h with a thymus-
dependent antigen, human gamma globulin (HGG)
(Sigma) at a dose of 35 mg/l (group A) or immersed in
water only and set up as control (group B). The fish were
maintained for 3 weeks (wk), thereafter half the number
of fish from each group was sampled. The remaining
fish in groups A and B were immunized with HGG as
before and a further non-immunized control group (C)
of fish from the same batch and age (relatively similar
size) was set up. All three groups were maintained for a
further 3 wk and then sampled. Hence, group A
received two immunizations of HGG to test for primary
and secondary response while group B acted as a
control to test that any heightened response in group A
following second immunization was a true memory
effect and not simply a reflection of the increasing
maturity of the response as the fish grew. Group C acted
as a control to test for any background or non-specific
reaction when assayed. The experiment was repeated
with the same batch of fish aged 4 and 6 wpf.
The experimental set-up for thymus-independent
antigen was similar as above except that formalin-
killed bacteria, Aeromonas hydrophila (strain ATCC
7966, grown 24 h in Tryptic Soy Broth at room
temperature) at a concentration of 109 cells/ml was
used as T-independent antigen to immunize fish
(group D) by immersion for 1 h. A non-immunized
control group (E) was also set up and both group D
and E were maintained for 3 wk before half the
number of fish from each group were sampled. The
remaining fish in group D and E were immunized in
ATCC 7966 as before and a further non-immunized
group F was set up as a control. All the three (D –F)
groups of fish were maintained for another 3 wk and
then sampled. The experiment was repeated with the
same batch of fish aged 4 and 6 wpf.
Groups of five fish were macerated and homogen-
ized at an approximate concentration of 0.1 g
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
tissue/100 ml of ice-cold PBS, centrifuged and the
whole fish extracts collected and stored as described
above. Although this method of sampling introduces a
dilution factor to any titer obtained, titers between
groups could be compared, as the dilution factor
remained constant throughout the experiments. It also
has the advantage of including the spleen, kidney, gut
and skin, other than the serum for assaying of any
antibodies resulting from the direct immersion immu-
nization. The appropriate samples were tested against
HGG and A. hydrophila, using the passive hemagglu-
tination and bacterial agglutination tests, respectively.
The passive hemagglutination assay using tanned rabbit
erythrocytes coated with HGG was performed on V-
bottom microtiter plates (Greiner), in triplicate, as
described by Garvey et al. [23]. The bacterial
agglutination assay was performed on U-bottom
microtiter plates (Greiner), in triplicate, as described
by Roberson [24]. Briefly, 25 ml of the whole fish
extract from each sample was serially diluted two-fold
in PBS in the appropriate microtiter plates before
mixing gently with another 25 ml of 2% (v/v) HGG-
coated rabbit red blood cells or 109 cells/ml formalin-
killed bacteria strain ATCC 7966. Appropriate negative
controls using PBS or non-coated rabbit red blood cell
were set up. Plates were incubated overnight at room
temperature before titers were recorded as the last well
in which visible agglutination occurred. Antibody titers
were expressed as 2 log2 ^ standard deviation (SD).
2.6. Statistical analysis
Significant differences of antibody titer between the
appropriate groups were either tested using Student’s t
test or Analysis of Variance (ANOVA) followed by
Duncan’s multiple comparison test. A value of
P , 0:05 was considered to be statistically significant.
3. Results
3.1. Expression profiles of immune-related genes
detected by conventional two-step RT-PCR and
quantitative real-time PCR
As detected by the conventional RT-PCR method,
the expression profiles of all the immune-related
genes, except Ikaros, showed a similar trend. Target
15
amplicons first appeared as a very faint band at a
certain early age group indicating very low expression
levels and subsequently the intensity of the ethidium
bromide stained bands gradually increased as the
expression level increased with age. Ikaros, however,
had a slightly different expression profile whereby the
target amplicon was first detected moderately at 1 dpf
and only slight increases of staining intensity were
observed from 21 dpf onwards after 30– 40 cycles of
amplification. These results suggest that the expression
level of Ikaros was moderate and only a relatively
small change of expression level occurs between the
early and later stages of development (Fig. 1a). PCR
amplification of other immune-related genes was
carried out for 30 cycles in order to avoid saturation
so that the qualitative differences of the staining
intensity between groups could be better observed. The
target amplicon for Rag-1 was detected with moderate
intensity as early as 3 dpf suggesting relatively high
expression level even at such an early age and by
14 dpf onwards the bands were intensely stained
(Fig. 1b). Low level of TCRAC expression was
detected as early as 3 dpf and the expression level
increased gradually as moderately intense bands were
observed between 7 and 17 dpf before highly intense
bands appeared from 21 dpf onwards (Fig. 1c). All
IgLC isotype target amplicons were detected weakly
by 3 dpf (Fig. 1d– f) while strong signals for all three of
the IgLC isotypes were observed only between 21 and
28 dpf and onwards. For b-actin, similar strong signals
represented by highly intense bands were observed
throughout all age groups tested, suggesting constant
expression levels at the different stages of development
used in this study (Fig. 1g). This was further confirmed
by quantitative real-time PCR which recorded a
relatively constant expression level for all the 36
samples used in this study, ranging from 1.1 £ 106 to
3.2 £ 106 copies with a mean ^ standard deviation
(SD) of (2.3 ^ 0.5) £ 106 copies, thus indicating the
suitability of this house-keeping gene as an internal
control and a reference for normalization.
In all quantitative real-time PCR runs, melting
curve analyses were performed and single specific
melting peaks were observed indicating amplification
specificity (data not shown). The specificity of the
quantitative real-time PCR products was further
confirmed by agarose gel electrophoresis, whereby
the amplicons were of the estimated size, similar to
16 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28

Fig. 1. Qualitative gene expression profile generated by conventional RT-PCR analysis of (a) Ikaros, (b) Rag-1, (c) TCRAC, (d) IgLC-1, (d)
IgLC-2, (e) IgLC-3 and (f) b-actin, from zebrafish aged 1–42 day post-fertilization (dpf). Two micrograms of total RNA from each age group
was DNase 1-digested, reverse transcribed with oligo (dT) as primer and equivalent aliquots of the first strand cDNA reactions were then
amplified using respective specific sets of primers by PCR (40 cycles for Ikaros and 30 cycles for all other genes). Respective amplicons were
detected by electrophoresis in 2.0–2.5% agarose gel stained with ethidium bromide. Expression profiles for the respective genes were similar
for three batches of total RNA. Positive control signals were amplified using the respective cDNA plasmids as templates and negative controls
(without reverse transcription and water blank) were negative (not shown).

the conventional PCR products (Fig. 2). Quantitative
real-time PCR confirms the moderate expression level
of Ikaros, with a mean ^ SD of (2.6 ^ 0.7) £ 101
copies/105 followed by a gradual increase of more than
two-fold between 28 and 42 dpf. However, in adult fish
aged 105 dpf, the Ikaros expression level had
decreased to the level equivalent to 1 dpf. In contrast,
Rag-1 expression level at 3 dpf was relatively high
with a mean ^ SD of (6.1 ^ 0.7) £ 101 copies/105
copies b-actin, rapidly increasing 10-fold by 17 dpf to
(6.5 ^ 1.1) £ 102 copies/105 copies b-actin and there-
after leveling between (7 and 10) £ 102 copies/105
copies b-actin by 42 dpf (Fig. 3). However, in adult fish
aged 105 dpf, the expression level of Rag-1 had
dropped markedly to (6.3 ^ 0.8) £ 101 copies/105
copies b-actin equivalent to the expression level at
3 dpf. The early expression level of TCRAC detected
at 3 dpf was low with a mean ^ SD of 2.0 ^ 0.7
copies/105 copies b-actin, rapidly increasing more
than 10-fold to (3.8 ^ 1.6) £ 101 copies/105 copies b-
actin by 7 dpf. Thereafter TCRAC expression levels
increased gradually by 10-fold to (3.2 ^ 0.9) £ 102
copies/105 copies b-actin by 35 dpf which is within the
expression level of adult fish aged 105 dpf
[(3.3 ^ 1.3) £ 101 copies/105 copies b-actin]. The
early expression levels of IgLC-1, IgLC-2 and IgLC-3
were low with a mean ^ SD of less than 1 copy/105
copies b-actin, 2.5 ^ 0.6 copies/105 copies b-actin
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
Fig. 2. Specific amplicons of Ikaros (156 bp), Rag-1 (144 bp),
TCRAC (155 bp), IgLC-1 (135 bp), IgLC-2 (160 bp), IgLC-3
(131 bp) and b-actin (201 bp) generated from Lightcyclere real-
time PCR system (Roche Applied Science) using the respective
cDNA plasmid as templates. The amplicons were detected by
electrophoresis in 2.5% agarose gel stained with ethidium bromide.
Note the specificity of each amplicon and compared them with Fig. 1
(positive control lane). Samples amplified from first strand cDNA
templates using the Lightcyclere real-time PCR system have
similar specificity (data not shown).
and 7.0 ^ 1.5 copies/105 copies b-actin, respectively,
at 3 dpf and thereafter increasing steadily. By 21 dpf,
the expression levels for all three IgL isotypes
increased more than 10-fold and by 42 dpf the
expression levels were within the adult range, about
100-fold above the respective initial expression levels.
The overall IgLC-1 expression level is about two- to
three-fold lower than IgLC-2 and IgLC-3 in compar-
able age groups. In summary, by 4 – 6 wpf, the
expression level of TCRAC and IgLC isotypes
resembled that of a mature functioning immune
system.
3.2. In situ hybridization detection of immune-related
genes mRNA
ISH was done using the six specific antisense RNA
probes on head kidney sections while only antisense
17
RNA IgLC isotype probes were performed on the
thymic sections. ISH localization of Ikaros, Rag-1 and
TCRAC mRNAs in developing zebrafish thymus has
been reported elsewhere (1, 6, 7, 18). Immunodetec-
tion was carried out with an antibody enhancer DIG
detection kit and by controlling the substrate incuba-
tion/color development duration, signal detection was
enhanced without increasing the background. The
staining was cell associated with little or no unspecific
background. Controls hybridized with DIG-labeled
sense RNA probes, incubated without probe or
developed without the specific anti-DIG antibodies,
were all negative (not shown).
An unequivocal hybridization signal for Ikaros
could not be demonstrated in the head kidney sections
from all age groups ðn ¼ 4Þ used in this study. As for
Rag-1, few scattered positive cells were detected in
head kidney at 2 wpf in all the fish examined (Fig. 4a).
Some background staining was observed in the head
kidney stromal cells indicating maturation of the
alkaline phosphatase-positive reticular cells which
had been reported to correlate with the appearance of
B-lymphocytes in the head kidney of rainbow trout
[3]. Between 3 and 4 wpf, clusters of Rag-1 positive
cells were observed and appeared to be increasing
markedly (Fig. 4b and c). The hybridization signal for
Rag-1 positive cells was strong indicating a high
abundance of transcripts present in these cells
suggesting that active V(D)J recombination is occur-
ring in the maturing head kidney lymphocytes during
this period. TCRAC-positive cells were difficult to
detect although occasionally, cells with weak and thin
cytoplasmic hybridization signals surrounding a
round nucleus and of lymphocyte-like morphology
were observed at 3 wpf onwards (not shown). IgLC
isotype-positive cells (all three isotypes) were
detected by 3 wpf, albeit with weak to moderate
hybridization signals (Fig. 4d). Clusters of IgLC
isotype-positive cells with moderate to strong hybrid-
ization signals were observed by 4 wpf, and positive
cells with intense signals suggestive of plasma cells
were observed by 6 wpf clustering near the vascular
system (Fig. 4e and f). Serial sections of the head
kidney hybridized with the three IgLC isotype probes
were examined with regard to the distribution of
positive cells expressing different constant light chain
isotypes. These comparisons revealed no distinct
distribution pattern of the respective IgLC-positive
18 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28

Fig. 3. Quantitative expression profiles of Rag-1, TCRAC, IgLC-1, IgLC-2 and IgLC-3, from zebrafish aged 1–42 day post-fertilization (dpf)
generated from the Lightcyclere real-time PCR system. Two micrograms of total RNA from each age group (n ¼ 3 batches of total RNA for
each age group) was DNase 1-digested, reverse transcribed with oligo (dT) as primer and equivalent aliquots of the first strand cDNA reactions
were then amplified (40 cycles) using respective specific sets of primers by the Lightcyclere real-time PCR system. Quantitative results are
presented as mean (^ standard deviation) copies of target gene per 105 copies of b-actin after normalization. Each sample was run in duplicate,
together with known dilutions of respective plasmid cDNA ranging from 106 to 102 copies and the appropriate non-template controls for each
Lightcycler run. In all quantitative real-time PCR runs, melting curve analyses were performed and single specific melting peaks were observed
indicating amplification specificity, and that non-template controls were negative (data not shown). Note that the Rag-1 expression profile can
generally be divided into three distinct phases (expanding, maintaining and declining) indicating the change in the population of maturing
lymphocyte from late embryo to adult fish while the increasing expression level of TCRAC and IgLC isotypes suggests an increasing population
of mature lymphocytes.

cells although some regions were observed to be
positive for one isotype and negative for another
(Fig. 4g –i). Although no quantification of the positive
cells was done, it was apparent that IgLC-2 and IgLC-
3-positive cells were more abundant than IgLC-1
positive cells. In the thymus, moderate to strong IgLC
hybridization signals were detected in the cortex
region at 3 wpf (Fig. 4j) but not at 1 –2 wpf. By 6 wpf,
moderate IgLC hybridization signals were detected in
the cortex –medulla boundary (Fig. 4k).
3.3. Detection of zebrafish immunoglobulin in serum
and whole fish extracts by affinity purification, SDS-
gel electrophoresis and Western blot analysis
Protein A-Sepharosew immuno-affinity column
chromatography has been shown to be effective
in isolating Ig from the serum of several fish species
[20,21,25 – 28] including cyprinids such as carp and
goldfish which are closely related to zebrafish. Using
this method, zebrafish Ig was successfully isolated.
The elution profile monitored at 280 nm indicated that
the protein from serum was eluted into fraction 2 and 3
while the protein from whole fish extract was eluted
into fraction 2 only (Fig. 5). The difference is likely due
to the higher load for serum. The protein was further
confirmed to be Ig by SDS-PAGE analysis when it was
separated into only two main bands with molecular
weights of about 84– 86 and 20– 24 kDa correspond-
ing to the heavy chain and light chain of typical teleost
Ig (Fig. 6a). Polyclonal rabbit antiserum against the
zebrafish Ig heavy chain was produced and was able to
detect, by Western blot analysis, the corresponding Ig
heavy chain band in protein A purified serum Ig and in
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28 19

Fig. 4. Expression of Rag-1 in head kidney (a –c), IgLC isotypes in head kidney (d–i) and IgLC isotypes in thymus (j, k). Tissue-section in situ
hybridization (ISH) with antisense RNA Rag-1 on head kidney aged (a) 2, (b) 4 and (c) 6 weeks post-fertilization (wpf). Note some background
staining observed in (a) suggesting maturation of the alkaline phosphatase-positive reticular cells in the head kidney which has been reported to
correlate with the appearance of B-lymphocytes [3]. Note increasing Rag-1 positive cells in (b) and (c). Tissue-section ISH with antisense RNA
of (d) IgLC-1 on head kidney aged 3 wpf, (e) IgLC-2 on head kidney aged 4 wpf, and (f) IgLC-3 on head kidney aged 6 wpf. Note (dotted box in
f) positive cells with intense signal clustering near the vascular system suggesting plasma cells. Serial sections of head kidney aged 4 wpf
probed with antisense RNA of (g) IgLC-1, (h) IgLC-2, and (i) IgLC-3. Note some regions (dotted box) which stained positive for one isotype but
negative for another. Tissue-section ISH with antisense RNA of (j) IgLC-3 on thymus aged 3 wpf, and (k) IgLC-2 on thymus aged 6 wpf. Note
the strong hybridization signal localized to the cortex region which is similar to the Rag-1 hybridization pattern [18]. Note the moderate and less
abundant hybridization signals which appear to localize near the cortex–medulla boundary. The sections in panels (a)–(c), (e), (f) were
mounted in DPXe (BDH) while the sections in panel (d), (g) –(k) were mounted in Aquamounte (BDH) and counterstained in either alcoholic
eosin or water-soluble eosin, respectively. Controls hybridized with DIG-labeled sense RNA probes, incubated without probe or developed
without the specific anti-DIG antibodies, were all negative (not shown) (scale bar ¼ 30 mm).
20 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28

Fig. 5. Elution profiles of zebrafish immunoglobulin (Ig) from serum and whole fish extracts (aged 2, 3, 4, and 6 weeks post-fertilization (wpf))
using Protein A affinity chromatography column. Note that Ig elution is mainly in the second and third (1 ml) fractions of the adult serum and
only in the second fraction for the whole fish extracts. The difference is likely due to the higher relative load for serum.

whole serum, although in the latter, other contami-
nating protein bands were also detected (not shown).
As the detection of these contaminating proteins could
generate false positives in immunohistochemical study
and enzyme-linked immunoassay (ELISA), these
assays were not further developed in this study.
Nevertheless, the antiserum could be used to reveal
and confirm the presence of Ig heavy chain in protein A
purified samples by Western blot analysis. Therefore,
using the antiserum in Western blot analysis, the
presence of Ig heavy chain was detected in all pooled
sample extracts ðn ¼ 3Þ from fish aged 4 and 6 wpf
(Fig. 6b). No detectable level of Ig was observed in all
the extracts from fish aged 2 and 3 wpf, even when the
blots were over-stained. The results were still negative
when the sample extracts from fish aged 2 and 3 wpf
were concentrated 10-fold, by freeze-drying and
reconstituting in 10-fold less volume (not shown). As
the whole process of Ig isolation employed in this
study, from extraction to affinity purification, favors
the isolation of secreted Ig, the results indicate that
secreted Ig was only detectable by Western blotting at
4 wpf onwards.
3.4. Onset of humoral immunity
As shown in Table 2, zebrafish aged 2 wpf which
had been immunized with either HGG (group A) or
formalin-killed bacteria Aeromonas hydrophila
(group D) showed no detectable antibodies to the
respective T-dependent and T-independent antigens
when tested at 5 wpf. Similarly, after reimmunization
with HGG at 5 wpf, no significant ðp . 0:05Þ
response to HGG was detected when tested at 8 wpf.
In zebrafish reimmunized with formalin-killed A.
hydrophila (group D) at 5 wpf, a weak response was
observed but was not significant ðp . 0:05Þ when
compared with the singly immunized control (group
E). Nevertheless, antibody titres for both groups D
and E were significantly increased compared to the
non-immunized group F, suggesting that a primary
response had occurred.
Zebrafish which received immunization with HGG
(group A) at 4 wpf did not show any significant ðp .
0:05Þ response compared to controls (group B) when
tested at 7 wpf (Table 3). Zebrafish immunized with
formalin-killed A. hydrophila (group D) at 4 wpf,
showed a weak positive primary response ðp , 0:05Þ
when compared to control (group E) when tested at
7 wpf. After immunization at 7 wpf with HGG, group
A and group B showed a heightened primary response
at 10 wpf, not significantly ðp . 0:05Þ different when
compared with each other but significantly ðp , 0:05Þ
different when compared to unimmunized controls
(group C). After a second immunization with
formalin-killed A. hydrophila at 7 wpf and testing at
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
Fig. 6. (a) SDS-PAGE analysis in 12.5% gel under reducing
conditions detected zebrafish serum Ig heavy chain (84–86 kDa)
and light chain (20–24 kDa) in elution fractions 2 and 3 from a
Protein A affinity chromatography column. The gel was silver
stained. (b) Western blot analysis showing detectable levels of
zebrafish Ig heavy chain in whole fish extract (aged 4 and 6 wpf)
after protein-A affinity purification. From left, prestained marker,
serum Ig after protein-A purification (positive), whole fish extract
from zebrafish aged 2, 3, 4, and 6 weeks post-fertilization (wpf).
Rabbit anti-zebrafish Ig heavy chain was used at 1:2000 dilution and
pre-immunized rabbit sera used at similar dilution was negative (not
shown).
10 wpf, group D demonstrated a true secondary
response with memory effect as it was significantly
ðp , 0:05Þ different from the primary response of
control group E which was immunized with formalin-
killed A. hydrophila at 7 wpf.
Table 4 shows that zebrafish receiving primary
immunization with either HGG (group A) or formalin-
killed A. hydrophila (group D) at 6 wpf, demonstrated
a significant ðp , 0:05Þ primary response compared to
their respective controls (group B and E), when tested
at 9 wpf. After a second immunization at 9 wpf with
HGG (group A) or formalin-killed A. hydrophila
21
(group D), a significant ðp , 0:05Þ secondary
response was obtained at 12 wpf when compared to
their respective control groups B and E which were
immunized with HGG and formalin-killed A. hydro-
phila, respectively, at 9 wpf.
4. Discussion
The expression of the immune-related genes used
in this study is of significance for indicating the state
of lymphopoiesis and maturation of the immune
system during development in zebrafish. The Ikaros
gene encodes a transcription factor, which in mice has
been shown to be essential for the correct differen-
tiation of B and T lymphocytes [29]. In zebrafish, it
has been shown to be a suitable marker for lymphoid
progenitors during early development [4,6,7]. The
early moderate expression level of Ikaros at 1 dpf
agrees with the report of Willett et al. [7] that early
Ikaros expression was first detected by ISH in the
lateral mesoderm at the 14 somite stage (about 16 h
post-fertilization). As Ikaros encodes a transcription
factor, its expression is likely well-regulated and
transient in a limited pool of lymphoid progenitors;
rapidly down-regulated when Ikaros-expressing cells
switch over to Rag-1-expression, hence limiting any
further increase of Ikaros expression. This may be the
reason for the relatively small change detected in the
expression level of Ikaros throughout the different age
groups used in this study. Willett et al. [7] reported
that in adult zebrafish, Ikaros can be detected in
thymus, kidney, spleen, intestine, heart and brain cells
as indicated by Southern blotting of RT-PCR products
from the tissues. The authors [7] did detect a low level
of Ikaros expression in the head kidney of a 96 h
embryo using whole mount ISH but indicated that
they had to digitally enhance the staining which
suggests that the signal was very weak. However, we
were not able to detect convincingly the presence of
Ikaros-positive cells in the head kidney sections,
although very weak signals were observed occasion-
ally, probably due to the less sensitive ISH technique
and low copy number of Ikaros mRNA per cell.
The relatively high onset expression of Rag-1
agrees well with the strong robust hybridization
signals reported in ISH studies [1,18] suggesting
that Rag-1 expression per cell is high and it is
22 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28

Table 2
Antibody response of zebrafish immunized at 2 weeks post-fertilization (wpf)

Type of antigen Immunization at 2 wpf. Response at 5 Immunization at 5 wpf. Response at 8

wpf. Mean titer wpf. Mean titer
(2log2 2 ^ SD) (2log2 2 ^ SD)

T-dependent antigen Group A (HGG 0 Group A (HGG second 0.58 ^ 0.49

human gamma first immunization) immunization)
globulin (HGG)
Group B (non-immunized) 0 (Group B) (HGG first 0.50 ^ 0.45
immunization)
Group C (non-immunized) 0.42 ^ 0.38
T-independent antigen Group D (AH first 0.7 ^ 0.3 Group D (AH second 2.7 ^ 0.8a
Aeromonas hydrophila (AH) immunization) immunization)
Group E (non-immunized) 1.0 ^ 0.9 Group E (AH first 1.9 ^ 0.6a
immunization)
Group F (non-immunized) 0.8 ^ 0.4b

SD, standard deviation and N ¼ 6 groups of five fish per group. a,bGroups D and E are significantly ðp , 0:05Þ different when compared to
group F at 8 wpf.

expressed in a relatively large pool of differentiating
lymphocytes in the thymus. The subsequent rapid
increase of Rag-1 expression levels between 1
and 3 wpf is contributed by the expansion of
the differentiating thymocyte population reported in
our previous study [18] as well as the expansion of
maturing (Rag-1 expressing) lymphocyte populations
in the head kidney observed between 2 and 4 wpf in
this study. Between 3 and 6 wpf Rag-1 expression
levels reached a maximum, suggesting that during this
period the expansion rate of the maturing lymphocyte
population equals the formation rate of mature
lymphocytes where Rag-1 expression is down-regu-
lated and genes coding for TCRs and Igs are
expressed. The increase of TCRAC and IgLC-positive
cells in the thymic medullary region and in head
kidney, respectively, from 3 to 4 wpf onwards (Ref.
[18] and present study), suggests that mature lym-
phocytes are increasing in the immune system during
this period. Furthermore, the expression levels of
TCRAC and IgLC isotypes were within the range of
adult (aged 15 wpf) levels by 4– 6 wpf, resembling

Table 3
Antibody response of zebrafish immunized at 4 weeks post-fertilization (wpf)

Type of antigen Immunization at 4 wpf Response at 7 Immunization at 7 wpf Response at 10

wpf. Mean titer wpf. Mean titer
(2log2 2 ^ SD) (2log2 2 ^ SD)

T-dependent antigen Group A (HGG first 0.50 ^ 0.55 Group A (HGG second 2.54 ^ 0.95a
human gamma immunization) immunization)
globulin (HGG)
Group B (non-immunized) 0.33 ^ 0.41 (Group B) (HGG first immunization) 1.80 ^ 0.85a
Group C (non-immunized) 0.44 ^ 0.25b
T-independent antigen Group D (AH first 1.03 ^ 0.17c Group D (AH second immunization) 3.00 ^ 0.38e
Aeromonas hydrophila (AH) immunization)
Group E (non-immunized) 0.67 ^ 0.18d Group E (AH first immunization) 2.20 ^ 0.68f
Group F (non-immunized) 1.08 ^ 0.25g

SD, standard deviation and N ¼ 6 groups of five fish per group. a,bGroups A and B are significantly ðp , 0:05Þ different compared to group
C at 10 wpf; c,dgroup D is significantly ðp , 0:05Þ different compared to group E at 7 wpf; e,f,gall groups are significantly ðp , 0:05Þ different
when compared to each other at 10 wpf.
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28 23

Table 4
Antibody response of zebrafish immunized at 6 weeks post-fertilization (wpf)

Type of antigen Immunization at 6 wpf Response at 9 Immunization at 9 wpf Response at 12

wpf. Mean titer wpf. Mean titer
(2log2 2 ^ SD) (2log2 2 ^ SD)

T-dependent antigen Group A (HGG first 1.38 ^ 0.58a Group A (HGG second 4.92 ^ 1.29c
human gamma immunization) immunization)
globulin (HGG)
Group B (non-immunized) 0.50 ^ 0.45b Group B (HGG first immunization) 3.62 ^ 0.52d
Group C (non-immunized) 1.60 ^ 0.77e

T-independent antigen Group D (AH first 1.97 ^ 0.70f Group D (AH second immunization) 3.94 ^ 0.27h
Aeromonas hydrophila (AH) immunization)
Group E (non-immunized) 0.75 ^ 0.35g Group E (AH first immunization) 3.10 ^ 0.38i
Group F (non-immunized) 1.33 ^ 0.31j

SD, standard deviation and N ¼ 6 groups of five fish per group. a,bGroup A is significantly ðp , 0:05Þ different compared to group B at
9 wpf; c,d,eall groups are significantly ðp , 0:05Þ different when compared to each other at 12 wpf; f,ggroup D is significantly ðp , 0:05Þ
different compared to group E at 9 wpf; h,i,jall groups are significantly ðp , 0:05Þ different when compared to each other at 12 wpf.

the expression profile of a mature immune system.
Hence, the increased presence of mature lymphocytes
indirectly suggests rapid down-regulation of Rag-1
has occurred and its rate must equal ‘up-regulation’
(contributed by the on-going expansion of the
maturing lymphocyte population) in order that the
expression level of Rag-1 is maintained at a steady
state between 3 and 6 wpf. The sharp decline of Rag-1
expression levels observed in the adult zebrafish (aged
15 wpf) indicates that the population of maturing
lymphocytes has dropped significantly and this is
consistent with our observation of the involuting
thymus (of similar age) which has a markedly reduced
percentage of Rag-1 positive signals [18]. However,
thymus involution, likely to begin at the onset of
sexual maturation (about 10 – 12 wpf in zebrafish) as
reported in other vertebrates [30,31], has little or no
effect on TCRAC and IgLC isotypes expression
levels.
ISH detection of Rag-1 and IgLC-positive cells in
head kidney at 2 and 3 wpf, respectively, and the
subsequent increase of these cells in the ensuing
weeks indicate that the head kidney is also an organ
for the development of B-lymphocytes, in addition to
the pancreas which has been reported by Danilova and
Steiner [32]. Similarly, these authors [32] reported
that Ig heavy chain (Igm) expression was detected in
head kidney aged 19 dpf, about 5 days later than the
detection of Rag-1 and 2 days earlier than IgLC
isotypes in this study. The temporal differences of
the expression of these genes are not contradictory as
Rag-1 is known to express earlier in maturing
lymphocytes while it is highly possible that IgLC
isotype-positive cells could have been detected earlier
than 3 wpf as fish aged 2 and 3 wpf, and not in
between these ages, were subjected to ISH in this
study. Similarly, although Willett et al. [2] reported
the presence of lymphoblasts and lymphocytes in the
zebrafish head kidney at 3 wpf but not at 2 wpf, their
study could not exclude the possibility of these
lymphoid cells appearing between 2 and 3 wpf as
fish between these ages were not studied. Moreover,
Hansen and Zapata [3] had interpreted Willett et al.’s
results [2] as indicating the appearance of lymphoid
cells in zebrafish head kidney from 2 wpf onwards.
Nevertheless, in an earlier report [1], Rag-1
expression could not be demonstrated in head kidney
at 2 – 3 wpf using whole-mount ISH technique and this
could be due to the inability of the Rag-1 probe to
penetrate sufficiently into the head kidney or the few
scattered Rag-1 positive cells at 2 wpf were unable to
generate sufficient signal to be recognized as positive
by whole mount ISH.
Significant quantitative and qualitative changes in
the expression of Ig mRNA occur during the
differentiation of a B-lymphocyte into an antibody-
secreting plasma cell. In mammals, the copy number
of Ig mRNAs is known to depend on the state of B-
lymphocyte differentiation and it ranges from low
copy numbers in B-cells producing predominantly
24 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
membrane-bound Ig to abundant copy numbers (50 –
300 fold increase) in plasma cells which produce
almost exclusively secreted Ig [33,34]. Because of the
low copy number of Ig transcripts in B-cells compared
to plasma cells, the ISH technique favors the detection
of plasma cells rather than mature B-cells which
produce predominantly membrane-bound Ig and are
detected weakly if at all [35 –37]. Thus the detection
of IgLC isotypes in head kidney cells at 3 wpf
onwards, especially the finding of positive cells with
moderate to strong hybridization signals, suggests that
intermediate B-cells differentiating into plasma cells
and plasma cells themselves may be present. The
detection of Ig in whole fish extracts from fish aged
4 wpf by Western blot analysis confirms the presence
of plasma cells in fish aged 3 wpf onwards. As it has
been shown in mammals [38], that translation of Ig
mRNA into protein follows closely after its transcrip-
tion, it is possible that low levels of secreted Ig may
have been produced by 3 wpf and were not detected
by the approach employed in this study. Similarly in
carp, C. carpio, a cyprinid closely related to zebrafish,
serum Ig levels increased from 3 weeks post-hatching
(wph) onwards coinciding with the appearance of
plasma cells between 3 and 4 wph [39 – 41]. The
increase of plasma cells must have contributed to the
rapid increase of IgLC isotypes expression at 4–
6 wpf, to levels close to the range found in a mature
functioning adult immune system. This is further
supported by the detection of a much stronger Ig
signal from extracts of whole fish aged 6 wpf
compared to fish aged 4 wpf using Western blot
analysis.
The early low expression of IgLC isotype mRNAs
at 3 –10 dpf detected by both RT-PCR methods, prior
to ISH detection of Rag-1 positive cells in head
kidney and IgLC-positive cells in head kidney and
thymus, may be associated with sterile/aberrant
transcripts which are reported to be abundant in
zebrafish [12] as well as in other teleosts [42 – 44].
While the actual basis for the presence of sterile
transcripts is not known, it has been proposed that
they may reflect a prerequisite for the initiation of
DNA rearrangement of TCR and Ig gene loci in
lymphocyte precursors, as is known to occur in
mammals. In mouse, it has been reported that
relatively high levels of Ig heavy and light chain
mRNAs were expressed in a large proportion of
differentiating fetal thymocytes, detected by ISH and
Northern blotting, but the respective proteins could
not be detected by immunohistochemical methods
indicating that the transcripts were sterile [45]. The
expression of these sterile transcripts occurred at a
time when the thymocytes are already actively
rearranging and expressing the T-cell receptor genes
[45]. Besides thymocytes, it has also been reported
that the production of these sterile transcripts is a
common feature of pre-B-lymphocytes and charac-
terizes an ontogenetic stage in early B-lymphocyte
development [46,47]. Pre-B-lymphocytes (as well as
T and myeloid precursors) have been detected very
early in fetal mouse in the aorta –gonad –mesonephros
(AGM) region [48] as well as very early in the fetal
thymus [49]. Furthermore, according to Hansen and
Zapata [3], an early source of B and/or T lymphocytes
may exist in trout before the appearance of thymus
and kidney, as suggested by Rag expression in 10 dpf
(, 23 day embryonic period) embryos [50,51]. In
zebrafish, after ikaros expression has begun during
early development, the presence of pre-B and pre-T-
lymphocytes in the intracellular cell mass (ICM) may
be the source of the early IgLC and TCRAC
transcripts. This evidence suggests that the early and
low expression levels of IgLC isotypes may be
associated with sterile transcripts generated by
differentiating lymphocyte precursors and/or thymo-
cytes. Moreover, Danilova and Steiner [32] have
shown, by PCR amplification of genomic DNA from
whole zebrafish with specific primers, that VDJ
rearrangement in B-lymphocytes has already been
initiated by 4 dpf, suggesting the early presence of
pre-B-lymphocytes. In addition, the authors [32] have
also reported that transcripts corresponding to mem-
brane and secreted IgM were detected in RNA
extracted in whole fish as early as 7 and 13 dpf,
respectively. Similarly, some of the early IgLC
expressions detected by both RT-PCR methods in
this study, may also be associated with membrane Ig
receptors which at low copy number per cell are
weakly detected or undetectable by ISH. The low to
moderate abundance of transcripts for the membrane
TCR also may account for the weak TCRAC
hybridization signal in the head kidney.
However, it is unclear whether the IgLC hybrid-
ization signals observed in the thymus, in this study,
are from differentiating thymocytes expressing sterile
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28 25

IgLC transcripts, differentiating pre-B-lymphocytes
expressing sterile transcripts or differentiating B
lymphocytes expressing functional transcripts.
While we are unable to answer with certainty, the
strong hybridization signal localized to the cortex
region (Fig. 4j) which is similar to Rag-1 hybridiz-
ation pattern [18] seems to suggest their association
with the differentiating thymocytes, while the weak to
moderate and less abundant hybridization signal
which appear to localize near the cortex – medulla
boundary (Fig. 4k) may suggest that they are B-
lymphocytes playing the role of self antigen present-
ing cells together with dendritic cells involved
in tolerance induction as has been reported in mice
[52,53]. In carp, membrane IgM-positive cells
detected by monoclonal antibodies which appeared

Fig. 7. Landmark events during ontogeny, development and maturation of the zebrafish immune system. Data represents the summation of
several studies (Refs. [1,2,6,7,18,32] and present study). The embryo, hatching and early larval periods follow after Kimmel et al. [57] while the
larval–juvenile transitory phase follows after Brown [58].
26 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
in the head kidney by 2 wph were detected in thymus
by 1 month post-hatch [38]. Moreover, it has also
been reported that B-lymphopoiesis occurs in the
thymus of mice and that a significant number of IgM
positive mature B cells developed intrathymically and
are exported from the thymus [54]. Thus, all the above
three possibilities may be occurring in the thymus and
the only way to be certain is to analyze these cells with
different specific cell surface markers which at present
is not possible in zebrafish.
Humoral responses to T-independent (formalin-
killed A hydrophila) and T-dependent antigen (HGG)
were observed in zebrafish immunized at 4 and 6 wpf,
respectively. Studies in rainbow trout and carp [22,55,
56] have shown that antibody production to T-
independent antigens (formalin-killed A salmonicida)
and to T-dependent antigens (HGG) develop at about
4 and 8 wph onwards, respectively. The slightly
earlier maturation of humoral responses observed in
this study may be species-related, as zebrafish
development is known to be rapid, or due to the
higher temperature in raising and immunizing the fish
as compared to the above studies. Temperature is
known to affect immune responses and development
in fish [15].
Finally, we have placed the ontogenetic develop-
ment and maturation events of the zebrafish immune
system in the context of its early life history (Fig. 7). It
was observed that while lymphopoiesis and lym-
phoid-organogenesis, are initiated at the middle to late
embryo period, they remained in their rudimentary
and immature form throughout the early larval stages.
The major maturation events leading to immunocom-
pentence occurred between 2 and 4 wpf (between 6.2
and 14.0 mm total length) coinciding with the larval
to juvenile transitory phase proposed by Brown [58].
This larval –juvenile transitory phase in zebrafish is at
a stage in its life cycle comparable to the anuran
metamorphic phase and both are dependent on thyroid
hormones [58]. The maturation of the immune system
coinciding with the larval –juvenile transitory phase
in zebrafish is likely a developmental strategy based
on the intrinsic link between nutrient availability,
metabolic efficacy, hormonal factors and the deve-
loping immune system. Nevertheless, the changes in
the immune system during the zebrafish larval to
juvenile transitory phase are neither similar nor as
dramatic as in the anuran amphibian [3].
In conclusion, we have shown some landmark
events leading to the maturation of the immune
system in zebrafish and that immunocompetence is
achieved by 4– 6 wpf. These maturation events in the
immune system will be useful for characterizing non-
lethal mutations that affect the development of the
immune system. In addition, the immune-related
genes expression profiles can be used as zebrafish
health reference markers and can further be developed
for use as indicators for developmental immunotox-
icology studies.
Acknowledgements
We thank the National University of Singapore for
providing the research scholarship, research grant (R-
154-00-091-112) and we also acknowledge the
support by a grant from Biomedical Research Council
of Singapore for this study. We also thank Dr Wen
Zilong of Institute of Molecular and Cell Biology,
National University of Singapore for providing the
Ikaros and Rag-1 plasmids. We are grateful to Drs
Gary W. Litman and Robert N. Haire of University of
South Florida, USA, for providing the TCRAC and
IgLC isotypes clones.
References
[1] Willet CE, Cherry JJ, Steiner LA. Expression of zebrafish rag
genes during early development identifies the thymus. Dev
Biol 1997;182:331 –41.
[2] Willet CE, Cortes A, Zuasti A, Zapata AG. Early hematopoi-
esis and developing lymphoid organs in the zebrafish. Dev
Dyn 1999;214:323–36.
[3] Hansen JD, Zapata AG. Lymphocytes development in fish and
amphibians. Immunol Rev 1998;166:199 –220.
[4] Trede NS, Zon LI. Development of T-cells during fish
embryogenesis. Dev Comp Immunol 1998;22:253–63.
[5] Schorpp M, Wiest W, Egger C, Hammerschmidt M, Schlake
T, Boehm T. Genetic dissection of thymus development. In:
Melchers F, editor. Lymphoid organogenesis. Berlin:
Springer; 2000. p. 119– 24.
[6] Trede NS, Zapata AG, Zon LI. Fishing for lymphoid genes.
Trends Immunol 2001;22:302 –7.
[7] Willett CE, Kawasaki H, Amemiya CT, Lin S, Steiner LA.
Ikaros expression as a marker for lymphoid progenitors during
zebrafish development. Dev Dyn 2001;222:694–8.
[8] Nasevicius A, Ekker SC. The zebrafish as a novel system for
functional genomics and therapeutic development appli-
cations. Curr Opin Mol Ther 2001;3(3):224–8.
 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
[9] Takeuchi H, Figueroa F, O’hUigin C, Klein J. Cloning and
characterization of class 1 Mhc genes of the zebrafish,
Brachydanio rerio. Immunogenetics 1995;42:77– 84.
[10] Sultmann H, Mayer WE, Figueroa F, O’Huigin C, Klein J.
Zebrafish Mhc class II a encoding genes: polymorphism,
expression and function. Immunogenetics 1993;38:420– 8.
[11] Sultmann H, Mayer WE, Figueroa F, O’Huigin C, Klein J.
Organization of Mhc class II b genes in the zebrafish
(Brachidanio rerio). Genomics 1994;23:1– 14.
[12] Haire RN, Rast JP, Litman RT, Litman GW. Characterization
of three isotypes of immunoglobulin light chains and T-cell
antigen receptor alpha in zebrafish. Immunogenetics 2000;
51(11):915–23.
[13] Danilova N, Hohman VS, Kim EH, Steiner LA. Immunoglo-
bulin variable-region diversity in the zebrafish. Immunoge-
netics 2000;52(1–2):81–91.
[14] Yoder JA, Mueller MG, Wei S, Corliss BC, Prather DM, Willis
T, Litman RT, Djeu JY, Litman GW. Immune-type receptor
genes in zebrafish share genetic and functional properties with
genes encoded by the mammalian leukocyte receptor cluster.
Proc Natl Acad Sci USA 2001;98(12):6771–6.
[15] Ellis AE. Ontogeny of the immune system in teleost fish. In:
Ellis AE, editor. Fish vaccination. London: Academic Press;
1988. p. 20–31.
[16] Zapata AG, Torroba M, Varas A, Jimenez A. Immunity in fish
larvae. Dev Biol Standardization 1997;90:22–32.
[17] Tatner MF. Natural changes in the immune system of fish. In:
Iwama GK, Nakanishi T, editors. The fish immune system:
organism, pathogen and environment. London: Academic
Press; 1996. p. 255 –87.
[18] Lam SH, Chua HL, Gong Z, Wen Z, Lam TJ, Sin YM.
Morphological transformation of the thymus in developing
zebrafish. Dev Dyn 2002;225:87–94.
[19] Jowett T. Tissue in situ hybridization: methods in animal
development. New York: Wiley; 1997.
[20] Scapigliati G, Romano N, Picchietti S, Mazzini M, Mastrolia
L, Scalia D, Abeli L. Monoclonal antibodies against sea bass
Dicentrarchus labrax (L.) immunoglobulins: immunolocali-
zation of immunoglobulin-bearing cells and applicability in
immunoassays. Fish Shellfish Immunol 1996;6:383–401.
[21] Crosbie PBB, Nowak BF. Production of polyclonal antisera
against barramundi (Lates calcarifer Bloch) serum immuno-
globulin derived from affinity columns containing mannan-
binding protein or staphylococcal protein A. Aquaculture
2002;211:49– 63.
[22] Tatner MF. The ontogeny of humoral immunity in rainbow
trout, Salmo gairdneri. Vet Immunol Immunopathol 1986;12:
93–105.
[23] Garvey JS, Cremer NE, Sussdorf DH. Methods in immu-
nology, 3rd ed. Reading: Benjamin Cummings; 1977.
[24] Roberson BS. Bacterial agglutination. In: Stolen JS, Fletcher
TC, Anderson DP, Robersonvan BS, Muiswinkel WB, editors.
Techniques in fish immunology. Fair Haven: SOS Publi-
cations; 1990. p. 81–6.
[25] Suzuki Y, Tanaka M, Aida K, Hanyu I. Affinity of fish
immunoglobulins to protein A. Nippon Suisan Gakkaishi
1990;56:831.
27
[26] Estevez J, Leiro J, Santamaria MT, Ubeira FM. Isolation and
partial characterization of turbot (Scophthalmus maximus)
immunoglobulins. Comp Biochem Physiol Part A Physiol
1993;41:353–66.
[27] Morrison RN, Nowak BF. Elution of snapper, Pagrus auratas
(Bloch and Schneider), immunoglobulin from a protein A
affinity column yields contamination from the binding ligand.
Bull Eur Assoc Fish Pathol 2000;20:174–8.
[28] Watts M. Isolation and partial characterization from southern
bluefin tuna Thunnus maccoyii Castelnau. Fish Shellfish
Immunol 2001;11:491 –503.
[29] Georgopoulos K, Winandy S, Avitathl N. The role of Ikaros
gene in lymphocyte development and homeostasis. Annu Rev
Immunol 1997;15:155 –76.
[30] Chilmonezyk S. The thymus in fish: development and possible
function in the immune response. In: Faisal M, Hetrick FM,
editors. Annual review of fish diseases, vol. 2. New York:
Pergamon Press; 1992. p. 181 –200.
[31] Turner RJ. Mammals. In: Turner RJ, editor. Immunology: a
comparative approach. Chichester: Wiley; 1994. p. 173–213.
[32] Danilova N, Steiner LA. B-cell develop in the zebrafish
pancreas. Proc Natl Acad Sci USA 2002;99:13711–6.
[33] Brown SL, Morrison SL. Regulation of the production of
secretory and membrane immunoglobulin during lymphocyte
development. Clin Immunol Immunopathol 1989;50(2):
155 –70.
[34] Perry RP, Kelley D. Transcriptional and post-transcriptional
control of immunoglobulin mRNA production during B-
lymphocytes development. Nucleic Acids Res 1986;14:
5431–47.
[35] Pringle JH, Ruprai AK, Primrose L, Keyte J, Potter L, Close P,
Lauder I. In situ hybridization of immunoglobulin light chain
mRNA in paraffin sections using biotinylated or hapten-
labelled oligonucleotide probes. J Pathol 1990;162(3):
197 –207.
[36] Schroder MB, Flano E, Pilstrom L, Jorgensen TO. Localiz-
ation of plasma cells in Atlantic cod (Gadus morhua L.)
tissues; identified by in situ hybridization. Fish Shellfish
Immunol 1998;8(8):565–76.
[37] Stenvik J, Schroder MB, Olsen K, Zapata A, Jorgensen TO.
Expression of immunoglobulin heavy chain transcripts (VH-
families, IgM, and IgD) in head kidney and spleen of the
Atlantic cod (Gadus morhua L.). Dev Comp Immnol 2001;25:
291 –302.
[38] Lamson G, Koshland ME. Changes in J chain and k chain
RNA expression as a function of B-cell differentiation. J Exp
Med 1984;160:488–96.
[39] Koumans-van Diepen JCE, Taverne-thiele JJ, van Rens
BTTM, Rombout JHWM. Immunocytochemical and flow
cytometric analysis of cells and plasma cells in carp (Cyprinus
carpio L.): an ontogenetic study. Fish Shellfish Immunol
1994;4:19–28.
[40] van Loon JJA, van Oosterom R, van Muiswinkel WB.
Development of the immune system in carp, Cyprinus carpio
L. In: Solomon JB, editor. Aspects of developmental and
comparative immunology. Oxford: Pergamon Press; 1981. p.
469–70.
28 S.H. Lam et al. / Developmental and Comparative Immunology 28 (2004) 9–28
[41] van Loon JJA, Secombes CJ, Egberts E, van Muiswinkel WB.
Ontogeny of the immune system in fish: role of the thymus.
Adv Exp Med Biol 1982;149:335–41.
[42] Daggfeldt A, Bengten E, Pilstrom L. A cluster type
organization of the loci of the immunoglobulin light chain
in Atlantic cod (Gadus morhua L.) indicated by nucleotide
sequences of cDNAs and hybridization analysis. Immunoge-
netics 1993;38:199– 209.
[43] Ghaffari SH, Lobb CJ. Structure and genomic organization of
immunoglobulin light chain in the channel catfish. J Immunol
1993;151:6900–12.
[44] Partula S, Schwager J, Timmusk S, Pilstrom L, Charlemagne
J. A second immunoglobulin light chain isotype in the rainbow
trout. Immunogenetics 1996;45:44–51.
[45] Gallagher PF, Miller JFAP. Immunoglobulin gene expression
is a normal differentiation event in embryonic thymocytes. Eur
Immunol 1988;18:183–6.
[46] Nelson KJ, Kelley DE, Perry RP. Inducible transcription of the
unrearranged k constant region locus is a common feature of
pre-B cells and does not require DNA or protein synthesis.
Proc Natl Acad Sci USA 1985;82:5305 –9.
[47] Staudt LM, Lenardo MJ. Immunoglobulin gene transcription.
Annu Rev Immunol 1991;9:373–8.
[48] Ohmura K, Kawamoto H, Fulimoto S, Ozaki S, Nakao K,
Katsura Y. Emergence of T,B, and myeloid lineage-committed
as well as multipotent hemopoietic progenitors in aorta–
gonad–mesonephros region of day 10 fetuses of the mouse.
J Immunol 1999;163:4788– 95.
[49] Kimoto H, Shirasawa T, Taniguchi M, Takemori T. B cell
precursors are present in the thymus during early develop-
ment. Eur Immunol 1989;19:97–104.
[50] Hansen JD, Kaattari SL. The recombination activating gene 2
(RAG2) of the rainbow trout Onchorhynchusmykiss. Immu-
nogenetics 1996;44:203–11.
[51] Hansen JD. Characterization of rainbow trout terminal
deoxynucleotidyl transferase structure and expression. TdT
and Rag1 co-expression define the trout primary and lymphoid
tissues. Immunogenetics 1997;46:365–7.
[52] Mazda O, Watanabe Y, Gyotuku J, Katsura Y. Requirement of
denritic cells and B cells in the clonal deletion of mls-reactive
T cells in the thymus. J Exp Med 1991;173:537 –9.
[53] Inaba M, Inaba K, Hosono M, Kumamoto T, Ishida T,
Muramatsu S, Masuda T, Ikehara S. Distinct mechanisms of
neonatal tolerance induced by dendritic cells and thymic B
cells. J Exp Med 1991;173:549 –59.
[54] Akashi K, Richie LI, Miyamoto T, Carr WH, Weissman IL. B
lymphopiesis in the thymus. J Immunol 2000;164:5221–6.
[55] Manning MJ, Grace MF, Secombes CJ. Ontogenetic aspects
of tolerance and immunity in carp and rainbow trout:
studies on the role of thymus. Dev Comp Immunol 1982;
Suppl 2:75–82.
[56] Mughal MS, Manning MJ. Antibody responses of young carp,
Cyprinus carpio, and grey mullet Chelon labrosus immunized
with soluble antigen by various routes. In: Manning MJ,
Tatner MF, editors. Fish immunology. London: Academic
Press; 1985. p. 313–26.
[57] Kimmel CB, Ballard WW, Kimmel SR, Ulkmann B, Schilling
TF. Stages of embryonic development of the zebrafish. Dev
Dyn 1995;203(3):253 –310.
[58] Brown DD. The role of thyroid hormone in zebrafish and
axolotl development. Proc Natl Acad Sci USA 1997;94:
13011–6.