Ikzf1 regulates embryonic T lymphopoiesis via Ccr9 and Irf4 in zebrafish

Abstract

Ikzf1 is a Krüppel-like zinc-finger transcription factor that plays indispensable roles in T and B cell development. Although the function of Ikzf1 has been studied extensively, the molecular mechanism underlying T lymphopoiesis remains incompletely defined during the embryonic stage. Here we report that the genetic ablation of ikzf1 in mutant zebrafish resulted in abrogated embryonic T lymphopoiesis. This was ascribed to impaired thymic migration, proliferation, and differentiation of hematopoietic stem/progenitor cells (HSPCs). Ccr9a and Irf4a, two indispensable factors in T lymphopoiesis, were the direct targets of Ikzf1 and were absent in the ikzf1 mutants. Genetic deletion of either ccr9a or irf4a in the corresponding mutant embryos led to obvious T cell development deficiency, which was mainly caused by disrupted thymic migration of HSPCs. Restoration of ccr9a in ikzf1 mutants obviously promoted HSPC thymus homing. However, the HSPCs then failed to differentiate into T cells. Additional replenishment of irf4a efficiently induced HSPC proliferation and T cell differentiation. Our findings further demonstrate that Ikzf1 regulates embryonic T lymphopoiesis via Ccr9 and Irf4 and provide new insight into the genetic network of T lymphocyte development.

Introduction

As a key cellular component in adaptive immunity, T lymphocytes are considered to originate from hematopoietic stem cells (HSCs)² through a tightly controlled hierarchy (1, 2). Although common lymphoid progenitors are regarded as the major sources of T cells (3), the identification of lymphoid-primed multipotent progenitors, which are endowed with both myeloid and lymphoid potential, has suggested that myeloid and lymphoid lineages share a common precursor (4). The seeding of the thymus by these hematopoietic progenitor cells is required for the commitment, proliferation, differentiation, and maturation of T lymphocytes (5). Although the processes of T cell differentiation and maturation have been investigated extensively, the molecular mechanisms governing the migration of hematopoietic progenitor cells to the thymus and their differentiation during embryonic stages are poorly understood.

Zebrafish (Danio rerio) are widely used as a model system for studying hematopoiesis because they present several unique characteristics, including extrauterine development and optical transparency. Zebrafish T lymphoid cells result from the definitive hematopoiesis that occurs in the ventral wall of the dorsal aorta (VDA), where hematopoietic stem/progenitor cells (HSPCs) emerge from the aortic endothelium through endothelial hematopoietic transition (6–8). These newly formed HSPCs subsequently migrate to the caudal hematopoietic tissue (CHT), thymus, and kidney, where they proliferate and differentiate to produce distinct blood lineages (9–12). At ∼60 h post-fertilization (hpf), VDA-born hematopoietic precursors appear to start colonizing the developing thymus (13), where they undergo rapid proliferation and differentiation and give rise to mature T cells (13).

The IKAROS family zinc finger 1 (Ikzf1) transcription factor, also known as Ikaros, was initially detected in mouse liver rudiment (14). The Ikzf1 gene generated various spliced isoforms that translated into a complex group of proteins characterized by a DNA-binding domain and protein–protein interaction domain in the N and C terminus, respectively (15, 16). Ikzf1 proteins are detected in various subsets of hematopoietic cells and are especially enriched in lymphocytes (15, 17, 18). Mice carrying distinct Ikzf1 mutant alleles display similar lymphopoiesis defects and other hematopoietic phenotypes (19–26). These studies indicate that Ikzf1 plays critical roles in multiple steps of hematopoietic cell growth, especially in T cell development. Ikzf1 targets the nucleosome–remodeling–deacetylase complex (NuRD) to promote lymphoid priming of HSPCs (27). Ikzf1 cooperates with other members of the IKAROS family (28, 29) to regulate several key molecules in T cell differentiation. Similar to the observations in mice, the zebrafish ikzf1 mutant alleles present a significant early T cell development deficiency (30, 31), suggesting an evolutionarily conserved role in T cell development. Although the function of Ikzf1 has been investigated extensively for decades, the corresponding molecular mechanisms in embryonic hematopoiesis, especially T lymphopoiesis, remain largely undefined.

In this study, two ikzf1 mutant alleles that affect the functional domains were generated. We show that loss of Ikzf1 function in zebrafish impairs thymus migration, proliferation, and differentiation of HSPCs, which results in failure of embryonic T lymphopoiesis. We further reveal that ccr9a and irf4a are directly regulated by Ikzf1 and are functionally indispensable in the T lymphopoiesis process.

Results

Embryonic T lymphopoiesis is abolished in ikzf1 mutants

To explore the function of ikzf1 in early hematopoiesis, including T lymphopoiesis, ikzf1 expression patterns and levels in different embryonic blood cells were examined in zebrafish. Ikzf1 clearly appeared in runx1-GFP⁺ (32) and cmyb-GFP⁺ (33) cells (Fig. S1A) in the VDA regions at 36 hpf. Ikzf1 was then detected in the larval hematopoietic organs (CHT, thymus, and kidney; Fig. S1B) (11). Ikzf1 transcript levels were highest in rag2-DsRed⁺ lymphoid cells (34), followed by CD41-GFP^lo HSPCs (10), mpx-GFP⁺ myeloid cells (35), and gata1-DsRed⁺ erythroid cells (36). In CD41-GFP^hi thrombocytes (37), a limited amount of ikzf1 transcript was detected (Fig. S1C). The hematopoietic expression of ikzf1 suggested its critical role during hematopoiesis, especially in T cell development. Therefore, the ikzf1 gene was edited (Fig. 1A). Two types of ikzf1 mutants were identified. One mutant allele was 1 bp short in exon 3 (deletion of 4 bp but addition of another 3 bp), and the other showed a 1-bp insertion in exon 9 (Fig. S1, D–F). Both mutations led to a premature stop codon in their coding sequence (Fig. 1B and Fig. S1F), which did not cause an obvious reduction in expression but resulted in synthesis of truncated Ikzf1 proteins in both mutant alleles (Fig. 1B and Fig. S1, G and H). This suggested loss of function in both alleles. Similar hematopoietic phenotypes were seen in both mutants (Fig. S2, A and B). Therefore, the ikzf1^{Δ4 + 3/Δ4 + 3} mutant was selected for intensive investigation.

Figure 1.

Disrupted T lymphopoiesis in ikzf1^{Δ4 + 3/Δ4 + 3} mutant larvae. A, the target sites for ikzf1 gene editing. The AlwI restriction enzyme was used to identify one mutant allele (purple). The AGG and GGG sequences (red) are the protospacer adjacent motifs. Red triangles indicate the mutation sites in two mutant alleles. B, the ikzf1^{Δ4 + 3} and ikzf1⁺¹ mutations lead to generation of truncated Ikzf1 proteins lacking zinc finger domains. The ikzf1^{Δ4 + 3} mutation (in amino acid (aa) 28) and ikzf1⁺¹ mutation (in aa 398) caused formation of 57- and 403-aa truncated proteins, respectively. The purple squares indicate the mismatched amino acid caused by the frameshift in the mutant compared with the WT. C, whole-mount in situ hybridization (WISH) of ccr9a, ccr9b, lck, and rag1 in the thymus of ikzf1^{Δ4 + 3/Δ4 + 3} and their siblings (Sib). The values in the bottom right corner of each panel indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). The red circles indicate the thymi.

The enrichment of the ikzf1 transcript in rag2-DsRed⁺ cells drew our attention to the T cell phenotypes. As expected, T cell formation, as indicated by the expression of ccr9a, ccr9b, lck, and rag1, was severely compromised in ikzf1^{Δ4 + 3/Δ4 + 3} mutants (Fig. 1C). We confirmed that this T cell deficiency phenotype in ikzf1^{Δ4 + 3/Δ4 + 3} mutants was caused by the ikzf1 mutation; forced expression of ikzf1 in hematopoietic cells, driven by the hematopoiesis-specific coro1a promoter and overexpressed in hematopoietic progenitors (Fig. S1I) and leukocytes (38, 39), was sufficient for restoring T cell development in mutants (Fig. 1C). The expression of ccl25a, a chemokine gene expressed in thymic epithelial cells (13), was not detectably altered (Fig. S1J). Collectively, these data demonstrate that Ikzf1 is required for T cell development, in accordance with previous findings (30).

Failed thymic migration of HSPCs in ikzf1^{Δ4 + 3/Δ4 + 3} mutants

The drastically compromised T lymphopoiesis in ikzf1 mutant zebrafish (Fig. 1C and Fig. S2A) (30, 31) and mice (19–21, 25) prompted a dissection of the molecular mechanisms involved. Zebrafish embryonic T lymphopoiesis initiated from HSPCs. However, the signature of cmyb⁺ cells showed no obvious change at 36 hpf and 48 hpf in the VDA and CHT regions between ikzf1 mutants and nonmutant siblings (Fig. S2, B–D), suggesting that ikzf1 is dispensable in HSPC production. At ∼60 hpf, HSPCs first appeared in the thymus (Fig. 2A) (13). However, almost no cmyb⁺ cells were detected in the thymus of ikzf1^{Δ4 + 3/Δ4 + 3} mutants compared with WT siblings at this stage, whereas a comparable number of cmyb⁺ cells was seen in the CHT at the same time point (Fig. 2, A and B). This result suggested that the initial thymic migration of progenitor cells was disrupted by ikzf1 deficiency. To test this hypothesis, we used time-lapse imaging to monitor the thymus homing of hematopoietic progenitors, from 60 hpf onwards, in ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:DsRed;lyz:GFP) transgenic mutants. To exclude the influence of myeloid cells, we focused on coro1a-Dsred⁺/lyz-GFP⁻ cells (40), located around the thymus, that featured a spherical morphology (presumably thymocytes and their progenitors). In control embryos, ∼0.7 spherical coro1a-DsRed⁺/lyz-GFP⁻ cells per hour entered the thymus between 60 and 72 hpf (Fig. 2, C and D, and Movie S1). In comparison, limited spherical coro1a-DsRed⁺/lyz-GFP⁻ cells were located around and entered the ikzf1^{Δ4 + 3/Δ4 + 3} thymus during the same period (Fig. 2, C and D, and Movie S2). These results indicated that movement to the thymus in hematopoietic progenitors was abrogated in ikzf1^{Δ4 + 3/Δ4 + 3} mutants. This conclusion was further validated by the results of another transient lineage-tracing assay, performed using the ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:Kaede) and ikzf1^{Δ4 + 3/Δ4 + 3};Tg(kdrl:Dendra2) lines, in which specific hematopoietic progenitors in the VDA (somites 8–10) and CHT (somites 15–17) were labeled (coro1a-Kaede⁺ or kdrl-Dendra2⁺ cells) and then followed after UV irradiation (41, 42) (Fig. S2, E and F). In WT siblings, the labeled coro1a-Kaede⁺ and kdrl-Dendra2⁺ cells were clearly detected in the thymus and kidney at 1–2 days post-UV irradiation (Fig. 2, E–I). In contrast, no red coro1a-Kaede⁺ or kdrl-Dendra²⁺ cells were detected in the thymus of ikzf1^{Δ4 + 3/Δ4 + 3} mutants, although the cells were clearly observed in the CHT and kidney (albeit at reduced numbers) (Fig. 2, E–K). These data support the conclusion that the thymus-homing of hematopoietic progenitors during early zebrafish development is compromised in ikzf1^{Δ4 + 3/Δ4 + 3} mutants.

Figure 2.

Compromised thymus homing of the hematopoietic progenitors in ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos. A and B, WISH of cmyb in the thymus (A) and CHT (B) of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. C, fluorescence image of thymi of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. D, calculation of spherical coro1a-DsRed⁺ lyz-GFP⁻ cells entering the thymus per hour (from 60–72 hpf; mean ± S.D.; Sib: 0.75 ± 0.21, n = 6; ikzf1^{Δ4 + 3/Δ4 + 3}: 0, n = 6). See also Movies S1 and S2. E, fluorescence images indicating labeled coro1a-Kaede⁺ cells in thymi of ikzf1^{Δ4 + 3/Δ4 + 3} mutants and siblings. F, calculation results of E (mean ± S.D.; Sib: 15.94 ± 6.68, n = 51; ikzf1^{Δ4 + 3/Δ4 + 3}: 0, n = 18). G, fluorescence images of labeled coro1a-Kaede⁺ cells in kidneys of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. H, fluorescence images indicating labeled kdrl-Dendra2⁺ cells in thymi of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. I, calculation results of H (mean ± S.D.; Sib: 3.75 ± 2.58, n = 36; ikzf1^{Δ4 + 3/Δ4 + 3}: 0, n = 14). J, fluorescence images indicating labeled kdrl-Dendra2⁺ cells in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3} mutants and siblings. K, calculation results of J (mean ± S.D.; Sib: 16.30 ± 7.09, n = 27; ikzf1^{Δ4 + 3/Δ4 + 3}: 14.79 ± 6.15, n = 14). Blue arrowheads indicate the WISH signals (B), red coro1a-Kaede⁺ cells (G), and red kdrl-Dendra2⁺ cells (J). The red (A) and white (C, E, and H) circles represent the thymi. The values in the bottom right corner in A, B, E, G, and H indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). OV, otic vesicle. Scale bars = 50 μm. ns, not significant. ***, p < 0.001.

Compromised HSPC expansion in ikzf1 mutants

The reduction of coro1a-Kaede⁺ cells in ikzf1^{Δ4 + 3/Δ4 + 3} mutant kidneys (Fig. 2G) suggested HSPC pool shrinkage at later stages. Indeed, the cmyb⁺ population in ikzf1 mutants was significantly decreased from 4 dpf compared with those of siblings (Fig. 3, A and B, and Fig. S2B). Consistently, the confocal microscopy images and statistical analysis indicated that the CD41-GFP^lo (10) and runx1-GFP⁺ (32) cells were significantly reduced in ikzf1^{Δ4 + 3/Δ4 + 3} mutants compared with WT siblings (Fig. S3, A–C). The ratio of EdU⁺ cells and expression levels of several cell cycle–related genes in the CD41-GFP^lo and runx1-GFP⁺ population were clearly decreased in ikzf1^{Δ4 + 3/Δ4 + 3} mutants compared with WT siblings (Fig. 3, C–F, and Fig. S3, D and E). However, TUNEL⁺ cells remained unchanged (Fig. S3, F and G). These results indicated that the reduction of HSPCs in ikzf1^{Δ4 + 3/Δ4 + 3} mutants was caused by impaired cellular proliferation rather than increased cell apoptosis. Collectively, ikzf1 was essential for thymic migration and proliferation of HSPCs at embryonic stages, which play indispensable roles in T cell development.

Figure 3.

Reduced proliferation and differentiation of HSPCs in ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos. A and B, WISH of cmyb in the CHT (A) and glomerulus (B) of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. C, double staining of CD41-GFP and EdU in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. Scale bar = 50 μm. D, the percentage of EdU⁺/CD41-GFP⁺ cells in C (mean ± S.D.; Sib: 26.55 ± 3.80, n = 16; ikzf1^{Δ4 + 3/Δ4 + 3}: 16.73 ± 5.42, n = 14). E, double staining of runx1-GFP and EdU in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. Scale bar = 50 μm. F, the percentage of EdU⁺/runx1-GFP⁺ cells in E (mean ± S.D.; Sib: 29.53 ± 7.49, n = 23; ikzf1^{Δ4 + 3/Δ4 + 3}: 15.08 ± 4.12, n = 13). The blue arrowheads point to the WISH signals in A. The blue arrows, purple arrows, and white arrowheads in C and E indicate the green, red, and double positive signals, respectively. The values in the bottom right corner in A and B indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). ***, p < 0.001. g, glomerulus.

Ccr9a was a key factor mediating thymus homing in HSPCs

The downstream factors accounting for HSPC thymus homing were investigated by focusing on the chemokine (C-C motif) receptor Ccr9, a critical factor involved in HSPC recruitment to the thymus (43–45). Two orthologues, ccr9a and ccr9b, exist in zebrafish. Both molecules were expressed in the thymic T lymphoid cells from 4 dpf onward in WT siblings, but neither was detected in the ikzf1 mutants (Fig. 1C and Fig. S2A). However, the ccr9a transcript was seen in the WT CHT much earlier than ccr9b (2.5–3 dpf) and was barely observed in ikzf1^{Δ4 + 3/Δ4 + 3} mutants (Fig. 4A). Genomic sequence analysis of the ccr9a promoter indicated that there were elements directly targeted by Ikzf1 (Fig. 4B). Notably, the results of ChIP-qPCR (46) demonstrated that the DNA fragment containing the binding site was enriched in the anti-HA immunoprecipitation group by using Tg(coro1a:HA-ikzf1) (Fig. 4C and Fig. S4, A–D). Furthermore, Dual-Luciferase reporter analysis (47) revealed that deletion of the Ikzf1 binding sites markedly reduced the activity of the ccr9a promoter (Fig. 4D), suggesting that ccr9a might be a critical mediator of ikzf1 in the early migration of HSPCs. Therefore, a ccr9a mutant allele with a 10-bp genomic fragment deletion in exon 3 was generated, which resulted in synthesis of a truncated protein (Fig. S5, A–D). The populations of cmyb⁺, rag1⁺, tcrb⁺, and lck⁺ cells in the thymus were notably reduced but not totally lost in the ccr9a^Δ10/Δ10 mutants compared with WT siblings (Fig. 4, E and F, and Fig. S5E). This phenomenon was reversed by restoration of ccr9a in coro1a⁺ hematopoietic cells (Fig. 4E). In contrast, cmyb⁺ and irf4a⁺ cells were unaltered in the CHT and kidney of ccr9a^Δ10/Δ10 mutants compared with siblings (Fig. 4F and Fig. S5, F and G). Thus, ccr9a function deficiency might impair the thymic trafficking of HSPCs. To prove this, similar time-lapse imaging and transient lineage-tracing assays utilized to study ikzf1 mutants were performed. Time-lapse imaging showed that the frequency of Kaede⁺ cells entering the thymus was strikingly decreased in ccr9a^Δ10/Δ10 mutants (Fig. 4, G and H, and Movies S3 and S4), and the numbers of red-Kaede⁺ and red-Dendra2⁺ cells were obviously decreased in the ccr9a^Δ10/Δ10 mutant thymus compared with siblings (Fig. 4, I and J, and Fig. S5, H and I). These data suggest defective homing of hematopoietic progenitors in ccr9a^Δ10/Δ10 mutants.

Figure 4.

Defective thymus migration of hematopoietic progenitors in ccr9a mutant embryos. A, WISH of ccr9a in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. B and C, ChIP-qPCR analysis (C) (mean ± S.D.; ccr9a-1F/1R, IgG: 1.00 ± 0.03, anti-HA: 4.73 ± 0.51; ccr9a-2F/2R, IgG: 1.00 ± 0.64, anti-HA: 15.13 ± 3.84; ccr9a-CF/CR, IgG: 1.00 ± 0.33, anti-HA: 1.14 ± 0.35) of the binding sites (green) in the ccr9a promoter for Ikzf1 (B). CF, control forward primers; CR, control reverse primers. D, luciferase assay showing the activities of ccr9a promoter (with the binding site (BS) and mutant binding site (MBS)) when regulated by the Ikzf1 protein (full-length CDS) (mean ± S.D.; ccr9a promoter: 1.00 ± 0.06, n = 3; ccr9a promoter + ikzf1 CDS: 3.07 ± 0.29, n = 3; ccr9a promoter (MBS1) + ikzf1 CDS: 2.17 ± 0.04, n = 3; ccr9a promoter (MBS2) + ikzf1 CDS: 2.51 ± 0.34, n = 3; ccr9a promoter (MBS1+MBS2) + ikzf1 CDS: 1.79 ± 0.04, n = 3). EV, pCS2 empty vector. E and F, WISH of tcrb and rag1 (E), and cmyb (F) in ccr9a^Δ10/Δ10 mutant larvae and their siblings. G, fluorescence image of thymi of ccr9a^Δ10/Δ10 mutant embryos and their siblings. H, calculation of spherical coro1a-Kaede⁺ cells entering the thymus per hour (from 60–72 hpf; mean ± S.D.; Sib: 0.63 ± 0.15, n = 6; ccr9a^Δ10/Δ10: 0.09 ± 0.06, n = 5). See also Movies S3 and S4. I, fluorescence images indicating labeled coro1a-Kaede⁺ cells in thymi of ccr9a^Δ10/Δ10 mutant embryos and their siblings. J, calculation results of I (mean ± S.D.; Sib: 11.27 ± 5.06, n = 26; ccr9a^Δ10/Δ10: 3.56 ± 2.04, n = 18). The blue arrowheads indicate the WISH signals (A and F). The red (E and F) and white (G and I) circles represent the thymi. The values in the bottom right corner in A, E, F, and I indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). OV, otic vesicle. Scale bars = 50 μm. ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001.

Restoration of Ccr9a rescued the thymus-settlement of HSPCs in ikzf1^{Δ4 + 3/Δ4 + 3} mutants

The importance of ccr9a in HSPC thymus-homing and their disappearance in ikzf1^{Δ4 + 3/Δ4 + 3} mutants prompted supply of this factor and an exploration of T cell phenotypes. To this end, the coro1a:ccr9a transgenic line was generated (Fig. S6, A and B). Overexpression of ccr9a had no detectable effects on hematopoietic development, as shown by the comparable expression of various hematopoietic lineage markers (ikzf1, irf4a, mpx, gata1, and mpl) in Tg(coro1a:ccr9a) (Fig. S6C). Restoration of ccr9a in ikzf1^{Δ4 + 3/Δ4 + 3} mutant hematopoietic cells efficiently promoted colonization of CD41-GFP⁺, even lyz-Dsred⁺, cells in the thymic anlage (Fig. 5, A and B, and Fig. S6D and Movies S5–S7). These recolonizing CD41-GFP⁺ cells should act as hematopoietic progenitors, as cmyb⁺ and lyz⁺ but not mpl⁺ signals were detected in the same situation (Fig. 5C and Fig. S6, D and E). However, they failed to differentiate into rag1⁺ and tcrb⁺ T cells in the thymus of ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:ccr9a) mutants (Fig. 5, D and E). Thus, ccr9a was critical in the initial settlement of HSPCs into the thymus but played a limited role in further differentiation, suggesting the requirement for additional factors.

Figure 5.

Restore of ccr9a in ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos effectively promotes thymus migration of hematopoietic progenitors. A, fluorescence images indicating CD41-GFP⁺ cells in thymi of ikzf1^{Δ4 + 3/Δ4 + 3}; Tg(coro1a:ccr9a) mutant embryos and their siblings. B, calculation results of A (mean ± S.D.; Sib: 21.57 ± 5.97, n = 14; ikzf1^{Δ4 + 3/Δ4 + 3}: 0, n = 10; Tg(coro1a:ccr9a): 22.29 ± 6.75, n = 17; ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:ccr9a): 10.5 ± 8.31, n = 16). C–E, WISH of cmyb (C), rag1 (D), and tcrb (E) in thymi of ikzf1^{Δ4 + 3/Δ4 + 3} mutant animals and their siblings upon replenishment of ccr9a. The white (A) and red (C–E) circles indicate the thymi. The values in the bottom right corner in A and C–E indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). Scale bars = 50 μm. ns, not significant; ***, p < 0.001.

Homing of hematopoietic progenitors was disrupted in irf4a^Δ18/Δ18 mutants

Irf4, a member of the interferon regulatory factor family that is essential in zebrafish T lymphoid determination (40), was then examined. Similar to ccr9a, the expression of irf4a was barely detected in ikzf1^{Δ4 + 3/Δ4 + 3} mutants (Fig. 6A). The data from ChIP and Dual-Luciferase reporter experiments showed that irf4a was directly regulated by Ikzf1 (Fig. 6, B–E), suggesting that irf4a may be the candidate factor. To this end, an irf4a^Δ18/Δ18 mutant line (harboring an 18-bp deletion) was created (Fig. S7, A and B). This mutation resulted in removal of six amino acids from the conserved DNA-binding domain of the protein (48) and, consequently, in a drastic reduction in Irf4a protein level (Fig. S7, C and D). Neither cmyb⁺, ikzf1⁺, or ccr9a⁺ cells in the CHT (Fig. S7, E and F) nor thymic structure/chemokines were affected by the irf4a mutation (Fig. S7G). However, most of the irf4a^Δ18/Δ18 larvae exhibited complete loss of ccr9a⁺ and ikzf1⁺ cells in the thymus at 3 dpf, and the remaining larvae only showed a trace amount of these cells, which suggests that early T cell development was interrupted (Fig. 6, F and G, and Fig. S7H). Intriguingly, the T cell number partially recovered in irf4a^Δ18/Δ18 mutants as the embryos grew. At 6 dpf, only a small group of irf4a^Δ18/Δ18 larvae showed complete lack of T cells, whereas a large proportion of the mutants displayed relatively smaller but detectable T cell foci in the thymus (Fig. 6, F and G, and Fig. S7H). The normal population of hematopoietic progenitors and the gradual recovery of T cells in the irf4a^Δ18/Δ18 mutants suggest that the T cell deficiency was probably caused by inefficient homing but was not due to the cell fate alteration (40) (Fig. S7I), as that found in the ikzf1^{Δ4 + 3/Δ4 + 3} and ccr9a mutants. Indeed, time-lapse imaging revealed the presence of only a limited number of spherical coro1a-DsRed⁺lyz-GFP⁻ cells in the irf4a^Δ18/Δ18 thymus, in contrast with that of the WT control (Fig. 6, H and I, and Movies S1 and S8). Furthermore, transient tracing of the coro1a-Kaede⁺ cells in the CHT (somites 15–17) showed that the marked cells presented notably compromised movement to the thymus of irf4a^Δ18/Δ18 mutants relative to that in the siblings (Fig. 6, J and K), recapitulating what occurred in the ikzf1 and ccr9a mutants. Collectively, these data suggested that the progenitors in the irf4a^Δ18/Δ18 mutants, as with those in the ikzf1 and ccr9a mutants, were compromised by a thymus homing blockade.

Figure 6.

Impaired thymus homing of the hematopoietic progenitors in irf4a^Δ18/Δ18 mutant larvae. A, WISH of irf4a in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3} mutant embryos and their siblings. B–D, ChIP-PCR (C) and ChIP-qPCR (D) (mean ± S.D.; irf4a-1F/1R, IgG: 1.00 ± 0.05, anti-HA: 23.34 ± 1.68; irf4a-CF/CR, IgG: 1.00 ± 0.30, anti-HA: 0.57 ± 0.15) analysis of the binding site (green) in the irf4a promoter (B) for Ikzf1. CF, control forward primers; CR, control reverse primers. E, luciferase assay showing irf4a promoter (with the BS and MBS) activity when regulated by the Ikzf1 protein (full-length CDS) (mean ± S.D.; Ikzf1-BS: ikzf1 mRNA, 2.55 ± 1.35, n = 12; ctrl = 1.00 ± 0.19; Ikzf1-MBS: ikzf1 mRNA, 1.00 ± 0.38, n = 11; ctrl = 1.00 ± 0.22, n = 10). F, WISH of ikzf1⁺ and rag1⁺ cells in thymi (red circles) of irf4a^Δ18/Δ18 mutants and their siblings at different time points. Light blue, green, and purple squares indicate normal, small, and no signal foci in the thymus, respectively. G, quantification of phenotype distribution of ikzf1⁺ and rag1⁺ cells in F or ccr9a⁺ and lck⁺ cells (Fig. S7H) at different stages. H, fluorescence images of thymi in irf4a^Δ18/Δ18 mutants and siblings at 60 hpf. I, calculation of spherical coro1a-DsRed⁺lyz-GFP⁻ cells entering the thymus per hour (from 60–72 hpf; mean ± S.D.; Sib: 0.75 ± 0.21, n = 6; irf4a^Δ18/Δ18: 0.07 ± 0.11, n = 6). See also Movies S1 and S8. J, fluorescence images indicating labeled coro1a-Kaede⁺ cells in thymi of irf4a^Δ18/Δ18 mutants and their siblings at 3 dpf. K, calculation results of J (mean ± S.D.; Sib: 10.50 ± 5.33, n = 32; irf4a^Δ18/Δ18: 0.91 ± 1.38, n = 35). The blue arrowheads indicate the WISH signals (A). The white circles (H and J) represent the thymi. The values in the bottom right corner in A and J indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). Scale bars = 50 μm. ns, not significant; ***, p < 0.001.

Supplementation of irf4a drastically promoted HSPC proliferation and T cell differentiation in ikzf1^{Δ4 + 3/Δ4 + 3}; Tg(coro1a:ccr9a) mutants

The irf4a mutant data prompted us to continue irf4a supplementation in ikzf1 mutants. The Tg(coro1a:irf4a) transgenic line (Fig. S8, A–C) partially promoted the production of rag1⁺ and lck⁺ cells (∼36%) in ikzf1^{Δ4 + 3/Δ4 + 3} mutants (Fig. 7, A–D). We detected an increase in cmyb⁺ cells, which were achieved by the recovered cell proliferation, as indicated by the EdU incorporation assay (Fig. S8, F and G), in the CHT of ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:irf4a) compared with ikzf1^{Δ4 + 3/Δ4 + 3} mutants (Fig. S8D). These rescued cmyb⁺ cells had lymphoid potential. They expressed high levels of the lymphoid (ccr9a) gene but limited myeloid (lyz) marker (Fig. S8E). However, achieving irf4a function in T lymphopoiesis partially required ccr9a-mediated thymic migration of HSPCs, as the rescue efficiency obviously reduced (below 20%) when ccr9a was mutated in ikzf1^{Δ4 + 3/Δ4 + 3};ccr9a^Δ10/Δ10;Tg(coro1a:irf4a) animals (Fig. 7, A and B). These data highlighted the roles of ccr9 in HSPC migration and the crucial function of irf4a in HSPC proliferation and T cell differentiation. Supporting this assumption, supplementation with both irf4a and ccr9a in ikzf1^{Δ4 + 3/Δ4 + 3} mutants effectively recovered great numbers of rag1⁺ and lck⁺ cell foci (∼65%) in the thymus compared with irf4a alone (∼36%) (Fig. 7, C and D). As a previous study suggested that ccr9a was directly regulated by Irf4a (40), Ikzf1 and Irf4a probably functioned together to regulate ccr9a activity. This prediction was validated by the results from the coimmunoprecipitation and Dual-Luciferase reporter experiments, which indicated that GFP-Ikzf1 or Ikzf1-GFP interacted with HA-Irf4a (Fig. 7E) and that cotransfection of both factors elevated the transcriptional activity of ccr9a more obviously than transfection with either one (although Ikzf1 played a major role) (Fig. 7F). Taken together, Ikzf1 regulated the thymic migration, proliferation, and differentiation of HSPCs via ccr9a and irf4a. These factors worked closely together to accomplish T lymphopoiesis in embryonic zebrafish (Fig. 7G).

Figure 7.

Additional replenishment of ikzf1^{Δ4 + 3/Δ4 + 3};Tg(coro1a:ccr9a) significantly rescues T lymphopoiesis. A–D, WISH of rag1 (A and C) and lck (B and D) in thymi of ikzf1^{Δ4 + 3/Δ4 + 3} mutants and siblings with various transgenic or mutant backgrounds. E, coimmunoprecipitation experiment showing coprecipitation of GFP-tagged Ikzf1 with HA-tagged Irf4a. F, schematic overview of the ikzf1 binding sites and irf4a binding site in the ccr9a promoter (top panel). A luciferase assay (bottom panel) shows the activities of the ccr9a promoter when regulated by Ikzf1 and Irf4a protein (full-length CDS) (mean ± S.D.; ccr9a promoter: 1.00 ± 0.09, n = 3; ccr9a promoter + ikzf1 CDS: 2.55 ± 0.18, n = 3; ccr9a promoter + irf4a CDS: 1.43 ± 0.08, n = 3; ccr9a promoter + ikzf1 CDS + irf4a CDS: 2.99 ± 0.28, n = 3). G, a model of our study. Ikzf1 promotes T lymphopoiesis via thymus migration, proliferation, and differentiation of HSPCs, which are mediated by the downstream targets irf4a and ccr9a. Loss of ikzf1 functions in mutant fish leads to compromised migration and arrested proliferation of HSPCs. Compromised production of T cells was then observed in ikzf1 mutant embryos. The red circles indicate the thymi (A–D). The values in the bottom right corner in A–D indicate counts with a typical appearance as presented (first number) in the total number of examined samples (last number). ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001.

T lymphopoiesis was impaired in adult ikzf1^{Δ4 + 3/Δ4 + 3} zebrafish

The survival of ikzf1^t24980 (30) led us to examine whether ikzf1^{Δ4 + 3/Δ4 + 3} mutants survive to adulthood. The homozygous mutant in the background of Tg(coro1a:DsRed) (39) was selected for rearing based on a lack of coro1a-DsRed⁺ cells in the thymus. The ikzf1^{Δ4 + 3/Δ4 + 3} mutant presented higher mortality from 9 dpf onward compared with the controls (Fig. S9A). Approximately 45% of the ikzf1^{Δ4 + 3/Δ4 + 3} mutants survived to 2.5 months, and no ikzf1^{Δ4 + 3/Δ4 + 3} mutants lived beyond 6 months (Fig. S9A). The ikzf1^{Δ4 + 3/Δ4 + 3} mutants that survived were considerably smaller than their siblings (Fig. S9B) and were less fertile. As in the larval stage, fewer T cells were detected in the thymus of adult ikzf1^{Δ4 + 3/Δ4 + 3} mutants, and the thymus was substantially smaller in the mutants than in their siblings (Fig. S9C). Therefore, T cell numbers could recover but were still reduced in adult ikzf1^{Δ4 + 3/Δ4 + 3} mutants.

Discussion

In addition to the compromised embryonic T lymphopoiesis that recovered at later stages and persisted for over 17 months in a reported zebrafish mutant (30), the newly created ikzf1 mutant alleles in this study presented a drastic reduction of T cells in the adult stage as well as diminished hematopoietic progenitors with impaired proliferation. The difference was ascribed to the more significant impairment of Ikzf1 function in mutants that lacked functional domains compared with previous mutants (30, 31) that carried a point mutation in the sequence encoding the C-terminal protein interaction domain. The mutants were not considered a dominant-negative mutation because no overt hematopoietic phenotype was detected in heterozygous mutant embryos.

T cells were completely absent in ikzf1 mutant zebrafish at the embryonic stage as a result of impaired HSPC thymus homing, proliferation, and differentiation. However, it is challenging to define whether the compromised homing of HSPCs was the primary cause of T cell loss or whether this simply reflected a defect in the specification or commitment of lymphoid cells in ikzf1 mutants. Although ikzf1 has been recognized previously as a lymphoid progenitor marker (13), our study indicated that ikzf1 was also expressed in hematopoietic progenitors for multiple lineages and plays a critical role in their proliferation. Consistently, Ikzf1 was seen to be highly expressed in mouse KTLS⁺ long-term HSCs, and Ikzf1^{Plastc/Plastc} mutant mice initially generated long-term HSCs but quickly failed to maintain and expand this pool at the embryonic stage (49). Similar HSPC phenotypes were observed in our ikzf1 mutant zebrafish, suggesting a functionally conserved role of Ikzf1 as a self-renewing regulator in HSC maintenance. However, it is difficult to exactly identify which HSPC populations were affected in ikzf1 mutant zebrafish because of the shortage of specific markers for different subtypes in zebrafish HSPCs.

Ccr9 was essential in the thymic seeding of hematopoietic progenitors in mice (43–45). One orthologue of ccr9, ccr9a, was absent in ikzf1^{Δ4 + 3/Δ4 + 3} mutant larvae. The obvious impairment of hematopoietic progenitor migration was seen in ccr9a mutants. Restoration of ccr9a in ikzf1 mutants efficiently promoted thymic migration of HSPCs but resulted in failed differentiation, implying that thymic migration of HSPCs was necessary but not sufficient for T cell development. Irf4a was another critical factor downstream of Ikzf1 in T lymphopoiesis. Irf4a has been reported to regulate lymphoid versus myeloid determination in zebrafish (40). In this study, irf4a expression was lost, and continual replenishment of irf4a only partially rescued the T cell deficiency in ikzf1^{Δ4 + 3/Δ4 + 3} mutants by promoting both CHT expansion and ccr9a-mediated thymic seeding of HSPCs, implying a requirement for additional factors and more complicated regulatory networks in this process.

Recently, IRF4 was reported to cooperate with Ikzf1 and serve as a transcriptional complex to target the zinc finger–IRF composite elements (ZICEs) that appeared in the promoters of genes that functioned essentially during plasma cell differentiation in mice (50). This report suggested the possibility that irf4a, when turned on by Ikzf1 in HSPCs with lymphoid potential, functioned together with Ikzf1 to further guide lymphoid differentiation in zebrafish. However, unlike the suppressive roles of the Ikzf1–IRF4 complex on the ZICE motif in mice (50), zebrafish Ikzf1 interacted with Irf4a and more effectively promoted ccr9a expression working together than singly. This difference was probably caused by the fact that there was no ZICE element in the ccr9a promoter. A previous study revealed the Ets-IRF and AP-1–IRF composite elements in the ccr9a promoter (40), which were located adjacent to the Ikzf1 binding sites. The highest transcriptional activity of Ikzf1/Irf4a in the ccr9a promoter, like the Ikzf1–IRF4 complex, acted as a transcriptional activator in the presence of E26 transformation-specific family members such as PU.1 (50), suggesting the involvement of other factors. This process should be a focus of future investigation.

Supplementation with both ccr9 and irf4 in ikzf1^{Δ4 + 3/Δ4 + 3} mutants restored thymic T cell phenotypes more efficiently than either one alone. Therefore, we propose that, during zebrafish embryonic T lymphopoiesis, ikzf1 activates irf4a expression to protect the progenitor pool and assure their lymphoid potential. Then ikzf1 works together with irf4a to produce sufficient amounts of ccr9a that efficiently accomplish thymic seeding of hematopoietic progenitors, where T cell differentiation is finally achieved. However, there was still approximately 30% T cell failure in ikzf1^{Δ4 + 3/Δ4 + 3} mutants supplied with both ccr9a and irf4a. One possibility for this phenomenon was that the ccr9a and irf4a levels, even in transgenic zebrafish lines, were insufficient for final thymic T cell formation. Concordant with this hypothesis, neither a transient supply of irf4a and ccr9a mRNA nor the weak transgenic Tg(coro1a:irf4a) allele rescued thymic T cells in ikzf1^{Δ4 + 3/Δ4 + 3} mutants (data not shown). In addition, it remains possible that other factors were involved. Transcription factors such as Bcl11b, E2A, and Hes1 are crucial for T cell commitment (2). The P-selectin/PSGL-1 axis plays a role in recruiting progenitors to the thymus (2, 51, 52), and CCR9, CCR7, and CXCR4 chemokine receptors function together to support cells found in the thymus (43–45). Therefore, achieving thymic T cell development under physiological conditions depends on the coordinated actions of an ensemble of transcriptional factors, cytokines, adhesion molecules, and chemoattractants. Overall, our findings extend the mechanistic understanding of embryonic T lymphopoiesis in zebrafish and are valuable for further elucidation of Ikzf1 function in hematopoiesis.

Experimental procedures

Fish lines

The AB, Tg(CD41:GFP) (37), Tg(runx1:GFP) (32), Tg(coro1a:DsRed)/Tg(coro1a:Kaede) (39), Tg(cmyb:GFP) (33), Tg(lyz:GFP)nz117/Tg(lyz:DsRed) (53), Tg(mpx-GFP)ill4 (35), Tg(rag2:DsRed) (34), Tg(gata1-DsRed) (36), and Tg(kdrl:Dendra2) strains were used. All zebrafish were maintained according to the guidelines of experimental animal welfare from the Ministry of Science and Technology of the People's Republic of China (2006).

Generation of mutants and transgenic lines

The ikzf1 and ccr9a mutants were created by the CRISPR/Cas9 system (54). The in vitro synthesized guide RNA and hCas9 mRNA were injected into one-cell WT embryos. The irf4a mutants were obtained using transcription activator-like effector nucleases (55). Constructs targeting 5′-CCTGCTCTTGACGACTgcaggcctgggctttatttaAGGGCAAATACAGGG-3′ were created. The mutants were identified by sequencing or restriction enzymes. To generate Tg(coro1a:HA-ikzf1), Tg(coro1a:ccr9a), and Tg(coro1a:irf4a), their complementary DNA was amplified with specific primers (Table S1) and constructed into the pTol2 vector with the coro1a promoter (39). To facilitate screening, a Cryaa-Cerulean-BGHpA element (56) was reversely inserted. The constructs were injected into one-cell WT embryos. The transgenic lines were identified based on both the fluorescent signals in the eyes and a similar ikzf1/ccr9a/irf4a expression pattern as that of coro1a.

Real-time qPCR

Approximately 250 GFP⁺ and DsRed⁺ cells in Tg(CD41:GFP), Tg(mpx:GFP), Tg(gata1:DsRed), and Tg(rag2:DsRed) were sorted out by using a Moflo XDP fluorescence-activated cell sorter (Beckman) according to the standard protocol. The harvested cells were dissociated to extract RNA by using a Qiagen REPLI-g WTA Single Cell Kit (150063) according to the manufacturer's protocols. Total RNA of the whole embryos was extracted using TRIzol reagent. The amplified mRNA was used for qPCR. Each sample was tested in triplicate. sep15 (selenoprotein F precursor 15) (57), β-actin, and ef1a (58) expression was measured and used to normalize signals for each queried transcript by using the ΔΔCt method. The primers of cell cycle–related genes (58) and other genes are listed in Table S2.

ChIP

For the ChIP assay, the tails of 300 Tg(coro1a:HA-ikzf1) embryos were collected at 3 dpf. After treatment with lysis buffer, the suspension was fragmentized by 0.5 units of MNase (37 °C for 45 s) to generate 300∼1000 fragments. Cross-linked chromatin was immunoprecipitated with anti-HA antibody (Abcam, ab9110) or anti-IgG antibody (Sigma, A6154, negative control), according to the procedure in Ref. 46. The resultant immunoprecipitation samples were subjected to semi-quantitative PCR or qPCR using primers (Table S1).

Reporter assay

The luciferase assays were performed with the Dual-Luciferase Reporter Assay System. The luciferase activity was measured with a GloMax® 20/20 luminometer (Promega) according to the manufacturer's instructions. The full length of ikzf1 CDS amplified with primers 5′- CGGAATTCAATGGAGACTGAGGAGGCACA-3′/5′-GCTCTAGAGTGGTTTACGTACACCATTG-3′ and the full length of irf4a CDS amplified with primers 5′-CCATCGATATGAACTTAGATGGGGACTG-3′/5′-GCTCTAGATCACTCTGTCAGGTGTTGTA-3′ were cloned into the pCS2 vector. The −2.8-kb ccr9a promoter was amplified with the primers (Table S1) and cloned into the pGL3-basic vector (Promega). The Ikzf1-binding sites were mutated by site-directed mutagenesis using the designed site-specific oligonucleotides primers 5′-GAAGGATCGGCAAAAGACCTCGATGAACTCGGGTCGCAGCAAG-3′/5′-CTTGCTGCGACCCGAGTTCATCGAGGTCTTTTGCCGATCCTTC-3′ for MBS1 (mutant Ikzf1 binding site 1) and 5′-CTTAAACAAACAAACCCAACTCAACTCAACCTTACTTAGT-3′/5′-ACTAAGTAAGGTTGAGTTGAGTTGGGTTTGTTTGTTTAAG-3′ for MBS2 (mutant Ikzf1 binding site 2) according to the manufacturer's instructions. To study the ccr9a promoter regulated by Ikzf1, Irf4a or both, the plasmids (200 ng of pGL3-ccr9a promoter, 200 ng of pGL3-ccr9a promoter-MBS1, 200 ng of pGL3-ccr9a promoter-MBS2, 200 ng of pGL3-ccr9a promoter-2MBS, 200 ng of pCS2-ikzf1 CDS, 200 ng of pCS2-irf4a CDS, 200 of ng pCS2, and 10 ng of pRL-CMV) were transfected into HEK293T cells by using Lipo8000^TM (Beyotime, C0533) with different combinations. pRL-CMV plasmids were added to each combinational group as an internal control. pCS2 empty vector was used as a negative control. HEK293T cells were cultured in DMEM with 10% fetal bovine serum in 24-well plates. After 36 h, cells were harvested for the luciferase assays according to the manufacturer's protocols. Analysis of the irf4a promoter was performed using zebrafish embryos. The irf4a promoter regions with the putative Ikzf1 binding site were amplified with the primers 5′-TCCCCCGGGCAGAAGAGCATCTCTGAACA-3′/5′-CCGCTCGAGCAGCTGAACCGCTGATACAA-3′ and cloned into the pGL3-Enhancer vector (Promega). The binding site was mutated by site-directed mutagenesis using the designed site-specific oligonucleotides primers 5′-GCCGCTCCTGTCAGCTAAGATGAGGCTACAATTCACACAG-3′/5′-CTGTGTGAATTGTAGCCTCATCTTAGCTGACAGGAGCGGC-3′ according to the manufacturer's instructions. The designed plasmids (30 pg of pGL3-irf4a promoter or pGL3-irf4a promoter-MBS1 together with 1 pg of pRL-CMV) and 150 pg of ikzf1 mRNA were injected into the embryos at the one-cell stage. Double-distilled H₂O was used to replace ikzf1 mRNA in the control groups at the same time. After 28 hpf, the embryos were collected for the luciferase assays according to the manufacturer's protocols and as described previously (47).

Western blotting

The N-terminal Myc-tagged Ikzf1 (6xMyc-Ikzf1), Ccr9a (6xMyc-Ccr9a), and Irf4a (HA-Irf4a) and their mutant forms were constructed in the pCS2⁺ vector. The mRNA was transcribed in vitro and injected into one-cell WT embryos. The protein was collected after 24 hpf to perform Western blotting according to standard procedures (59). The following antibodies were utilized: rabbit anti-Myc (Sigma, C3956, 1:500), mouse anti-β-tubulin (Invitrogen, MA5-11732, 1:1000), and rabbit anti-HA (Santa Cruz Biotechnology, sc-805, 1:200). Detection was performed with goat anti-mouse or anti-rabbit HRP-conjugated secondary antibodies (31430 and 32260, Thermo, 1:5000) and ECL Plus reagent.

Coimmunoprecipitation assay

The plasmids (2 μg of pCS2-HA-irf4a CDS, 2 μg of pCS2-GFP-ikzf1 CDS, 2 μg of pCS2-ikzf1-GFP CDS, and 2 μg of pCS2-GFP CDS) were transfected into HEK293T cells using Lipo8000^TM with different combinations. After 36 h, cells were harvested and proteins were purified using radioimmune precipitation assay buffer (50 mm Tris, 150 mm NaCl, 10% glycerol, 5 mm MgCl₂, 0.5% NP40, and Roche protease inhibitor mixture). Then proteins were mixed with anti-GFP–agarose beads (Smart Life Sciences, SM03801) for 2–3 h at 4 °C. Then the beads were collected to perform Western blotting using goat anti-GFP (Abcam, ab6658, 1:2000) and rabbit anti-HA (Santa Cruz Biotechnology, sc-805, 1:200).

Histological analysis, WISH, immunostaining, double staining, TUNEL staining, and EdU incorporation

H&E staining was performed on 10-μm paraffin sections as described previously (60). After H&E staining, images were taken under an Axio Imager.Z2 Vario (Carl Zeiss). Antisense RNA probes and WISH were performed using a standard protocol (61). Single-color FISH and double staining were performed as described previously (38). TUNEL staining was performed according to the manufacturer's instruction or as reported previously (62). For the EdU incorporation assay, the Click-iT® EdU Imaging Kit (Invitrogen, C10340) was used. Zebrafish larvae at different developmental stages were injected intravenously with 10 mm EdU and incubated for 2 h. The subsequent experiments followed the standard protocol. Goat anti-GFP antibody (Abcam, ab6658, 1:250, 4 °C, overnight) and Alexa Fluor 488 donkey anti-goat secondary antibody (Invitrogen, 1:400, 4 °C, overnight) were used successively to visualize the signals. The WISH signals were observed under a SteREO Discovery.V20 microscope, and the fluorescence signals were imaged under an LSM700 confocal microscope (Carl Zeiss).

Time-lapse live imaging and transient lineage tracing

The fish embryos were anesthetized, mounted in 1% agarose, and subsequently imaged under an LSM700 confocal microscope (Carl Zeiss) with a ×20 objective. For live imaging, the targeted cells (DsRed⁺, GFP⁺, or Kaede⁺ cells) located around the thymi were imaged and calculated. To achieve transient tracing, cells in the targeted regions of Tg(coro1a:Kaede) and Tg(kdrl:Dendra2) embryos (somites 8–10 in the VDA and 15–17 in the CHT) were selected and stimulated using the ROI mode via a 405-nm laser for around 15–20 s. The labeled cells were revealed and calculated 1–2 days later in the CHT, kidney, and thymus regions.

Quantification, calculation, and statistical methods

The positive signals in larval CHT were manually scored and double-confirmed blindly. All quantified data (mean ± S.D.) were analyzed by GraphPad Prism 6. Student's t test (one-tailed) was used.

Author contributions

Y. Huang and L. Li conceptualization; Y. Huang, Y. He, and H. L. data curation; Y. Huang, M. M., and L. Li validation; Y. Huang, Y. L., Y. He, Z. F., Y. Z., X. H., and L. Li investigation; Y. Huang and Y. L. visualization; Y. Huang, Y. L., Y. He, Z. F., Y. Z., H. H., and L. Luo methodology; Y. Huang and L. Li project administration; Y. L. formal analysis; Q. L., J. Z., H. L., H. H., M. M., L. Luo, and L. Li resources; Q. L., J. Z., and L. Li funding acquisition; L. Li supervision; L. Li writing-original draft; L. Li writing-review and editing.