Research ArticlePhysiology

Ca2+ concentration–dependent premature death of igfbp5a−/− fish reveals a critical role of IGF signaling in adaptive epithelial growth

See allHide authors and affiliations

Science Signaling  18 Sep 2018:
Vol. 11, Issue 548, eaat2231
DOI: 10.1126/scisignal.aat2231

Stressing out about low Ca2+

Under low Ca2+ conditions, an epithelial cell population in zebrafish known as NaR cells proliferate so that they can take up more Ca2+. Liu et al. found that Igf-binding protein 5A (Igfbp5a) was required for this adaptive proliferation and for survival under low Ca2+ conditions. The defect in NaR adaptive proliferation in igfbp5a−/− zebrafish was rescued by expression of wild-type human or zebrafish IGFBP5 but not two cancer-associated IGFBP5 mutants. Mice deficient in the various IGFBPs lack obvious phenotypes, and these results reveal the importance of using the appropriate physiological conditions to determine the function of IGFBPs.

Abstract

The phenotype gap is a challenge for genetically dissecting redundant endocrine signaling pathways, such as the six isoforms in the insulin-like growth factor binding protein (IGFBP) family. Although overexpressed IGFBPs can inhibit or potentiate IGF actions or have IGF-independent actions, mutant mice lacking IGFBP-encoding genes do not exhibit major phenotypes. We found that although zebrafish deficient in igfbp5a did not show overt phenotypes when raised in Ca2+-rich solutions, they died prematurely in low Ca2+ conditions. A group of epithelial cells expressing igfbp5a take up Ca2+ and proliferate under low Ca2+ conditions because of activation of IGF signaling. Deletion of igfbp5a blunted low Ca2+ stress–induced IGF signaling and impaired adaptive proliferation. Reintroducing zebrafish Igfbp5a, but not its ligand binding–deficient mutant, restored adaptive proliferation. Similarly, adaptive proliferation was restored in zebrafish lacking igfbp5a by expression of human IGFBP5, but not two cancer-associated IGFBP5 mutants. Knockdown of IGFBP5 in human colon carcinoma cells resulted in reduced IGF-stimulated cell proliferation. These results reveal a conserved mechanism by which a locally expressed Igfbp regulates organismal Ca2+ homeostasis and survival by activating IGF signaling in epithelial cells and promoting their proliferation in Ca2+-deficient states. These findings underscore the importance of physiological context when analyzing loss-of-function phenotypes of endocrine factors.

INTRODUCTION

Zebrafish live in hypo-osmotic environments and continuously lose Ca2+ and other ions into the surrounding water. They constantly take up Ca2+ from the aquatic habitat to maintain body fluid Ca2+ homeostasis. This vital task is accomplished by a group of epithelial cells called Na+-K+-ATPase–rich (NaR) cells in zebrafish (1). NaR cells are functionally equivalent to human intestinal epithelial cells, and they contain all major molecular components of the transcellular Ca2+ transport machinery, including the epithelial calcium channel Trpv5/6 (1, 2). Although distributed in the adult intestine and gills (a major osmoregulatory organ in fish), NaR cells are located on the yolk sac skin during the embryonic and larval stages, making them easily accessible for experimental manipulation and observation (1, 3, 4). When faced with low-[Ca2+] challenge, preexisting NaR cells reenter the cell cycle to yield more cells (3, 4). This is considered to be an adaptive response that allows fish to acquire adequate Ca2+ to keep their body fluid [Ca2+] constant to survive under low-[Ca2+] environments (3, 4). Ca2+ deficiency also stimulates the proliferation of mammalian and human intestinal epithelial cells (58), suggesting an evolutionarily conserved mechanism. We have shown that low-[Ca2+] stress activates insulin-like growth factor (IGF) receptor (IGF1R)–mediated phosphoinositide 3 kinase (PI3K)–Akt–target of rapamycin (Tor) signaling specifically in NaR cells and the increased IGF signaling promotes NaR cells to reenter the cell cycle and proliferate (3, 4). How the IGF signaling is activated specifically in NaR cells under low-[Ca2+] stress is not fully understood.

IGFs are evolutionarily ancient polypeptides structurally related to insulin. IGF binds to and activates the IGF1R, a receptor tyrosine kinase that is structurally related to the insulin receptor. Unlike insulin, however, IGFs are bound to one of six types of IGF binding proteins (IGFBPs) in extracellular fluids. Binding to an IGFBP increases the half-life of IGF in the circulation and prevents the potential interaction of IGFs to the insulin receptor (9, 10). In addition to these endocrine roles, gain-of-function studies have suggested that IGFBPs can also act locally, both negatively and positively modulating IGF action in target tissues; in addition, some IGFBPs have IGF-independent biological activities (912). Some IGFBPs are present in the nucleus and have nuclear activity (12). Unexpectedly, mice lacking IGFBP-encoding genes do not exhibit major phenotypes. IGFBP2 knockout mice are phenotypically normal with the exception of minor gender-specific changes in bone structure and minor changes in the spleen and liver in adult males (13, 14). IGFBP3 knockout mice are normal in size (15). Deletion of the IGFBP4 gene in mice results in a mild 10 to 15% reduction in prenatal growth, which is somewhat paradoxical given that overexpression of IGFBP4 also reduces growth (15). IGFBP5 knockout mice are phenotypically normal (15). The IGFBP3, IGFBP4, and IGFBP5 triple knockout mice are viable and only show a 25% reduction in body growth (15). The mechanisms underlying the lack of phenotypes are poorly understood.

We have previously reported that a zebrafish igfbp gene, igfbp5a, is specifically expressed in NaR cells (3). Knockdown of Igfbp5a using antisense morpholinos decreases the low-[Ca2+] stress–induced Akt signaling, suggesting that locally expressed Igfbp5a may be important (3). Because morpholinos only knock down target genes transiently, this approach could not be used to investigate the possible role of Igfbp5a in adaptive NaR cell proliferation nor the long-term biological consequence. Furthermore, we now understand that morpholinos have many limitations, including high toxicity and “off-target” effects (16). Here, we sought to investigate the functional role of Igfbp5a in NaR cells and its long-term functional role by deleting Igfbp5a with TALEN (transcription activator-like effector nuclease) and CRISPR-Cas9 (clustered regularly interspaced short palindromic repeats–CRISPR-associated 9) technology. The igfbp5a−/− mutant fish appeared indistinguishable from their wild-type siblings when raised in Ca2+-rich embryo solutions. When raised in regular water or low Ca2+ solutions, however, they died prematurely. Further analyses showed that deletion of Igfbp5a impaired the low-[Ca2+] stress–induced NaR cell proliferation and diminished IGF signaling. The diminished IGF signaling and NaR cell proliferation were restored by reintroducing Igfbp5a or human IGFBP5. Further analyses suggested that Igfbp5a acts by binding to IGF ligand and activating Akt-Tor signaling in epithelial cells under Ca2+-deficient states and that this action is conserved in human colon cells.

RESULTS

Two independent F1 lines were obtained using a TALEN-based approach (Fig. 1A and fig. S1, A and B). Both are predicted to be null alleles. The igfbp5aΔ4 line harbors a 4–base pair (bp) deletion allele that introduces a 23–amino acid sequence shortly after the signal peptide. The igfbp5aΔ11 line contains an 11-bp deletion allele, which truncates Igfbp5a after the signal peptide. F2 progeny from these lines were raised in the standard embryo solution (containing 0.33 mM [Ca2+] and referred to as E3 solution hereafter) to 5 days postfertilization (dpf) and transferred to the standard fish system water. Genotyping of adults showed a complete lack of homozygous adult fish in both lines. To determine whether the mutant fish died prematurely, we intercrossed each of the two F1 lines and randomly genotyped their F2 progeny at the larval and adult stages. Whereas 27% of larvae (5 dpf) in the igfbp5aΔ4 group were homozygous, adult progeny were either heterozygous or wild-type (Fig. 1B). Similarly, none of homozygous igfbp5aΔ11 fish (20% detected at 3 dpf) survived to 3 months (Fig. 1B). Among the survived igfbp5aΔ11 adult fish, 43% were heterozygous (Fig. 1B).

Fig. 1 Deletion of Igfbp5a leads to premature death.

(A) Structure of wild-type (WT) Igfbp5a, Igfbp5aΔ4, and Igfbp5aΔ11. SP, signal peptide. N, L, and C indicate the N-, L-, and C-domain of Igfbp5a. Asterisk (*) indicates a stop codon. The underlined amino acids were introduced by a frameshift. aa, amino acids. (B and C) F2 progeny of igfbp5aΔ4+/− intercrosses (left panel) and igfbp5aΔ11+/− (right panel) intercrosses were raised in E3 solution until 3 or 5 dpf and transferred into either fish system water (B) or 5PPT high [Ca2+] solution (C). The fish were sampled randomly and genotyped at the indicated stages. The genotype distributions are shown. ****P < 0.0001 by χ2 test. ns, not statistically significant. Total number of fish in each group is shown above the bar.

Because igfbp5a mRNA is specifically expressed in NaR cells, which are crucial for Ca2+ uptake, we speculated that the mutant fish died because of deficiency of this essential ion. If this notion were correct, then raising the progeny in embryo solutions containing high [Ca2+] should rescue the mutant fish because Ca2+ can enter through paracellular transport. When progeny of igfbp5aΔ4 intercrosses and igfbp5aΔ11 intercrosses were raised in 5PPT solution (containing 1.6 mM [Ca2+] and referred to as high [Ca2+] solution hereafter), many mutant fish grew to adulthood (Fig. 1C). The adult mutant fish in both lines were fertile and morphologically indistinguishable from their siblings. There were no notable differences in the growth rate or speed of development (Fig. 2, A to D). The igfbp5aΔ11 line, referred to as igfbp5a−/− mutant fish hereafter, was used in subsequent studies unless otherwise stated.

Fig. 2 igfbp5a−/− mutant fish develop and grow normally in Ca2+-rich solutions.

(A) Progeny of igfbp5a+/− intercrosses were raised in E3 solution until 5 dpf and transferred into high [Ca2+] 5PPT solution. The vertebrate columns of 12 dpf larvae were visualized by calcein staining and quantified. Data shown are means ± SEM. n = 18 to 19 fish per group. No statistically significant differences were found among the groups using one-way analysis of variance (ANOVA), followed by Tukey’s multiple comparison test. (B to D) Embryos of the indicated genotypes were raised in E3 solution. Their body length (B) and somite number (C) were measured at 24 hours postfertilization (hpf), and head-trunk angle (D) was measured at 72 hpf. Data shown are means ± SEM. n = 12 to 73 fish per group. No statistically significant differences were found using one-way ANOVA, followed by Tukey’s multiple comparison test. (E) Progeny of igfbp5a+/− intercrosses were sampled at the indicated stages. After NaR cells were visualized by in situ hybridization using a trpv5/6 complementary RNA (cRNA) probe and quantified, the fish were genotyped individually. Data shown are means ± SEM. n = 6 to 14 fish per group. No statistically significant differences were found by one-way ANOVA, followed by Tukey’s multiple comparison test.

To investigate whether Igfbp5a plays a local role in NaR cell development, we randomly sampled progeny from an igfbp5a+/− intercross at various stages. NaR cells were visualized and quantified, and each fish was genotyped individually. There was no difference in the NaR cell number between the igfbp5a−/− mutants and their siblings at all stages examined (Fig. 2E). A heteroduplex mobility assay (HMA) confirmed the expression of igfbp5aΔ11 mutant mRNA in igfbp5a−/− mutants (fig. S2A). The expression of igfbp5a was similar between igfbp5a−/− mutant fish and their siblings (fig. S2B), suggesting that nonsense-mediated decay of igfbp5a mRNA did not occur. Furthermore, the expression pattern (fig. S3, A and B) and mRNA levels (fig. S3C) of the paralogous igfbp5b were also similar between wild-type and igfbp5a−/− mutant fish. Finally, low-[Ca2+] treatment did not change igfbp5b mRNA levels in either wild-type or igfbp5a−/− mutant fish (fig. S3D).

To confirm that igfbp5a−/− fish were prone to premature death, we performed a low-[Ca2+] challenge test with 5-dpf larvae obtained from an igfbp5a+/− intercross. igfbp5a−/− mutant fish died earlier and at a significantly greater rate (Fig. 3A). On 2 and 3 days after the low-[Ca2+] challenge, igfbp5a−/− mutants comprised 66 and 38% of the dead larvae. These values were much greater than the expected 25% according to Mendelian ratio. Similarly, igfbp5aΔ4 mutant fish died at a significantly greater rate under the low-[Ca2+] challenge (fig. S4). We hypothesized that the premature death of igfbp5a−/− fish was due to Ca2+ deficiency in the body. Alizarin red staining showed that calcified bone mass in mutant fish raised in a solution containing 0.2 mM [Ca2+] (referred to as regular [Ca2+] solution hereafter because its [Ca2+] is similar to fish system water) was comparable to that of wild-type fish raised in a solution with 10-fold lower Ca2+ levels (Fig. 3B). Calcified bone mass in igfbp5a−/− fish was further reduced by exposure to the lower [Ca2+] solution (Fig. 3, B and C). Increased expression of trpv5/6, which encodes a channel that fluxes cations including Ca2+, is a hallmark of body calcium deficiency (1, 4). igfbp5a−/− mutants had increased trpv5/6 expression in NaR cells when raised in regular [Ca2+] solution (Fig. 3D). Low-[Ca2+] challenge treatment increased trpv5/6 mRNA levels in both wild-type and igfbp5a−/− fish, but the magnitude of increase was greater in igfbp5a−/− fish. These results suggest that deletion of Igfbp5a leads to body Ca2+ deficiency, and low-[Ca2+] treatment further exacerbates the Ca2+ deficiency.

Fig. 3 Deletion of Igfbp5a increases mortality and impairs adaptive NaR cell proliferation.

(A) Progeny of igfbp5a+/− intercrosses were raised in E3 solution until 5 dpf and transferred into low-[Ca2+] solution. Dead larvae were collected daily and genotyped. The survival curve is shown, and the numbers of fish are indicated. P < 0.0001 by log-rank test. (B and C) Fish of the indicated genotypes were raised in solutions containing the indicated Ca2+ concentration to 11 dpf and stained by Alizarin red. Representative images are shown in (B), and the average number of calcified vertebral columns is shown in (C). n = 19 to 29 fish per group. ***P < 0.001, ****P < 0.0001, unpaired two-tailed t test. Scale bar, 0.5 mm. (D) Embryos of the indicated genotypes were raised in E3 solution until 72 hpf and transferred to solutions containing the indicated Ca2+ concentration. The expression of trpv5/6 mRNA was measured by quantitative polymerase chain reaction (qPCR) and normalized to the NaR cell number for the group. Values are means ± SEM of three independent experiments, each containing RNA isolated from a pool of 20 larvae of the indicated stages. Different letters indicate statistically significant differences (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test). (E) Progeny of igfbp5a+/− intercrosses were raised in E3 solution to 72 hpf and transferred to solutions containing the indicated Ca2+ concentration. The fish were sampled at 120 hpf. NaR cells were visualized by in situ hybridization using the trpv5/6 cRNA probe and quantified. Each larva was genotyped afterward. Data shown are means ± SEM. n = 13 to 40 fish per group. Different letters indicate statistically significant differences (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test). (F) Tg(igfbp5a:GFP) fish of the indicated genotypes were raised in E3 solution until 72 hpf and transferred to low-[Ca2+] solution. The number of GFP-labeled NaR cells was quantified at 72, 96, 104, and 120 hpf. Data shown are means ± SEM. n = 7 to 36 fish per group. Different letters indicate statistically significant differences (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test).

When acclimated to low-[Ca2+] environments, NaR cells undergo proliferation to yield more NaR cells to ensure sufficient Ca2+ uptake for survival (3, 4). We tested whether deletion of Igfbp5a impaired this adaptive proliferation response. Visualization of NaR cells by in situ hybridization using a trpv5/6 probe showed that 48 hours of low-[Ca2+] treatment significantly increased the NaR cell number in wild-type and heterozygous larvae, an increase that was blunted in igfbp5a−/− larvae (Fig. 3E). We next crossed igfbp5a−/− fish with Tg(igfbp5a:GFP) fish, a transgenic line in which NaR cells are labeled by green fluorescent protein (GFP) expression, allowing real-time visualization and quantification of NaR cell division (4). Low-[Ca2+] treatment resulted in significant increases in NaR cells in Tg(igfbp5a:GFP) larvae at 96, 104, and 120 hpf (Fig. 3F). This increase was significantly reduced in igfbp5a−/−;Tg(igfbp5a:GFP) larvae (Fig. 3F). Next, we performed transient knockout experiments in Tg(igfbp5a:GFP) fish using a CRISPR-Cas9–based approach (fig. S5A). Transient deletion of Igfbp5a using two different synthetic guide RNAs (sgRNAs) significantly decreased NaR cell proliferation under low-[Ca2+], but not under normal [Ca2+] (fig. S5B). Therefore, either permanent or transient deletion of Igfbp5a impairs the adaptive proliferation response of NaR cells.

RNA expression analysis showed that among members of the IGF and insulin gene family, igf2b was most abundantly expressed in NaR cells, followed by igf2a (fig. S6A). The expression of igf1, igf3 (also known as igf1b), insulin a, and insulin b was minimal (fig. S6A). The strong expression of igf2b mRNA in NaR cells was confirmed by in situ hybridization analysis of Tg(igfbp5a:GFP) larvae (fig. S6B). The two igf1r genes and two insr genes were all expressed in NaR cells with igf1ra being the most abundant (fig. S6A). Low-[Ca2+] treatment did not alter the expression of these genes (fig. S6, A and B).

Our previous studies have shown that low-[Ca2+] treatment induces the phosphorylation of Akt exclusively in NaR cells (3, 4). Accordingly, immunostaining analysis revealed that there were few phosphorylated Akt–positive NaR cells on the yolk sac in wild-type larvae under regular [Ca2+] conditions. Low-[Ca2+] treatment resulted in a robust increase in NaR cells positive for phosphorylated Akt (Fig. 4, A and B). In comparison, low-[Ca2+] treatment resulted in significantly fewer NaR cells that were positive for phosphorylated Akt in igfbp5a−/− larvae (Fig. 4, A and B), suggesting that deletion of Igfbp5a impairs Akt signaling in NaR cells. Similar results were observed with phosphorylated S6, an indicator of Tor signaling, although basal phosphorylated S6 levels were higher (Fig. 4, C and D). In contrast, phosphorylated Erk levels were similar between wild-type and igfbp5a−/− larvae under both regular and low-[Ca2+] conditions (Fig. 4E and fig. S7). The phosphorylated Erk signal was authentic because it was abolished by MAPK (mitogen-activated protein kinase) kinase (MEK) inhibitor treatment (Fig. 4E and fig. S7). Deletion of the paralogous igfbp5b gene using CRISPR-Cas9 did not change the induction of phosphorylated Akt in NaR cells by low-[Ca2+] treatment (Fig. 4F).

Fig. 4 Deletion of Igfbp5a but not its paralog Igfbp5b impairs low-[Ca2+] stress–induced Akt-Tor signaling in NaR cells.

(A to D) Fish of the indicated genotypes raised in E3 solution were transferred to solutions containing the indicated Ca2+ concentration at 72 hpf. After 24 hours, they were stained for phosphorylated Akt (pAkt) (A and B) or phosphorylated S6 (pS6) (C and D). Representative images are shown in (A) and (C). Scale bar, 0.1 mm. The numbers of cells positive for phosphorylated Akt or phosphorylated S6 in each fish were quantified, and the quantitative results are shown in (B) and (D). Values are means ± SEM. n = 6 to 12 fish per group. Different letters indicate statistically significant differences between groups (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test). (E) Fish of the indicated genotypes were raised in E3 solution to 72 hpf and transferred to solutions containing the indicated [Ca2+] with or without U0126 (10 μM) or PD98059 (10 μM). After 8 hours, they were stained for phosphorylated Erk (pErk). The phosphorylated Erk signal was scored following a previously published scoring system (3). Total numbers of fish are shown above the bar. (F) Larvae of the indicated genotypes were raised in E3 solution to 72 hpf and transferred to solutions containing the indicated Ca2+ concentration. After 8 hours, they were stained for phosphorylated Akt. The number of cells positive for phosphorylated Akt was quantified. Values are means ± SEM. n = 13 to 18 fish per group. Different letters indicate statistically significant differences (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test).

To test whether expression of Igfbp5a was sufficient to activate IGF signaling in NaR cells, we randomly expressed zebrafish Igfbp5a, its ligand binding domain (LBD), or nuclear localization signal (NLS) mutants in NaR cells in igfbp5a−/−;Tg(igfbp5a:GFP) larvae using a Tol2 transposon bacterial artificial chromosome (BAC)–mediated genetic mosaic assay (Fig. 5, A and B). NaR cells expressing Igfbp5a proliferated under low-[Ca2+] stress and were positive for phosphorylated Akt (Fig. 5, C and D). NaR cell proliferation in igfbp5a−/−;Tg(igfbp5a:GFP) larvae under low-[Ca2+] stress was also restored by the expression of the Igfbp5a NLS mutant, but not by that of the LBD mutant (Fig. 5C), suggesting that Igfbp5a acts in an IGF binding–dependent manner. If Igfbp5a indeed stimulates NaR cell proliferation by binding to IGFs and activating Akt signaling, then expression of a constitutively active Akt should rescue the defective NaR cell proliferation in igfbp5a−/− larvae. Expression of myr-Akt restored NaR cell proliferation in response to low-[Ca2+] stress (Fig. 5, D and E). This approach does not rescue the lethality phenotype, presumably because it targets Igfbp5a expression transiently only in a small number of NaR cells.

Fig. 5 Reintroduction of Igfbp5a or constitutively active Akt in NaR cells restores adaptive proliferation.

(A) Structure of zebrafish Igfbp5a and its mutants. SP, N-, L-, and C-domains are indicated. Mutated residues are shown. (B and C) NaR division scoring system. Zebrafish igfbp5a−/−;Tg(igfbp5a:GFP) embryos injected with BAC(igfbp5a:Igfbp5a-IRESmCherry) constructs containing the indicated genes were raised in E3 solution until 72 hpf and transferred to low-[Ca2+] solutions. Igfbp5a-expressing NaR cells are labeled by both GFP and mCherry. The NaR division scoring system is shown in (B). During the 48-hour treatment period, NaR cells that divide 0, 1, or 2 times were scored as −, +, and ++. Scale bar, 0.05 mm. The quantified data are shown in (C). χ2 test was used for statistical analysis. **P < 0.01, ***P < 0.001. Total numbers of cells are shown above the bar. (D) The larvae described in (B) and (C) were collected after 24 hours in low-[Ca2+] solution and stained for phosphorylated Akt and mCherry. Scale bar, 0.02 mm. (E) BAC(igfbp5a:myr-Akt-mCherry) DNA was injected into igfbp5a−/−;Tg(igfbp5a:GFP) embryos. NaR division was scored as described in (B). χ2 test was used for statistical analysis. ****P < 0.0001. Total numbers of cells are shown above the bar.

To test whether IGFBP5 acts in a similar manner in human epithelial cells, human colon carcinoma (LoVo) cells were transfected with a validated small interfering RNA (siRNA) plasmid (pSuperBP5) targeting human IGFBP5 L-domain (17). In cells transfected with the empty pSuper plasmid, IGF2 treatment resulted in a significant increase in the percentage of S-phase cells. This increase was abolished in the pSuperBP5 transfected cells (Fig. 6A), suggesting that the endogenous IGFBP5 is required in the proliferative response of these human cells to IGF2 stimulation. To demonstrate the specificity of the siRNA knockdown effect and the sufficiency of IGFBP5, we cotransfected LoVo cells with pSuperBP5 and IGFBP545, which is not targeted by pSuperBP5 because it contains the IGFBP5 N- and C-domain but the L-domain from IGFBP4 (17). Expression of IGFBP545 restored the proliferation response to IGF2 stimulation (Fig. 6A). To further investigate the function of human IGFBP5 and its mechanism of action in an in vivo setting, we expressed human IGFBP5 and two mutants (G223R and W242*) in NaR cells in igfbp5a−/−;Tg(igfbp5a:GFP) embryos using the Tol2 transposon BAC genetic mosaic assay (Fig. 6B). G223R and W242* are two cancer-associated mutations that have been speculated to have lost IGF binding ability (18). Ectopic expression of wild-type human IGFBP5, but not these mutants, significantly increased NaR cell proliferation under low-[Ca2+] stress (Fig. 6C). These results suggest that human IGFBP5 acts by binding to the IGF ligand and promoting IGF action in human LoVo cells.

Fig. 6 IGFBP5 but not two cancer-associated IGFBP5 mutants promotes the mitotic response of human epithelial cells to IGF stimulation.

(A) Human LoVo epithelial cells were transfected with the indicated siRNA construct with or without pCMV-BP545. After 24 hours, they were treated with human IGF2 (400 ng/ml) for 3 days and analyzed by flow cytometry. The percentage of S-phase cells is shown. Values shown are means ± SEM. n = 3 independent experiments. Groups labeled with different letters are statistically significantly different from each other (P < 0.05 by one-way ANOVA, followed by Tukey’s multiple comparison test). (B) Structure of human IGFBP5 and its two cancer-associated mutants. SP, N-, L-, and C-domains are indicated. Mutated residues are shown. Asterisk (*) indicates a stop codon. (C) Zebrafish igfbp5a−/−;Tg(igfbp5a:GFP) embryos injected with the indicated human IGFBP5 construct were raised in E3 solution until 72 hpf and transferred to low-[Ca2+] solution. Forty-eight hours later, GFP and mCherry double-positive cells were quantified using the scoring system shown in Fig. 5B. χ2 test was used for statistical analysis. **P < 0.01, ***P < 0.001, ****P < 0.001. Total numbers of cells are shown above the bar.

DISCUSSION

IGFBP5 is the most conserved member of the IGFBP family (11, 12) and has been reported to inhibit or potentiate IGF or acts as a growth factor itself in mammalian cells (912). Despite these findings, IGFBP5-deficient mice do not have overt phenotypes (19). Here, we showed that genetic deletion of zebrafish Igfbp5a causes body Ca2+ deficiency and premature death under regular laboratory conditions. The igfbp5a−/− mutant larvae exhibited clear and specific defects in NaR cell adaptive proliferation under Ca2+-deficient conditions. These results reveal a critical role of Igfbp5a in regulating organismal survival and calcium balance by promoting NaR cell proliferation under low-calcium states.

The lack of phenotype in IGFBP5-null or other IGFBP-null mice has been postulated to be due to the genetic redundancy in the mouse model (12). The use of CRISPR-Cas9 and TALEN to delete and edit genes suggest that the phenotype gap is also present in zebrafish. Many cases have been reported in which permanent deletion of gene(s)-encoding various hormones and growth factors, including many that were previously thought to be critical or even essential, does not result in notable phenotypic changes (16, 2023). Genetic studies in zebrafish have also found major discrepancies in phenotypes between permanent knockout and transient knockdowns (2427). Several additional mechanisms, including genetic compensation and/or transcriptional adaptation, nonsense-mediated decay, and altered mRNA processing, have been proposed to account for the lack of phenotypes in permanent genetic mutants (16, 27). These mechanisms cannot fully explain the findings of this study. Because the zebrafish genome contains nine igfbp genes due to a teleost-specific genome-wide duplication event (28), one could argue that zebrafish may have a greater genetic redundancy than mice. For example, there are two igfbp5 genes in zebrafish compared to a single IGFBP5 gene in mice and humans (29). Nevertheless, our genetic data showed that permanent deletion of Igfbp5a compromised the adaptive NaR cell proliferation response and led to low body Ca2+ content and premature death. We did not detect nonsense-mediated mRNA decay of igfbp5a mRNA or observe compensatory increases in the paralogous igfbp5b expression in the igfbp5a−/− mutant fish. Furthermore, transient deletion of Igfbp5a resulted in similar defects in NaR cells. The phenotypes of the igfbp5a−/− mutant fish were context-dependent. When raised in [Ca2+]-rich solutions, igfbp5a−/− mutant fish grew to the adult stage and appeared indistinguishable from their siblings. However, when raised in low-[Ca2+] solutions, they showed specific defects in adaptive NaR cell proliferation and died prematurely. These context-dependent phenotypes were specific because these changes were not found in igfbp5b−/− mutant fish. Our findings underscore the importance of physiological context in genetic dissection of the highly redundant endocrine signaling network.

In addition to regulating global growth and development (30, 31), IGF signaling also exerts many cell type–specific and context-dependent actions (11, 23). How the IGF signaling system exerts these cell type–specific and context-dependent actions is not well understood. Here, we showed that genetic deletion of Igfbp5a, but not the closely related Igfbp5b, reduced IGF1R-mediated Akt and Tor signaling specifically in NaR cells and in a Ca2+ concentration–dependent manner. Reintroducing Igfbp5a was sufficient to activate Akt signaling in mutant NaR cells and restore their proliferation. Moreover, knockdown of IGFBP5 abolished the proliferation of human LoVo cells in response to IGF2 stimulation, an effect that was rescued by expression of a siRNA-resistant form of IGFBP5. These results reveal a conserved mechanism by which a locally expressed IGFBP activates the IGF signaling in a cell type–specific and context-dependent manner.

IGFBP5 appears to have several modes of actions. When added to cultured human or rodent cells together with IGF1 or IGF2, IGFBP5 inhibits IGF-induced cell proliferation and/or differentiation (18, 29, 3234). IGFBP5 can also potentiate IGF-induced cell proliferation and/or survival (3539). IGFBP5 also has IGF-independent activities (1012). Mammalian IGFBP5 is detected in the nucleus and has nuclear activity (40, 41), properties also observed in zebrafish (29). How IGFBP5 acts in vivo under a physiological context has not been reported. Here, we found that Igfbp5a acted by binding to IGF and activating Akt-Tor signaling in NaR cells and that nuclear localization is not important for this action of Igfbp5a. Our gene expression analysis indicated that both igf1r genes were detected in NaR cells with igf1ra being the more abundant isoform. Future studies will be needed to determine whether Igfbp5 acts by binding to Igf2b to activate Igf1ra and/or Igf1rb in NaR cells. Mammalian cell culture and biochemical studies suggest that IGFBP5 can potentiate IGF action by bringing IGF to the IGF1R through the proteolysis of IGFBP5 or through association with the cellular membrane or extracellular matrix proteins (37, 42, 43). Whether these mechanisms are involved in Igfbp5a action in NaR cells need to be elucidated with future studies.

Aberrant regulation or mutations of key components in the IGF-PI3K-AKT-TOR signaling pathway have been linked to major human diseases including cancers (44). In colon cancers, for example, several genes belonging to this pathway are mutated (45). These mutations result in increased IGF2 expression, increased PIK3CA and PIK3R1 activity, and reduced PTEN expression or activity (46). Ding et al. (18) have identified more than 20 nonsynonymous mutations for IGFBP5 in cancer, including several frameshift and nonsense mutations. Because the IGF-PI3K-Akt-Tor pathway is highly conserved and activation of this pathway is both required and sufficient for NaR cell proliferation, the zebrafish larvae–NaR cell system provides an excellent in vivo platform for functional analysis of these cancer-associated mutations. As a proof-of principle experiment, we assessed two cancer-associated IGFBP5 mutants, G223R and W242*. Unlike wild-type IGFBP5, neither mutant could stimulate NaR cell proliferation in vivo. These results provided in vivo evidence supporting the idea that IGFBP5 G223R and W242* have lost IGF binding ability (18).

In summary, the findings in this study reveal a conserved mechanism by which a locally expressed Igfbp regulates organismal Ca2+ homeostasis and survival by activating IGF signaling in epithelial cells and promoting their proliferation under Ca2+-deficient states. This regulatory mechanism may not be limited to zebrafish or IGFBP5. Although IGFBP5 knockout mice exhibit normal mammary gland development under normal conditions, these animals show delayed mammary gland involution and enhanced alveolar bud formation after ovariectomy, followed by estrogen treatment (19). In breast cancer cells, doxorubicin-induced senescence increases IGFBP3 expression and cleavage of IGFBP3 prevents cell senescence (9). In human vascular endothelial cells, hypoxia induces the expression of IGFBP6, which in turn suppresses angiogenesis as a negative feedback during tumor growth (47). Future studies are needed to identify additional stress or disease conditions to which different IGFBPs respond to and to determine whether regulation of IGF signaling locally by IGFBPs is a widespread mechanism.

MATERIALS AND METHODS

Ethics statement

All experiments were conducted in accordance with the guidelines approved by the Institutional Committee on the Use and Care of Animals, University of Michigan.

Zebrafish husbandry and embryo rearing media with different Ca2+ concentrations

All zebrafish lines were raised and crossed according to zebrafish husbandry guidelines (48). Embryos were obtained by natural cross and staged according to Kimmel et al. (49). Several embryo rearing media were used. A standard solution containing 0.33 mM [Ca2+], referred to as E3 solution, was prepared as described in The Zebrafish Book: A Guide for the Laboratory Use of Zebrafish (Danio rerio) (48). A solution containing high concentration of Ca2+ (1.6 mM [Ca2+]), referred to as 5PPT high [Ca2+] solution, was prepared by dissolving 5 g of Instant Ocean salt in 1000 ml of distilled water. Two embryo rearing solutions containing either 0.2 mM [Ca2+] (referred to as regular [Ca2+] solution) or 0.001 mM [Ca2+] (referred to as low-[Ca2+] solution) were prepared as previously reported (3). To inhibit pigmentation, 0.003% (w/v) N-phenylthiourea was added to these solutions.

Chemicals and reagents

Reagents were purchased from Thermo Fisher Scientific unless stated otherwise. The MEK inhibitors (U0126 and PD98059), phospho-Akt, phospho-S6, and phospho-Erk antibodies were purchased from Cell Signaling Technology. Restriction enzymes were purchased from New England BioLabs. Primers, cell culture media and supplements, the TRIzol reagent, Moloney murine leukemia virus (M-MLV) reverse transcriptase, and the Alexa Fluor 488 Tyramide SuperBoost Kit were purchased from Invitrogen. Cas9 protein was purchased from ToolGen. Calcein and Alizarin red S were purchased from Sigma-Aldrich. Anti-Digoxigenin-POD and Anti-Digoxigenin-AP were purchased from Roche.

Generation of igfbp5a mutant line using TALEN

TALEN target sites were designed following published criteria (50) and are shown in fig. S1A. TALEN constructs were assembled using a Golden Gate cloning strategy following a published protocol (51). The assembled plasmids were determined by Stu I and Afl II digestion and confirmed by DNA sequencing.

TALEN-capped mRNA was synthesized by in vitro transcription using T3 polymerase (Ambion mMessage mMachine kit) and injected into embryos at the one-cell stage. After confirming indels using T7E1 assays in a subset of pooled F0 embryos, the remaining F0 embryos were raised to adulthood and crossed with wild-type fish. F1 fish were raised, and caudal fin was severed and genotyped as described below. Positive F1 fish were subjected to DNA sequencing. The F1 adult fish were intercrossed, and the progeny were raised and genotyped as described below. To obtain the viable homozygous fish, progeny from the F1 intercrosses were raised in the 5PPT high [Ca2+] solution in the first month and then transferred into the fish system.

Transient and stable knockout of igfbp5a and 5b genes using CRISPR-Cas9

Two sgRNAs targeting igfbp5a exon 1 and two sgRNAs targeting igfbp5b exon 1 were designed using the online tool ZiFiT Targeter V4.2 (http://zifit.partners.org/ZiFiT). Their sequences are as follows: igfbp5a-sgRNA-1, 5′-GGTGCGGAACGGGTGTCAGGTGG-3′ (sense strand, +114 to +136); igfbp5a-sgRNA-2, 5′-GTTCCGCACCGGAGGACACATGG-3′ (anti-sense strand, +123 to +101); igfbp5b-sgRNA-1, 5′-GGTGGGCTGTCAGCTAGTGAAGG-3′ (sense strand, +114 to +136); and igfbp5b-sgRNA-2, 5′-GTGCGAGCCGTGCGATCAGAAGG-3′ (sense strand, +66 to +88). These sgRNAs were synthesized by in vitro transcription following a published method (52). sgRNAs (30 ng/μl) were mixed with Cas9 protein (700 ng/μl) or mRNA (250 ng/μl) and coinjected into Tg(igfbp5a:GFP) or wild-type embryos at the one-cell stage. The injected embryos were raised in E3 solution until 72 hpf and transferred to either the regular or low-[Ca2+] solution. NaR cells were quantified at 120 hpf as previously reported (4). To generate the igfbp5b mutant line, F0 fish were screened using T7E1 assays with the following primers: igfbp5b-gt-F, 5′-AACCCCAACCTGTGTTTTCA-3′; igfbp5b-gt-R, 5′-GATACAAACCACCGCACCCA-3′. igfbp5b homozygous mutants were obtained following the cross strategy described above.

Genotyping

To isolate genomic DNA, pooled embryos or individual adult caudal fin was incubated in 50 μl of NaOH (50 mM) at 95°C for 10 min, which was neutralized by adding 5 μl of 1 M tris-HCl (pH 8.0). T7E1 assay and HMA were carried out as previously described (53). For T7E1 assays, PCR was performed using the following primers: igfbp5a-TAL-nF, 5′-GAACGCTGTTCGCTTGAT-3′; igfbp5a-TAL-nR, 5′-CGCAGTCCTCGTCCACAT-3′. For HMA assays, PCR was performed using the following primers: igfbp5a-HRMA1-F, 5′-GTGGCATTTGTATGGGGACT-3′; igfbp5a-HRMA2-R, 5′-AGGTCAAGCAGCATCCGCA-3′.

Individual genotyping was carried out as previously reported (3). Briefly, after immunostaining or in situ hybridization analysis, each embryo or larva was washed with 1× phosphate-buffered saline with Tween 20 (PBST) twice and methanol three times. They were air-dried overnight. Then, each embryo or larva was digested in 100 μl SZL buffer [50 mM KCl, 2.5 mM MgCl2, 10 mM tris-HCl (pH 8.3), 0.45% NP-40, 0.45% Tween 20, 0.01% gelatine] containing proteinase K (100 μg/ml) for 1 hour at 60°C. The reaction was stopped by 15-min heat treatment (95°C). After spinning at 10,000 rpm for 1 min, the supernatant was used as template for PCR, followed by HMA assay as previously described (53).

Zebrafish larva survival experiments

For survival experiments, embryos or larvae of different genotypes were raised in E3 solution to 72 or 120 hpf. They were then transferred to the low-[Ca2+] solution. Dead larvae were collected daily and genotyped individually.

Morphology analysis

Body length is defined as the curvilinear distance from the head to the end of caudal tail. Somite number and head-trunk angles were measured manually (54, 55). To visualize the bone morphology, calcein and Alizarin red staining were performed following a published protocol (56). For calcein staining, juvenile fish were immersed in the calcein staining solution for 10 min. Then, they were washed three times in E3 solution to remove residual calcein. Images were captured with a stereomicroscope (Leica MZ16F, Leica) equipped with a QImaging QICAM camera (QImaging).

qPCR, whole-mount in situ hybridization, and immunostaining

Total RNA was isolated using TRIzol reagent (Invitrogen). RNA samples were treated with deoxyribonuclease I to remove DNA. One microgram of RNA was reverse-transcribed to single-strand complementary DNA using M-MLV (Promega) and oligo(dT)18 primer. Reverse transcription PCR was performed using Taq DNA polymerase (New England BioLabs). qPCR was carried out using SYBR Green (Bio-Rad) on a StepONE PLUS real-time thermocycler (Applied Biosystems). Primers for qPCR used in this study are shown as follows: igfbp5a, 5′- GCTGCACGCTCTGCTTTAC-3′ and 5′-AATGGAACCTTGGCCTGAG-3′; igfbp5b, 5′-GGGAGTGTGTACGAACGAGAA-3′ and 5′-TCCTGTCACAGTTAGGCAGGTA-3′; trpv5/6, 5′-GGACCCTACGTCATTGTGATAC-3′ and 5′- GGTACTGCGGAAGTGCTAAG-3′; β-actin, 5′-GATCTGGCATCACACCTTCTAC-3′ and 5′-CCTGGATGGCCACATACAT-3′.

Whole-mount in situ hybridization and immunostaining were performed as previously reported (3). For simultaneous in situ hybridization and immunostaining, mCherry or igfbp2b mRNA was detected using digoxigenin (DIG)–labeled antisense RNA probes (57). The mCherry signal was detected using anti–DIG-POD antibody (Roche), followed by Alexa 488 Tyramide Signal Amplification (Invitrogen). The igfbp2b signal was detected using anti–DIG-AP antibody (Roche), followed by BCIP-NBT (bromochloroindolyl phosphate–nitro blue tetrazolium) staining. After in situ hybridization analysis, the stained larvae were washed in 1× PBST and incubated with phosphorylated Akt or GFP antibody overnight at 4°C. Next, they were stained with a Cy3-conjugated goat anti-rabbit immunoglobulin G antibody (Jackson ImmunoResearch). Fluorescent images were acquired using a Nikon Eclipse E600 Fluorescence Microscope with PMCapture Pro 6 software.

Plasmid and BAC constructs

The construction of the Igfbp5a expression plasmid has been reported (3). This plasmid was used as a template to generate Igfbp5a NLS and LBD mutants. For the NLS mutant, the 236KGRK239 motif in Igfbp5a C-domain was changed to LDGQ, the corresponding sequence of zebrafish Igfbp1a (58) by site-directed mutagenesis (table S1). For the LBD mutant, the sequence 86KPLLL92 in Igfbp5a was mutated to NQQQQ (59). Human IGFBP5 G223R and IGFBP5 W242* were generated by site-direct mutagenesis (table S1). The resulting Igfbp5a/IGFBP5 DNAs were cloned into the pIRES2-mCherry using the Eco RI and Bam HI sites. The pIRES2-mCherry plasmid was engineered by replacing the DsRed-Express2 cassette in the pIRES2-DsRed-Express2 plasmid with mCherry using the Bst XI and Not I sites. The Igfbp5a/IGFBP5-IRES2-mCherry-KanR cassette DNAs were amplified by PCR using primers listed in table S1. They were inserted into the igfbp5a BAC construct (4) to replace the igfbp5a sequence from the start codon to the end of the first exon through homologous recombination. The resulting BAC DNA was validated by DNA sequencing using primers shown in table S1. The validated BAC(igfbp5a:Igfbp5a/IGFBP5-mCherry) DNA constructs were injected into igfbp5a−/−;Tg(igfbp5a:GFP) embryos at the one-cell stage. The embryos were raised in the E3 solution until 72 hpf and then subjected to the low-[Ca2+] challenge test by transferring into low-[Ca2+] solution. Cells coexpressing mCherry and GFP were defined as igfbp5a/IGFBP5-expressing NaR cells. They were scored using the scoring system shown in Fig. 5B. A similar cloning strategy was used to generate BAC(igfbp5a:myr-Akt-mCherry) plasmids. Constructs were confirmed by DNA sequencing at the University of Michigan DNA Sequencing Core Facility.

Cell culture, transfection, and flow cytometry analysis

Human colorectal carcinoma (LoVo) cells were purchased from American Type Culture Collection and cultured in Dulbecco’s modified Eagle’s medium/F-12 supplemented with 10% fetal bovine serum, penicillin, and streptomycin in a humidified-air atmosphere incubator containing 5% CO2 at 37°C. LoVo cells were seeded in six-well plates (Falcon) and transfected with 2 μg of pSuperBP5 or pSuper and 2 μg of pCMV-BP545 or pCMV plasmids using Lipofectamine 2000 (Invitrogen). Twenty-four hours after transfection, the cells were washed and incubated with fresh serum-free medium for 8 hours. IGF2 (400 ng/ml) was then added. Cells were collected after 3 days of IGF2 treatment. Cell cycle analysis was carried out using an Attune Acoustic Focusing Cytometer (Applied Biosystems, Life Technologies), as previously described (4). Briefly, a two-parameter dot plot of forward scatter (FSC) versus side scatter was constructed along with a two-parameter dot plot of FL2 [propidium iodide (PI)] area versus FSC area. In addition, a single-parameter FL2 (PI) area histogram was constructed to illustrate relative DNA content in each cell cycle phase.

Statistical analysis

Values are shown as means ± SEM. Statistical significance among experimental groups was determined using unpaired two-tailed t test, one-way ANOVA, followed by Tukey’s multiple comparison test, log-rank test, or χ2 test. Statistical significances were accepted at P < 0.05 or greater.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/548/eaat2231/DC1

Fig. S1. Generation of igfbp5a mutant fish.

Fig. S2. igfbp5a mRNA levels in mutant fish.

Fig. S3. Deletion of Igfbp5a does not alter igfbp5b expression.

Fig. S4. igfbp5a Δ4 mutant fish are prone to dying under low Ca2+ stress.

Fig. S5. CRISPR-Cas9–mediated transient deletion of Igfbp5a impairs adaptive NaR cell proliferation.

Fig. S6. Expression of IGF and insulin ligand and receptor genes in NaR cells.

Fig. S7. Deletion of Igfbp5a has no effect on phosphorylated Erk signaling.

Table S1. Primers for cloning and sequencing of Igfbp5a, IGFBP5, and their mutants.

REFERENCES AND NOTES

Acknowledgments: We thank J. Allard for critically reading and editing of the initial draft of the manuscript. This work was supported by NSF grant IOS-1557850 and MCube2.0 Project U0496246 to C.D. Author contributions: C.D. conceived the study. C.D., C.L., and Y.X. designed the experiments and wrote the paper. C.L., Y.X., Y.B., and G.L. performed the experiments and carried out the analysis. G.H. and K.M. analyzed the data. All authors have contributed to and approved the final manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article