In Xenopus laevis embryos, kidney field specification is dependent on retinoic acid (RA) and coincides with a dramatic increase of Ca2+ transients, but the role of Ca2+ signaling in the kidney field is unknown. Here, we identify TRPP2, a member of the transient receptor potential (TRP) superfamily of channel proteins encoded by the pkd2 gene, as a central component of Ca2+ signaling in the kidney field. TRPP2 is strongly expressed at the plasma membrane where it might regulate extracellular Ca2+ entry. Knockdown of pkd2 in the kidney field results in the downregulation of pax8, but not of other kidney field genes (lhx1, osr1 and osr2). We further show that inhibition of Ca2+ signaling with an inducible Ca2+ chelator also causes downregulation of pax8, and that pkd2 knockdown results in a severe inhibition of Ca2+ transients in kidney field explants. Finally, we show that disruption of RA results both in an inhibition of intracellular Ca2+ signaling and of TRPP2 incorporation into the plasma membrane of kidney field cells. We propose that TRPP2-dependent Ca2+ signaling is a key component of pax8 regulation in the kidney field downstream of RA-mediated non-transcriptional control of TRPP2.
The vertebrate kidney originates from the dorso-lateral mesoderm and develops through the formation of three successive renal structures; the pro-, meso- and meta-nephros. These three structures represent organs of increasing complexity, which contain a similar basic unit of filtration, the nephron. The structure of the nephron and the molecular mechanisms involved in its formation are evolutionarily conserved (Dressler, 2009; Saxen, 1987; Vize et al., 2003). In Xenopus laevis tadpoles, the functional pronephros is made of a single nephron, thus providing a simple system to study how pluripotent mesodermal cells are committed to a renal fate (Jones, 2005).
Specification of mesodermal cells to a pronephric fate is achieved at late gastrula and early neurula stages. Dorso-lateral mesodermal explants isolated at these stages differentiate into the glomus and tubule (Brennan et al., 1998; Brennan et al., 1999). Several genes encoding transcription factors whose murine orthologs are known to play an important function during renal development (Dressler, 2009) are expressed in this territory at the onset of neurulation. They include pax8, lhx1, osr1 and osr2 (Carroll and Vize, 1999; Tena et al., 2007). Loss of function of osr1, osr2 (Tena et al., 2007) and lhx1 (Chan et al., 2000) impair the development of the pronephros, whereas overexpression of pax8 and/or lhx1 results in the development of ectopic and enlarged pronephroi (Carroll and Vize, 1999). The mesodermal territory where pax8 and lhx1 expression overlap has been defined as the pronephric field or kidney field.
The mechanisms controlling kidney field emergence from the mesodermal layer at gastrula stages are only partially understood. Retinoic acid (RA) signals during gastrulation are absolutely required. Disruption of RA signaling results in a loss of pax8 and lhx1 expression in the kidney field, and in a loss of glomus and tubule development at later stages (Cartry et al., 2006). Wnt-11b induces pronephric structures in unspecified lateral mesoderm explants, and also acts as a potential inducer (Tételin and Jones, 2010). In contrast, FGF signaling needs to be downregulated to allow mesodermal cells to adopt a pronephric fate (Colas et al., 2008; Le Bouffant et al., 2012).
Ca2+ signaling is also an important regulator of pronephric tubule differentiation. Previously we have recorded transient increases in intracellular Ca2+ concentrations ([Ca2+]i) in the dorso-lateral mesoderm with maximum of Ca2+ activities at early neurula stage, that is, coinciding with kidney field specification. The spatial distribution of Ca2+ transients in cultured kidney field explants are concentrated within the area where the pronephric tubule will later differentiate. Inhibition of these transients with Ca2+ chelators at late gastrula or mid-neurula stages results in a defective development of the tubule (Leclerc et al., 2008). The origin of these Ca2+ transients has not been attributed to specific channels. However, the inhibitory effect of lanthanum chloride (La3+), a potent antagonist of most transient receptor potential (TRP) channels (Clapham et al., 2005), on tubulogenesis strongly suggests that TRP channels are involved in the process (Leclerc et al., 2008). One likely candidate is TRPP2.
TRPP2 (also known as polycystin-2 and encoded by the pkd2 gene), is a member of the TRP superfamily of channels, which are known to function as non-selective cation channels that are permeable to Ca2+ (Köttgen, 2007). TRPP2 can function at several subcellular locations, including the basolateral plasma membrane of epithelial cells (Ma et al., 2005), endoplasmic reticulum (ER), where it might operate as a Ca2+-release channel (Koulen et al., 2002) and primary cilia, where it is thought to participate in mechanosensation (Nauli et al., 2003). At primary cilia, the function of TRPP2 has been linked to kidney cysts formation in vertebrates. Human PKD2 mutations are responsible for a large number of cases of autosomal dominant polycystic kidney disease (ADPKD) (Mochizuki et al., 1996) and null mutation of the murine ortholog Pkd2 is embryonic lethal. Pkd2 mutants exhibit severe cardio-vascular defects, and form cysts in developing nephrons and pancreatic duct (Wu et al., 2000). Morpholino-based loss of function of pkd2 also results in the formation of pronephric cysts during zebrafish development (Obara et al., 2006; Sun et al., 2004) and in the formation of severe edema and of dilated pronephric tubules in Xenopus (Tran et al., 2010).
Here, we address the potential role of TRPP2-dependent Ca2+ signaling during early pronephros development in Xenopus embryos and show that TRPP2 is involved in the formation of the kidney field. Pkd2 loss of function causes a severe inhibition of pax8 expression in the kidney field. Interestingly, expression of other kidney field genes such as osr1, osr2 and lhx1 is unaffected. Furthermore, morpholino (MO)-mediated pkd2 depletion results in a dramatic decrease of the number of Ca2+ transients in kidney field explants, showing that TRPP2 is directly involved in the generation of Ca2+ signaling in this region of the embryo. Using a photo-inducible Ca2+ chelator, we show that kidney field expression of pax8 is decreased when Ca2+ signaling is inhibited from the midgastrula stage. This function of TRPP2 is associated with the localization of TRPP2 to the plasma membrane and not to the primary cilium. We further show that both Ca2+ signaling and TRPP2 incorporation into the plasma membrane are both dependent on RA signaling, suggesting that RA might play an important function in the control of Ca2+ signaling in the kidney field. Taken together, these data provide evidence that TRPP2-dependent Ca2+ signaling is acting very early during development to specifically control the expression of pax8, a master gene regulator of pronephric kidney development.
TRPP2 expression peaks at the late gastrula stage when the [Ca2+]i increase reaches its maximum in the emerging kidney field
Our previous data suggest that Ca2+ signaling during pronephros induction ex vivo by activin and RA involves TRP family channels (Leclerc et al., 2008). Therefore, to determine whether TRPP2 might be required during kidney field formation in vivo, we analyzed its expression during development using a cross-reacting antibody directed against human TRPP2 (supplementary material Fig. S1). TRPP2 expression was analyzed by western immunoblotting on whole embryo extracts. TRPP2 was detected throughout development (Fig. 1A). Although TRPP2 was only barely detected during segmentation stages and early gastrulation, expression levels raised at midgastrula stage [Nieuwkoop and Faber (NF) stage 11] to reach a maximum at the end of gastrulation. TRPP2 expression then dropped during neurula and early tailbud stages (NF stage 14–22), and increased again from mid-tailbud stage (NF stage 25) onwards, to reach high levels in the early tadpole (NF stage 35/36). The first period of high TRPP2 expression at the late gastrula stage therefore coincides with the maximum increase of [Ca2+]i in the kidney field of whole embryos (Leclerc et al., 2008). It is also interesting to note that the second period of high TRPP2 expression is correlated with tubule differentiation, where TRPP2 is known to play an important function (Tran et al., 2010).
We have further studied the localization of TRPP2 in cells of kidney field explants isolated at the late gastrula stage (NF stage 13), using immunofluorescence staining and confocal microscopy. In explants isolated from embryos expressing membrane-targeted GFP, anti-TRPP2 immunoreactivity colocalized with GFP, suggesting that TRPP2 is expressed at the plasma membrane (Fig. 1B).
TRPP2 has been found to be located to primary cilia in mammalian cells (Pazour et al., 2002; Yoder et al., 2002). In Xenopus neurula, TRPP2 is expressed in cilia of the gastrocoel roof plate (GRP), a ciliated epithelium forming the dorsal surface of the archenteron. TRPP2 in GRP plays a crucial role in the establishment of left–right asymmetry (Schweickert et al., 2007). At the tailbud stage, TRPP2 is also associated with cilia in another epithelium, the pronephric nephron tubule (Tran et al., 2010). We tested for the presence of cilia in cells from kidney field explants isolated at early (NF stage 13–14) or late (NF stage 18) neurula stage using anti-acetylated tubulin antibodies (supplementary material Fig. S2). GRP and kidney field are different embryonic tissues. As a positive control for the presence of cilia, GRP explants were taken at the same stages. No cilia-like structures were observed in kidney field explants, whereas they were readily detected in the GRP. In order to further attempt to detect cilia, we expressed an mCherry-tagged version of the axoneme protein arl13b (Borovina et al., 2010; Caspary et al., 2007; Chung et al., 2012), and analyzed its distribution in kidney field and GRP cells at NF stage 14. mCherry–arl13b was clearly localized in GRP cilia but never in kidney field cells (supplementary material Fig. S2). Our data strongly suggests that the function of TRPP2 at this early stage of kidney development is not linked to the primary cilium.
pkd2 knockdown results in the downregulation of pax8 in the kidney field without affecting other kidney field genes
To determine whether TRPP2 is required for kidney field specification, we analyzed whether pkd2 knockdown affects pax8 expression using previously validated pkd2-specific translation-blocking antisense MOs (Pkd2-MO) (Tran et al., 2010). Pkd2-MO was injected at the four-cell stage into the two left blastomeres (2 pmoles per blastomere) in order to target the presumptive dorso-lateral mesoderm containing kidney field precursors on the left side of the embryo. The right side of the embryo provided an internal control. Embryos were then cultured until early neurula stage (NF stage 14–15) for in situ hybridization (ISH) analysis of pax8 expression. Comparison of the left injected side and the right control side showed that pkd2 knockdown resulted in a strong inhibition of pax8 expression. Phenotypes ranged from a total extinction to a slight decrease of pax8 expression (Fig. 2A). With Pkd2-MO alone, the strongest phenotypes (total extinction plus strong pax8 inhibition) represented 91% of analyzed embryos (n = 91). Control morpholino (CMO) injection (2 pmoles per blastomere) had no effect upon pax8 expression (n = 37) (Fig. 2A). In order to assess pkd2 knockdown specificity, we tested whether co-injection of a Pkd2-MO-resistant version of Xenopus pkd2 mRNA (1 ng) was able to rescue pax8 expression. Co-injection of pkd2 mRNA indeed resulted in a strong decrease of the strongest phenotypes (62%, n = 98), and an increase of normal patterns of pax8 expression in the kidney field (Fig. 2A). The difference observed between frequency of strongest phenotypes in Pkd2-MO injected embryos, and in Pkd2-MO plus pkd2 mRNA-injected embryos is significant (P<0.01, Pearson's chi-square test of independence), showing that pkd2 knockdown specifically causes pax8 downregulation.
The induction of the kidney field has been proposed to depend on signals emanating from the paraxial mesoderm (Seufert et al., 1999; Tételin and Jones, 2010). In order to rule out the possibility that pkd2 knockdown indirectly interferes with pax8 expression in the kidney field by affecting paraxial mesoderm, we analyzed myod expression in pkd2 morphant embryos. Pkd2-MO was injected at the four-cell stage into the two left blastomeres (2 pmoles per blastomere), and pax8 and myod expression were investigated at early neurula stage (NF stage 14–15). Although pax8 was readily downregulated on the injected side (n = 18/20), the Pkd2-MO did not affect myod expression (n = 34/37). Analysis of mlc expression in somitic mesoderm in later morphant tailbud stage embryos (NF stage 28) further confirmed that Pkd2-MO did not affect development of the somites (n = 17/17) (supplementary material Fig. S3)
To study whether pkd2 knockdown affects kidney field formation in a general manner, or whether it principally impacts upon pax8, we compared expression levels of pax8, osr1, osr2 and lhx1 in kidney field explants isolated from CMO- or Pkd2-MO-injected embryos. Explants were isolated at NF stage 13, and were further cultured until early (NF stage 14) or late (NF stage 18) neurula stages for quantitative real-time PCR (RT-QPCR) analysis. As expected from ISH data on whole embryos, pkd2 knockdown resulted in a dramatic reduction of pax8 expression at NF stage 14 (Fig. 2B). At NF stage 18, Pkd2-MO still caused a substantial inhibition of pax8 expression, but this was less pronounced than that at NF stage 14 (supplementary material Fig. S4). In contrast, pkd2 knockdown did not affect osr1, osr2 or lhx1 expression at NF stage 14. At NF stage 18 pkd2 knockdown might cause a slight increase of lhx1 expression, but osr1 and osr2 remain unaffected (Fig. 2B).
This shows that pkd2 loss of function specifically targets pax8 expression in the kidney field.
Disruption of intracellular Ca2+ signaling during gastrulation results in pax8 inhibition in the kidney field
Intracellular Ca2+ increases are required for pronephric tubule differentiation ex vivo and in intact Xenopus laevis embryos (Leclerc et al., 2008). Inhibition of Ca2+ signaling with a photo-inducible Ca2+ chelator targeted to the kidney field results in defective pronephros tubulogenesis. This effect is observed either when the chelator is uncaged at late gastrula stage (NF stage 11–12.5), that is, prior to kidney field specification, or at neurula stage (NF stage 16–17) when the kidney field is already specified (Leclerc et al., 2008). However, it remains unclear whether the disruption of Ca2+ signaling affects pax8 expression in the emerging kidney field (Carroll and Vize, 1999). To test this hypothesis, we injected the photo-inducible Ca2+ buffer diazo-2 (Adams et al., 1989) into the left V2 blastomere at the eight-cell stage in order to target the presumptive dorso-lateral mesoderm containing kidney field precursors on the left side of the embryo. The right side of the embryo provided an internal control. Injected embryos were cultured until NF stage 11, at which time diazo-2 was uncaged, and were then further cultured until early neurula stage (NF stage 14) for ISH analysis of pax8 expression (Fig. 3A). Comparison of control and injected sides revealed a clear inhibition of pax8 expression in the kidney field on the left side injected with diazo-2. Expression was totally abolished in 52% of cases (n = 17/33), and strongly reduced in 33% of cases (n = 11/33) (Fig. 3B). These observations show that disruption of Ca2+ signaling not only affects tubulogenesis at later stages of pronephric development (Leclerc et al., 2008), but also early pax8 expression in the kidney field. Given that diazo-2 and pkd2 knockdown both cause a similar inhibition of pax8 expression, we also tested whether pkd2 knockdown could affect tubulogenesis at later stages. Embryos injected with Pkd2-MO in the two left blastomeres as above (2 pmole per blastomere) were cultured until NF stage 39-40 to analyze proximal tubule differentiation using the 3G8 antibody, which marks the proximal pronephric tubule, the nephrostomes and otic vesicle (Vize et al., 1995). Comparison of 3G8 immunoreactivity on the control and injected sides clearly showed that tubulogenesis was impaired in all examined embryos (n = 14) (Fig. 3C). These data show that disruption of Ca2+ signaling and pkd2 knockdown produce very similar effects, and further argue for a role of Ca2+ signaling in controlling pax8 expression during kidney field formation at neurula stage.
Pkd2 knockdown results in an inhibition of Ca2+ signals in kidney field explants at neurula stage
To determine whether the Ca2+ signals observed in the kidney field at neurula stage (Leclerc et al., 2008) require functional TRPP2 Ca2+-conducting channels, we depleted pkd2. Ca2+ recordings were performed using the bioluminescent Ca2+ reporter aequorin (Shimomura, 1991). Embryos were co-injected at the eight-cell stage into the V2 blastomere to target the kidney field with mRNA encoding aequorin fused to GFP, and Pkd2-MO (2 pmoles) (Tran et al., 2010) or CMO (2 pmoles). Embryos properly expressing GFP–aequorin in the kidney field were selected for dissection and Ca2+ recording at NF stage 12.5 (Fig. 4A). [Ca2+]i measurements were carried out as previously described (Leclerc et al., 2008) for a period of 4 h, during which sibling controls evolved from the late gastrula (NF stage 12.5–13) to late neurula stage (NF stage 16–17). Recording of pairs of CMO and Pkd2-MO-expressing explants were always performed simultaneously.
Comparison of pairs of recordings always revealed a dramatic reduction of Ca2+ transients in kidney field explants expressing Pkd2-MO. Fig. 4B shows representative photo-multiplier tube (PMT) traces (n = 4) obtained from pairs of kidney field explants dissected from CMO and Pkd2-MO injected embryos. A series of Ca2+ transients that appeared as vertical spikes on the time scale used, are observed on CMO kidney field data (Fig. 4B). The onset of Ca2+ transients occurred within 1 h after the beginning of the experiment, which corresponds to NF stage 13.5, when the kidney field is already specified. The maximum Ca2+ transient activity was reached at ∼3 h, at about NF stage 15. The Ca2+ dynamics recorded from such kidney field explants were very similar to those observed previously (Leclerc et al., 2008). Interestingly, Ca2+ dynamics recorded from the Pkd2-MO kidney field were inhibited (Fig. 4B). The Ca2+ recordings were further analyzed by calculating the number, duration and the luminescence ratio (L/L0) of the Ca2+ transients from four pairs of recordings over time for CMO and Pkd2-MO kidney fields (supplementary material Table S1). A histogram (Fig. 4C) shows that pkd2 loss of function results in the significant reduction (P = 0.014, paired Student's t-test) in the number of the Ca2+ transients without affecting the luminescence ratio of the transients (supplementary material Table S1). In order to control the specificity of Pkd2-MO effect, we attempted to rescue Pkd2-MO-inhibited Ca2+ signals by co-injecting an mRNA encoding a Pkd2-MO-resistant form of pkd2 mRNA. Comparison of pairs of recording indicated that injection of the mutated form of pkd2 mRNA (1 ng) was able to partially restore Ca2+ transients in kidney field explants (Fig. 4D, n = 3). The average number of Ca2+ transients increased from 419 (n = 3 explants) in explants expressing Pkd2-MO alone to 874 (n = 3 explants) in explants expressing Pkd2-MO and pkd2 mRNA (Fig. 4C; supplementary material Table S2). Although this increase is not statistically significant, a careful examination of the Ca2+ transients parameters revealed a significant increase in the duration of these Ca2+ transients (supplementary material Table S2; paired Student's t-test of three experiments, P = 0.016) indicating that the overall Ca2+ level is higher in the explants expressing Pkd2-MO and pkd2 mRNA. Taken together, these results show that pkd2 knockdown results in an inhibition of Ca2+ signals in kidney field explants. This confirms that TRPP2 is involved in the generation of these signals.
In a further attempt to address the question of a direct role of TRPP2-dependent Ca2+ signaling in the regulation of pax8 in the kidney field, we studied the possibility of rescuing pax8 expression in Pkd2-MO morphants by artificially increasing intracellular Ca2+ with ionomycin. Embryos injected with Pkd2-MO in the two left blastomeres (1.6 pmole per blastomere) were cultured until late gastrula stage, at which stage beads soaked in ionomycin solution or in DMSO (controls), were implanted into dorso-lateral mesoderm, and embryos were further cultured for analysis of pax8 expression. Beads can stay where they have been implanted for about 3 h by which time embryos have reached neurula stage (NF stage 14-15). Upon longer times of culture, we observed that beads are either expelled out of the embryo or fall into the archenteron, probably because of constraints resulting from neural tube closure. When pax8 expression was analyzed 3 h after ionomycin bead implantation, no rescue could be observed. However, when cultured until late neurula stage (NF stage 19), a significant increase of pax8 expression was observed in the ionomycin-soaked bead series relatively to the control bead series (supplementary material Fig. S4). This shows that increasing intracellular Ca2+ with ionomycin in the kidney field cells at early neurula stage can rescue, at least partly, pax8 expression.
Disruption of RA signaling results in a strong decrease of kidney field intracellular Ca2+ signaling
Given that RA triggers an increase in [Ca2+]i during pronephric tubule differentiation in vitro (Leclerc et al., 2008), we wanted to test whether RA is also important for the generation of Ca2+ transients during pronephros formation in the kidney field. To test this, Ca2+ transients were recorded in kidney field overexpressing the RA-catabolizing enzyme Cyp26. Embryos were either injected into the V2 blastomere with GFP–aequorin mRNA (control) or were co-injected with GFP–aequorin and Cyp26 mRNA (Cyp26) (350 pg). Ca2+ recordings were performed as previously on pairs of kidney field from NF stage 12.5 embryos. Disruption of RA signaling leads to the inhibition of Ca2+ signals as shown by the representative PMT traces for a pair of control and Cyp26 kidney field explants (Fig. 5A, n = 3) and by the histogram plot of the average number of Ca2+ transients (Fig. 5B). The difference observed between the average number of Ca2+ transients in control- and Cyp26-mRNA-injected kidney fields is significant (P = 0.008, paired Student's t-test of three pairs of experiments). These results raised the question of how RA is acting.
Therefore, we first tested the possibility that RA induces Ca2+ transients in the kidney field through the regulation of pkd2 expression by analyzing whether RA signals disruption affects pkd2 expression in lateral mesoderm explants. Expression of Cyp26 was targeted to the mesoderm by equatorial mRNA injection at the four-cell stage. Lateral marginal zone (LMZ) explants were dissected at early gastrula stage (NF stage 10), and cultured until early neurula stage (NF stage 14) for RT-QPCR analysis of pkd2 expression (Le Bouffant et al., 2012). Although RA disruption readily causes pax8 and lhx1 downregulation, expression of pkd2 was not affected (Fig. 6A). This shows that RA is not affecting TRPP2-dependent Ca2+ signaling in lateral mesoderm through the transcriptional regulation of pkd2.
Disruption of RA signaling causes a strong reduction in the amount of TRPP2 channel at the plasma membrane
We next asked whether RA signaling induces Ca2+ transients in the kidney field by regulating TRPP2 incorporation into the plasma membrane. To test this, we used total internal reflection fluorescence (TIRF) microscopy to visualize GFP-tagged TRPP2 proteins that are exclusively localized at the plasma membrane or in the 120 nm close to the membrane (Fig. 6A). mRNA coding a human GFP-tagged TRPP2 (hTRPP2–GFP) was injected into the V2 blastomere either alone (control) or in presence of Cyp26 mRNA to reduce RA signaling. Kidney field explants comprising ectoderm and mesoderm, and expressing hTRPP2–GFP were dissected at the end of gastrulation (stage 12.5–13), and the surface of mesodermal cells was analyzed by TIRF microscopy (Fig. 6B). The area of fluorescent spots per field in the presence and absence of Cyp26 mRNA (Fig. 6C) gives the proportion of TRPP2 channels localized close to the membrane. In the control condition, the calculated area of GFP spots was 220 µm2 (n = 51, four independent experiments). This area is reduced to 94.9 µm2 (n = 71, four independent experiments) in presence of 350 pg of Cyp26 mRNA (Fig. 6D), these results are in accordance with the reduction of Ca2+ signals observed in Fig. 5. The effect of RA disruption on the TRPP2 relocation is dose dependent; the co-injection of a 700 pg of Cyp26 mRNA results in a greater reduction in the area of GFP spots (28.9 µm2, n = 64, four independent experiments). These results show that RA signaling is affecting the incorporation of TRPP2 channels into the plasma membrane.
Several important conclusions can be drawn from this study. First, TRPP2 is expressed in the kidney field during specification and is not associated with cilia structures. Second, we report for the first time that pkd2 loss of function is affecting early steps of Xenopus kidney development, that is, the establishment of the kidney field. Third, our data show that the Ca2+ transients observed in the kidney field during late gastrula and early neurula stages are due to the activation of TRPP2 channels. Furthermore, they strongly suggest that pax8 expression is regulated by Ca2+ through a TRPP2-dependent mechanism acting upstream of the activation and/or maintenance of pax8 expression in the kidney field. Fourth, we show that pkd2 knockdown in the kidney field specifically affects pax8 expression, but not lhx1, osr1 and osr2, which implies that intracellular Ca2+ signaling is an intermediate step between the general mechanisms controlling kidney field emergence within latero-dorsal mesoderm and the control of pax8 expression. Finally, we show that both intracellular Ca2+ signaling and incorporation of TRPP2 in the plasma membrane are both impaired upon disruption of RA signaling, suggesting that RA might control intracellular Ca2+ signaling in the kidney field.
Is TRPP2-dependent Ca2+ signaling control of pax8 in renal precursors a specific feature of early Xenopus pronephros development?
Although several features of pkd2 loss-of-function phenotypes are conserved among vertebrates, such as cystic kidneys and laterality defects, others already appear to be restricted to a species, as for example body curvature in zebrafish (Obara et al., 2006). Here, we report for the first time that pkd2 loss of function affects early steps of Xenopus kidney development, that is, the emergence and/or the maintenance of the kidney field. This differs significantly from kidney pkd2 loss-of-function phenotypes described in mouse (McGrath et al., 2003; Wu et al., 2000) or zebrafish (Bisgrove et al., 2005; Obara et al., 2006; Schottenfeld et al., 2007; Sun et al., 2004), where early kidney development does not appear to be affected. However, during mouse pronephric development it is unclear whether Pkd2 does not have any function. Like in Xenopus, the commitment of mouse intermediate mesoderm to a kidney fate involves Lhx1, Osr1, Osr2, Pax2 and Pax8 (Dressler, 2009). Therefore, it is possible that, in the mouse, the pronephric development is impaired, but that further mesonephric development proceeds normally. Indeed, it has been shown, for example, that mouse pronephric development is severely compromised in Raldh2−/− embryos, but that Pax2-expressing prospective mesonephric collumns still develop (Cartry et al., 2006). In zebrafish, pan-embryonic and localized intercellular Ca2+ waves do affect the embryo during gastrulation. They are intense in the dorsal region of the embryo, but whether they have any function during kidney specification is unknown (Webb and Miller, 2006). Kidney specification in zebrafish might involve mechanisms differing from those underlying this process in Xenopus. It is noteworthy that RA disruption does not affect early kidney development in zebrafish as it does in Xenopus (Cartry et al., 2006; Wingert et al., 2007). Moreover, the consequence of pkd2 loss-of-function upon pax8 expression has not been investigated in zebrafish. Differences between zebrafish and Xenopus might also be the consequences of the respective roles played by the two closely related pax genes, pax2 (pax2a in zebrafish) and pax8. In Xenopus only pax8 is expressed in renal precursors of the kidney field, whereas both genes are expressed during early pronephric development in zebrafish (Majumdar et al., 2000). If TRPP2-dependent Ca2+ signaling is also acting upstream of pax8 in zebrafish, pkd2 loss-of-function would not cause early renal defects because of the redundancy between pax2a and pax8.
Ca2+ and other inputs regulating pax8 in the kidney field
Our results show that Ca2+ signaling is required for the proper emergence and/or maintenance of the kidney field. It is likely that Ca2+-signaling-dependent regulation of pax8 takes place downstream of the signals potentially responsible for kidney field specification during gastrulation, such as RA (Cartry et al., 2006) or Wnt11b (Tételin and Jones, 2010). In this work we show that a TRPP2-dependent Ca2+ signaling is not acting upstream of lhx1, osr1 or osr2. It is thus possible that some of these genes might be regulated more directly by the kidney field specification signals than pax8. For example, lhx1 upregulation in response to exogenous RA at the neurula stage occurs within 20 min, and is observed in the absence of protein synthesis, suggesting a direct regulation, whereas pax8 upregulation is only observed after 2 h, suggesting an indirect response to RA (Cartry et al., 2006). In addition, kidney field expression of pax8 is also likely to be dependent on other inputs, such as those mediated by bone morphogenetic proteins (BMPs), whose disruption during neurulation results in a loss of pax8 expression (Bracken et al., 2008). Although we are not addressing the mechanisms by which Ca2+ controls pax8 expression, an interesting possibility would be that the [Ca2+]i increase acts as a release of pax8 transcriptional repression. Regulation of Pax8 during murine thyroid cell differentiation has indeed been shown to be dependent on a Ca2+-dependent release of repression involving the Ca2+-binding protein downstream response element-antagonist modulator (DREAM, also known as calsenilin, KCHIP3 or KCNIP3) (D'Andrea et al., 2005).
How might kidney field inductive signals act upon intracellular Ca2+ signaling?
How the [Ca2+]i increase is related to more general mechanisms of kidney field induction remains elusive. Activation of non-canonical pathways by Wnt11b might act as a potential mechanism triggering the increase of [Ca2+]i in the kidney field. Some Wnts are acting through the activation of phospholipase C and inositol 1,4,5-trisphosphate (IP3) production. IP3 in turn activates Ca2+ release from internal stores through the IP3 receptor in the ER (De, 2011; Sheldahl et al., 2003; Slusarski et al., 1997). It is interesting to note that TRPP2 has been shown to amplify Ca2+ signals in mouse renal epithelial cells in a Ca2+-induced Ca2+ release mechanism necessitating interaction between the IP3 receptor and TRPP2 (Sammels et al., 2010). We cannot rule out the possibility that such a potential amplification mechanism might be involved in the generation of Ca2+ transients in kidney field cells. However, we show that disruption of RA signaling results in a dramatic decrease in TRRP2 at the plasma membrane of kidney field cells. This shows that RA can regulate TRPP2 trafficking to the plasma membrane and thus might increase intracellular Ca2+ signaling. Previous studies showing that RA-induced differentiation of SH-SY5Y cells (Toselli et al., 1991) or of the human teratocarcinoma cell line into NT2N neurons (Gao et al., 1998) is associated with the incorporation of new Ca2+ channels in the plasma membrane further support this hypothesis. Data obtained previously in animal caps also point to a role of RA in the generation of Ca2+ transients. Treatment of animal cap explants with activin causes induction of neural, mesodermal and endodermal derivatives. However, kidney structures are not induced. When RA is added to activin, pronephros tissue differentiates (Moriya et al., 1993), and RA, but not activin, triggers the [Ca2+]i increase observed during pronephric tubule formation. Using this assay, it has also been shown that the RA-dependent induction of tubules is inhibited by the Ca2+ chelator BAPTA-AM, or by lanthanum, an inhibitor of TRP channels. Conversely, experimentally increasing [Ca2+]i with caffeine, ionomycin or NH4Cl can substitute to some extent for RA in inducing pronephric tubules. Treatment of animal caps with RA alone is sufficient to elicit an increase of [Ca2+]i within 45 min. This process is abolished in the absence of protein synthesis (Leclerc et al., 2008).
We show that pkd2 knockdown in the kidney field specifically affects pax8 expression but not lhx1, osr1 and osr2. It is therefore likely that lhx1 or osr1, osr2 and pax8 are differently regulated by the signals potentially responsible for kidney field specification during gastrulation, such as RA (Cartry et al., 2006) and Wnt11b (Tételin and Jones, 2010). As outlined above, some of these genes are directly regulated by specification signals. The mechanism by which a TRPP2-dependent Ca2+ signaling controls pax8 expression in the kidney field during Xenopus gastrulation is presented in Fig. 7 as a working model supported by our data showing that disruption of RA signals both results in an inhibition of intracellular calcium signaling (Fig. 5) and TRPP2 incorporation in the plasma membrane of kidney field cells (Fig. 6).
MATERIALS AND METHODS
Cloning of pkd2 cDNA was performed from NF stage 35–36 cDNA pools. A 599-nucleotide (nt) fragment was amplified by PCR using primers corresponding to conserved regions (forward, 5′-AGGTTATTGGTTGAATTCCC-3′; reverse 5′-CGGAATTGGGTGAAGATACA-3′). This sequence was expanded by 5′ and 3′ RACE PCR using the SMARTerTM RACE cDNA amplification kit (Clontech). A 2953-nt cDNA containing the full coding sequence (accession number HG421008) was obtained by end-to-end PCR. It encodes a 947-amino-acid protein with 67.7% identity with human TRPP2 (supplementary material Fig. S1). PCR mutagenesis was performed to introduce silent mutations downstream of the start AUG in order to avoid binding to Pkd2-MO in rescue experiments (5′-AUGAAUCCCAGCAGAAUCAAA-3′ to 5′-AUGAACCCGUCACGUAUCAAG-3′). The mutated fragment was subcloned into pSP64TBX for synthesis of functional mRNA. Coding sequences of GFP–aequorin (Baubet et al., 2000) and human TRPP2 with GFP inserted after amino acid 157 (Hoffmeister et al., 2011) were both subcloned into pCS2+.
mRNA and morpholino microinjection
Synthesis of capped mRNA was performed as previously described (Umbhauer et al., 2000). Plasmids linearization was performed as follows: pSP64TBX-mutpkd2 with SalI, pCS2+ memGFP (Moriyoshi et al., 1996) and pCS2+ GFP-aequorin with NotI, SP6nucβGal encoding LacZ (Smith and Harland, 1991) and PCS2+ arl13b-cherry (Borovina et al., 2010) with XhoI, pβSRN3-GFP with SfiI (ZernickaGoetz et al., 1996), pCS2XCyp26 (Hollemann et al., 1998) with EcoRI, and pCS2+GFP-hPKD2 with Asp718. All plasmids were transcribed with SP6 RNA polymerase. Pkd2-MO (Tran et al., 2010) and the standard control MO (CMO) were purchased from Gene Tools. Microinjections were performed at the two- to eight-cell stage as described previously (Colas et al., 2008). Morpholino and mRNA doses are given in the results section.
In situ hybridization and LacZ staining
Embryo protein extraction was performed as previously reported (Le Bouffant et al., 2012). Immunoblotting with anti-α-tubulin has been previously described (Le Bouffant et al., 2012). Anti-polyclonal anti-human TRPP2 (Novus Biochemicals #NB100-92215), was used at 1:1000. This antibody was produced against a peptide corresponding to a highly conserved region of the C-terminal tail of the protein (supplementary material Fig. S1). Cross-reactivity with Xenopus TRPP2 was controlled by western blotting. A major band of 110 kDa was detected on embryo extracts in accordance with human TRPP2 data (Giamarchi et al., 2010), together with two much fainter bands of lower mobility. Overexpression of Xenopus TRPP2 caused by pkd2 mRNA injection resulted in a strong increase of this 110 kDa component, showing that it corresponds to Xenopus TRPP2 (supplementary material Fig. S1).
Explants dissection and immunofluorescence
For intracellular Ca2+ measurements and RT-QPCR analyses, archenteron wall explants corresponding to the pronephric pax8 expression domain were dissected at late gastrula stage (NF stage 12.5–13). They were cultured in 1× modified Barth's solution (MBS).
Ionomycin- and DMSO-soaked AG1-X2 beads (BioRad) were prepared as described previously (Papanayotou et al., 2013). They were implanted at late gastrula stage (NF st.12.5) beneath the ectodermal layer overlying the kidney field. Embryos were placed in 1× MBS. A small slit was cut into epidermis with platinum wire and loop. The bead was inserted into the slit and pushed laterally with fine forceps. Implanted embryos were immediately transferred into 0.1×MMS where they were further cultured.
For immunofluorescence staining, only mesodermal and ectodermal layers were dissected. GRP explants were dissected as previously described (Schweickert et al., 2007). In all cases, explants were immediately transferred into fixative-containing Petri dishes. For anti-TRPP2 staining, explants were fixed in formaldehyde-glutaraldehyde (FG) fixative and processed for immunofluorescence staining as described (Nandadasa et al., 2009). Rabbit anti-TRPP2 antibody and mouse anti-GFP antibody (Roche) were used at 1:100 and 1:1000, respectively. Alexa-Fluor-568-conjugated anti-rabbit and Alexa-Fluor-488-conjugated anti-mouse IgG antibodies (InVitrogen) were both used at 1:1000. After washing, explants were dehydrated and cleared in benzyl benzoate in benzyl alcohol (2∶1) (Dent et al., 1989) prior to confocal microscopy analysis. Monoclonal anti-acetylated tubulin staining (Sigma clone 6-11B-1, 1:500) of primary cilia was carried out as described previously (Stubbs et al., 2008), except that Alexa-Fluor-488-conjugated anti-mouse IgG antibodies were used. mCherry–arl13b-expressing explants were fixed as for anti-acetylated tubulin staining, washed and processed for confocal microscopy. Observations were carried out on Leica SPE and SP5 inverted confocal microscopes. Whole-mount immunofluorescence for 3G8 antibody was carried out as described previously (Vize et al., 1995) with Alexa-Fluor-568-conjugated anti-mouse IgG antibodies (InVitrogen) (1:1000).
Lateral marginal zone explants from embryos injected with GFP or a mix of GFP and Cyp26 mRNA were dissected at early gastrula stage and cultured until early neurula stage for RT-QPCR analysis as previously described (Le Bouffant et al., 2012).
Real-time quantitative PCR
Microinjection and flash photolysis of caged-compounds
Injection was performed into the lateral marginal zone of four-cell stage embryos. 10 nl of Diazo-2 (Diazo-2 tetrapotassium salt is a cell impermeant caged BAPTA; 50 mM dissolved in distilled water, Molecular Probes) was co-injected with lacZ-encoding RNA (250–500 pg) as a lineage tracers. UV photolysis was performed as described previously (Leclerc et al., 2008).
Ca2+ measurements, temporal data acquisition and analysis
Intracellular Ca2+ measurements were performed with the bioluminescent Ca2+ probe GFP–aequorin (the GFP–aequorin construct is a gift from Jean-René Martin, CNRS, Gif-sur-Yvette, France). 10 nl of a mixture of GFP–aequorin mRNA (200 pg) and either Pkd2-MO or CMO was injected into the left V2 blastomere. Kidney field explants corresponding to the pronephric pax8-expression domain were dissected at the late gastrula stage (NF stage 12.5–13). Functional aequorin was reconstituted by incubating the kidney field explants with 1 ng/ml of cp-coelenterazine (Molecular Probes; stock prepared in ethanol). The kidney field explants were then processed for bioluminescence photon counting using a PMT as described previously (Leclerc et al., 2000). The activity of aequorin in vivo is not modified by fusion with EGFP (Baubet et al., 2000). Background noise is less than 5 photons/s. Data were collected every 1 s over a period of at least 4 h. Burn-out experiments using Triton X-100 to lyse the explant showed that when signals were not observed aequorin was not a limiting factor. Data are expressed in relative arbitrary units, proportional to the photon number. Data were selected for detailed analysis according to the following criteria: (1) explants did not dissociate during the course of the experiment; (2) intact control embryos developed normally.
Kidney field explants expressing hTRPP2–GFP were dissected at the end of gastrulation (NF stage 12.5–13) and immediately fixed in 4% paraformaldehyde. For the observations, kidney field explants were mounted in PBS. GFP was imaged using a Nikon Eclipse TIRF microscope with a 60× objective oil immersion NA 1.49 (Fig. 6A). Excitation was performed at 488 nm and the emitted fluorescence was recorded with an Andor iXON EMCCD camera. TIRF penetration depth was set to 120 nm. Observations were performed on mesoderm. Background-subtracted fluorescent pixels were processed with a custom-made analysis routine written for Image J. The area of expression of TRPP2 channels labelled with GFP was calculated for 10 to 20 fields of 50×50 µm on four different explants overexpressing either hTRPP2–GFP mRNA alone or hTRPP2–GFP mRNA in presence of Cyp26 mRNA at 350 pg or 700 pg (Hoffmeister et al., 2011).
We thank S. Authier and E. Manzoni for excellent technical assistance in the maintenance of the Xenopus animal facility. We also want to thank C. Vesque for invaluable advice and criticisms concerning primary cilia experiments, Brian Ciruna for the mCherry–arl13b construct, J. R. Martin for the GFP–aequorin construct, R. Witzgall for the GFP–human TRPP2 construct, and E. Jones for the 3G8 antibody. We are very grateful to S. Bolte, R. Schwartzmann and J. F. Gilles for confocal imaging (core facility cell imaging, IFR83 CNRS, UPMC). C.L., I.N. and M.M. are members of the GDRE 731 “Ca2+ toolkit proteins as drug targets in animal and plant cells”.
M.F. was responsible for design, execution and interpretation of experiments, and preparation of the article; C.L. was responsible for conception, design, execution and interpretation of experiments, and preparation of the article; R.L.B. was responsible for execution and interpretation of experiments; I.B. was responsible for execution of the experiments; I.N. was responsible for execution of the experiments; M.U. was responsible for execution of the experiments; M.M. was responsible for conception, design, execution and interpretation of the experiments and preparation of the article; J.F.R. was responsible for conception, design, execution and interpretation of the experiments and preparation of the article.
This work was supported by grants from CNRS and from Université Pierre et Marie Curie (UPMC). We acknowledge funding from Émergence-UPMC-2009 research program. M.F. is financed by a 2010–2013 contract doctoral from the ‘Complexité du Vivant’ doctoral school.
The authors declare no competing or financial interests.