|
Fig. 1.
aqp3bis expressed in the marginal zones of gastrula embryos. aqp3b expression was characterized by in situ hybridization in gastrula Xenopus embryos. (A-C) Animal pole views of aqp3b expression in stage 10.5 embryos. Expression of aqp3b was strong in the entire animal cap (A) and otx2 expression marks the dorsal lip of the blastopore (B). (C) Double in situ hybridization showed aqp3b (bracket) and otx2 (arrow) expression. The aqp3b expression domain extended to the region of otx2 expression. (D-I) Hemisected embryos. (D) aqp3b is expressed in the sensorial layer of the blastocoel roof and in the marginal zones at early gastrula stage 10. (F) aqp3b expression (bracket) extends to the dorsal lip of the blastopore (marked by otx2, arrow; stage 10.25 embryo). The strong staining lining the blastocoel was also present in otx2 single in situ hybridizations and therefore due to the otx2 probe (not shown). Stage 11 (G) and stage 12 (I) embryos showed diminishing levels of aqp3b expression in involuted mesoderm (arrowheads). In situ hybridization with aqp3b sense probe demonstrated low levels of background in the areas of interest (E, H). Remnants of brown pigment are due to the use of bleached wild type embryos. The dorsal lip of the blastopore is at the top in (A-C) and to the right in (D-I). D, dorsal; V, ventral.
|
|
Fig. 2.
Knockdown ofaqp3bresults in defects during gastrulation.Xenopus embryos were pre-injected at the 1-cell stage with lacZ RNA (injection control) (A-H) or aqp3bMM RNA, which is not recognized by either aqp3b MO (I, J). At the 4-cell stage, the embryos were additionally injected into one dorsal blastomere with control MO1 (A, E), aqp3b MO1 (C, G, I), control MO2 (B, F), or aqp3b MO2 (D, H, J) together with the injection tracer rhodamine-dextran. (Each embryo set is paired with the corresponding fluorescent images.) Blastopore closure proceeded normally in control MO-injected embryos (A, B, E, F). aqp3b MOs did not have gross effects on early blastopore closure in stage 10.5 embryos (C, D), but clear defects developed by stage 12 (G, H), which were rescued by aqp3bMM RNA (I, J). (K) Percentage of embryos showing gastrulation defects at stage 12 (injection as above with aqp3b MOs, control MOs, or rescue of aqp3b MO defects with aqp3bMM RNA). Error bars represent standard error of the mean. Seven independent clutches were scored with a total number of embryos as indicated. (L) In four independent experiments, 40–90 embryos were scored per clutch for the types of phenotypic defects resulting from aqp3b MO1 injection and rescue, compared to control MO1 injections.
|
|
Fig. 3.
Impairedaqp3bexpression results in abnormal tissue architecture.aqp3b MO1-injected embryos (stage 10.5) were analyzed by confocal microscopy to visualize phalloidin (green) and co-injected rhodamine-dextran tracer (red). Embryos that received control MO1 in their dorsal marginal zone (A, B) showed a highly organized boundary between ectoderm (e) and mesendoderm (me), which was revealed by actin cytoskeletal staining (A). In contrast, aqp3b MO1-injected embryos (C, D) displayed irregularities in this boundary and jagged protrusions were found along the ectoderm layer (arrows, C). In higher magnification images, control MO1-injected embryos (E, F) showed highly organized tissue architecture and regularly shaped cells, while in aqp3b MO1-injected embryos (G, H), the ectoderm/mesendoderm boundary was irregular and cells seemed less tightly packed (arrows). dlb, dorsal lip of the blastopore; e, ectoderm; me, mesendoderm. Scale bars: 50um.
|
|
Fig. 4.
The fibronectin (FN) matrix is defective next to cells with compromisedaqp3bexpression within the DMZ. Gastrula embryos (stage 10.5) that had been co-injected with MOs and membrane-bound mem-EGFP tracer RNA (green) were hemisected sagittally through the dorsal blastopore lip, stained with FN antibodies (red), and evaluated by confocal microscopy. (A) In a control MO1-injected gastrula embryo, the FN matrix extended along the inner surface of the blastocoel roof and into the dorsal and ventral marginal zones. (B-H) Control MO-injected embryos had an uninterrupted FN matrix. The dashed box in the lower magnification image (B) indicates the area shown in (C, F) from a different optical plane. Continuous FN matrix was also present in embryos injected with control MO2 (D, G) and both control MOs (E, H). The green channel (mem-EGFP) was omitted in the lower panels to clearly show the extent of the FN matrix (red), which was continuous between non-involuted ectoderm and involuted mesendoderm. (I-O) aqp3b morphants showed interruption of the FN matrix wherever aqp3b MOs were present (marked by green mem-EGFP tracer). The dashed box in (I) shows an equivalent area as in (J, M). In the area sandwiched between ectoderm and mesendoderm, the fibrillar FN matrix is highly disrupted by the presence of aqp3b MO1 (I, J, M), aqp3b MO2 (K, N), and synergistically in embryos injected with lower amounts of both aqp3b MOs (L, O). DMZ, dorsal marginal zone; VMZ, ventral marginal zone; BCR, blastocoel roof; DLB, dorsal blastopore lip; e, ectoderm; me, mesendoderm; bc, blastocoel cavity; the dotted lines demarcate the floor of the blastocoel. Scale bars: 50 um.
|
|
Fig. 5.
Cells with disruptedaqp3black a continuous fibrillar FN matrix across the face of the ectoderm. Confocal image stacks of hemisected embryos were rotated to show the surface of ectoderm cells (B-E, G-J). Control MO1-injected embryos (A-E) showed an extensive fibrillar FN matrix across the face of the ectoderm in Brachet's cleft (B, C) and along the BCR (D, E). aqp3b MO1-injected embryos (F-J) were nearly devoid of fibrillar FN matrix across the face of the ectoderm in Brachet's cleft (G, H), but did possess FN matrix lining the face of the BCR (I, J). The boxes in (A) and (F) indicate the areas that were used for the rotated projections (dashed box = BCR: B, C, G, H, and solid box = Brachet's cleft: D, E, I, J). The green channel was removed in (C, E, H, J). The dotted lines delineate the floor of the blastocoel. Scale bars: (A, F) 100 um; (B-E, G-J) 10 um.
|
|
Fig. 6.
Cells protruding past the defective tissue boundary. (A, B, E, F) Control MO and mem-EGFP (green) injected cells showed an intact boundary between ectoderm and mesendoderm cells, which was respected by all cells. (FN matrix is labeled red.) (C, D, G, H) Cells injected with aqp3b MOs lacked FN matrix, which allowed abnormal behavior of cells at the tissue border. Cells failed to respect the tissue boundary, forming protrusions that extend past the border between ectoderm and mesendoderm (arrows). Scale bars: 30 um.
|
|
Fig. 7.
Embryos that receivedaqp3bMO localized non-fibrillar FN matrix around all cells. Embryos co-injected with both aqp3b MOs (4 ng each) lacked a continuous fibrillar fibronectin (FN) matrix (C, D) but exhibited the same level of pericellular FN matrix (arrowheads) as control MOs-injected cells (A, B), suggesting that the presence of non-fibrillar FN was not affected by the absence of Aqp3b. Asterisks in (C-F) mark cells that did not receive aqp3b MOs in otherwise aqp3b MOs-injected embryos. These cells had adjacent fibrillar FN matrix (white arrows), unlike neighboring cells, which received aqp3b MOs and where the fibrillar FN matrix was lacking. Scale bars: 10 um.
|
|
Fig. 8.
Convergence but not radial intercalation movements were affected in embryos that receivedaqp3bMO. Embryos co-injected with both aqp3b MOs (B, D) showed equal thinning of the blastocoel roof as embryos that received both control MOs (A, C); stage 9 (A, B), stage 11 (C, D). Similarly, stage 11–11.25 embryos injected with control MOs (E, F) or with aqp3b MOs (G, H) showed equally thinned DMZs (arrowheads), indicating normal radial intercalation. Embryos were stained with antibodies against β-catenin (red) to outline cells (E-H) and with anti-EGFP antibodies (green) to indicate areas that received the aqp3b MOs (A-H). In situ hybridization with chordin probe of mid-gastrula embryos (stage 11; I-L) and after completion of gastrulation (stage 13; M-P) demonstrated that uninjected (I, M) and control MOs-injected (J, N) embryos showed normal convergent extension movements. Wider chordin expression patterns in aqp3b MOs-injected embryos (K, L, O, P) were consistent with defects in convergence movements. The defects ranged from mild (O) to strong (P), which is quantified in (Q). Four independent batches were scored with a total number of embryos as indicated. Tailbud embryos (R-T) showed shortened body axes when injected with aqp3b MOs (T). Arrows indicate the injected sides of the embryos. The embryo in (L) contained remnants of wild type pigment. Thus, brackets indicate the extent of chordin expression. Scale bars: 25 um (A-D); 100 um (E-H).
|
|
Fig. 9.
Models for the mechanisms by which Aqp3b may affect fibrillar FN matrix assembly. (A) We have shown that knockdown of Aqp3b expression (indicated by red X) caused cells to be less well organized, particularly at the tissue boundary between mesendoderm and ectoderm in Brachet's cleft in the dorsal marginal zone. This resulted in cell autonomous loss of the fibrillar FN matrix in Brachet's cleft. Cells with unaffected Aqp3b formed normal fibrillar FN matrix. (B) Aqp3b may influence fibrillar FN matrix formation by several mechanisms. Membrane protrusions are important for cell-mediated FN fibril assembly (Davidson et al., 2008) and Aqps have been localized to the leading edge of cell protrusions (Saadoun et al., 2005). Further, integrin clustering is essential for FN fibril assembly (Wu., et al., 1993), and Aqps have been shown to bind to the β1 integrin subunit, which is required for integrin endocytosis (Chen et al., 2012). Endocytosis of the α5β1 integrin receptors is necessary for FN fibril assembly (Spicer et al., 2010). Tension imparted by cadherins is required for fibrillar FN matrix formation (Dzamba et al., 2009), which has not yet been shown to be affected by aquaporins. Finally, Aqp3b may engage in cellular signaling to influence fibrillar FN matrix formation, and Aqp3 overexpression was found to upregulate FN expression in cancer cells (Chen et al., 2014). Finally, Aqp3b appears to be required for normal convergence movements during gastrulation, which may be fibrillar FN matrix-dependent or âindependent. Solid arrows: relationships that have been documented; dashed arrows: possible relationships or alternate pathways.
|
|
Figure S1. aqp3b MO prevents Aqp3b translation. Embryos were pre-injected at the 1-cell stage with either HA-tagged aqp3b sense RNA (aqp3bHA; panel A) or with aqp3bMM sense RNA (D). These pre-injected embryos were then co-injected with aqp3b MO1 and mEGFP tracer into one cell at the 4-cell stage. Antibody staining showed that aqp3b MO1 was able to inhibit aqp3bHA RNA translation (A-C), but that translation of aqp3bMM RNA proceeded undisturbed (D-F). The bracket indicates a region of aqp3b MO1 and Aqp3bMM overlap.
|
|
Figure S2. C-Cadherin and ï¢1-integrin expression was not affected in aqp3b morphants. Embryos were co-injected with both aqp3b MOs or both control MOs and then stained with either C-cadherin (A, B) or ï¢1-integrin (C, D) antibodies (both red) at stage 11. The companion images indicate the presence of the MOs by staining for the mEGFP tracer (green). Inhibition of Aqp3b translation changed neither C-cadherin (B) nor ï¢1-integrin (D) levels. Scale bars: 50 um.
|
