Drosophila von Hippel-Lindau Tumor Suppressor Gene Function in Epithelial Tubule Morphogenesis

Abstract

Mutations in the human von Hippel-Lindau (VHL) gene are the cause of VHL disease that displays multiple benign and malignant tumors. The VHL gene has been shown to regulate angiogenic potential and glycolic metabolism via its E3 ubiquitin ligase function against the alpha subunit of hypoxia-inducible factor (HIF-α). However, many HIF-independent functions of VHL have been identified. Recent evidence also indicates that the canonical function cannot fully explain the VHL mutant cell phenotypes, although it is still unclear how many of these noncanonical functions relate to the pathophysiological processes because of a lack of tractable genetic systems. Here, we report the first genomic mutant phenotype of Drosophila melanogaster VHL (dVHL) in the epithelial tubule network, the trachea, and show that dVHL regulates branch migration and lumen formation via its endocytic function. The endocytic function regulates the surface level of the chemotactic signaling receptor Breathless and promotes clearing of the lumen matrix during maturation of the tracheal tubes. Importantly, the regulatory function in tubular morphogenesis is conserved in the mammalian system, as conditional knockout of Vhl in mouse kidney also resulted in similar cell motility and lumen phenotypes.

Establishing and maintaining the integrity of tubular epithelium is critical for embryonic development, organogenesis, and tissue homeostasis. Dysregulation of tubular epithelia also underlies a majority of human cancers, such as those of mammary glands, lung airways, and kidney tubules. The von Hippel-Lindau (VHL) tumor suppressor gene mutations are the genetic cause of VHL disease. The germ line mutations in VHL predispose the patients to several benign and malignant tumors, including renal cell carcinoma, hemangioblastoma (overgrowth of blood vessels in the retina and central nervous system), and pheochromocytoma (tumor in the adrenal glands). VHL protein has been shown to function as an E3 ubiquitin ligase. Among its best-documented targets is the alpha subunit of the hypoxia-inducible factor (HIF-α). Therefore, the canonical tumor suppressor function of VHL is mediated through modulation of the normal oxygen-sensing mechanism that regulates angiogenic response and metabolic switch to glycolysis (29). However, recent evidence suggests that VHL is a multifunctional protein. It exhibits functions in matrix deposition (47), integrin processing (17), endocytosis (9, 27, 66), kinase adaptor (72), senescence (73), protein stability (10, 52), epithelial cell junction formation (4, 7, 23), and microtubule stability (24, 36), among many others (18). While some of these noncanonical functions involve novel HIF targets, most others are HIF independent and appear to facilitate protein stability or activity, contrary to the expected ubiquitin ligase activity (19). The study of VHL gene functions has been hampered by the lack of tractable genetic models. It is yet unclear how the loss of many of these chaperone/adaptor functions, mostly identified in cell culture systems, contributes to VHL tumor formation.

VHL is evolutionarily conserved from worm to mammal. In this report, we describe the first genomic mutant phenotype of Drosophila melanogaster VHL (dVHL) in the epithelial tubule system, the trachea, and its dual role in regulating tracheal branching morphogenesis and lumen formation, via modulation of endocytosis. We also demonstrate the same phenotypic outcome of the mammalian Vhl mutation in kidney tubule cells. Such in vivo function may underlie the tumor suppressor function of this enigmatic gene.

MATERIALS AND METHODS

Drosophila strains and genetics.

The dVHL¹ mutant was generated by replacing the wild-type copy, via homologous recombination (53), with a deletion construct that removes 81 codons, encompassing the first two in-frame AUG codons. The mutant stock is maintained with the balancer chromosome CyO, Dfd-YFP that carries a YFP reporter gene directed from the Deformed (Dfd) promoter (expressed in the head region) (34). The dVHL full-length cDNA (1) and head-to-head duplex of the open reading frame (dVHLi) was cloned in the pUAST or pCaSpe-hs vectors for Gal4- or hsp70-directed expression, respectively. These expression vectors were used to transform the y w flies. The hsp-dVHL-transgenic flies were combined with the dVHL¹ allele. For testing the specificity of the dVHL knockout, progenies from dVHL¹/CyO; hsp-dVHL intercrosses were analyzed as follows. The newly laid embryos were heat treated twice daily at 37°C for 30 min each until adults eclosed. The adults were counted based on gender and wing phenotypes.

The btl-Gal4 fly strain (57) was provided by T. Kornberg (UC San Francisco) and used for UAS-driven expression of the dVHL wild type and duplex. awd^j2A4 (13, 33), pnt^Δ88 (6), and shi² mutant fly stocks (31) were obtained from the Bloomington Stock Center, and btl^H82Δ3 mutant fly stock (32) was a gift from D. Montell (Johns Hopkins University). awd^j2A4, pnt^Δ88, and btl^H82Δ3 are homozygous lethal and are maintained by pairing with the balancer TM3, P{ry t7.2 = HZ^2.7}DB2 Sb. The DB2 reporter gene shows β-galactosidase (β-Gal) expression in the maxillary tissues. shi² is a temperature-sensitive allele of shibire, which encodes Drosophila dynamin. shi² embryos were collected at 25°C for 7 h, followed by incubation at 34°C for another 7 h. Genetic recombination was used to incorporate the 1-eve-1-transgenic line, which expresses the lacZ reporter gene in all tracheal cells at all stages from the trachealess gene promoter (49) (a gift from A. Brand, University of Cambridge). Transgenic flies carrying wild-type UAS-YFP-Rab5 fusion genes and constitutively active (CA) and dominant negative (DN) variants of them were obtained from Bloomington Stock Center and have been described previously (74).

The y w; btl-Gal4; UAS-btl::GFP transgenic line expresses a Breathless (Btl)::GFP fusion protein (where GFP is green fluorescent protein) in btl-expressing tissues, including the trachea, and is a gift from T. Kornberg (55) (UC San Francisco). The UAS-dEGFR::GFP fly strain is a gift from J. Duffy (Worcester Polytechnic Institute).

Disease-related VHL mutations.

The dVHL full-length cDNA (1) and the dVHL^YH and dVHL^RQ variants were cloned in the UAS-based pGateway vector tagged with three hemagglutinin (HA) epitopes at the 3′ end. The expression vectors were used to transform the y w flies. These transgenic flies carrying a single copy of the UAS-dVHL-HA transgenes were crossed with hsp-dVHLi, btl-Gal4; 1-eve-1 flies. Embryos collected over 12 hours were heat treated at 37°C for 30 min and allowed to continue development for 6 h. They were then processed for immunostaining for β-Gal and HA. Within the collected embryos, dVHLi, btl-Gal4 was paired with the dVHL transgenes, which served as internal controls for the efficiency of the dVHL knockdown-induced phenotypes (ectopic migration and abnormal lumen), or were left unpaired. See Results for the description of the VHL mutations.

S2 cell culture.

S2 cells (obtained from American Type Culture Collection) were maintained in Schneider's medium supplemented with 10% heat-treated fetal bovine serum (FBS; Cambrex). For immunohistochemistry, the cells were grown on cover slides coated with poly-l-lysine and were transfected using Lipofectamine 2000 (from Invitrogen) with pUC-HygMT-btl::GFP without clonal selection (13, 26). Transfection with control (pCaSpe-hs), hsp-dVHL, or hsp-dVHLi plasmids was performed using Lipofectamine 2000 according to the manufacturer's protocol. For inducing dVHL or dVHLi expression, transfected S2 cells were incubated at 29°C for 2 days. Btl::GFP fusion protein expression from the metallothionein promoter was then induced with 10 μM CuSO₄ for 2 h.

Immunohistochemistry.

Standard protocol was followed for immunostaining whole-mount embryo and S2 cells (13, 26). The commercially available primary antibodies used were mouse monoclonal antibody 2A12 (1:3; Developmental Studies Hybridoma Bank [DSHB]), rabbit polyclonal anti-GFP (1:200; Abcam), chicken polyclonal anti-β-Gal (1:500; Abcam), mouse anti-Crb (1:100; DSHB), and mouse anti-VHL (1:100; Cell Signaling).

Rabbit polyclonal anti-dVHL and anti-Pnt antibodies were generated against full-length bacterially expressed His-tagged proteins and purified by protein A adsorption. Rat polyclonal anti-Trachealess (anti-Trh) was generated against full-length Trh expressed in bacteria and purified by protein A/G adsorption. Fluorescein isothiocyanate (FITC) or rhodamine-conjugated chitin-binding probes (fusion proteins containing a chitin-binding domain of chitinase and maltose-binding protein) were from New England Biolabs. For colorimetric reactions, VectaStain (Vector) reagents were used. For immunofluorescence, Alexa Fluor 488- or 546-conjugated secondary antibodies (Invitrogen/Molecular Probes) were used at a 1:200 dilution, and incubation was performed for 2 h at room temperature. Stained samples were examined with an Olympus BH20 bright-field microscope or an Olympus IX70 microscope equipped with the FluoView 300 confocal capability. Digital images were processed using Photoshop software, without biased manipulations.

Surface biotinylation assay.

Transfected S2 cells, as described above, were grown in 10-cm dishes. A total of 4 × 10⁷ cells for each condition were used. For surface labeling and extraction, instructions from the cell surface protein isolation kit from Pierce were followed. Briefly, cells were washed with ice-cold phosphate-buffered saline (PBS; pH 7.4) and incubated with 3 ml/well PBS containing 0.25 mg/ml sulfo-NHS-LC-biotin [sulfonsuccinimidyl-6-(biotinamido)hexanoate] for 30 min at 4°C. Unincorporated cross-linkers were then quenched by washing the cells twice with cold PBS-50 mM Tris-HCl, pH 7.0. Cells were collected and lysed with 200 μl/well of 1% Triton-radioimmunoprecipitation assay (RIPA) buffer containing 1× protease inhibitor cocktail (Roche). Half of the protein extracts were incubated with NeutrAvidin biotin-binding protein gel (Pierce) overnight at 4°C and then washed with RIPA buffer. The other half of the total lysate was reserved for measuring total protein levels and for loading controls. The total and surface fractions were eluted by boiling in 1× SDS sample buffer and were run on a gradient (10% to 15% to 20%) SDS-polyacrylamide gel, followed by Western blotting. The equivalent of one-fourth of the lysate used for surface fractionation was used for total protein control.

Kidney tubule cell isolation.

The protocol for kidney tubule cell isolation was based on published reports (46, 56). Briefly, 3 pairs of mouse kidneys were dissected and the medullas were cut away. The cortex-enriched tissues were homogenized lightly (5 strokes) in a loose-fit Dounce homogenizer. The tissue suspension was centrifuged at a very low speed (50 × g for 2 min). The precipitates (enriched for the tubules) were incubated with collagenase (10.68 mg collagenase I [Worthington] and 15 mg soybean trypsin inhibitor [Sigma] in 20 ml Hanks solution). Large tissue debris were allowed to settle for 1 min without centrifugation, and the suspension was decanted and centrifuged at 50 × g for 2 min again. The precipitates were resuspended and loaded onto a preformed gradient of 40% Percoll-60% medium (the gradient was preformed by centrifugation at 36,000 × g for 20 min). The cushioned sample was centrifuged at 400 × g for 10 min. The tubule-containing band (the opaque, middle band) was collected, and the fraction was washed two times with Dulbecco's modified Eagle's medium (DMEM) and centrifuged at 50 × g for 2 min. The tubule fraction was then resuspended in DMEM-F12-based medium with supplements (46) and placed in culture dishes. For the 3-dimensional culture of kidney tubule cells, the published protocol was followed (39).

RESULTS

dVHL mutant.

The genomic dVHL mutant allele was generated using the homologous recombination strategy (53). The targeting sequence contains an ∼4-kb genomic fragment encompassing the dVHL locus with a deletion of 243 bp that removes the first two in-frame AUG codons (Fig. 1 A, left panel). Homozygous dVHL mutants are sluggish after hatching but can feed and survive till the end of the first-instar larval stage. No homozygous mutants survive past the second-instar stage. The allele is named dVHL¹. The stock is maintained over a balancer chromosome containing the YFP transgene expressed from the Dfd gene promoter (34). Homozygous dVHL, heterozygous dVHL, and homozygous balancer (“wild-type”) embryos can be distinguished visually based on the extent of the yellow fluorescent protein (YFP) expression in the head region (Fig. 1A, middle panels). Homozygous balancer embryos express a band of YFP about twice as thick as that of heterozygotes, while homozygous dVHL embryos express no YFP. The visual identification was confirmed by genomic analysis of handpicked embryos (Fig. 1A, right panel).

FIG. 1.

Characterization of the dVHL¹ allele. (A) The left panel shows a schematic representation of the genomic region of the dVHL open reading frame. The two in-frame AUG codons are marked. Arrowheads depict the PCR primers used to distinguish between the wild-type and the deletion mutant fragments. The expected fragment lengths are indicated. The distinction between heterozygous dVHL¹ (dVHL¹/CyO, Dfd-YFP), homozygous dVHL¹, and homozygous balancer (CyO, Dfd-YFP/CyO, Dfd-YFP) is shown in the middle panels. Homozygous balancer (Cy/Cy, containing two copies of Dfd-YFP) expresses an expanded YFP pattern in the head region (thick bracket) compared to that expressed by heterozygotes (dVHL/Cy), which contain only one copy of Dfd-YFP (thin bracket). The visual identification is confirmed by the genotyping (right panel) of hand-sorted embryos. (B) Progeny counts from dVHL¹/CyO; hsp-dVHL parents. The adult progenies were counted based on gender and wing phenotypes. The exogenously expressed dVHL can fully rescue the dVHL¹ lethal phenotype, since the genotypes of the progenies follow the expected Mendelian ratio. (Note: CyO/CyO is lethal at the late embryonic/early larval stages.) No external morphological phenotypes were observed in heterozygous (Cy⁻) or homozygous (Cy⁺) dVHL¹ (right panel). (C) Embryos of the indicated genotypes and stages were stained with rabbit anti-dVHL and mouse anti-GFP antibodies. In y w (representing wild-type) embryos, dVHL protein is ubiquitous but enriched after stage 10 in the tracheal cells (arrowheads in insets). dVHL¹ heterozygous embryos (dVHL¹/CyO, Dfd-YFP) are identifiable by modest expression of YFP in the head region (arrow). Homozygous dVHL¹ embryos exhibit a very low level of dVHL staining. Bars, 50 μm.

The lethal phenotype can be rescued by expressing wild-type dVHL cDNA under the control of the hsp70 promoter (twice daily at 37°C for 30 min each, from embryonic stage to eclosion), and the rescued homozygous adults show no external morphological defects (Fig. 1B). Therefore, dVHL gene truncation is the only major genetic defect in the dVHL¹ allele. Polyclonal antibody raised against the full-length dVHL protein detects ubiquitous expression of dVHL in the embryos (Fig. 1C). However, the protein is enriched in the trachea during tracheal morphogenesis, beginning in the tracheal pits before branch migration at stage 11 and persisting in the extended tubules at stage 16. The expression level is reduced in the heterozygotes and nearly completely disappears in homozygotes. Residual expression in the homozygotes may indicate maternal contribution.

Tracheal phenotypes in the dVHL mutant.

Monoclonal antibody 2A12 against a lumen antigen was used in the initial examination of the tracheal network. Two distinct classes of phenotypes were observed. The first involves abnormal tubule migration, and the second exhibits enlarged lumina and tortuous tracheal tubes. The former phenotype resembles those resulting from overactive chemotactic signaling (2, 20, 63), while the latter has recently been shown to result from defective production and secretion of the lumen structural protein chitin (3, 14, 37, 43, 61), septate junction defects leading to defective secretion (37, 65, 69-71), and defective resorption of lumen material (5, 62). We set out to analyze how these two phenotypes might be regulated by an underlying dVHL function.

Compared to the wild type (y w) (Fig. 2 A), dVHL mutants (Fig. 2B to D) show misdirected tubules and ectopic branching in addition to disruption of the dorsal trunk. This represents the first phenotypic class. In a heterozygous example (Fig. 2B), a section of the dorsal trunk does not fuse laterally with its neighboring subunit (arrowhead) but turns dorsally. In another position, an ectopic side branch is observed (arrow). In more severe phenotypes seen in homozygotes, tracheal fragments are detached from the main trunk (sharp arrow in Fig. 2C) and ectopic loops and extra secondary branches are observed (filled arrows in Fig. 2C and D). In the same embryos, convoluted dorsal trunks (the second phenotypic class) are also obvious (empty arrowheads in Fig. 2C and D), indicating that the two phenotypic classes are not mutually exclusive.

FIG. 2.

The dVHL loss-of-function phenotype results in abnormal tracheal structure during embryonic development. (A to J) Stage 15 and 16 embryos were stained with a tracheal lumen-specific antibody 2A12 and anti-GFP antibody. (A) y w stage 16 embryo. Six primary tracheal tubes are marked. DB, dorsal branch; DT, dorsal trunk; GB, ganglionic branch; TC, transverse connector; VB, visceral branch; VT, ventral trunk. (B) dVHL¹ heterozygote. The bracket shows the partially overlapped image of YFP expression from two sides of the embryo. The arrowhead points to a section of the DT that does not connect with the neighboring branch but turns dorsally. The arrow points to an abnormal branching point in the dorsal trunk. (C and D) Two dVHL¹ homozygotes. The sharp arrow in panel C points to isolated fragments of DT. The filled arrows in panels C and D point to ectopic branches, and the empty arrowheads point to convoluted DT. (E) An awd^j2A4 homozygous embryo exhibiting tracheal phenotypes similar to those in panels B to D. The arrowhead points to a section of the DT that turns 90°. The arrow points to an ectopic branch. (F and G) Two independent transgenic lines expressing dVHL duplex (dVHLi) directed by the btl promoter. The arrow in panel F points to an upturned DT tip sprouting abnormal branches (enlarged view). The arrows in panel G point to ectopic branches. (H and I) Two independent transgenic lines expressing full-length dVHL sequences directed from the btl promoter. Empty arrows point to gaps in the DT. These gaps are not accompanied by ectopic branching. (J) Homozygous btl^H82Δ3 embryo showing the typical lack-of-migration phenotype: a gap in the DT without ectopic branching. (K) Quantification of phenotypes. “Moderate” phenotype denotes up to two defects (breaks in the trunk, abnormal branches, etc.) found in the tracheal tree. “Severe” denotes more than two identified defects, including embryos showing severe destruction of the tracheal tree. Bars, 50 μm.

The ectopic branching phenotypes are similar to those observed in the abnormal wing discs (awd) and shibire (shi)/dynamin mutants (Fig. 2E) (13), which result in surface overaccumulation of the fibroblast growth factor receptor (FGFR) (Breathless) due to defective endocytosis. Btl-mediated signaling is the major chemotactic system in tracheal branching morphogenesis (2, 20, 63). The guidance cue is provided by the FGF homolog Branchless (Bnl), expressed in target tissues distal to the migrating tips (51, 60), and is received by the receptor Btl expressed in the tracheal cells (21, 32, 35, 48).

The ectopic branching phenotypes in dVHL loss-of-function mutants are reproduced by independent lines of transgenic flies expressing a dVHL-specific RNA duplex (dVHLi) using the UAS/Gal4 binary system directed by the tracheal cell-specific btl gene promoter (Fig. 2F and G). Also, importantly, when full-length dVHL is overexpressed using the btl promoter in the trachea (two independent lines tested), a disconnected dorsal trunk without ectopic branching is observed (Fig. 2H, I, and K). Such a phenotype is typical of the lack-of-migration phenotype observed in the btl signaling pathway mutants (Fig. 2J; also see below). Thus, dVHL loss of function and overexpression generate opposite phenotypes reminiscent of those resulting from elevated and reduced Btl signaling, respectively. In addition, btl-directed expression of dVHL can rescue dVHL branching phenotypes (Fig. 2K). This also demonstrates that the tracheal branching phenotypes are a specific, cell-autonomous effect of dVHL loss of function in the tracheal cells.

To examine whether maternal contribution of dVHL may account for the survival of dVHL homozygotes to 2nd-instar larvae, we generated germ line clones of the dVHL¹ mutant (dVHL^1germ). Embryos from females carrying germ line mutations crossed with heterozygous males do not hatch and show no development of tracheas. Crumbs (Crb; an apical marker of epithelial cells) staining highlights the apical (lumen-facing) surface of wild-type trachea at all stages. In dVHL^1germ mutants, Crb-positive cells are seen spreading out in the midsection of the embryos without forming a tracheal network (Fig. 3 A and B). The identities of these unattached tracheal cells were further confirmed by staining with a trachea-specific marker, Trachealess (Trh) (Fig. 3C and D). Therefore, without maternal contribution of dVHL, tracheal epithelium is completely disrupted, resulting in prelarval lethality.

FIG. 3.

dVHL¹ germ line mutants show complete failure in forming trachea. y w embryos and embryos from females containing germ line clones of dVHL¹ ({ovoD1-18}, FRT^G13/dVHL¹, FRT^G13; ovo-FLP/+) crossed with heterozygous dVHL¹ males were stained for Crumbs (Crb) (A and B) or Trachealess (Trh) (C and D). y w embryos show Crb staining on the apical surface of the tracheal tubes facing the lumen (A) and Trh in the nuclei (C). (B) In dVHL¹ germ line mutants, scattered Crb-positive cells are detected, but no tracheal tubes are formed. (D) The identity of the scattered cells is further confirmed by staining for Trh, the tracheal cell-specific transcription factor. Bars, 50 μm.

Ectopic tracheal cell migration in the dVHL mutant.

Lumen staining using monoclonal antibody 2A12 is useful for observing the overall structure of the tracheal network, but the cellular behavior underlying the branching defects cannot be discerned. Completion of the dorsal trunk requires correct directional movement and fusion of the neighboring tracheal subunits, as illustrated in Fig. 4 A'; therefore, either ectopic migration or lack of migration can result in breakage of the tracheal tree. In order to examine the tracheal cell behavior, we combined the dVHL mutant with the lacZ transgene expressed from the trachealess (trh) gene promoter (the 1-eve-1-transgenic fly), which marks all tracheal cells. Embryos expressing 1-eve-1 alone are normal and do not exhibit major tracheal defects (Fig. 4A) (13, 26), while significant tracheal phenotypes are observed in both heterozygous (Fig. 4B and C) and homozygous (Fig. 4D and E) embryos. Consistent with the phenotypes revealed in lumen staining by 2A12 (Fig. 2), gaps in the dorsal trunk are associated with misallocation of tracheal cells. In some cases, an entire section of the dorsal trunk migrates dorsally or ventrally (Fig. 4B). Disorganized tubules and ectopic branching are frequent (Fig. 4C). Abnormal sprouting of filopodium-like projections can be seen in isolated cells (insets in Fig. 4B and E). In very severe phenotypes, the entire tracheal system is disrupted, with an abundance of ectopic branching and rounded tracheal cells detached from the epithelium (Fig. 4D and E). Consistent with the results shown in Fig. 2, these severe phenotypes can also be found in the awd and shi mutants (Fig. 4F and G). Conversely, these phenotypes are opposite to the lack-of-migration phenotypes observed in the btl and the btl signaling effector pnt mutants (Fig. 4H and I). Consistent with the idea that dVHL and the endocytic pathway components may function cooperatively, transheterozygous dVHL and awd embryos show very severe phenotypes resembling those found in individual awd and dVHL homozygotes (compare Fig. 4J with 4E and F; also see Fig. 5 E).

FIG. 4.

dVHL mutant tracheal cells exhibit ectopic migration phenotypes. Mutant strains were combined with the 1-eve-1-transgenic marker that expresses β-Gal in the trachea (from the trachealess promoter). Stage 15 and 16 embryos were stained with anti-β-Gal and anti-GFP antibodies. (A) Stage 16 1-eve-1 control embryo. (A') Two tracheal subunits from a stage 14 1-eve-1 embryo. Arrows indicate the directional movements of the branches that form the dorsal trunk. (B and C) dVHL¹ heterozygotes. Brackets mark the YFP expressed in the head region from the balancer chromosome. The arrowhead in panel B points to a section of the DT turning 90° ventrally. The further enlarged area shows an isolated cell sprouting multiple filopodia. Arrows in panel C point to ectopic branches. (D and E) Two dVHL¹ homozygous embryos showing severely disrupted tracheal trees with rounded, isolated tracheal cells. Filled arrows in panel D point to ectopic branches. Empty arrowheads in panel E point to abnormal sprouts of filopodia. Sharp arrows point to isolated, rounded cells. The two enlarged views in panel E are different focal planes of the same area. (F) An awd^j2A4 homozygote showing a severely disrupted tracheal tree. The empty arrowhead points to abnormal sprouts of filopodia. The sharp arrow points to rounded cells. (G) A shi² temperature-sensitive mutant grown at a restrictive temperature showing abnormal sprouts (empty arrowheads) and isolated cells (sharp arrows). (H) btl^H82Δ3 homozygote showing the lack-of-migration phenotype with gaps (empty arrows) in the DT. (I) pnt^Δ88 homozygote showing the lack-of-migration phenotype with gaps (empty arrows) in the DT. (J) dVHL¹/+; awd^j2A4/+ transheterozygote showing a severely disrupted tracheal tree with ectopic filopodia (empty arrowheads) and isolated, rounded cells (sharp arrows). Bars, 50 μm.

FIG. 5.

Elevated Btl signaling in the dVHL mutant. 1-eve-1 (A and C) and dVHL¹ mutant (B and D) embryos were stained for β-Gal and dually phosphorylated MAPK (dp-MAPK; A and B) or Pnt (C and D). In 1-eve-1 embryos, dp-MAPK and Pnt are modestly expressed and localized to the migrating tip cells. In dVHL¹ mutant embryos, dp-MAPK and Pnt expression levels are elevated. They are also ectopically expressed in multiple cells. (E) Quantification of phenotypes showing genetic interactions between dVHL¹ and endocytic pathway components and btl signaling components. Phenotype designations are as described in the legend to Fig. 2K. The pairwise statistical analysis is shown on the left, using a two-sample t test for percentages. *, significant changes (P < 0.05). Bars, 20 μm.

Overactive FGFR/Btl signaling in the dVHL mutant.

To further test whether increased FGF signaling is the cause of the ectopic branching phenotype in the dVHL mutant, we first examined the level of dually phosphorylated mitogen-activated protein kinase (dp-MAPK), which is generated downstream of activated FGFR (64), and the level of the ETS family of transcription factor Pointed (Pnt), which is activated by Btl signaling (54). In wild-type cells (Fig. 5A and C), dp-MAPK and nuclear Pnt are expressed at moderate levels mainly in the migrating tip cells. In dVHL mutant tracheal cells (Fig. 5B and D), not only are the levels of these two markers increased, but also more cells become positive. We next performed genetic interaction analyses. Since the dVHL mutant and many of the btl signaling pathway mutants show haploid insufficiency in tracheal development (Fig. 2 and 4) (13, 26), we examined whether transheterozygous combinations could modify the dVHL phenotypes. As shown in Fig. 5E, 62% of heterozygous dVHL mutants exhibit moderate (56%) or severe (6%) ectopic migration defects, while 81% (71%, moderate; 10%, severe) of heterozygous awd mutants and 91% (72%, moderate; 19%, severe) of heterozygous/hemizygous shi mutants (shi is X linked) exhibit abnormal phenotypes. When dVHL is combined with awd or shi, the ectopic migration defects are significantly increased, especially the severe defects (from 6% to 35% or 40%, respectively). In contrast, 82% of heterozygous btl mutants and 59% of pnt mutants exhibit the lack-of-migration phenotypes. When dVHL is combined with either btl or pnt, both the ectopic and lack-of-migration phenotypes are significantly reduced. Therefore, further impairing internalization of Btl in the dVHL mutant background by the awd and shi mutants exacerbates the ectopic migration defects, while reducing the dosage of FGF signaling components btl and pnt in the dVHL mutant rescues the severity of both phenotypes. These results strongly suggest that in the dVHL mutant, FGF signaling is elevated, resulting in ectopic branching.

Defects in Btl internalization in dVHL mutants.

We have shown that the branching phenotypes in the dVHL mutant correlate with overactive chemotactic signaling and that dVHL genetically interacts with btl signaling and endocytic components. Since our observations so far suggest a cell-autonomous function of dVHL, we examined the expression pattern of Btl in the tracheal cells. A btl::GFP fusion protein gene was expressed from the UAS enhancer under the control of the btl promoter-driven Gal4. As shown in Fig. 6 A, in wild-type (1-eve-1) embryos, Btl::GFP is expressed at a low level on the cell periphery, in contrast to the high-level, generalized cytoplasmic expression of UAS-GFP from the same btl-Gal4 driver (Fig. 6A'). This indicates that the posttranscriptional expression level and localization of the receptor are tightly regulated. In contrast, in dVHL homozygotes (Fig. 6B), Btl::GFP is highly accumulated on or near the cell surface in both migrating tip cells and in nonmigrating cells located in the tracheal trunks. The surface accumulation phenotype is to a certain extent specific, since the ectopically expressed Drosophila epidermal growth factor receptor (dEGFR)-GFP fusion protein from the same btl promoter shows no difference in the wild type or in the dVHL mutant (Fig. 6C). The mechanism for such specificity is currently unknown but is consistent with previous findings in the mammalian systems (9, 27).

FIG. 6.

Surface accumulation of Btl in dVHL mutant cells. (A and B) 1-eve-1 and dVHL¹ mutant late-stage-12 embryos carrying the btl-Gal4; UAS-btl::GFP transgenes were double stained for β-Gal and GFP. (A) The 1-eve-1 embryo shows modest expression of Btl::GFP on the cell surface. (B) The dVHL¹ mutant embryo shows extensive accumulation of Btl::GFP at or near the cell periphery in all tracheal cells. (A') Stage 11 y w embryo expressing GFP (not btl fusion) from the btl promoter. The GFP itself shows strong, diffused expression throughout the cell bodies. (C) 1-eve-1 and dVHL¹ mutant stage 13 embryos carrying the btl-Gal4; UAS-dEGFR::GFP transgenes were double stained for GFP. There is no difference in the expression levels or patterns of dEGFR::GFP. (D) S2 cells are transfected with metallothionein>btl::GFP plasmids and hsp70 vectors expressing randomized small interfering RNA (control), dVHL duplex (dVHLi), or full-length dVHL and processed for staining with anti-GFP and anti-dVHL antibodies. In control cells, the Btl::GFP fusion protein is localized mostly in vesicles also containing dVHL (D'). With dVHL knockdown (dVHLi), Btl::GFP is highly enriched at the cell periphery (arrows). Reduction of surface Btl::GFP expression is observed with dVHL overexpression. (E) Surface biotinylation assay showing surface fractions and total lysates of Btl::GFP. Total cell lysates (left panel) were Western blotted with the indicated antibodies. Surface fractions were probed with anti-GFP. Numbers under the protein bands are the intensities of the signals normalized against the actin control. Red bars, 20 μm; white bars, 5 μm.

This Btl accumulation phenotype is verified in cultured Drosophila S2 cells. The cells were stably transfected with a btl::GFP-expressing plasmid (from the inducible metallothionein promoter) and double transfected with a heat-shock promoter-directed expression vector without the transgene (control), with the dVHL coding sequence, or with the dVHL duplex (dVHLi for knockdown). As shown in Fig. 6D, in control cells expressing btl::GFP, the receptor is present at a modest level in vesicles throughout the cell body. Interestingly, most of these vesicles contain both Btl::GFP and endogenous dVHL (Fig. 6D'), indicating that dVHL is part of the endosomal components that internalize Btl. Overexpression of dVHL slightly reduces the level of Btl::GFP, while knockdown of dVHL dramatically increases the surface and peripheral accumulation of Btl::GFP. To quantify the changes of surface Btl::GFP levels, we performed a surface biotinylation assay on these cells, in which surface proteins were conjugated with biotin moieties under nonpermeable conditions. The surface proteins were then pulled down by streptavidin-agarose beads. As shown in Fig. 6E, total Btl::GFP levels change only modestly with increased or reduced dVHL levels. However, the surface level of Btl::GFP is increased by 70% in dVHL knockdown cells compared to that in control cells. Conversely, overexpression of dVHL reduces the surface level of Btl::GFP by 40%. Since in the S2 cell culture system btl::GFP is expressed from a noncognate promoter (metallothionein) but shows the same surface accumulation in dVHL knockdown cells as in the mutant embryos, it is unlikely that this phenotype is mediated at the transcriptional level. Therefore, we suspect that the Drosophila HIF (Sima)-mediated transcription mechanism is not a major factor in Btl surface accumulation.

dVHL protein interacts with Awd.

In order to verify the endocytic function of dVHL, we performed an immunoprecipitation assay on extracts from embryos expressing the dVHL-hemagglutinin tag (dVHL::HA) fusion protein from the btl promoter. Canonical endocytic markers such as dynamin and Rab5 are not detected in the dVHL-containing protein complex (data not shown). Interestingly, the endocytic factor Awd is readily detected (Fig. 7 A). A reciprocal experiment using anti-Awd antibody for immuno-pulldown confirms that dVHL and Awd can interact in vivo (Fig. 7A). As a first step toward resolving the functional relationship between dVHL and Awd, we show that the Awd protein level is reduced in step with the knockdown of endogenous dVHL (Fig. 7B). Immunofluorescence analyses also confirm that the Awd protein levels are greatly reduced in the dVHL¹ mutant compared to the wild type (1-eve-1) (Fig. 7C). Therefore, one likely function of dVHL is to promote the stability and activity of the endocytic factor Awd, which in turn may be involved in stabilization of Rab5 (68).

FIG. 7.

dVHL forms a complex with Awd in vivo. (A) Embryos carrying btl-Gal4; UAS-dVHL::HA were collected at 29°C, and extracts were prepared in RIPA buffer. Extracts (50 μg) were incubated with rabbit polyclonal anti-Awd, rabbit polyclonal anti-HA, or rabbit IgG, as indicated. Immunoprecipitation was performed using protein A-agarose beads. The pulldown fraction was then Western blotted using rabbit anti-dVHL antibody, chicken anti-human Nm23 (which cross-reacts with Awd), or mouse antiactin antibodies. (B) y w embryos or y w embryos containing one copy or two copies of hsp-dVHLi were left untreated or were heat treated twice a day for 30 min each at 37°C for 3 days. Embryonic extracts were Western blotted for the indicated proteins. dVHL levels are knocked down by dVHLi in a dosage-dependent manner, and Awd levels are reduced in step with the reduced dVHL levels. (C) 1-eve-1 or dVHL¹ embryos were processed for immunofluorescence detection of the proteins indicated. The Awd level is diminished in the dVHL mutant. Bars, 50 μm.

Lumen defects in the dVHL mutant.

The potential endocytic function of dVHL brought to our attention the second phenotypic class of the dVHL mutant mentioned earlier. These dramatic phenotypes, including tortuous paths of tracheal tubules (Fig. 8 A and C to G) and dilated dorsal trunks (Fig. 8B, C, and G), are seen in dVHL (Fig. 8A to D), shi (Fig. 8F), and awd (Fig. 8G) mutants, as well as in dVHL duplex-induced knockdown in the trachea (Fig. 8E). Also, clearing of the lumen antigen is delayed in dVHL mutant first-instar larvae (compare Fig. 8I to H). These phenotypes have been shown to also result from defects in the endocytic activity of the tracheal cells (5, 62). As pointed out earlier, these lumen defects are not mutually exclusive with the branch migration defects, since tortuous tracheal tubes can be found in the same embryos that show ectopic branching as well (Fig. 8A, C, and E to G). However, lumen phenotypes are not observed in the FGF signaling pathway mutants, such as btl and pnt. Indeed, while awd or shi can exacerbate the lumen phenotype when combined with dVHL¹, btl and pnt have no modifier effects (Table 1). This indicates that the underlying endocytic function of dVHL can work on different aspects of tracheal morphogenesis, one involving internalization of the Btl receptor and the other possibly involving the clearing of lumen materials.

FIG. 8.

Lumen phenotypes in the dVHL mutant. (A to I) Late-stage embryos or first-instar larvae were stained with the 2A12 and GFP antibodies. (A) dVHL¹ heterozygote showing a convoluted tracheal trunk (arrow) and ectopic, tortuous secondary branches (arrowheads). (B) A dVHL¹ homozygote showing a dilated dorsal trunk (arrow). (C and D) Two dVHL¹ homozygotes showing a dilated and convoluted dorsal trunk (arrow) and ectopic branches (arrowheads). (E) Embryo expressing dVHL duplex (dVHLi) from the btl promoter showing a convoluted dorsal trunk (arrow) and tortuous branches (arrowhead). (F and G) shi² and awd^j2A4 mutants showing the same lumen phenotypes as in the dVHL¹ mutant, including a dilated dorsal trunk (arrow) and ectopic, tortuous branches (arrowheads). (H) y w 1st-instar larva showing little 2A12 staining in the trachea. (I) dVHL¹ homozygous 1st-instar larva showing strong residual 2A12 staining (arrow). (J and K) 1-eve-1 and dVHL¹; 1-eve-1 embryos were stained with anti-β-Gal antibody and FITC-labeled chitin-binding probe (CBP). At stage 14 (J), both 1-eve-1 and dVHL¹; 1-eve-1 embryos show proper deposition of chitin in the lumen. At stage 15 (K), lumen chitin begins to fade in the 1-eve-1 embryo but remains abundant in the dVHL¹ mutant. In addition, unlike the wild-type lumen chitin, the mutant lumen chitin is filled up against the apical surface of the tracheal cells (compare insets). The arrowhead in the panel K inset points to the small gap between the chitin material and the tracheal cell apical surface in the wild type. (L) A stage 15 embryo from transgenic flies carrying btl-Gal4; UAS-dVHLi was stained with anti-β-Gal antibody and rhodamine-CBP. The tracheal lumen shows abundant chitin accumulation similar to that found in the dVHL¹ mutant. (M to O) Stage 15 embryos from transgenic flies carrying btl-Gal4; UAS-control (as a negative control) or btl-Gal4; UAS-dVHLi were combined with transgenic flies carrying the wild type (Rab5^wt) or the constitutively active (Rab5^CA) or dominant negative (Rab5^DN) variants of UAS-YFP::Rab5. These were stained with anti-GFP antibody and rhodamine-CBP. Rab5^wt cannot rescue (M) but Rab5^CA can reverse (N) the dVHL knockdown phenotypes (increased lumen chitin and a tortuous dorsal trunk). (O) Rab5^DN alone causes delayed chitin clearing and a convoluted dorsal trunk, resembling that of the dVHL¹ mutant. The phenotype is exacerbated when combined with the dVHL duplex. Bars, 50 μm.

TABLE 1.

Genetic interactions between dVHL and endocytic and btl signaling pathway components^a

Genotype	No. of embryos	% normal (Pb)	% with dorsal trunk phenotypes (convoluted + dilated)
1-eve-1	25	100	0
dVHL1/+; 1-eve-1/+	27	37	63
awdJ2A4, 1-eve-1/+	21	52	48
shi2/+ or Y; 1-eve-1/+	16	25	75
dVHL1/+; awdJ2A4, 1-eve-1/+	23	9 (0.025)	91
shi2/+ or Y; dVHL1/+; 1-eve-1/+	12	0 (0.020)	100
btlH82Δ3, 1-eve-1/+	23	100	0
pntΔ88/+; 1-eve-1/+	25	100	0
dVHL1/+; btlH82Δ3, 1-eve-1/+	24	42 (0.72)	58
dVHL1/+; pntΔ88, 1-eve-1/+	22	45 (0.57)	55

^aQuantification is based on embryos stained with 2A12, as shown in Fig. 2.

^bStatistical analyses used two-sample t tests for percentages. P values are based on comparisons to the dVHL¹/+; 1-eve-1/+ genotype. A P value of <0.05 is considered significant.

To test the lumen function further, we examined the deposition and clearing of luminal chitin, visualized by using a fluorescence-labeled chitin-binding probe (CBP). In the wild type and dVHL mutant at stage 14, when lumen begins to form, there is no difference in the timing of the appearance of the luminal chitin (Fig. 8J). At stage 15, wild-type embryos (Fig. 8K, left panel) show a dispersed pattern of luminal chitin with a small gap between the central lumen matrix and the apical surface of the tracheal cells (arrowhead in inset). Such lumen architecture has been noted before (5, 62). Also, the amount of lumen material begins to fade at stage 15 because of resorption by the tracheal cells (5, 62). In the dVHL mutant, the chitin material is filled up to the tracheal surface and remains abundant in the mature dorsal trunk (Fig. 8K, right panel). These phenotypes have been linked to defective endocytosis in the wurst mutant, which affects uptake of lumen materials (5). To further establish the endocytic function of dVHL, we expressed variants of the Rab5 proteins (the wild type and constitutively active and dominant negative variants fused at the N termini with the YFP coding sequence) and analyzed their functional interaction with dVHL. For this set of genetic interaction analyses, dVHL is knocked down by the dVHL duplex (dVHLi) expressed from the btl promoter, resulting in delayed CBP clearing (Fig. 8L). Wild-type Rab5 (Rab5^wt) overexpressed in the trachea does not generate the lumen phenotype and does not rescue the dVHL phenotype in CBP resorption (Fig. 8M and Table 2). This indicates that dVHL function may be required for normal Rab5 activity. Constitutively active Rab5 (Rab5^CA), on the other hand, can moderately rescue dVHL phenotypes (Fig. 8N and Table 2), including convoluted dorsal trunk and defective chitin resorption. In contrast, dominant negative Rab5 (Rab5^DN) can generate dVHL-like lumen phenotypes (Fig. 8O), suggesting that the two genes affect the same process in a similar manner. In agreement with this assertion, Rab5^DN can exacerbate the phenotype when combined with the dVHL knockdown duplex (Fig. 8O and Table 2).

TABLE 2.

Genetic interactions between dVHL and Rab5 in the lumen formation phenotype^a

Genotype	No. of embryos	% normal (Pb)	% with dorsal trunk lumen phenotype (convoluted + dilated)
UAS-lazZ/+; UAS-dVHLi, btl-GAL4/+	24	63	37
UAS-YFP::Rab5wt/+; UAS-FLP, btl-GAL4/+	16	100	0
UAS-YFP::Rab5wt/+; UAS-dVHLi, btl-GAL4/+	18	61	39
UAS-YFP::Rab5CA/+; UAS-FLP, btl-GAL4/+	17	100	0
UAS-YFP::Rab5CA/+; UAS-dVHLi, btl-GAL4/+	23	87 (0.065)	13
UAS-YFP::Rab5DN/+; UAS-FLP, btl-GAL4/+	17	76	24
UAS-YFP::Rab5DN/+; UAS-dVHLi, btl-GAL4/+	21	33 (0.066)	67

^aUAS-lacZ and UAS-FLP are used as neutral UAS transgenes for normalizing the effective Gal4 protein dosage.

^bStatistical analyses used two-sample t tests for percentages. P values are based on comparisons to the UAS-lazZ/+; UAS-dVHLi, btl-GAL4/+ genotype. A P value <0.1 indicates modest significance.

To begin to address the question of whether the tracheal function of dVHL is HIF dependent, we tested the rescuing capability of human disease-related mutations that have been shown to partially retain HIF degradation activity: a Y-to-H mutation at human residue 98 (Drosophila residue 51) and an R-to-Q mutation at human residue 167 (Drosophila residue 132) (12, 25, 30, 45). The mutant dVHL fusion proteins (as well as the wild-type control) were expressed from the btl promoter in combination with the hsp-dVHLi allele. The mutant phenotypes (ectopic migration and abnormal lumen) in HA-expressing embryos (containing rescue constructs) as well as the internally controlled non-HA-expressing embryos (no rescue constructs) were tabulated. While wild-type dVHL::HA can significantly rescue the loss-of-function phenotypes, no rescue of the phenotypes was observed for either mutation (Fig. 9 A and B). Importantly, the ectopically expressed Sima (Drosophila HIF-1α)-GFP fusion protein is not overtly upregulated in dVHL knockdown trachea (Fig. 9C), indicating that either the ectopically expressed Sima protein can overwhelm the endogenous oxygen-sensing system or such system is not efficient in the embryonic trachea or both. In addition, in a transheterozygous genetic interaction analysis, we observed that sima heterozygosity can only modestly suppress the branch migration phenotypes in the heterozygous dVHL mutant (Table 3). Although more detailed genetic analyses involving additional VHL mutations are needed for a definitive conclusion, it is likely that the HIF-dependent function of dVHL does not play a major role in tracheal development during embryogenesis (see Discussion).

FIG. 9.

VHL disease-related mutations cannot rescue the tracheal phenotypes. (A) There is no rescue of the dVHL phenotypes by either of the dVHL mutations in this assay. (B) Examples of the expressed dVHL transgenes. There is no difference in the cellular expression patterns of the dVHL variants. (C) The Sima::GFP fusion protein was expressed in the btl-directed system, with or without dVHLi-mediated knockdown. Sima::GFP is detected strongly in the posterior portion of the dorsal trunk (green dots). There is no increase in the expression areas or levels of the fusion protein in dVHL knockdown embryos. Bars, 50 μm.

TABLE 3.

Modest genetic interactions between dVHL and sima^a

Genotype	No. of embryos	% normal (Pb)	% with ectopic migration phenotype		% with lack-of-migration phenotype
			Moderate	Severe	Moderate	Severe
dVHL1.4/+	31	35 (0.09)	55	10	0	0
sima/+	33	67 (0.42)	0	0	33	0
dVHL1.4/+; sima/+	30	57	43	0	0	0

^aTracheal phenotypes were analyzed using 2A12 staining, as described in the legend to Fig. 2. Transheterozygous dVHL¹ and sima genotypes only modestly rescued the severity of phenotypes exhibited in the individual heterozygotes (not statistically significant). Moderate and severe phenotypes are as defined in the legend to Fig. 2.

^bStatistical analyses used two-sample t tests for percentages. P values are based on comparisons to the dVHL^1.4/+; sima/+ genotype. A P value <0.1 indicates moderate significance.

Vhl knockout in mouse kidney tubules generates lumen and branching defects.

To examine whether the novel function of dVHL described here is relevant in the mammalian system, we generated a conditional Vhl-null knockout in the kidney tubules using the existing Vhl loxP allele (22) and the Hoxb7-driven Cre strain (59). Remarkably, in the 3- to 4-month-old knockout mice (8 of 8), prominent kidney lesions containing distorted and dilated kidney tubules were observed in kidney sections (compare Fig. 10 B and C with A). To confirm the Vhl mutant status, kidney tubule cells were isolated from Vhl knockouts (Hoxb7-Cre^/+; Vhl^loxP/loxP) and their wild-type littermates (Vhl^loxP/loxP). As shown in Fig. 10D, tubule cells from knockout mice show no dVHL expression, either by Western blotting or by immunohistochemistry. Importantly, they exhibit overaccumulated FGFR on the cell surface. Photometric analysis of 10 randomly chosen cells from each genotype shows that the Vhl mutant cells exhibit about a 6-fold-higher ratio of surface FGFR to total FGFR expression than wild-type cells. It has been shown that in VHL mutant cells, the main signaling event downstream of FGFR is mediated by extracellular signal-regulated kinase (ERK) and its target Ets1 transcription factor (Fig. 5) (9, 27). We Western blotted for activated ERK (dually phosphorylated) and phosphorylated Ets1 and demonstrated the activation of this signaling pathway in Vhl knockout cells (Fig. 10E). We note that these cells show substantial nuclear localization of FGFR1. The mechanisms for nuclear localization of FGFR in mammalian cells have been studied for some time (38, 44, 50), although the functional significance is not yet clear. Recently, however, nuclear localization of FGFR1 has been shown to depend on ligand binding and participate in gene transcription (16). These cells were then cultured in a 3-dimensional matrix (Fig. 10F). Wild-type cells form epithelial acini with organized lumina (Fig. 10Fa). In contrast, Vhl knockout cells form distorted multicellular structures. In one example, a group of cells protrudes from the epithelium, reminiscent of the ectopic branching seen in the dVHL mutant trachea (Fig. 10Fb). The lumen is also enlarged and distorted. In another example (Fig. 10Fc), the mutant cells cannot form proper epithelia. Instead, cells are observed breaking away from the cluster, with prominent actin stress fibers extending from the migrating front (inset). We conclude that the VHL function in regulating epithelial tubule morphogenesis is conserved.

FIG. 10.

Mouse conditional Vhl mutants exhibit dilated kidney tubules and ectopic migration phenotypes. Kidneys from 4-month-old Vhl^loxP/loxP (wild-type) or Hoxb7-Cre/+; Vhl^loxP/loxP (Vhl^−/−) mice were dissected. The kidneys were fixed and sectioned for hematoxylin-eosin staining (A to C) or were processed for tubule cell isolation (D to F). (A) The wild-type kidney shows normal tubules (examples are outlined with dashed lines). G, examples of glomeruli. (B and C) Kidneys from two knockout animals, showing convoluted and dilated tubules. (D) Left panels show cultured primary kidney tubule cells immunostained for VHL and FGFR1. The mutant shows little VHL protein expression and accumulation of FGFR1 on the cell surface at the migrating front (arrows). Bars, 50 μm. The right panel shows quantification of surface FGFR1 versus total cellular FGFR1 levels using the ImageJ photometric analysis program. (E) Lysates from isolated kidney tubule cells were Western blotted for VHL and FGFR signaling mediators: dually phosphorylated ERK (dp-ERK), total ERK, phosphorylated Ets1 (p-Ets1), total Ets1, and actin control. (F) Primary tubule cells were cultured in a 3-dimensional matrix (2% Matrigel) and stained for F-actin (phalloidin), E-cadherin, and DNA (To-Pro3). (a) Wild-type cells form an organized acinus with a defined lumen in the center. (b) Vhl mutant cells form an irregular cluster with outgrowth and a dilated lumen. (c) Another mutant example shows cells migrating away from the irregular cluster with prominent actin fibers in the migrating tips (insets). Bars, 20 μm.

DISCUSSION

In this report, we show that dVHL regulates tracheal tubule development in two aspects—branch migration and lumen formation. These two aspects of tracheal morphogenesis are regulated by the same endocytic function; one involves internalization of the Btl signaling receptor, and the other involves resorption of lumen materials.

We show that dVHL genetically interacts with endocytic pathway components such as awd, shi, and Rab5 in both branch migration and lumen formation phenotypes but interacts with btl signaling pathway components only in branch migration phenotypes. The lumen defects in the dVHL mutant are similar to those found in the wurst mutant. wurst encodes a transmembrane protein that promotes clathrin-mediated endocytosis of lumen material (5). wurst function is specific for lumen maturation and, unlike dVHL, has little effect on branch migration. This indicates that wurst may be specific for pinocytosis (uptake of extracellular materials) and not involved in internalization of surface proteins, while the dVHL- and awd-regulated dynamin-Rab5 pathway is necessary for both (42).

Recently, it was reported that some of the tracheal tubule migration defects generated by dVHL RNA interference could be attributed to upregulation of Sima, the Drosophila homolog of HIF-α, which in turn upregulates btl transcription (41). Hypoxia-induced, Sima-dependent btl transcription has also been demonstrated in terminal branching in the larval trachea (8). It is possible that the endocytic and the HIF-dependent functions of VHL are not mutually exclusive. On the other hand, it has been shown that the stereotypic tracheal branch migration pattern in the embryo (as opposed to the terminal branching in larvae) is normally not dependent on sima function (48). Also, overexpression of the wild-type btl transgene in the embryonic trachea using the cognate btl promoter, although exhibiting ectopic elevation of the btl transcript level, could not lead to embryonic tracheal defects (48). This indicates that posttranscriptional regulation is the major mechanism for modulating active Btl level and thus Btl signaling. Indeed, we show here that the exogenously expressed Btl::GFP protein is under stringent control at the protein level. Also, we show that sima can only modestly suppress the dVHL¹ branch migration phenotype. It is interesting to point out that Mortimer et al. (41) reported a robust rescue of phenotypes in the btl promoter-driven dVHL RNA duplex by sima heterozygotes. This may be because the exogenous btl promoter, presumably positively regulated by sima, is itself downregulated in the sima mutant, thereby diminishing the dVHL duplex expression, resulting in apparent rescue.

It should be noted that primary and secondary tracheal tubule development during embryogenesis is a stereotypic, genetically programmed process (40). Tissue microenvironmental factors such as hypoxia do not play a significant role in this process. On the other hand, in late embryonic and larval stages, when the trachea system extends to connect with internal organs, localized hypoxia in the target tissue is critical in inducing FGF/bnl expression and thus promoting terminal branching (28, 67). More recently, HIF-α/Sima has been shown to also play an important role in the terminal tracheal tubule cells in sensing hypoxia and inducing FGFR/btl expression (8). This mechanism, in the absence of dVHL mutation, can contribute to tracheal branching probably because it is sensitized by the overproduced Bnl ligand in hypoxic conditions (8). It is conceivable that dVHL loss of function during embryogenesis can result in similar upregulation of the btl gene transcription. However, our results clearly indicate that in nonhypoxic development of trachea in the embryonic stage, the additional function of dVHL in regulated protein internalization plays the major role in tracheal morphogenesis.

Most interestingly, we demonstrate that the ectopic branching of the tubule epithelial cells and the malformation of lumen phenotypes are reproducible in the kidney tubules of mice with conditional Vhl knockout and in organoid culture using primary tubule cells. This is highly significant, since dilated tubules (minicysts) have been documented as preceding renal cell carcinoma (11, 58). Thus, our findings provide a plausible mechanistic explanation, involving increased cell motility and disruption of tubule epithelium, for the etiology of VHL mutant kidney cancer. In addition, our recent study also implicated dVHL in the morphogenesis of organ-associated epithelium, the follicle cells in the ovary (15). This function is mediated by another HIF-independent activity of dVHL that stabilizes microtubule bundles. Future studies should exploit further the Drosophila genetic system for elucidating how various VHL functions and a myriad of disease-related VHL mutations may differentially affect the pathophysiological roles of this interesting tumor suppressor gene.