Research ArticleDevelopmental Biology

The scaffolding protein Cnk binds to the receptor tyrosine kinase Alk to promote visceral founder cell specification in Drosophila

See allHide authors and affiliations

Sci. Signal.  24 Oct 2017:
Vol. 10, Issue 502, eaan0804
DOI: 10.1126/scisignal.aan0804

Identifying downstream components of Alk

Aberrant activation of the receptor tyrosine kinase ALK contributes to the progression of several human cancers. In developing Drosophila larvae, Alk and its ligand Jeb are critical for the specification of visceral muscle founder cells. Wolfstetter et al. found that the scaffolding protein Cnk interacted with Alk and was required for Alk signaling in the visceral mesoderm of this model organism. In addition, the Cnk binding partner Ave promoted the function of Cnk in this pathway. Mutants deficient in Cnk or Ave or overexpressing the Alk binding region of Cnk (which would be expected to disrupt the interaction between Alk and Cnk) lacked founder cells and failed to develop functional gut musculature. Together, these results identify components downstream of Alk. Future studies are necessary to determine whether human homologs of Cnk also participate in ALK signaling.


In Drosophila melanogaster, the receptor tyrosine kinase (RTK) anaplastic lymphoma kinase (Alk) and its ligand jelly belly (Jeb) are required to specify muscle founder cells in the visceral mesoderm. We identified a critical role for the scaffolding protein Cnk (connector enhancer of kinase suppressor of Ras) in this signaling pathway. Embryos that ectopically expressed the minimal Alk interaction region in the carboxyl terminus of Cnk or lacked maternal and zygotic cnk did not generate visceral founder cells or a functional gut musculature, phenotypes that resemble those of jeb and Alk mutants. Deletion of the entire Alk-interacting region in the cnk locus affected the Alk signaling pathway in the visceral mesoderm and not other RTK signaling pathways in other tissues. In addition, the Cnk-interacting protein Aveugle (Ave) was critical for Alk signaling in the developing visceral mesoderm. Alk signaling stimulates the MAPK/ERK pathway, but the scaffolding protein Ksr, which facilitates activation of this pathway, was not required to promote visceral founder cell specification. Thus, Cnk and Ave represent critical molecules downstream of Alk, and their loss genocopies the lack of visceral founder cell specification of Alk and jeb mutants, indicating their essential roles in Alk signaling.


Receptor tyrosine kinase (RTK) signaling plays an essential role in development by transducing external signals into the nucleus and other cellular compartments, thereby altering gene expression and promoting intracellular responses. The hallmarks of RTK signaling are conserved among eukaryotic organisms and involve ligand-dependent activation of a transmembrane receptor protein tyrosine kinase and the recruitment of canonical intracellular signaling modules and cascades, such as the mitogen-activated protein kinase/extracellular signal–regulated kinase (MAPK/ERK) pathway. Alk activation stimulates this pathway through the guanosine triphosphatase Ras and the serine-threonine kinases Raf [a MAPK kinase kinase (MAPKKK)], MEK (a MAPK kinase), and MAPK/ERK (13). Other factors that contribute to or modulate the activity of this pathway have been identified, such as the kinase suppressor of Ras (Ksr), which was identified by mutagenesis screens in Ras-sensitized genetic backgrounds in Drosophila melanogaster and Caenorhabditis elegans (46). Because of inconsistent findings regarding the catalytic activity of its kinase domain, the role of Ksr has remained controversial. Different models have proposed distinct roles for Ksr as an activator of Raf in parallel to or downstream of Ras or as a scaffolding protein for the assembly of Raf-MEK protein complexes (48). There is no evidence for a direct interaction between Ksr and Ras (9), but dimerization between Ksr and Raf can stimulate Raf activity in a manner that is independent of the kinase activity of Ksr (10), suggesting that Ksr may act as a scaffold in the context of RTK signaling.

A genetic modifier screen using ectopic expression of a dominant-negative, chimeric version of Ksr (11) in the fly eye led to the identification of another critical factor for Ras-ERK signaling named connector enhancer of kinase suppressor of Ras (Cnk) (12). The cnk locus encodes a large protein of 1557 amino acids containing an N-terminal sterile α motif (SAM), followed by a conserved region in Cnk (CRIC), a PDZ domain, proline-rich motifs, and a pleckstrin homology (PH) domain. The protein structure suggests that Cnk acts as a multidomain protein scaffold. Like Ksr, Cnk functions downstream of various RTK signaling events including epidermal growth factor receptor (EGFR)–mediated patterning of wing disc territories and fibroblast GFR (FGFR)– and EGFR-dependent air sac development in the dorsal thorax (13, 14). Ectopic expression of the Cnk N-terminal region enhances the effects of activated RasV12 independently of MAPK/ERK activation in the Drosophila eye. The C-terminal region contains a Raf inhibitory region (RIR) that binds to and represses Raf, which is released upon phosphorylation of Cnk by the Src family kinase Src42A (12, 1517). Thus, Cnk functions as a molecular scaffold to support Ksr-mediated Raf activation and to recruit and integrate additional signaling components such as Src42A.

During embryonic development in D. melanogaster, the visceral mesoderm (VM) gives rise to a lattice of midgut muscles that ensheaths the larval midgut. The VM consists of naïve myoblasts that become specified as either founder cells (FCs) or fusion competent myoblasts (FCMs). Subsequently, the FCs fuse one-to-one with FCMs and eventually form the binucleate visceral myotubes (1822). Specification of VM cells requires the Drosophila ortholog of the receptor anaplastic lymphoma kinase (ALK), initially identified as part of a chimeric protein created by the 2;5 (p23:q35) translocation in human anaplastic large cell lymphoma cell lines (23). Drosophila Alk is expressed in the segmental clusters of the embryo that segregate from the dorsal trunk mesoderm to form the VM (24). Alk protein can be detected at the membrane of all VM cells, but only the distal arch within each cluster comes into direct contact with a secreted, small low-density lipoprotein domain ligand named jelly belly (Jeb) (2427). Binding of Jeb to the extracellular part of Alk activates a downstream signaling cascade that results in ERK phosphorylation and triggers expression of an FC-specific subset of genes including Hand, optomotor-blind-related-gene-1 (org-1), and kin of irre (kirre; also referred to as dumbfounded or duf) (2631). Jeb-Alk signaling is crucial for visceral myoblasts to commit to the FC fate. In the absence of either ligand or receptor, neither ERK phosphorylation nor the expression of FC-specific marker genes in the VM occurs. Moreover, visceral cells fail to undergo myoblast fusion, and the VM subsequently disintegrates in jeb and Alk mutant embryos (2628, 32).

Here, we identified the multidomain scaffolding protein Cnk as a potential Alk binding partner and essential component in the Alk signaling pathway. Cnk bound to the intracellular part of Alk by its C-terminal region. Loss of cnk function or expression of dominant-negative cnk constructs in Drosophila interfered with Alk signaling in multiple developmental contexts. Moreover, germline clone–derived embryos lacking maternal and zygotic Cnk failed to specify visceral FCs and did not develop a functional midgut. In agreement with its proposed function, epistasis experiments revealed that Cnk operates between Ras and Raf in the Alk signaling pathway. Further targeted deletion of a minimal Alk interaction region (AIR) in Cnk resulted in a specific decrease of Jeb-Alk–induced ERK phosphorylation within the visceral FC row. Deletion of the larger Alk interacting region blocked specification of visceral FCs in response to Alk activation. Although the SAM domain containing Cnk binding partner Aveugle (Ave) was essential for Alk signaling, we found that Ksr is not essentially required to drive Alk signaling in the developing VM. Thus, Cnk and its binding partner Ave serve as critical components for Alk signaling in Drosophila.


Cnk binds to Alk

To identify components of the Alk signaling pathway in Drosophila, we performed a yeast two-hybrid (Y2H) screen using the whole intracellular domain (corresponding to amino acids Tyr1128-Cys1701) of the Drosophila Alk protein (subsequently referred to as AlkICD) as bait. Thirty-six prey clones encoding portions of Cnk were identified as AlkICD interactors (Fig. 1A). All identified Cnk prey clones shared a minimal region encoding 42 amino acids (Ala1384-Ser1425). This region (hereafter referred to as the minimal Cnk-Alk interaction region or CnkAIR) was located in the functionally unannotated C-terminal region of Cnk and was sufficient to bind to Alk in the Y2H assay (Fig. 1B). Thus, the CnkAIR contained the minimal requirements to mediate the interaction of Cnk with the AlkICD.

Fig. 1 Alk and Cnk interact through an AIR.

(A) Schematic of the cnk locus indicating the Y2H AlkICD-interacting Cnk prey constructs. PDZ, postsynaptic density protein (PSD95), Drosophila discs-large tumor suppressor (Dlg1), and zonula occludens–1 protein (zo-1); RIM, Raf-interacting motif; IS, inhibitory sequence; pYELI, phosphorylation site for Src42A. Untranslated regions are noted in gray. The minimal AIR (Ala1384-Ser1425) is shown in red. (B) Yeast growth on double-selective transformation-mating control (−Leu, −Trp) and triple-selective (−Leu, −Trp, −His) media plates. Each streak corresponds to diploid yeast colonies coexpressing the indicated Gal4_AD Cnk fusion proteins and either Gal4_BD-AlkICD or Gal4_BD (empty vector control). (C) Late-stage female Drosophila pupae expressing pan-neuronal UAS-jeb under the control of elavC155-Gal4 in control, cnkk16314/+, cnk63F/+, and Df(2R)rl10a/+ genetic backgrounds. Pupae carrying no UAS transgene were used as a negative control. Quantification of the Cnk-dependent modification of body size (right). Outliers from the 2.5 to 97.5 percentile (whiskers) appear as circles, ***P < 0,001; n.s., not significant. One-way analysis of variance (ANOVA) and Tukey’s multiple comparisons test were used. n = 559 pupae, >100 pupae for each genotype; values were pooled from four experimental replicates. (D to I) In situ hybridization showing cnk expression in cnk63F(m/z)/BSC161 embryos (D) and in embryos expressing cnk under control of the en2.4-Gal4 driver (E). Maternal cnk transcripts in a stage 2 w1118 embryo (F), ubiquitous zygotic cnk expression in cnk63F(m/z)/+ embryos at stages 5 (G) and 10 (H). Cnk expression in the brain (br), ventral nerve chord (vnc), and peripheral nervous system (arrowheads) at the end of embryogenesis (I). n ≥ 100 embryos for each genotype. (J and K) Cnk and Alk protein localization in stage 10/11 fTRG1248 (cnk.SGFP) control (J) and Alk1/Exel7144; cnk.SGFP transheterozygous (K) embryos. Dotted yellow lines indicate VM clusters; n ≥ 100 embryos for each genotype. Scale bars, 500 bp (A), 500 μm (C), 50 μm (D to I), and 10 μm (J and K).

Cnk modulates Jeb-Alk signaling in the central nervous system

To determine whether the interaction between Alk and Cnk was relevant in vivo, we tested the ability of Cnk to modify the previously described pupal size reduction phenotype caused by increased Alk signaling (33). The Drosophila Alk ligand Jeb was expressed using the pan-neuronal elavC155-Gal4 driver, resulting in a pupal size reduction of about 20 to 25% when compared to elavC155-Gal4/+ controls (Fig. 1C). Introducing a single copy of the loss-of-function alleles cnkk16314 or cnk63F (fig. S1A) in this Jeb overexpression background caused a significant increase in pupal size, indicating that loss of Cnk counteracts the effect of enhanced Jeb-Alk signaling (Fig. 1C). As a control, we reduced the dose of the Drosophila ERK homolog rolled, using the chromosomal deletion Df(2R)rl10a, which resulted in a comparable increase in pupal size as observed with reduction of Cnk. Together, these results support a role for Cnk as a downstream component of Jeb-Alk signaling.

Cnk localization in the VM does not depend on Alk

A prominent function for Alk signaling in Drosophila is the specification of visceral muscle FCs in the developing embryo (2628). We investigated whether Cnk is present in the embryonic VM during Jeb-Alk–dependent FC specification by first performing in situ hybridizations using cnk-specific antisense probes. In situ probe specificity was tested on cnk63F(m/z)/BSC161 germline clone–derived embryos that lacked cnk transcripts (Fig. 1D) and on embryos exhibiting en2.4-Gal4–driven cnk misexpression with the P{GSV6}GS9388 gene trap insertion (Fig. 1E). Our analysis of endogenous cnk expression revealed a strong maternal component, suggesting that all cells in the embryo are supplied with Cnk (Fig. 1F). Zygotic cnk transcription in progeny of cnk63F germline clone–producing females crossed to w1118 control males was detectable in all cells at the blastoderm stage (Fig. 1G) and remained ubiquitously expressed during germ band elongation (Fig. 1H). After germ band retraction, ubiquitous expression had declined, and cnk was specifically expressed in the brain, the ventral nerve cord, and the peripheral nerves (arrowheads) at the end of embryogenesis (Fig. 1I). In addition to in situ hybridization, we used the multi-tagged FlyFos TransgeneOme (fTRG) library line fTRG1248 (subsequently referred to as cnk.SGFP), which carries an extra copy of the cnk locus encoding a C-terminally green fluorescent protein (GFP)–tagged variant of Cnk expressed under the control of its endogenous regulatory elements (34). Notably, cnk.SGFP rescued the lethality caused by cnk loss-of-function mutations (fig. S2, A to C, and table S1), indicating that Cnk.SGFP can functionally substitute for the wild-type protein. At stage 10/11, Cnk.SGFP was detected in all cells of the ectoderm and the underlying mesoderm including Alk-positive cells (Fig. 1J). The majority of the protein was localized in close proximity to the plasma membrane (Fig. 1J), and this localization of Cnk.SGFP was maintained in Alk1/Df(2R)Exel7144 (Exel7144, an Alk deficiency) embryos that express one copy of a truncated Alk protein, suggesting that Cnk localization does not depend on Alk (Fig. 1K).

Ectopic expression of CnkAIR interferes with Alk signaling

To further examine a possible function for the interaction between Alk and Cnk, we generated UAS constructs encoding a tandem array of five copies of CnkAIR, C-terminally tagged with 3xHA (UAS-5xcnkAIR.3xHA). Because Cnk was localized close to the plasma membrane, we used the N-terminal myristoylation sequence of Src42A to generate a membrane-associated variant (UAS-Myr::5xcnkAIR.3xHA). We first studied the effects of 5xCnkAIR on ectopic expression of Alk in the developing eye with the sevEP-Gal4 driver. Expression of 5xCnkAIR (Fig. 2, A and B) did not obviously affect eye morphology, whereas expression of Alk alone (Fig. 2C) led to severe defects in ommatidia formation in the anterior part of the eye. The combined expression of Alk and 5xCnkAIR substantially restored a wild-type eye morphology (Fig. 2, D and E). In agreement with our assumption that the membrane localization of Cnk might influence its function, the Myr::5xcnkAIR.3xHA construct suppressed the effects of ectopic Alk signaling (Fig. 2E) more effectively than 5xcnkAIR.3xHA (Fig. 2D). To assess whether 5xCnkAIR interfered with endogenous Alk signaling during VM development in the embryo, we expressed 5xCnkAIR with the mesodermal P{GAL4-twi.2xPE} driver (hereafter referred to as 2xPE-Gal4) and used rP298-lacZ or HandC-GFP as FC reporters (35, 36). Using antibody staining against Alk or Fasciclin III (FasIII) to reveal VM and muscle morphology (Fig. 2F), we found that ectopically expressed 5xCnkAIR.3xHA exhibited a cytoplasmic expression pattern (Fig. 2G), whereas the Myr::5xCnkAIR.3xHA construct was membrane-localized (Fig. 2H). Expression of the rP298-lacZ FC marker was severely reduced in the VM of 2xPE>5xcnkAIR.3xHA and 2xPE>Myr::5xcnkAIR.3xHA embryos when compared with controls (Fig. 2, F to H). Moreover, the chambered midgut surrounded by HandC-GFP– and FasIII-positive visceral myotubes (Fig. 2I) was not formed at the end of embryogenesis (Fig. 2, J and K) upon ectopic expression of either 5xCnkAIR.3xHA or Myr::5xCnkAIR.3xHA. rP298-lacZ expression in somatic FCs (Fig. 2, G and H, arrows) and the presence of body wall muscles (Fig. 2, J and K, asterisks) in late-stage 2xPE>5xcnkAIR embryos further indicated that ectopic expression of 5xCnkAIR affected visceral FC formation but not somatic muscle development, which suggests a specific effect on Alk signaling in the embryonic VM.

Fig. 2 Ectopic expression of CnkAIR inhibits Alk signaling.

(A to E) Heads of adult female flies expressing hemagglutinin (HA)–tagged 5xCnkAIR (cnkAIR.HA) (A), HA-tagged 5xMyr-CnkAIR (Myr::cnkAIR.HA) (B), UAS-Alk alone (C), and UAS-Alk in combination with either HA-tagged 5xCnkAIR (D) or HA-tagged 5xMyr-CnkAIR (E) under the control of the sev-Gal4 driver. n ≥ 400 flies for each genotype. (F to H) Alk antibody staining (blue) in twist.2xPE (2xPE) driver control (F), 2xPE>cnkAIR.HA (G), and 2xPE>Myr::cnkAIR.HA (H) embryos at stage 11/12. rP298-LacZ [β-galactosidase (β-Gal), red] reporter gene expression in visceral FCs and somatic FCs (arrows in G and H). Construct expression is revealed by HA tag antibody staining (HA, green). (I to K) Stage 16 embryos of the indicated genotype stained for HandC-GFP reporter gene expression (GFP, green), FasIII (red), and HA (blue). Asterisks in (J) and (K) indicate somatic muscles labeled by HA staining (blue). n ≥ 200 embryos for each genotype and staining. Scale bars, 50 μm.

Cnk is required for VM development in the Drosophila embryo

The interaction between Cnk and the AlkICD in the Y2H assay and the effect of CnkAIR ectopic expression on Alk signaling led us to ask whether endogenous Cnk function might be critical for Alk signaling in the Drosophila embryo. As Alk, jeb, and cnk are located on the right arm of the second chromosome, we confirmed that the cnk alleles used in this study complemented the loss-of-function mutations Alk1, Alk10, jebweli, and jebk05644 and the jeb deficiency Df(2R)BSC199 (table S2) (37). Further complementation tests using the cnk deficiency Df(2R)BSC161 revealed that transheterozygous cnk mutant animals appeared to be viable until late larval or pupal stages, contradicting our assumption for a critical role of cnk in visceral FC specification. Moreover, analysis of VM development in control siblings (Fig. 3, A and B) and zygotic cnk mutant embryos (Fig. 3, C and D) using rP298-lacZ and HandC-GFP as FC markers in combination with antibody staining against Alk or FasIII did not reveal a loss of FCs or the absence of visceral muscles. Given the maternal cnk component revealed by our in situ analysis, we hypothesized that maternally derived Cnk masked a potential visceral phenotype in zygotic cnk mutants, prompting us to examine germline clone–derived embryos lacking maternal and zygotic (m/z) cnk. In agreement with an important role for Cnk in Jeb-Alk signaling, we observed a VM phenotype characterized by complete loss of FCs at stage 11/12 (Fig. 3E) and the absence of differentiated midgut muscles at the end of embryogenesis (Fig. 3F). This fully penetrant phenotype was observed in maternal and zygotic (m/z) embryos of all cnk alleles analyzed (fig. S1, B to F) and was reminiscent of the visceral defects characteristic of jeb (Fig. 3, G and H) and Alk (Fig. 3, I and J) mutant embryos (2628). Finally, Cnk.SGFP restored visceral FC formation and the development of midgut muscles in maternal and zygotic (m/z) cnk embryos (Fig. 3, K and L), providing further evidence that cnk specifically functions in the Jeb-Alk signaling pathway.

Fig. 3 Loss of maternal and zygotic cnk genocopies Alk and jeb mutants.

(A, C, E, G, I, and K) VM specification in stage 11/12 Drosophila embryos with the indicated genotypes revealed by rP298-lacZ (β-Gal, red) and Alk (green) expression. In (I), embryos are stained for FasIII (green) instead of Alk. (B, D, F, H, J, and L) Stage 16 embryos carrying the HandC-GFP reporter stained for FasIII and GFP to reveal the presence of visceral muscles. GFP expression in garland cells (gc) confirms the presence of the HandC-GFP reporter in (F), (H), and (J). n ≥ 600 embryos for (A) to (J); n = 50 embryos for (K) and (L). Scale bars, 50 μm.

The semang (sag) complementation group consists of previously unknown cnk mutations

With a modifier screen, Zhang et al. (38) have identified suppressors of the Src42ASu(Raf)1 allele that suppressed the lethality of the hypomorphic RafC110 [also referred to as polehole (phl)C110 and Raf1] mutation. Germline clone–derived embryos of one of the identified suppressor mutations named semang (sag) have a midgut morphology reminiscent of the phenotypes observed in jeb, Alk, and cnk (m/z) mutant embryos (Fig. 3 and fig. S1, E and F). We therefore analyzed the sag mutations with respect to a possible role in VM formation. sag maps to 2R(54), a region that also harbors the cnk locus, and, like cnk, sag is involved in the cell autonomous specification of certain photoreceptor cells in the Drosophila eye, where it functions in the Ras-Raf-ERK pathway downstream of EGFR (39). Complementation tests revealed that the semang alleles sag13L and sag32–3 failed to complement either the loss-of-function mutation cnkk16314 or Df(2R)BSC161, which deletes the entire cnk locus (table S2 and fig. S1, E and F). In addition, the lethality of sag13L/sag32–3 animals was rescued by cnk.SGFP (fig. S2C). Sequencing of the cnk locus in both semang alleles confirmed them to be cnk mutants at the molecular level. In sag13L mutants, we detected a point mutation (2R:C17415781T) that introduces a stop codon in place of Gln1156 of the Cnk protein, and in sag32–3 mutants, a nucleotide exchange (2R:T17415826A) led to a premature stop in place of Lys1141. Notably, both stops were located between the RIR and the Src42A binding motif (pYELI) that had been suggested to relieve the inhibitory effect of Cnk on Raf-dependent MEK phosphorylation (fig. S1) (16, 17). Therefore, the sag mutations represent previously unrecognized alleles of cnk and are subsequently referred to as cnksag13L and cnksag32–3.

Cnk functions between Ras and Raf in the MAPK/ERK pathway

Given their possible role as active suppressors of Raf downstream of Alk, we analyzed germline clone–derived cnk (m/z) mutant embryos in the background of activated signaling pathway components. To do this, we used bap3-Gal4–driven expression of the Alk ligand Jeb (UAS-jeb), an activated variant of the Drosophila Ras oncogene at 85D (UAS-Ras85DV12) and a constitutively active form of the Drosophila Raf kinase polehole (UAS-Raf.gof) in the VM. As hypothesized and in agreement with previous studies (27, 29, 40, 41), expression of these factors in the VM resulted in an increase in HandC-GFP expression and ERK phosphorylation [phosphorylated ERK (pERK)] within the VM, suggesting that most Alk-positive visceral myoblasts were converted into FCs (Fig. 4, A to D). Maternal and zygotic loss of Cnk function in cnksag13L(m/z)/BSC161 embryos was sufficient to abolish both ERK phosphorylation and HandC-GFP expression in the VM (Fig. 4E). In agreement with a role for Cnk downstream of activated Alk, cnksag alleles suppressed the overexpression effects of Jeb and RasV12 in HandC-GFP; cnksag13L(m/z); bap3>jeb, and HandC-GFP; cnksag13L(m/z); bap3>RasV12 embryos (Fig. 4, F and G), but not that of activated Raf in HandC-GFP; cnksag13L(m/z); bap3>Raf.gof embryos (measured as ERK phosphorylation or HandC-GFP expression; Fig. 4H). These data suggest that Cnk functions downstream of Jeb, Alk, and Ras but upstream of Raf in the Alk signaling pathway.

Fig. 4 Cnk functions between Ras and Raf in the Ras-ERK pathway.

Stage 11/12 embryos were stained for pERK (blue), HandC-GFP reporter gene expression (GFP, green), and Alk (red). Dorsal views are shown. (A) bap3-Gal4 (bap>) driver control. Arrow marks pERK- and GFP-positive visceral FCs. (B to D) pERK staining and HandC-GFP reporter expression in HandC-GFP, bap3>UAS-jeb (B), HandC-GFP bap3>UAS-Ras85DV12 (C), and HandC-GFP bap3>UAS-Raf.gof (D) embryos. (E) Maternal and zygotic HandC-GFP; cnksag13L(m/z)/BSC161 mutant. (F to H) pERK and GFP staining in cnksag13L(m/z)/BSC161 embryos that carry the HandC-GFP reporter and express either UAS-jeb (F), UAS-Ras85DV12 (G), or UAS-Raf.gof (H) under the control of the bap3-Gal4 driver. n ≥ 40 embryos for each genotype. Scale bar, 50 μm.

Binding of Cnk to Alk is required for robust activation of ERK in visceral FCs

The interaction between Alk and the C-terminal region of Cnk in the Y2H analysis defined a minimal AIR of 42 amino acids. Ectopic expression of CnkAIR tandem repeats resulted in the specific loss of visceral FC identity, suggesting that this interaction is important for Alk signaling. To test this hypothesis, we used clustered regularly interspaced short palindromic repeats/CRISPR-associated 9 (CRISPR/Cas9) (42) in combination with a cnkΔAIR donor construct to enforce homologous recombination with the cnk locus. This approach resulted in cnkΔAIR mutants specifically lacking the 126 base pairs (bp) that encode for the minimal Cnk AIR (Fig. 5A). cnkΔAIR mutants were viable and could be maintained as a fertile stock, suggesting that the minimal AIR is not a critical requirement for Alk signaling in the VM. In agreement, analysis of the FC markers rP298-lacZ or HandC-GFP and Alk or FasIII immunofluorescence did not show a loss of visceral FCs in cnkΔAIR mutants, which were similar to control embryos (Fig. 5, B to E). We also used antibody staining against pERK because activation of Alk by its ligand Jeb results in Ras-Raf-ERK pathway activation (24, 26, 27). In contrast to the Alk signaling output observed with the rP298-lacZ and HandC-GFP reporters, pERK signals were considerably decreased in the VM of cnkΔAIR mutants when compared to wild-type controls (Fig. 5, F and G). Similar to AlkKO mutants (Fig. 5H), ERK activation was only affected in the VM of cnkΔAIR mutant embryos (arrows in Fig. 5, F to H), whereas other tissues such as the tracheal pits (arrowheads in Fig. 5, F and H) still exhibited robust pERK staining (quantified in Fig. 5I).

Fig. 5 CnkAIR is required for robust activation of ERK in visceral FCs.

(A) Schematic of the cnk locus indicating deletions in cnkΔAIR and cnkΔY2H and the cnkCC9-110B allele. cnkΔAIR flies harbor a 126-bp in-frame deletion in the endogenous cnk locus, which removes the region encoding the 42–amino acid AIR (Ala1384-Ser1425) of the Cnk protein. cnkΔY2H flies harbor a 357–amino acid deletion (His1182-Asn1538). (B to C) Stage 11/12 balanced siblings (B) and cnkΔAIR(m/z)/BSC161 (C) embryos expressing the rP298-LacZ FC reporter were stained for β-Gal (red) and Alk (green). (D and E) Stage 16 embryos stained for HandC-GFP reporter gene expression (green) and FasIII (red). β-Gal (blue) staining reveals balancer-associated LacZ expression in a sibling embryo (D). n ≥ 100 embryos for (B) to (E). (F to H) pERK (black) in visceral FCs (arrows) and tracheal pits (arrowheads) of ventrally orientated stage 11 control (F), cnkΔAIR(m/z) (G), and AlkKO (H) embryos. The bottom panel in (F), (G), and (H) depicts a [rainbow RGB color lookup table (LUT)] heat map representation of close-ups from the embryo in the upper half. (I) Quantification of (background-)corrected total fluorescence intensities (CTF) in the VM relative to pERK CTFs of the adjacent tracheal pits. One-way ANOVA with Dunett’s multiple comparisons test was used to reveal statistical significance (***P < 0.001; n = 40 measurements from three different staining replicates for each genotype). (J) VM development in a stage 11/12 cnkΔY2H(m/z)/BSC161 embryo, expressing the rP298-LacZ FC reporter stained for β-Gal (red) and Alk (green). (K) Stage 16 HandC-GFP and cnkΔY2H(m/z)/BSC161 embryo stained for GFP (green) and FasIII (red). n ≥ 50 embryos for (J) and (K). (L and M) pERK (black) in visceral FCs (arrows) and tracheal pits (arrowheads) of ventrally orientated stage 11 cnkΔY2H(m/z) (L) and cnk63F(m/z)/BSC161 (M) embryos. n ≥ 30 embryos. Bottom panels in (L) and (M) depict a (rainbow RGB color LUT) heat map representation of close-ups from the embryo in the top panels. Scale bars, 50 μm.

Thus, the VM-specific reduction of pERK indicated a role for the Cnk AIR in Alk-dependent ERK activation. However, the observation that visceral FC specification proceeds in the context of reduced ERK activation in the VM of cnkΔAIR mutant embryos suggested the presence of (an) additional or extended binding interface(s) between Alk and Cnk or an additional indirect interaction with another factor. A Cnk fragment comprising all Alk interacting regions from our initial Y2H screen (corresponding to His1182-Asn1538 of Cnk) but lacking the CnkAIR (CnkY2H/ΔAIR) still interacted with AlkICD in our Y2H analysis (fig. S3), suggesting that additional residues within Cnk mediate the Alk-Cnk interaction. Thus, we generated a larger deletion in the cnk locus by CRISPR/Cas9 that removed the entire region encoding for His1182-Asn1538 (referred to as cnkΔY2H). Although cnkΔY2H flies were viable, they exhibited a rough eye phenotype (fig. S4, A to C) and could not be maintained as homozygous stock. Furthermore, crossing cnkΔY2H over the cnk63F null allele resulted in lethality during hatching, suggesting that the deletion of His1182-Asn1538 created a hypomorphic cnk allele. We also obtained a CRISPR/Cas9-generated cnk mutant exhibiting a truncation of Cnk at Gly1420 (Fig. 5A). This cnkCC9-110B mutant also exhibited a rough eye phenotype (fig. S4D) but was viable and fertile and did not display any defects in VM development, suggesting that the C-terminal region of Cnk after the 42–amino acid AIR is not essential for Alk signaling. In contrast, embryos derived from homozygous cnkΔY2H flies or cnkΔY2H (m/z) germline clone–producing females revealed a loss of rP298-lacZ–positive visceral FCs and the later absence of HandC-GFP– and FasIII-positive visceral muscles (Fig. 5, J and K). pERK staining was undetectable in the VM of cnkΔY2H (m/z) embryos and also slightly reduced in the tracheal pits (Fig. 5L), suggesting that other signaling pathways might be weakly affected by the deletion. On the other hand, cnk63F (m/z) embryos lacked any detectable pERK signals except for early Torso signaling, which was reduced but not absent (fig. S5, A and B). In accordance, Torso-induced expression of the terminal gap genes tailless and huckebein and the formation of ectodermal and endodermal derivatives of the alimentary tract were mildly affected in cnk63F (m/z) mutants (fig. S5, C to H). Further morphological analysis of cnk63F(m/z)/BSC161 embryos revealed various phenotypes connected to restricted or impaired RTK signaling (fig. S5, G to X). Severe heart and dorsal muscle formation defects, the absence of even-skipped–positive pericardial cells, and impaired longitudinal visceral muscle migration and tracheal development in cnk63F (m/z) embryos indicated a critical function for Cnk in FGFR signaling (fig. S5, K, N, and Q). Notably, tracheal phenotypes of cnk63F (m/z) embryos were distinct from cnksag (m/z) germline clones, indicating that different domains of Cnk are required to mediate the signaling events involved in trachea formation (fig. S5T). Terminal muscle tendon differentiation was also affected in cnk63F (m/z) embryos, suggesting involvement of Cnk in EGFR signaling (fig. S5W). In contrast to the phenotypes observed in the cnk63F(m/z)/BSC16 null mutant, cnkΔY2H(m/z)/BSC16 embryos exhibited even-skipped–positive pericardial cells and only mild defects during heart and dorsal somatic muscle development as well as longitudinal visceral muscle migration (fig. S5, L, O, and R). In addition, the effects of the cnkΔY2H deletion on alimentary tract and tracheal development as well as on tendon cell differentiation were milder compared to the phenotypes of cnk63F(m/z)/BSC16 embryos (fig. S5, I, U, and X), suggesting a specific function for the deleted region in Alk signaling. Together, the hypomorphic nature of the cnkΔY2H allele and the alleviated EGFR-, FGFR-, and Torso-related phenotypes indicate that removal of the CnkY2H region allowed us to dissect an Alk-specific function of Cnk from its more general role in RTK signaling.

The small SAM domain protein Ave is essential for Cnk activation downstream of Alk

To identify potential Cnk-interacting partners, we used the Cnk full-length protein as bait in a Y2H screen. Several Cnk binding partners were identified (Fig. 6A), including the small, SAM domain–containing protein Ave (also referred to as Hyphen or HYP), which is important for MAPK/ERK signaling (43, 44). Notably, we did not observe interactions with Raf, Ksr, or Src42A, which had previously been identified as binding partners of truncated Cnk variants (12, 16, 17, 43). To functionally analyze the role of the Cnk-Ave interaction in more detail, we generated a series of sequence aberrations in the ave locus using CRISPR/Cas9 genome editing. Two classes of molecular lesions were obtained: (i) deletions removing the ATG start codon of ave and (ii) smaller sequence aberrations that introduced shifts in the open reading frame (Fig. 6B). Because the ave locus partially overlaps with one isoform of the minus-orientated Rpn6 gene, we confirmed that the generated ave alleles complemented the lethal, hypomorphic Rpn620F mutation (45) and the lethal P-element insertion Mi{ET1}Rpn6MB09493 (table S3). Homozygous and transheterozygous ave mutants survived until pupal stages, suggesting that possible embryonic phenotypes are masked by a maternal component, which had been reported for ave (44, 46). We therefore analyzed the visceral morphology of embryos lacking both maternal and zygotic Ave [aveCC9 (m/z)] using HandC-GFP as visceral FC marker and FasIII immunofluorescence to label the VM. In contrast to sibling embryos that had received one functional copy of ave and displayed wild-type VM morphology (Fig. 6, C to E), aveCC9 (m/z) embryos exhibited similar visceral phenotypes as those observed in cnk (m/z), jeb, or Alk mutants, such as loss of visceral FC identity (Fig. 6F), scattering of the VM after germ band retraction (Fig. 6G, arrowheads), and the absence of a functional gut musculature at late embryonic stages (Fig. 6H). Therefore, we concluded that Ave could indeed serve as critical activator of Cnk in the Jeb-Alk pathway.

Fig. 6 Ave, but not Ksr, is required for Cnk-mediated Alk signaling in the VM.

(A) Table summarizing the binding partners and numbers of interacting preys obtained by Y2H screening using full-length Cnk as bait. (B) Schematic representation of the ave locus. The first exon of the minus-orientated Rpn6 gene is shown below; CRISPR/Cas9-induced molecular lesions of the aveCC9-20A and aveCC9-36A null alleles are indicated. (C to K) Antibody staining against HandC-GFP reporter gene expression (green), FasIII (red), and balancer-associated β-Gal expression [blue in (C) to (E)] in stage 11/12 (C, F, and I), stage 13/14 (D, G, and J), and stage 16 (E, H, and K) embryos of the indicated genotypes. n ≥ 100 embryos for (C) to (H); n ≥ 40 embryos for (I) to (K). Scale bars, 50 μm.

Cnk function in the VM is independent of Ksr

To identify additional factors that are recruited by Ave-activated Cnk and might function in Jeb-Alk signaling, we investigated the role of Ksr. In agreement with previous findings, we did not observe obvious somatic clones of either ksrS-638 or ksrS-627 in germline clone–producing females (6), and only a small number of ksr (m/z) embryos could be obtained. Our analysis of ksrS-638 and ksrS-627 germline clone–derived ksr (m/z) embryos carrying the HandC-GFP reporter did not reveal any loss of visceral FCs (Fig. 6I). In contrast to cnk (m/z) and ave (m/z) embryos, ksr (m/z) mutants developed visceral muscles, although they exhibited milder phenotypes such as myotube stretching defects at stage 13/14 and an incomplete midgut constriction at the end of embryogenesis (Fig. 6, J and K). To our surprise, terminal defects were present but variable and not as severe as previously reported for homozygous hkb and tll embryos (40), suggesting that Torso signaling and the expression of its downstream effectors hkb and tll are not entirely Ksr-dependent. Notably, terminal defects as well as the visceral muscle phenotype were more severe in HandC-GFP; ksrS-638(m/z)/ksrS-627 embryos (fig. S6, A and B) when compared to HandC-GFP; ksrS-627(m/z)/ksrS-627. Sequencing analysis revealed that ksrS-627 is caused by a premature stop codon likely resulting in a truncated Ksr protein that lacks its entire kinase domain and the Raf-dimerization surface (fig. S6A) (10). Therefore, it could be possible that the point mutations Ala696→Val and Ala703→Thr in ksrS-638 (6) create a protein that dominantly interferes with Torso signaling. Although not essentially required for Alk signaling in the VM, Ksr could contribute to the robustness of ERK activation in a similar way to CnkAIR. When we analyzed ksrS-627 zygotic and maternal zygotic ksrS-627 (m/z) embryos with pERK antibody staining, we observed irregularities in the pERK staining pattern and a decrease in pERK signals in patches of the VM (fig. S6, C and D). These defects were stronger in ksrS-627 (m/z) embryos, but uneven signals within the pERK-positive tissues in addition to the morphological defects of ksr mutant embryos did not permit quantification. Together, our analysis of embryos lacking Ksr, which exhibit HandC-GFP–positive FCs and the formation of a midgut, suggests that although pERK signals are reduced, Ksr is not a critical component of Jeb-Alk signaling in the VM.


Here, we uncovered an essential function for the protein scaffold Cnk in Alk signaling. The identification of multiple Cnk preys as AlkICD interactors in our Y2H analysis revealed a region in Cnk that likely mediates this interaction and allowed definition of a minimal AIR that was sufficient to bind Alk. The importance of Cnk in Alk signaling was supported by the loss of FCs in the VM of cnk (m/z) mutants, which genocopied the embryonic Alk loss-of-function phenotype and the dominant-negative effect of ectopic CnkAIR expression on visceral FC specification. The tissue-specific decrease of ERK phosphorylation in the visceral FC row of cnkΔAIR mutants and the loss of visceral FCs upon deletion of the entire Alk interacting region identified by our Y2H approach further support a direct interaction between Cnk and Alk.

Various interactions between Cnk and membrane-associated factors (although not other RTKs) have been reported. In cultured mammalian cells, CNK1 promotes insulin signaling by binding to and localizing cytohesins at the plasma membrane (47), and binding of CNK1 to the transmembrane ligand EphrinB1 links fibronectin-mediated cell adhesion to EphrinB-associated JNK signaling (48). Moreover, mammalian CNK2 (also called MAGUIN), the mammalian CNK homolog most similar to Drosophila Cnk, binds to various members of the membrane-associated MAGUK family proteins (49) and Densin-180 (50). It will be interesting to determine whether the RTK binding capacities of Cnk are limited to Alk.

Cnk localizes close to the plasma membrane (Fig. 1J) (12, 51). Although the Alk-Cnk interaction appears to be important for ERK activation and visceral FC specification in the VM, Alk does not appear to be required for the subcellular localization of Cnk. Whether the Alk-Cnk interaction depends on Alk activity, potentially resulting in posttranslational modification of Cnk, will be interesting to pursue in further studies. Notably, Drosophila Cnk is tyrosine-phosphorylated upon coexpression with the activated form of the RTK Sevenless (SEVS11) in S2 cells (12, 17), and activation of the platelet-derived growth factor receptor induces tyrosine phosphorylation of mammalian CNK1, leading to changes in CNK1 subcellular localization (52).

Cnk is generally abundant in the Drosophila embryo where its function is required in various RTK signaling pathways (fig. S5). However, the CnkAIR appears to be critical only for Alk signaling. Our morphological analysis further reveals a minor contribution of Cnk to embryonic Torso signaling, supporting the finding that Torso signals are processed by three or more parallel branches (53). This finding also agrees with earlier analyses of embryonic Torso signaling in ave (m/z) mutants (44), which form terminally derived structures. Notably, the differences in tracheal phenotypes caused by the cnk63F null allele and the strong, Raf-repressing cnksag alleles (38), which have C-terminal nonsense mutations and abrogate Alk signaling in the VM, indicate that distinct Cnk domains are involved in specific RTK-signaling pathways. This notion is also supported by the observation that the strong, Heartless-related phenotypes exhibited by cnk (m/z) null mutants were barely apparent in cnkΔY2H (m/z) embryos, which, on the other hand, displayed a complete loss of Alk-driven FC specification in the VM.

Selectivity for a requirement of CNK by different RTKs has also been observed in mammalian cells because CNK2 appears to be required for nerve growth factor, but not EGF-induced ERK activation in PC12 cells (54). Therefore, Cnk contributes to multiple signaling events, but its importance to different RTKs varies, perhaps reflecting differential wiring of downstream signaling in different developmental processes (55), an aspect that will be interesting to explore in future studies.

Cnk has been described as a protein scaffold that facilitates Ras-Raf-MAPK signaling at the plasma membrane, allowing signal integration to enhance Raf and MAPK activation (9). The epistatic analysis presented here shows that Cnk is required downstream of activated Alk and RasV12 but upstream of activated Raf in the VM (Fig. 7), which agrees with previous studies in Drosophila (12, 56). Thus, activated Ras seems to require Cnk to transmit signaling to Raf in the VM.

Fig. 7 Model incorporating a role for Cnk and Ave in Alk signaling during visceral FC specification.

Alk activation at the membrane of a prospective visceral FC by the ligand Jeb, which is secreted from the neighboring somatic mesoderm, induces the Raf/MAPK/ERK signaling cascade, eventually leading to the transcriptional activation of downstream targets including kirre, org-1, and Hand. Cnk and Ave are core components of the Alk signaling pathway that are required downstream of the receptor and upstream of Raf to mediate visceral FC specification. Although not essential for Jeb-Alk signaling, Ksr appears to be required for full activation of ERK.

Ave directly binds to Cnk through an interface formed by their SAM domains (57). This interaction is thought to be necessary to recruit Ksr to a complex that in turn promotes Raf activation in the presence of activated Ras (43). Although we identified Ave as a critical component for Cnk function downstream of Alk, the single Ksr in Drosophila was not required for Alk-mediated FC specification (Fig. 7). Cnk was originally identified in a Ksr-dependent genetic screen in Drosophila (12), and its function has been proposed to mediate the association between Ksr and Raf (43), suggesting that Ksr should also play an important role in Alk signaling. However, the role of Ksr is unclear, with early reports suggesting an inhibitory function rather than an activating potential in RTK signaling (58, 59). Ksr requires the presence of additional factors such as 14-3-3 proteins or activated Ras (11, 60), and loss of Ksr-1 suppresses RasE13-induced but not wild-type signaling during C. elegans vulva formation, suggesting altered affinities of Ksr for different variants of Ras (5). Because our analyses revealed a function for ksr in driving robust ERK phosphorylation, it is plausible that, although nonessential, Ksr might enhance Alk signaling by integrating signals from activated Ras to Cnk-associated Raf.

The importance of ERK activation in RTK-mediated signaling in the Drosophila embryo is difficult to address directly because the rolled locus (encoding the only MAPK/ERK1-2 ortholog in Drosophila) is located close to the centromere and therefore not accessible for standard germline clone analysis. The removal of the 42–amino acid AIR from Cnk by genomic editing of the cnk locus suggests that ERK phosphorylation, even at decreased amounts, is sufficient to promote Alk-induced specification of visceral FCs in vivo. Whereas the AIR is sufficient to mediate the Alk-Cnk interaction in vitro, the ability of the cnkΔAIR mutant to support Alk-driven FC specification in the developing embryo highlights additional requirements. Our Y2H analysis supports a direct interaction between Alk and Cnk mediated by a more extensive binding interface within the Alk-Y2H binding region of Cnk. Accordingly, deletion of this region in the cnk locus (cnkΔY2H) blocks VM specification in vivo. However, we cannot exclude the possibility that the Y2H region of Cnk may form interactions with additional Alk binding proteins, which could mediate the Alk-Cnk interaction in an indirect manner.

In summary, we identified an interaction between Alk and Cnk mediated by an Alk binding region in Cnk. This region was specifically required for the activation of ERK and formation of FCs downstream of the activated Alk receptor in the Drosophila VM. Together with Ave, Cnk represents an important signaling module that is required for Alk-mediated signaling during embryogenesis. Cnk and Ave represent molecules identified downstream of Alk, whose loss genocopies the lack of visceral FC specification of Alk and jeb mutants. Further work should allow a better understanding of the importance of Cnk in Alk signaling and whether this is conserved in mammalian systems.


Drosophila husbandry and fly genetics

Standard Drosophila husbandry procedures were followed (61). Stocks were maintained on a potato-mash–based diet at room temperature. Crosses were performed at 25°C. LacZ or GFP balancer chromosomes [Bloomington Drosophila Stock Center (BDSC) at Indiana University] were used to distinguish the progenies of a cross. white1118 or balanced, heterozygous sibling embryos were used as wild-type controls. Image stacks of adult flies were acquired with a Zeiss Axio Zoom.V16 stereo zoom microscope equipped with a light-emitting diode ring light and an Axiocam 503 color camera and further processed using the EDF (enhanced depth of focus) module of the ZEN Blue edition software.

Fly stocks

Fly stocks were obtained from the BDSC (NIH P40OD018537), the Kyoto Stock Center [Drosophila Genetic Resource Center (DGRC)] at the Kyoto Institute of Technology, and the Vienna Drosophila Resource Center (VDRC) at the Campus Science Support Facilities GmbH (CSF). Other lines used were as follows: rP298-lacZ (35), which is an enhancer trap in the kin of irre/dumbfounded locus (62), HandC-GFP (36), bap3-Gal4 (63), UAS-jeb (29), UAS-Alk (24), the ethyl methanesulfonate alleles Alk1 and Alk10 as well as the Alk deficiency Df(2R)Alk-21 (32), and jebweli (28), which carries a stop codon instead of Gln74 in the Jeb coding region (2R:C12117841T). In the AlkKO allele, exon 8 of the endogenous Alk locus has been replaced with an attP landing site following the protocol of Baena-Lopez et al. (64). Alk exon 8 encodes for Asn1197-Cys1701 that corresponds to most of the AlkICD (Tyr1128-Cys1701). cnk14C (also referred to as cnkC14), cnk63F, and cnk116C were previously identified in a mosaic genetic screen (65). Sequencing of these alleles revealed nucleotide exchanges in the coding region of cnk, resulting in premature stops in place of Trp17 (2R:G17419418A in cnk63F), Gln253 (2R:C17418490T in cnk14C), and Trp806 (2R:G17416829A in cnk116C). The semang alleles sag32–3 and sag13L (38) were identified in this paper as previously unknown cnk alleles and are therefore referred to as cnksag32–3 and cnksag13L. cnkCC9-110B carries a 19-bp genomic deletion (2R:T17414990-A17414972). The insertion of four nucleotides (CTAA) at the deletion site introduces a premature stop in place of Gly1420. The aveCC9-20A allele carries a five-nucleotide deletion (2R:C14752633-A14752637) that induces a frameshift resulting in a premature protein truncation [predicted sequence, MGEETINSTQNKTRNYATEGSVPVDS* (underlined amino acids differ from the corresponding residues in wild-type Ave protein)]. In aveCC9-36B, 103 bp of the ave locus (2R:C14752530-A14752632) including the ATG start codon are deleted, and eight nucleotides (AAACTACG) have been inserted between the Cas9 cutting sites. ave108V (44) was used for complementation tests (table S3). The molecularly characterized allele ksrS-638 (6) and the previously uncharacterized allele ksrS-627 (66) were used to generate germline clone–derived embryos. Sequencing of ksrS-627 revealed a nucleotide exchange (3R:C5483151T) indicating a premature stop in place of Gln163 that results in a truncated protein lacking its kinase domain and Raf dimerization surface (fig. S6A).

Generation of somatic clones and germline mosaics

Germline clones were generated combining the Flippase (FLP) recombinase/FLP recombination target system (67) and the “dominant-female-sterile” technique as described by Chou et al. (68) and Chou and Perrimon (69).

Whole-mount in situ hybridization and fluorescent antibody staining

Embryo staining was carried out according to Müller (70). For staining of phosphorylated MAPK/ERK (pERK), we followed the protocol from Gabay et al. (71). Whole-mount in situ hybridization was done according to Lécuyer et al. (72), with modifications adapted from Pfeifer et al. (73). The following antibodies were used in the specified dilutions: guinea pig anti-Alk and rabbit anti-Alk (1:1000) (24), rabbit and guinea pig anti–β3-tubulin (1:2500) (74), chicken anti–β-galactosidase (1:200; Abcam, ab9361), rabbit anti–β-galactosidase (preabsorbed on fixed embryos, 1:1500; Cappel, 0855976), rabbit anti-GFP (1:500; Abcam, ab290), chicken anti-GFP (1:300; Abcam, ab13970), mouse anti–activated MAPK/diphosphorylated ERK1 and 2 (1:250; Sigma-Aldrich, #M8159), and mouse 16B12 anti-HA.11 (1:500; Covance, #MMS-101P). The monoclonal antibodies 7G10 anti-FasIII (1:50), 2A12 anti-GASP (gene analogous to small peritrophins) (1:5), 3C10 anti–even-skipped (1:50), 2B10 anti–cut protein product (1:100), DCAD2 anti–DE-cadherin (1:100), and CF.6G11 anti-βPS integrin (1:500) were obtained from the Developmental Studies Hybridoma Bank (DSHB). Alexa Fluor–, Cy–, biotin-SP–, and horseradish peroxidase–coupled secondary antibodies as well as animal sera were purchased from Jackson ImmunoResearch. For signal amplification, we used the VECTASTAIN Elite ABC kit (Vector Laboratories) in combination with the Tyramide Signal Amplification Plus Fluorescein or Cyanine 3 systems (PerkinElmer). Stained embryos were dehydrated in an ascending ethanol series before clearing and mounting in methyl salicylate. Samples were analyzed under a Zeiss Axio Imager.Z2 microscope, and images were acquired with a Zeiss LSM800 confocal microscope or an Axiocam 503 color camera using the ZEN Blue edition software.

DNA constructs

UAS-5xcnkAIM.3xHA, UAS-Myrsrc42A::5xcnkAIM, UAS-Myrsrc42A::5xcnkAIM.3xHA, and the cnkΔY2H donor DNA sequences were assembled by GenScript. UAS-5xcnkAIM encodes for five tandem repeats of the CnkAIR (CnkAIR; Ala1384-Ser1425). The myristoylation signal (Met1-Lys10) from Drosophila Src42A was used to force membrane localization of the respective constructs, whereas a C-terminal, triple HA (3× HA) tag was used to detect construct expression. The assembled sequence was cloned into the pUAST transformation vector (75) using Eco RI and Xba I restriction sites. To generate the cnkΔAIR gene editing donor construct, we polymerase chain reaction (PCR)–amplified ~1-kb homology arms from genomic DNA of w*; P{FRT(whs)}G13 flies using the following primer combinations: CCAGAGTCCCAGTAGCAAGTCGAGT and AATTAGTGGTCTGCTGCTGATGGTGATGG as well as AGCAGACCACTAATTTGTGCTCG in combination with TTAGGTCTTTGAATAAGTTGCGTGC. This introduced a 15-bp overlapping sequence (underlined in the primer sequences), which was further used as “internal primer” in an assembly PCR with the external (nonunderlined) primer pair. The obtained cnkΔAIR donor sequence was subcloned in pCRII (Invitrogen) and shuttled into pBluescript II KS(−) using Hind III and Xho I restriction sites. All DNA constructs were verified by Sanger sequencing (GATC Biotech).

Generation of cnk mutants using CRISPR/Cas9-facilitated homologous recombination

To induce double-strand breaks in the region of the Cnk AIR (CnkAIR), we used the single guide RNA (sgRNA) target sequences GCACAAATTAGTGGTCGAGG (cnkΔAIR) or GATGATGGTGGTGCTCTTGC and CATCACACATCTAGGTCTGA (cnkΔY2H). Target sequences were cloned into the pBFv-U6.2 or pU6-BbsI-chiRNA gRNA expression vectors (GEPC, respectively; Addgene). To achieve molecularly defined deletions, we forced homologous recombination by simultaneously injecting (BestGene Inc.) sgRNA(s) and the respective pBluescript gene editing donor construct into embryos carrying the P{FRT(whs)}G13 insertion and expressing Cas9 in their germ line (M{vas-Cas9}ZH-2A). Balanced stocks were established from single flies carrying the FRT insertion and a possible recombination event and further analyzed by PCR for the desired modification in the cnk locus.

Generation of ave mutants by CRISPR/Cas9-mediated genome editing

To induce CRISPR/Cas9-mediated lesions in the ave locus, we used the following sgRNA sequences: CGGTCGCGTAGTTTTCGTTC and AGCAACAAAACAAATAGTGA. pBFv-U6.2 expression vectors containing these sequences (GEPC) were injected into M{vas-Cas9}ZH-2A/+(or Y); P{FRT(whs)}G13/+ embryos by BestGene Inc. Balanced stocks were established from single flies carrying the FRT insertion and further screened by PCR for molecular lesions in the ave locus.

Databases and bioinformatics

Information about Drosophila genetics is available on the FlyBase Database of Drosophila Genes and Genomes ( (76). Genomic coordinates refer to the Dmel_Release_6 sequence assembly (77). The FancyGene v1.4 application (78) and the MyDomains image creator (79) were used to create scale models of gene loci and proteins. For sgRNA validation and CRISPR/Cas9 genome editing strategies, we used the flyCRISPR resource page ( Fluorescence intensity measurements were acquired with the Fiji distribution of ImageJ (80). Briefly, area, mean fluorescence, and integrated density values were acquired from regions of interest (ROIs) selected in confocal stacks of stage 11/12 embryos. The ROIs corresponded to a nonstained area (background), an arch of the VM (revealed by Alk staining in the AlkKO mutant), and the adjacent tracheal pit for each measurement. Three to four measurements were taken from each analyzed embryo. The (background-)corrected total fluorescence [CTF = integrated density ROI – (area ROI × mean fluorescence background)] of the VM arch relative to the adjacent tracheal pit was calculated to minimize staining-dependent or stage-dependent fluctuations. For statistical analysis, we used the GraphPad Prism 6 software.

Y2H experimental analysis

Y2H expressed sequence tag library screening was conducted by Hybrigenics. The Matchmaker Gold Yeast Two-Hybrid System (Clontech) was used to validate the interactions of the Drosophila Alk pB66 bait (Tyr1128-Cys1701, N-terminally fused to the Gal4 DNA binding domain) with Drosophila Cnk pB6 prey clones (encoding for Gly1303-Ser1425 or Gln1383-Asn1538, respectively; provided by Hybrigenics) and to determine the minimal AIR. Standard cloning techniques were used to generate the pP6 prey–CnkAIR and pP6 prey–CnkY2H/ΔAIR constructs.


Fig. S1. Characterization of the cnk alleles used in this study.

Fig. S2. The fTRG library line fTRG1248 (Cnk.SGFP) rescues the lethality of cnk mutations.

Fig. S3. CnkY2H/ΔAIR binds to the AlkICD in Y2H.

Fig. S4. Eye morphology of mutants with CRISPR/Cas9-modified cnk alleles.

Fig. S5. Morphological analysis of cnk germline clone–derived embryos.

Fig. S6. Visceral phenotypes caused by mutations in ksr.

Table S1. The fTRG library line fTRG1248 (Cnk.SGFP) rescues the lethality of cnk mutants.

Table S2. Complementation tests for cnk alleles.

Table S3. Complementation tests for aveCC9 alleles.


Acknowledgments: We thank M. Frasch, A. Holz, A. Nose, S. Önel, A. Paululat, M. Therrien, and J. Treisman for sharing fly stocks and reagents. We acknowledge the DGRC (NIH 2P40OD010949-10A1), the DSHB, created by the National Institute of Child Health and Human Development of the NIH and maintained at the University of Iowa, the BDSC (NIH P40OD018537), the Kyoto Stock Center (DGRC) at the Kyoto Institute of Technology, and the VDRC at the Campus Science Support Facilities GmbH (CSF) for the reagents used in this study. We also thank B. Hallberg, A. Holz, and G. Kao for helpful suggestions and critical comments. Funding: This work was supported by the Swedish Cancer Society (2015/391 to R.H.P.), the Swedish Childhood Cancer Foundation (15-0096 to R.H.P.), the Swedish Research Council (2015-04466 to R.H.P.), the Swedish Foundation for Strategic Research (RB13-0204 to R.H.P.), the Göran Gustafsson Foundation (RHP2016), the American Cancer Society (RPG-96-13504-DDC to X.L.), and the NIH (1 P50 DK57301-01 to X.L.). G.W. was supported by postdoctoral fellowships from the Swedish Childhood Cancer Foundation (NC2014-0045) and the Sven and Lily Lawski Foundation. K.P. was supported by a postdoctoral stipend from the Carl Tryggers Foundation (CTS KF15:15). Author contributions: G.W. and R.H.P. conceived the project, designed the experiments, and wrote the first draft of the manuscript. G.W., K.P., J.R.V, F.H., and X.L. conducted the experiments. G.W., K.P., X.L., and R.H.P. analyzed the data. All authors contributed to the final version of the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article