FKBP14 is an essential gene that regulates Presenilin protein levels and Notch signaling in Drosophila

Presenilins (PSNs) are highly conserved multi-pass transmembrane proteins that are synthesized as highly unstable 50 kDa precursor proteins that undergo tightly regulated endolytic processing to generate stable PSN C-terminal and N-terminal fragments that form the catalytic core of the γ-secretase complex (Thinakaran et al., 1996; Guo et al., 1999; Kimberly et al., 2000). γ-secretase cleaves a number of type I transmembrane proteins associated with a wide array of developmental processes, including the Notch receptor and Amyloid precursor protein (APP). In the case of APP, γ-secretase cleavage results in the generation of small amyloid β peptides, including Aβ40 and Aβ42. Mutations associated with Alzheimer's Disease (AD) result in the selective enhancement of Aβ42, which accumulates in the endoplasmic reticulum (ER) of neuronal cells and extracellularly as toxic plaques and is thought to initiate the AD pathogenic cascade (Hashimoto et al., 2003; Vetrivel et al., 2006). In the case of Notch, cleavage by γ-secretase results in release of the intracellular domain, which translocates to the nucleus to initiate transcription of several target genes involved in a broad array of developmental decisions, including cell fate specification, progenitor cell maintenance, boundary formation, cell proliferation and apoptosis (Bray, 2006). Despite the clear importance of the γ-secretase complex and, in particular, the role of PSN in development and disease, little is known about how PSN itself is regulated.

Each of the identified components of the γ-secretase complex, including PSN, are functionally conserved in Drosophila (Greeve et al., 2004). Further, many of the known γ-secretase targets and pathways are also conserved, making Drosophila a useful model with which to identify novel PSN interactors and regulatory mechanisms. We previously carried out a screen for modifiers of Drosophila Presenilin (Psn) and APP (van de Hoef et al., 2009) and found that mutations in FKBP14 could suppress the phenotype associated with overexpression of Psn (van de Hoef et al., 2009).

FKBP14 belongs to a family of highly conserved proteins known as immunophilins that bind to the immunosuppressive drugs FK506, rapamycin and cyclosporin A, and often exhibit peptidyl-prolyl cis-trans isomerase (PPIase) activity (Barik, 2006; Kang et al., 2008). The FK506-binding proteins (FKBPs) are a subfamily of these immunophilins, the smallest members of which are composed almost entirely of a single PPIase domain, whereas larger FKBPs are composed of modular domains that are functionally independent (Barik, 2006). FKBPs are found in a broad range of organisms and have been implicated in various biochemical processes, including protein folding, receptor signaling, protein trafficking and transcription (Barik, 2006; Kang et al., 2008). FKBP family members also exhibit distinct subcellular localizations and bind specific protein targets (Galat, 2008; Kang et al., 2008). Despite the growing evidence for the important roles of various FKBPs, knowledge of their specific functions remains vague. In particular, there is little information on the roles that FKBPs play in the development of multicellular organisms. Studies in yeast have demonstrated that all cyclophilins and FKBPs are individually and collectively dispensable for viability (Dolinski et al., 1997) and no FKBPs have been shown to be essential for viability in any multicellular model.

We have investigated the function of one member of this family, FKBP14, in Drosophila development. We show that FKBP14 is an ER resident protein that is broadly expressed throughout development. Null mutations in FKBP14 are lethal throughout the larval and pupal stages of development, with escapers exhibiting defects in eye, wing and sensory bristle development. We show that FKBP14 mutants do not appear to induce the unfolded protein response, indicating that phenotypes are not likely to be due to ER stress, but are instead associated, at least in part, with the dysregulation of Notch signaling. FKBP14 mutants genetically interact with components of the Notch pathway, including Notch, Delta and Psn. We also find that although trafficking of Notch to the membrane appears unaffected in FKBP14 mutants, the levels of several Notch targets, including Cut, Wingless and Enhancer of split, are significantly reduced, suggesting that FKBP14 acts downstream of Notch activation at the membrane. Consistent with this model, we find that Psn protein levels and γ-secretase activity are reduced in FKBP14 null mutants. Altogether, our data demonstrate that FKBP14 plays an essential role in development, and we propose that one of its primary functions is to stabilize Psn protein in the ER, which is essential for formation of the γ-secretase complex that is required for effective Notch signal transduction.

Flies were maintained on standard media. An excision screen was performed on FKBP14^EP2019 to generate FKBP14 alleles as described (Robertson et al., 1988). The P-element lies 101 bp downstream of the first exon and 1639 bp upstream of the translational start site within exon 2. Sara lies 58 bp downstream of exon 5 and CG10496 is 1691 bp upstream of the first exon of FKBP14-RA, as described in FlyBase (Wilson et al., 2008). An imprecise excision line, FKBP14^D58, has a 2405 bp deletion that includes residues 105-145 of exon 1 and the entire exon 2. Breakpoints were determined by DNA sequence analysis (ACTG). A precise excision line, FKBP14^D34, was also generated and confirmed by DNA sequence analysis. This line is used as a genetic control. The balancer CyO-GFP was used to identify homozygotes. Wild-type Psn overexpression and Psn^w6rp have been described previously (Guo et al., 1999).

To achieve RNAi-mediated inactivation of Psn, UAS-Psn-RNAi⁴³⁰⁸² transgenes were expressed at 29°C during larval and pupal stages of development using daughterless-GAL4 (da-GAL4) or from late embryogenesis until adulthood using pannier-GAL4 (pnr-GAL4) drivers.

To generate clones, the imprecise excision line FKBP14^D58 was recombined onto an FRT(42B) chromosome and confirmed by PCR. Recombination was induced in second and early third instar larvae heterozygous for hs-flp; FRT(42B) FKBP14^D58 and FRT(42B) ubi-GFP by heat shock at 37°C for 60 minutes for 2-3 days consecutively and were analyzed in late third instar larvae imaginal discs (Xu and Rubin, 1993).

The da-GAL4 (5460), Df(1)N-8/FM7c (729), Dl⁷/TM2 (485), w¹¹¹⁸,hs-flp; Adv¹/CyO (6), FRT(42B) (1956) and pnr-GAL4 (3039) lines were obtained from the Bloomington Stock Center. FKBP14^EP2019 (P[EP]Fkbp13^EP2019) is available from the Szeged Stock Center. The UAS-Psn-RNAi⁴³⁰⁸² (43082) line is available from the Vienna Drosophila RNAi Center (VDRC) (Dietzl et al., 2007).

GMR-GAL4 was double balanced with UAS-APP C99^J6 (Finelli et al., 2004) and males were crossed to virgin FKBP14^D34, FKBP14^EP2019, FKBP14^D58 and Psn^w6rp. Crosses were carried out at 29°C and progeny were raised to 10- to 14-day-old adults prior to ELISA.

Immunostaining was performed on embryos, larval imaginal discs and embryonic cell culture as described (Patel, 1994; Yeh et al., 2000; Kim et al., 2006; Commisso and Boulianne, 2007). The primary antibodies used were mouse anti-Ct (1:100), rat anti-Elav (1:500), mouse anti-Notch extracellular domain, EGF repeats #12-20 (C458.2H; 1:100) and mouse anti-Wg (1:500) [all from Developmental Studies Hybridoma Bank (DSHB), University of Iowa, IA, USA], mAb323 [E(spl); gift of Dr S. Bray, University of Cambridge, UK; 1:1], rabbit anti-GFP (Invitrogen; 1:1000), mouse anti-KDEL (MBL; 1:100), mouse anti-p120 (Calbiochem; 1:500), guinea-pig anti-Sens (gift of Dr H. Bellen, Baylor College of Medicine, Houston, TX, USA; 1:1000), rabbit anti-Srp (Sam et al., 1996) (1:500), rabbit anti-Psn^NTF (Guo et al., 1999) and a rabbit anti-Psn^CTF peptide antibody corresponding to amino acids 426-441 (CG18803-PB; NCBI), which was affinity purified (Antibodies Incorporated, Davis, CA, USA; 1:200). Rat anti-FKBP14 (cells, 1:100; ovaries, embryos, larval discs, 1:500) was generated as follows: a Drosophila FKBP14 cDNA [GH08925, amino acids 298-828; Berkeley Drosophila Genome Project (BDGP)] was cloned into pGEX-4T-1/His 6C (Novagen) to produce an FKBP14 fusion protein for polyclonal antibody production in rats (Antibodies Incorporated) and preabsorbed using fixed S2 cells. A488 and Cy3 secondary antibodies were used (Jackson ImmunoResearch; 1:1000). Phalloidin-TRITC (Sigma-Aldrich) was used at 1:100 and DAPI (Invitrogen) was used at 1:5000. Third instar larval CNS and discs were dissected in PBS, incubated in 0.25 mg/ml Acridine Orange (Invitrogen) in PBS, rinsed in PBS and mounted immediately in PBS prior to fluorescence microscopy.

Images were acquired at room temperature using either an LSM510 META confocal microscope (Carl Zeiss), 40′/1.2 and 100′/1.3 objectives and standard fluorescence filters, or a DMRA2 fluorescence microscope (Leica Microsystems) equipped with a Hamamatsu Orca-ER digital camera and Improvision Openlab software, 20′/0.5, 40′/1.25-0.75 and 100′/1.4 objectives, brightfield and standard filters, or a DMLB fluorescence microscope (Leica Microsystems), 5′/0.12 objective, brightfield and CoolSNAP software. Micrographs were analyzed using an XL30 scanning electron microscope (FEI, Hillsboro, OR, USA) equipped with XL Docu software. Images were processed in Photoshop CS and Illustrator CS (Adobe).

Fly lysates were prepared using standard procedures and analyzed with mouse anti-β-tubulin (DSHB; 1:1000), mouse anti-actin (GeneTex, Irvine, CA, USA; 1:1000), mouse anti-Csp-2 (DSHB; 1:25), mouse anti-Fz (Santa Cruz Biotechnology; 1:100), guinea-pig anti-Cad87A (gift from D. Godt, University of Toronto, ON, Canada; 1:2000), rabbit anti-Psn^NTF (1:1000) and rat anti-FKBP14 (1:2000). HRP secondary antibodies were used (Jackson ImmunoResearch; 1:10,000). All blots were performed in triplicate. Quantitation was performed using a Fluorchem 8000 Gel Documentation System and Alpha Innotech software (Alpha Innotech, San Leandro, CA, USA).

RNA was extracted using Trizol (Invitrogen). For reverse transcription (RT)-PCR, RNA was reverse transcribed using the Superscript First-Strand System (Invitrogen). For quantitative (q) RT-PCR, the results were normalized to an internal control, Rp49 (RpL32 - FlyBase) (three separate experiments, at least ten animals per sample). Primer sequences (5′-3′) were: FKBP14, AGCTGATCAACATCGGCAAT (+) and CCAAGAACCCCCATTATTGA (-); CG10496, ATAAAGGGGAAGGAGCTGGA (+) and GGCGCCATATAGCTTTGGTA (-); Sara, CACCGACGATCAGAGTGAGA (+) and CTCCGCAATCCGTGTTATCT (-); Hsc3, GCGAACAAGATACCGATGCT (+) and GTTATCGGAGGCAGTGGAGA (-); and Rp49, AGTGCGTCGCCGCTTCAAGG (+) and AGAACGCA GGCGACCGTTGG (-).

The levels of Aβ40 and Aβ42 were determined using commercially available human Aβ-specific ELISA kits (Invitrogen) according to the manufacturer's instructions. Heads from 10- to 14-day-old adult progeny were lysed in RIPA buffer (50 mM Tris, 150 mM NaCl, 1% SDS, 1% NP40, 0.5% sodium deoxycholate, pH 8.0) containing Complete Protease Inhibitor Cocktail (Roche). Lysates were diluted 1:5 or 1:10 with PBS containing protease inhibitors, followed by a further 1:1 dilution with the kit dilution buffer containing protease inhibitors prior to analysis.

There are eight known FKBPs in Drosophila that share homology with the archetypal human FKBP12 (FKBP1A - Human Gene Nomenclature Committee). Sequence analysis of one of these, Drosophila CG9847, reveals a signal peptide (SS), peptidyl-prolylcis-trans isomerase (PPIase) domain (FKBP_C), an EF-hand motif (EFh) and an ER retention signal known as HDEL (Fig. 1A; supplementary material Fig. S1A). Although it was originally identified as Fkbp13 (Spradling et al., 1999), its closest mammalian ortholog is FKBP14, with 37% overall sequence identity (supplementary material Fig. S1B). By contrast, mammalian FKBP13 shares 26% sequence identity with Drosophila CG9847 and 57% sequence identity with another Drosophila FKBP, CG14715 (supplementary material Fig. S1B). This indicates that Drosophila CG9847 encodes an ortholog of mammalian FKBP14. Hereafter, we will refer to the protein encoded by CG9847 as Drosophila FKBP14.

To determine its subcellular localization, we raised antibodies against a Drosophila FKBP14 fusion protein. Immune serum detected a 26 kDa band in western blots of fly protein lysates, which was not observed with pre-immune serum (data not shown). The specificity of the primary antibody was confirmed using hypomorphic and null FKBP14 mutants (Fig. 1B). Using this anti-FKBP14 antibody, we examined the subcellular localization of FKBP14 in Drosophila Schneider 2 (S2) cells, in which it is endogenously expressed. Drosophila FKBP14 colocalized with KDEL, an ER marker (supplementary material Fig. S2A-A′), but not with p120 (p120ctn - FlyBase), a Golgi marker (supplementary material Fig. S2B-B′).

We then examined FKBP14 expression in Drosophila embryos and larval imaginal tissues. We found that FKBP14 is broadly expressed throughout development. In mid- and late-stage embryos, FKBP14 colocalizes with a number of pro-hemocyte and plasmatocyte markers, including the GATA transcription factor Serpent (Tepass et al., 1994; Sam et al., 1996), which colocalized with FKBP14 in pro-hemocytes that circulate from cephalic mesoderm (supplementary material Fig. S3B-B′) and partially colocalized with FKBP14 in differentiated pro-hemocytes in the dorsal region of 16- to 18-hour embryos (supplementary material Fig. S3C-C′). FKBP14 also colocalized in plasmatocytes with the collagen type IV factor Viking (Yasothornsrikul et al., 1997) at late stages of embryogenesis (supplementary material Fig. S3D-D′). In third instar larvae FKBP14 was broadly expressed in the wing disc (supplementary material Fig. S3E), eye disc (supplementary material Fig. S3F), CNS optic lobes (supplementary material Fig. S3G) and in the ventral nerve cord (supplementary material Fig. S3H). Altogether, these data suggest a role for FKBP14 in multiple tissue types during development.

To determine the function of Drosophila FKBP14, we characterized a lethal P-element insertion, EP(2)2019, located within the first intron of FKBP14 (Fig. 1A). We also identified an independent P-element insertion in FKBP14, FKBP14²²⁰⁶, which failed to complement FKBP14^EP2019. We also generated additional alleles of FKBP14 through imprecise excision of the EP(2)2019 insertion, including FKBP14^D58, which removes part of the first exon and completely removes the second exon, including the translation start site (Fig. 1A). FKBP14^D58 failed to complement both FKBP14^EP2019 and FKBP14²²⁰⁶. FKBP14^D58 and FKBP14^EP2019 mutants are homozygous lethal, with escapers reaching late pupal development as pharate adults. In addition to the deletion allele, we obtained a revertant line, FKBP14^D34, which precisely removed the original P-element insertion. FKBP14^D34 homozygotes are viable, demonstrating that FKBP14^D58 lethality is not due to a second site mutation. FKBP14^D34 was used as a genetic control for all subsequent experiments.

To further characterize the FKBP14 alleles, we examined FKBP14 protein expression in wild-type and mutant flies by western blot. We observed up to 80% reduction of FKBP14 in pharate adults from the original insertion line, FKBP14^EP2019, demonstrating that this is a strong hypomorphic allele (Fig. 1B). No protein was detected in pharate adults from the deletion mutant FKBP14^D58, confirming that this allele represents a null mutation (Fig. 1B). These data also demonstrate that our antibody is specific to Drosophila FKBP14.

To ensure that neither P-element insertion nor deletion of FKBP14 affected the expression of neighboring genes, we performed qRT-PCR analysis. The neighboring genes Sara and CG10496 lie 58 bp downstream from exon 5 and 1691 bp upstream of FKBP14-RA exon 1, respectively (Fig. 1A). We observed asignificant reduction in FKBP14 expression from FKBP14 homozygous mutant larvae, as compared with controls (P≤0.001), and no significant changes in the expression levels of CG10496 or Sara (Fig. 1C). Together, these data demonstrate that both the P-element insertion and the deletion significantly affect FKBP14 transcription alone.

PPIases catalyze the cis-trans isomerization of peptidyl-prolyl amide bonds and are implicated in multiple intracellular processes, including protein folding (Göthel and Marahiel, 1999). FKBP14 contains a single PPIase domain, suggesting that reduced FKBP14 expression could result in the accumulation of misfolded proteins, a condition that leads to ER stress and activation of the unfolded protein response (UPR) (Ryoo and Steller, 2007). Once activated, the UPR helps to reduce ER stress by attenuating protein synthesis, enhancing degradation of misfolded ER proteins and inducing expression of ER resident chaperones (Ryoo and Steller, 2007). The ER chaperone BiP is a well-known member of the UPR, and its transcription is upregulated when UPR is active (Ryoo and Steller, 2007). To determine whether loss of FKBP14 causes a global ER stress response, we examined the transcript levels of a Drosophila BiP homolog, Hsc3 (Hsc70-3 -FlyBase), by qRT-PCR (Hirota et al., 2006). As a positive control, FKBP14^D34 larvae were placed on food containing Tunicamycin (Tm), which inhibits N-glycosylation and causes accumulation of unfolded proteins in the ER (Kaufman, 1999). This resulted in a significant increase in Hsc3 levels compared with the vehicle-treated control (Fig. 1D). However, we found similar Hsc3 transcript levels in FKBP14^EP2019 and FKBP14^D58 third instar larvae as compared with control FKBP14^D34 (Fig. 1D). Taken together with the mutant qRT-PCR results, these data indicate that the phenotypes observed in 2019 and D58 mutants are due to loss of FKBP14 and are likely independent of ER stress.

We previously demonstrated that FKBP14 genetically interacts with Psn, such that mutations in FKBP14 can suppress phenotypes associated with overexpression of Psn (van de Hoef et al., 2009). One of the essential developmental roles for Psn is as a core member of the Notch signaling pathway (Bray, 2006). To determine whether FKBP14 interacts with the Notch pathway, we performed genetic epistasis experiments using FKBP14, Notch and the Notch ligand Delta. We also examined a mutation in Psn that we have previously shown to modulate both Notch and Delta phenotypes in transheterozygotes (Guo et al., 1999). We did not observe any significant defects in wing development in either heterozygous mutant FKBP14 or Psn lines on their own (Fig. 2C,D).

We then analyzed whether FKBP14 mutants genetically interact with a Delta (Dl) mutation that causes wing vein defects (Fig. 2E) and a Notch deficiency that causes notching of the distal wing margin due to haploinsufficiency (Fig. 2I). We found that transheterozygotes of Dl⁷ and FKBP14^D58 exhibit enhanced wing vein thickening and ectopic deltas (Fig. 2G), compared with controls (Fig. 2C,E,F). Similarly, transheterozygotes of FKBP14^D58 and the Notch deficiency Df(1)N-8 exhibit enhanced wing notching (Fig. 2K), compared with controls (Fig. 2C,I,J). Interestingly, transheterozygotes of Psn^w6rp and Dl⁷ displayed a similar enhancement of wing vein thickening as observed for FKBP14^D58 and Dl⁷ transheterozygotes (Fig. 2H), relative to controls (Fig. 2D,E). Both phenotypes were fully penetrant.

These data demonstrate that FKBP14 genetically interacts with members of the Notch pathway during wing development and that the phenotypes observed in FKBP14 mutants, including the ability of loss-of-function mutants in FKBP14 to suppress the dominant phenotype caused by overexpression of Psn (van de Hoef et al., 2009), may be due, at least in part, to defects in Notch sigaling.

Although FKBP14 mutants are homozygous lethal during a broad range of larval development, we also observed ‘escapers’ that survived to become pharate adults, probably owing to perdurance of a maternal contribution. In pharate adults, loss of FKBP14 function during wing development causes mild wing margin notching (Fig. 3B), loss of bristles (Fig. 3B′) and bristle doublets(Fig. 3B′), compared with the control (Fig. 3A,A′). These defects are phenotypically consistent with some Notch loss-of-function mutants (de Celis et al., 1996; Guo et al., 1999) and with a role for FKBP14 in Notch regulation during wing development, as seen in our genetic epistasis experiments (Fig. 2). Patterning of the adult wing begins at the larval stage and requires activity of the transcription factors Cut (Ct) and Wingless (Wg) to define the presumptive wing margin (de Celis and Bray, 1997; Micchelli et al., 1997). Notch signaling maintains Ct expression at the margin and Ct may be required to maintain wg transcription (Micchelli et al., 1997). To examine the effects of FKBP14 on Notch signaling in third instar wing disc development we analyzed the levels of Ct and Wg expression and found that both were reduced in FKBP14^D58 mutants (Fig. 3D,F), compared with the control (Fig. 3C,E). The reduction in Ct expression is more extreme than that in Wg, similar to what has been shown in Notch loss-of-function alleles (Micchelli et al., 1997). These data indicate that loss of FKBP14 may affect Notch-dependent wing margin specification.

Notch signaling in presumptive wing tissues activates the expression of proteins encoded by the E(spl) complex in cells targeted for an epidermal cell fate that surround sense organ precursors (SOPs) (de Celis et al., 1996). To further define the potential role of FKBP14 in mediating a Notch signal, we analyzed the expression of E(spl) in FKBP14^D58 mutant wing discs. We found that E(spl) levels are significantly reduced in cells at the presumptive wing margin and hinge region in FKBP14 mutant wing discs (Fig. 3H), compared with control (Fig. 3G). These data, together with the mild wing margin notches observed in FKBP14 null mutants, suggest that FKBP14 function is required to maintain Notch signaling in third instar presumptive wing tissues. However, we also observed reduced numbers of cells marked by the nuclear protein Senseless (Sens) (Nolo et al., 2000) at the presumptive wing margin in FKBP14 mutants, indicating a loss of SOPs (Fig. 3H). This is in contrast to typical defects in Notch, which result in supernumerary SOPs (Logeat et al., 1998; Kidd and Lieber, 2002).

In FKBP14 mutant pharate adult nota, the numbers of microchaetae and macrochaetae are significantly reduced in FKBP14^EP2019 (data not shown) and FKBP14^D58 (Fig. 4B,B′) mutants, compared with control (Fig. 4A,A′). Once again, these phenotypes are highly reminiscent of a defect in Notch signaling (Frise et al., 1996; Yaich et al., 1998; Lai and Rubin, 2001). We therefore examined whether FKBP14 was modulating Notch signaling during sense organ patterning by analyzing the effects of FKBP14 loss on the expression of the Notch downstream target E(spl) in third instar presumptive notum tissue. As in the presumptive wing tissue, Notch signaling in the presumptive notum activates the expression of proteins that are encoded by the E(spl) complex in cells targeted for an epidermal cell fate, which surround SOPs (Jennings et al., 1994; Castro et al., 2005). In control larvae, SOPs marked by Sens are surrounded by cells expressing E(spl) proteins, particularly in supraalar (Fig. 4C,C′) and postalar (Fig. 4C,C′) regions. We observed reduced E(spl) expression in FKBP14^D58 presumptive nota, particularly in supraalar (Fig. 4D,D′) and postalar (Fig. 4D,D′) regions, compared with the control. This reduction in E(spl) expression surrounding SOPs suggests a role for FKBP14 in modulating Notch signal transduction during SOP determination. However, we again observed a decrease in SOP levels as marked by Sens in FKBP14^D58 larval presumptive nota (Fig. 4D′,D′; quantified in supplementary material Fig. S4), similar to that seen for FKBP14^D58 presumptive wing tissue. This indicates that FKBP14 may have multiple functions during SOP specification, some of which reflect an effect on Notch signaling, whereas others appear to be independent of Notch.

Since our FKBP14 mutants display Notch-like phenotypes in both the notum and wing, and we observed defects in downstream targets of the Notch pathway in both tissues, we sought to determine how FKBP14 affects the Notch pathway. In vertebrates, Notch is synthesized in the ER and then processed in the Golgi, leading to the formation of a heterodimeric receptor at the plasma membrane, whereas in Drosophila the majority of Notch protein at the plasma membrane is uncleaved (Logeat et al., 1998; Kidd and Lieber, 2002). FKBP14 is an ER resident protein that might be involved in protein folding; therefore, we examined membrane trafficking of the Notch receptor in FKBP14 mutants. We found that the Notch receptor appears predominantly localized at the plasma membrane in FKBP14^D58 presumptive nota tissues (Fig. 5B), similar to the control (Fig. 5A). Moreover, Notch expression at the membrane is intact in cells that border third instar larval presumptive wing margins in control and FKBP14^D58 mutant tissues (supplementary material Fig. S5A,B).

Although we cannot rule out subtle defects in Notch trafficking, based on these findings we conclude that trafficking of the Notch receptor to the plasma membrane is not grossly affected in our FKBP14 mutants. As a control, we examined the surface levels of Notch ligand in Psn null mutants, which interrupt Notch signaling downstream of the Notch receptor reaching the cell surface. As in our FKBP14 mutants, trafficking of the Notch ligand to the cell surface was not grossly affected in the Psn null mutant w6rp (Fig. 5C).

Although trafficking of Notch to the membrane appears largely unaffected in FKBP14 mutants, the levels of Notch target genes are reduced suggesting that FKBP14 is likely to act downstream of Notch activation at the membrane. The γ-secretase complex plays a key role in Notch activation by cleaving Notch, leading to translocation of the intracellular domain to the nucleus where it activates downstream genes (Selkoe and Kopan, 2003). The γ-secretase complex is made up of four essential components: Anterior pharynx defective 1 (Aph-1), Nicastrin (Nct), Presenilin enhancer 2 (Pen-2) and the catalytic core Psn (Selkoe and Kopan, 2003). To assess whether FKBP14 interacts with members of the γ-secretase complex, we examined the subcellular localization of endogenous FKBP14, Psn and transiently transfected Aph-1, Nct and Pen-2 in Drosophila cells. We detected colocalization of FKBP14 with transiently transfected Aph-1 (supplementary material Fig. S6A′) but not Nct or Pen-2 (supplementary material Fig. S6B′-C′). We also found that although Psn is broadly distributed throughout the cytoplasm, it colocalizes with endogenous FKBP14 within ER-localized puncta (supplementary material Fig. S6D′,E′), suggesting that these proteins might interact within the early secretory pathway.

Given that we identified FKBP14 as a Psn modifier (van de Hoef et al., 2009), and observed that Psn mutants modified Notch mutant wing phenotypes, similar to FKBP14 mutants, we examined whether loss of Psn might be partially responsible for the Notch-like traits that we observe in our mutant escapers. To this end, we utilized RNAi-mediated knockdown of Psn during larval and pupal development (UAS-Psn-RNAi⁴³⁰⁸² from VDRC) (Dietzl et al., 2007) driven either ubiquitously (da-GAL4) or in the medial presumptive notum (pnr-GAL4). In Psn knockdown flies, we observed a reduction in bristle sense organs, similar to what was observed in FKBP14 mutants (supplementary material Fig. S7A-C). This was not due to a loss of FKBP14 function because FKBP14 levels were unaffected by Psn knockdown (supplementary material Fig. S7E).

To determine how the loss of FKBP14 affects Psn, we examined Psn protein levels in FKBP14 mutants using an anti-Psn^NTF antibody. We found that levels of endogenous Psn are reduced in our FKBP14 mutants. In the most extreme cases, Psn appears reduced by up to 90% in FKBP14 mutants as compared with controls (Fig. 5D); however, the effect on Psn is highly variable. The effects on Psn levels are post-transcriptional, as levels of Psn mRNA are not reduced in FKBP14 mutants compared with controls (Fig. 5E). We also examined whether FKBP14 levels wereaffected by Psn loss and found no significant changes (Fig. 5D), suggesting that FKBP14 acts upstream of Psn. The effect of loss of FKBP14 on Psn protein levels appears relatively specific, as we also examined the levels of several other single and multipass transmembrane proteins in FKBP14 mutants and found no gross differences in protein levels to controls (supplementary material Fig. S7D).

To assess whether Psn-dependent cleavage by γ-secretase is affected in our mutants, we expressed the C99 fragment of human APP (Finelli et al., 2004), a direct γ-secretase target, in wild-type and heterozygous FKBP14 mutant backgrounds using the GMR-GAL4 driver. Specifically, extracts were prepared from heads of 10- to 14-day-old flies and the levels of the γ-secretase-dependent cleavage fragments Aβ40 and Aβ42 were measured by ELISA. A significant reduction in the Aβ42 fragment was observed in flies expressing APP-C99 in an FKBP14 heterozygous mutant background compared with controls (Fig. 5F). The levels of Aβ40 were also reduced compared with controls, although these results were not statistically significant. Importantly, the reduction in the levels of Aβ40 and Aβ42 observed in FKBP14 mutants was similar to that observed in heterozygous Psn mutants (Fig. 5F). These results indicate that γ-secretase activity is indeed reduced in FKBP14 mutants relative to controls, probably owing to the loss of Psn protein observed in our mutants.

Taken together, these data indicate that FKBP14 is likely to function, directly or indirectly, to stabilize Psn in the ER and that, in the absence of FKBP14, Psn protein levels are reduced resulting in reduced γ-secretase activity and Notch-related developmental defects.

By analyzing the phenotypes of FKBP14 escapers we have identified a role for FKBP14 in the Notch pathway. However, the broad expression pattern of FKBP14 suggests that it might play additional roles in development. Moreover, FKBP14 mutants are lethal throughout development, and most die during mid-pupal stages. Defects include ventral nerve cord retraction defects, small optic lobes and misshapen imaginal discs (supplementary material Fig. S8B,D,F), compared with controls (supplementary material Fig. S8A,C,E); however, the predominant mutant phenotype is a loss of imaginal discs, with 80% of FKBP14^D58 homozygotes lacking imaginal discs (supplementary material Fig. S8G; n=250). Since we hypothesized that this variation in lethality might be due to perdurance of maternal contribution, we examined FKBP14 expression during oogenesis and detected FKBP14 expression in unfertilized eggs of control FKBP14^D34 flies (Fig. 6A). In mid-stage egg chambers, FKBP14 is expressed in nurse cells and oocytes (Fig. 6B). At later stages of oogenesis, FKBP14 protein is maintained in nurse cells, oocytes and somatic follicle cells (data not shown), and in syncytial blastoderm embryos FKBP14 expression is detected throughout the cytoplasm (supplementary material Fig. S3A-A′).

To determine the phenotype associated with a complete loss of FKBP14 and to further define its role during development we first attempted to generate maternal-zygotic null animals using the FLP/FRT mosaic clone technique (Perrimon, 1998). Initially, we generated transheterozygotes of FRT(42B)FKBP14^D58 and FRT(42B)ovoD, a dominant female-sterile mutation (Chou and Perrimon, 1996). Heat shock-mediated FLP/FRT recombination produced apparently sterile females that did not lay eggs, indicating a possible requirement for FKBP14 during oogenesis (data not shown). To further determine the effect of a complete loss of FKBP14 we generated clones in somatic tissues. Using the ubiquitous GFP marker allele FRT(42B)ubi-GFP, we induced recombination and assessed clones in imaginal discs. In control wing discs, we induced recombination between FRT(42B)ubi-GFP and FRT(42B), resulting in clones that lack GFP staining (Fig. 6D) adjacent to GFP twin spots (Fig. 6D). By contrast, following recombination between FRT(42B)ubi-GFP and FRT(42B)FKBP14^D58, we observed GFP twin spots (Fig. 6E) without any associated FKBP14 null clones. These data indicate that FKBP14 is required for cell viability. This is in contrast to mutations in Notch and Psn, for which it is possible to generate both maternal-zygotic and somatic null clones, but is similar to what has been observed for aph-1 (López-Schier and St Johnston, 2002; Cooper et al., 2009). Whether this result is due to cell death or to a defect in cell competition has yet to be determined.

Psn is synthesized within the ER and is rapidly cleaved, generating N- and C-terminal fragments that are essential for γ-secretase catalytic activity (Thinakaran et al., 1996; Kimberly et al., 2000). We identified a novel genetic interactor of Psn, Drosophila FKBP14, which is a member of the large FKBP family of immunophilins (van de Hoef et al., 2009). FKBP14 has a single PPIase domain, a calcium-binding EF hand, an N-terminal signal sequence and C-terminal endoplasmic HDEL motif. Here we have further defined a potential role of FKBP14 in Notch signaling and begun exploring the broader role of FKBP14 in Drosophila development.

We have shown that FKBP14 genetically interacts with Notch and Delta during wing patterning, similar to what has been shown in Psn and Notch or Delta transheterozygotes (Guo et al., 1999). Our data show that FKBP14 mutations give rise to developmental defects, a subset of which are similar to those previously identified in Notch loss-of-function mutants. In particular, FKBP14 escaper pharate adults exhibit bristle phenotypes in the notum and wing tissues that suggest defects in Notch signaling. In keeping with a role for FKBP14 in Notch signaling, we observe a decrease in the expression of Notch target genes, such as ct and E(spl), in third instar wing disc tissues.

Notch signaling can be regulated at a number of steps, including Notch receptor synthesis, transport to the membrane and proteolysis. In Drosophila, the majority of Notch protein at the plasma membrane is uncleaved (Kidd and Lieber, 2002). Activation of Notch occurs at the cell surface upon interaction with its ligands, and subsequent γ-secretase proteolysis of Notch within the endocytic pathway leads to Notch signaling events (Gupta-Rossi et al., 2004). We found that loss of FKBP14 does not affect trafficking of Notch to the membrane, yet it does result in reduced expression of Notch target genes. Although we cannot rule out the possibility that subtle effects on Notch receptor trafficking contribute to our FKBP14 null phenotype, these results suggest that FKBP14 is likely to affect components of the Notch pathway downstream of Notch localization at the plasma membrane. Consistent with this model, we find that Psn protein but not RNA levels are often reduced in FKBP14 mutants. Further, we found that FKBP14 and Psn colocalize in the ER, where the majority of Psn protein is uncleaved, suggesting that FKBP14 might interact with Psn prior to endoproteolytic cleavage. The holoprotein, unlike the N- and C-terminal fragments, is highly unstable and must incorporate into a larger complex for stabilization (Ratovitski et al., 1997). A putative role for FKBP14 might therefore be to properly fold or stabilize Psn in the ER, allowing for the formation of a functional γ-secretase complex that is required for cleavage and activation of Notch. Consistent with this hypothesis, we found that γ-secretase activity is reduced in flies heterozygous for mutations in FKBP14, in a manner similar to those heterozygous for Psn mutations.

Although we observed specific defects in Notch signaling associated with the loss of FKBP14, it is likely that FKBP14 also plays a broader role in development. FKBP14 is broadly expressed in many tissues and loss-of-function mutations in FKBP14 are lethal throughout development. Perdurance of maternal contribution likely leads to the broad time span of lethality observed, depending on the amount of maternal protein and transcript each animal received. Moreover, FKBP14 mutants show a range of defects during imaginal disc development, some of which are reminiscent of Notch defects whereas others appear to be independent of Notch. FKBP14 mutants, for instance, show fewer SOPs than controls. Defective Notch signaling at this stage of SOP specification results in supernumerary SOPs (Mummery-Widmer et al., 2009). We also found that FKBP14 clones are not viable, neither in the germline nor in somatic tissue, in contrast to mutations in both Notch and Psn (Lopez-Schier and St Johnston, 2002; Cooper et al., 2009). Interestingly, mutations in one of the γ-secretase components, Aph-1, show a similar defect in cell viability, and this effect appears to be independent of its role in regulating γ-secretase activity (Cooper et al., 2009). Indeed, we observe partial colocalization of FKBP14 with Aph-1, but not with the other γ-secretase components Nct and Pen-2, in Drosophila cells. It is currently thought that Aph-1 and Nct might form a subcomplex that stabilizes Psn early in γ-secretase assembly, prior to Psn endoproteolysis and incorporation of Pen-2 (Hu and Fortini, 2003). Whether loss of FKBP14 affects Aph-1 function, however, is not yet known. It is also unclear whether the cell lethality phenotype observed in somatic FKBP14 clones is due to apoptosis or the inability of these cells to successfully compete with wild-type neighbors. Further analysis using recessive cell-lethal (Menut et al., 2007) and cell competition-defective (Marygold et al., 2007) marker alleles will enable us to distinguish between these possibilities.

Previous studies in yeast have demonstrated that all cyclophilins and FKBPs are individually and collectively dispensable for viability (Dolinski et al., 1997) and their roles in multicellular organisms have yet to be determined. Of note, a recent study has linked mutations in human FKBP14 to recessive developmental disorders with various congenital symptoms (Baumann et al., 2012). Our study demonstrates that FKBP14 is required for metazoan development and describes, among other potential roles, a novel requirement for FKBP14 in development, cell viability, Psn protein stability and Notch signaling.

We thank David Knight and Michael Garrori for assistance in tissue collection and fly crosses; all members of the G.L.B. laboratory for helpful discussions and comments; and the Bloomington, Szeged and Vienna RNAi Stock Centers for stocks.