Hairless-binding deficient Suppressor of Hairless alleles reveal Su(H) protein levels are dependent on complex formation with Hairless

Cell fate choices during metazoan development are driven by the highly conserved Notch signalling pathway. Notch receptor activation results in release of the Notch intracellular domain (NICD) that acts as transcriptional co-activator of the DNA-binding protein CSL. In the absence of signal, a repressor complex consisting of CSL bound to co-repressors silences Notch target genes. The Drosophila repressor complex contains the fly CSL orthologue Suppressor of Hairless [Su(H)] and Hairless (H). The Su(H)-H crystal structure revealed a large conformational change within Su(H) upon H binding, precluding interactions with NICD. Based on the structure, several sites in Su(H) and H were determined to specifically engage in complex formation. In particular, three mutations in Su(H) were identified that affect interactions with the repressor H but not the activator NICD. To analyse the effects these mutants have on normal fly development, we introduced these mutations into the native Su(H) locus by genome engineering. We show that the three H-binding deficient Su(H) alleles behave similarly. As these mutants lack the ability to form the repressor complex, Notch signalling activity is strongly increased in homozygotes, comparable to a complete loss of H activity. Unexpectedly, we find that the abundance of the three mutant Su(H) protein variants is altered, as is that of wild type Su(H) protein in the absence of H protein. In the presence of NICD, however, Su(H) mutant protein persists. Apparently, Su(H) protein levels depend on the interactions with H as well as with NICD. Based on these results, we propose that in vivo levels of Su(H) protein are stabilised by interactions with transcription-regulator complexes.


Introduction
Cell-to-cell communication is essential to development in higher animals and relies in part on the Notch signalling cascade which specifies the lineage of cells in a multitude of tissues [1,2]. As a consequence of activating Notch signalling-and subsequently Notch target genes-cells are directed into specific cell fates. The Notch signalling pathway is highly conserved, it is found from worms to insects and mammals, and defects in the Notch pathway are involved in a number of congenital human diseases and several types of human cancers [3].
The principles of Notch signal transduction are rather simple: interactions of the Notch receptor with its ligands result in cleavage of Notch, nuclear translocation of its intracellular domain (NICD, for Notch Intracellular Domain), and assembly of a transcriptional activator complex with the DNA-binding protein CSL and the co-activator Mastermind (Mam) [4,5]. CSL is the initialism for the respective protein orthologues from human, D. melanogaster, and C. elegans, respectively (human C-promoter Binding Factor 1 [CBF1] also named RBPJ, Suppressor of Hairless [Su(H)] and lin-12 and Glp-1 phenotype [Lag1]). In the absence of a Notch signal, a repressor complex consisting of CSL and co-repressors silences transcription from Notch target genes. Most mammalian co-repressors compete with NICD for binding to a hydrophobic pocket within the beta-trefoil domain (BTD) of CSL [5][6][7]. In contrast, in Drosophila a protein named Hairless (H) binds the C-terminal domain (CTD) of Su(H) at sites that are distinct from NICD binding [8][9][10][11]. H recruits the general co-repressors Groucho and C-terminal binding protein, resulting in transcriptional silencing of Notch target genes [12][13][14][15].
The model system Drosophila melanogaster has been extensively used for systematic analysis of Notch signalling, and in particular, our group has used Drosophila to study the Notch repressor complex in vivo. Recently, the crystal structure of the core Notch repressor complex, containing Su(H) and H bound to DNA, was determined [11]. Strikingly, H binds deeply into the hydrophobic core of the CTD of Su(H), thereby causing a major conformational change in Su(H) that is incompatible with NICD binding. The contacts between Su(H) and H are mostly hydrophobic in nature. In order to disrupt formation of the repressor, but not the activator complex, several alanine substitutions were introduced into Su(H). A double mutation (L445A/L514A) and two triple mutations (L434A/L445A/L514A and L445A/L514A/F516A) were shown to specifically disrupt H binding while still allowing NICD interactions as well as activator complex formation. The effects these mutations had on the biological activity of Su (H) were investigated in cell culture assays and overexpression studies [11].
In this work we addressed the question, how is Notch signalling affected in vivo when Su (H) is incapable of binding to H, and therefore unable to act as a repressor, but is capable in activator complex formation. Hence, we set out to analyse the effects of the three H-binding deficient Su(H) mutations on normal fly development. To this end, we replaced the Su(H) locus with mutant versions of Su(H) by genome engineering. The resultant Su(H) alleles Su (H) LL , Su(H) LLF , and Su(H) LLL were subsequently analysed in detail. As expected for Su (H) alleles that are unable to bind to H, the three alleles are recessive lethal, demonstrating that H interactions are imperative to Su(H) function in the fly. Despite quantifiable and statistically significant differences in protein binding strength in vitro [11], the three alleles behave very similarly in vivo. Due to the inability to form repressor complexes, Notch signalling activity is strongly increased in homozygotes, similar to a complete loss of H activity. Unexpectedly, we note a dependence of the abundance of Su(H) protein on H protein binding. Specifically, the protein levels of the H-binding deficient mutants appeared strongly reduced in larval tissue, which was similar to the results observed for the wild type isoform in the absence of H. These data suggest that Su(H) protein is stabilised by forming a complex with H. An analogous stabilisation of Su(H) was observed when Su(H) was engaged in the activator complex. Taken together, as both, H and NICD, are involved in nuclear shuttling of Su(H), the stability of Su (H) may depend on subcellular localisation, which will be addressed in future studies.

Generation of H-binding deficient Su(H) alleles by genome engineering
Previously, we have shown that specific mutations in the C-terminal domain (CTD) of Su(H) significantly perturb binding interactions between Su(H) and H, thereby affecting repressor complex formation [11]. With the goal to study Su(H) proteins that are deficient in H binding in vivo, we generated mutant fly lines by genome engineering as outlined previously [16,17]. In the founder line Su(H) attP , most of the coding region was replaced by an attP landing site that served the introduction of constructs coding for the Su(H) replacement mutants Su(H) LL

H-binding deficient Su(H) alleles show characteristics of a gain of Notch activity
Su(H) attP is expected to be a complete null allele and no Su(H) protein was detected in homozygous mutant larvae (Fig 1A and 1C). Accordingly, Su(H) attP homozygotes died before pupation. Mutant larvae displayed small wing imaginal discs typical of a Su(H) loss of function [18], The Su(H) locus is located on the left arm of chromosome 2; it is flanked by the genes CIAPIN1 and yellow-C on the distal and crinkled (ck) on the proximal side (according to http://www.flybase.org, released 2016_02). The respective transcripts are depicted: exons are boxed, coding sequences are in dark, introns are presented as a dash; direction of transcription is indicated. Above is a restriction map with the relevant restriction sites used almost completely lacking the presumptive wing blade and the dorso-ventral boundary expression of Wingless (Wg) protein (Fig 2A' [19,20]. Wg staining appeared normal or even enhanced along the dorso-ventral boundary ( Fig 2C'-2F'). Overall, the mutant wing discs matched those of the homozygous H attP null allele (Fig 2G'), indicative of a gain of Notch activity. This phenotype might be expected as a consequence of a loss of H binding, entailing the inability to repress Notch target genes. In this case, Notch signalling should generally be overactive in the Su(H) LL , Su(H) LLF or Su(H) LLL mutants, which we addressed by analysing sensory organ precursor (SOP) formation.
SOPs are singled out by Notch mediated lateral inhibition from proneural fields. In the absence of Notch activity, too many SOPs form [18,21,22]. The SOP, however, is protected by Su(H)-H mediated repression from epidermal fate [23]. This was exactly the pattern observed -SOP clusters in Su(H) attP mutant discs (Fig 2B"), and no or sporadic SOPs in discs of any of the three H-binding deficient mutants compared to controls (Fig 2A"-2F"). The phenotypes again matched those of the H attP mutant (Fig 2G"). We frequently observed more SOPs in Su (H) LL compared with the other two mutants, suggesting some residual activity with regards to H binding and in agreement with the biochemical assays [11].
We next induced homozygous mutant cell clones in an otherwise heterozygous background and analysed the expression of Cut (Fig 3) serving as a read out for Notch signalling activity [24][25][26]. Cut is normally expressed along the dorso-ventral boundary in a stripe about three cells wide (Fig 3A) [27]. While a loss of Su(H) activity resulted in the absence of Cut expression (Fig 3B), cells homozygous for any of the three H-binding deficient Su(H) alleles turned on Cut expression ectopically (Fig 3C-3E). Similarly, increased Cut expression was observed in H attP or H LD mutant cells, the latter specifically affecting H-Su(H) binding [8,17] (Fig 3F and  3G). The expression pattern of the Notch target gene wingless (wg) [24][25][26] was altered in a similar way (S3 Fig). These data support the notion that the three alleles Su(H) LL , Su(H) LLF and Su(H) LLL gain Notch activity to a similar degree likely due to a lack of H-binding and repressor complex formation.

Su(H) protein appears to be stabilised within the repressor complex
Curiously, despite its function as a transcriptional regulator Su(H) relies on NICD or H for nuclear import [28][29][30][31][32]. Our new Su(H) mutants allow to directly address how the lack of H for cloning. Two genomic fragments, covering the 5'arm and the 3'arm, were used for homologous replacement of the locus, eventually resulting in the allele Su(H) attP . This served as founder line for the integration of wild type and mutant constructs as shown for Su(H) gwt .(B) The Su(H) protein consist of 594 amino acids and contains three important domains, NTD (N-terminal domain), BTD (beta-trefoil domain) and CTD (C-terminal domain). The scheme shows their relationship to the coding exons: numbers represent the amino acid position upstream of each intron. Below, the three mutants are shown that were generated within the CTD, affecting H protein binding. (C) Protein extracts derived from homozygous larvae of the given genotype reveal expression of Su(H) protein which, however, is reduced in the mutants. The lowest band presumably stems from degradation (open arrowhead). Beta-tubulin (b-tub) served as loading control. M, prestained protein ladder; approximate size is given in kDa.  [29][30][31][32], we expected a shift from the nuclear to the cytoplasmic compartment. In cells homozygous for either Su(H) LL , Su(H) LLF , or Su(H) LLL , however, the Su(H) staining was remarkably weaker and appeared less distinct than in the neighbouring heterozygous or homozygous wild type cells, where Su(H) protein was concentrated within the nucleus (Fig 4A-4E). In the control Su(H) gwt no differences between homo-or heterozygous cells were noted, whereas in Su (H) attP mutant cells Su(H) signals were nearly absent as expected for a null mutant (Fig 4A and  4B). The reduced levels of mutant Su(H) protein suggested to us that Su(H) protein is stabilised while bound to H in the repressor complex (Fig 4C-4E). In this case, we would expect a likewise altered staining pattern of wild type Su(H) protein if H protein was absent or deficient for Su(H) binding. To this end, we generated cell clones homozygous for either H attP or H LD mutant-the former lack H protein, whereas the latter express a Su(H)-binding defective isoform of H [8,17]. As shown in Fig 4F and    As we were puzzled by the lowered signal intensity, we went ahead to directly follow Su(H) protein using mCherry as a protein tag [34]. Three further Su(H) fly lines were generated expressing the wild type, Su(H) LLF , and Su(H) LLL proteins tagged with mCherry (S2B and S2E Fig). Using antibodies directed against mCherry, the results resembled those described above: the wild type Su(H) gwt-mCh protein was conspicuously enriched within the nucleus, whereas Su (H) LLF-mCh and Su(H) LLL-mCh signals were weak and less distinct (Fig 5A-5C [30]. These data emphasize that stability of Su(H) protein depends on the binding to H. Perhaps Su(H) protein is protected from degradation when present in the nucleus whilst assembled in the repressor complex. Hence, H may have an additional role with regard to Su(H) availability during Notch signalling processes, apart from its function as a co-repressor.

Su(H) protein is stabilised by the binding to Hairless and to Notch
If we hypothesize that Su(H) is protected from degradation when nuclear and assembled within the repressor complex, it might likewise be protected within the activator complex when bound to NICD. Notch signalling is specifically activated along the dorso-ventral boundary of the wing imaginal disc [24,35]. Indeed, Su(H) LLL-mCh and Su(H) LLF-mCh proteins appeared specifically enriched in these cells (Fig 5D-5F).
To further substantiate this idea, we turned to RNAi analyses. Knock down of H protein levels along the antero-posterior border of wing imaginal discs resulted in a strong decrease of Su(H) protein levels-yet, a weak signal was still present specifically along the dorso-ventral boundary ( Fig 6A). As predicted by our model, the Su(H) boundary-accumulation was completely lost, when we knocked down Notch at the same time (Fig 6B), presumably because none of the complex forming proteins was present any longer.  In order to test the influence of NICD on Su(H) protein accumulation, we specifically induced the pathway in an ectopic location by misexpressing the ligand Serrate (Ser) along the anteroposterior boundary in the wing imaginal disc. In this setting, the Notch signalling cascade is induced specifically within the ventral domain of the wing disc [35][36][37][38]. Moreover, cis-inhibition resolves the Notch response to two stripes straddling the ventral Ser expression domain [38,39], which was visualized by Cut protein expression as a read out (Fig 7). In a control disc little difference in Su(H) protein abundance was detected due to high endogenous levels ( Fig  7A). In the homozygous mutant background of Su(H) LL , Su(H) LLF , or Su(H) LLL , however, where Su(H) protein levels are low, the signal enhancement was clearly seen at places of the strongest Notch activation, i.e. exactly along the border of Cut induction (Fig 7B-7D). Next we overexpressed NICD along the antero-posterior border of the mutant wing imaginal discs: indeed accumulation of the mutant Su(H) protein was observed demonstrating that NICD itself is sufficient for Su(H) protein stabilisation (Fig 8). These data strongly indicate that Hbinding deficient Su(H) protein is bound and stabilised by NICD protein, supporting the notion of a stabilisation of Su(H) within the nucleus when assembled in the activator complex.

Discussion
In this work, three H-binding deficient Su(H) alleles, Su(H) LL , Su(H) LLF , and Su(H) LLL , were generated by genome engineering. These mutations were based on our previous findings that alanine substitutions at these positions specifically affected repressor but not activator complex formation [11]. In vivo, we noted only little differences amongst the three mutant alleles. With regards to fly viability and SOP formation, however, Su(H) LL appeared a somewhat weaker allele than the other two, in agreement with the residual H-binding activity this mutant displayed in yeast two-hybrid and overexpression assays [11]. All three alleles are larval to early pupal lethal, clearly demonstrating the pivotal role of Su(H) as a repressor during fly development. For example, maintenance of the SOP strictly depends on activity by the Su(H)-H repressor complex. In the case of low repressor activity, the SOP is prone to the epidermal fate through the presence of high levels of proneural proteins that activate E(spl) genes if not hindered [23]. Similar mechanisms likely occur in many other tissues, as E(spl) genes are Notch targets in almost every instance of Notch signalling, including the nervous system, the mesoderm, and the germ line (reviewed in [40]). Our expectation for the H-binding deficient Su(H) alleles, namely gain of Notch activity, was confirmed by our data. In fact, the three alleles were nearly indistinguishable from a complete loss of H in homozygosis. This allows the conclusion that the mutant Su(H) proteins indeed fail to form a repressor complex in vivo. Moreover, these results show that the primary function of H during imaginal development is repression of Notch signalling. During embryogenesis, however, H may take part in additional repressor activities [41]. Originally Su(H) was identified as a dominant suppressor of the H heterozygous bristle loss phenotype [42,43]. Positive and negative autoregulation of Su(H) in the context of mechano-sensory organ formation in addition to several feedback loops built into the Notch signalling pathway, however, complicate genetic analyses [2,4,13,28,[42][43][44][45]   Importantly, our studies revealed a novel additional regulatory mechanism of Notch signal transduction at the level of Su(H) protein stability. Our data demonstrate that Su(H) protein is stabilised when in complexes with either H or NICD, but appears to be unstable when unbound. Accordingly, mutant cells lacking H protein or expressing H protein deficient in Su (H) binding show low Su(H) protein levels (Figs 4-6). Likewise H-binding deficient Su(H) protein is barely detected apart from regions of highest Notch activity (Figs 6-8). Apparently, Su(H) protein levels are ruled by the amount of H or NICD within a cell, consistent with the normal appearance of the heterozygotes. Whereas H protein is unaffected by Su(H) levels (S8 Fig) [30], there are indications for a mutual inter-dependence of overall Notch and Su(H) protein levels during Drosophila embryogenesis [46]. Taken together, our data now implicate a novel role for H in the regulation of Su(H) availability apart from its role as co-repressor. The mouse homologue RBPJ is a rather unstable protein with a half life of about 2 hours [47]. RBPJ degradation involves both the proteasome and the lysosome and is regulated by p38 MAPK phosphorylation and Presenilin-2 [47]. Possibly, Su(H) protein is protected from degradation as long as it is bound within protein complexes, be it activator or repressor complexes. It Fig 2E' and 2F'). Moreover, Su(H) LLF-mCh and Su(H) LLL-mCh protein is strongly reduced overall, yet appears in a stripe where Notch activity is highest (arrow). Size bar represents 50μm in (A-C), and 100μm in (D-F).

wild type sister cells (asterisk). Heterozygous cells display reduced levels and appear yellow in the merge. (B, C) In contrast, Su(H) LLF-mCh or Su(H) LLL-mCh protein accumulates at much lower levels compared to GFP, and is less distinct within nuclei. Intensity of the red channel was increased to visualize the staining. (D-F) mCherry (red) was detected in wing imaginal discs of homozygous animals. As read out for Notch activity, Wg protein is shown (green). Whereas Su(H) gwt-mCh discs appear normal in shape as well as in Wg and Su(H) protein expression (D), Su(H) LLF-mCh and Su(H) LLL-mCh mutant discs are hypertrophied (E, F) similar to the untagged version (compare with
https://doi.org/10.1371/journal.pgen.1006774.g005 Su(H) protein levels depend on the binding of Hairless remains to be established, whether NICD or co-repressors protect RBPJ from degradation similar to Su(H). As both H and NICD are involved in the nuclear shuttling of Su(H), stability may likewise (or in addition) depend on subcellular localisation. To date, we cannot distinguish between these two aspects, as this requires further and more detailed analyses of Su(H) stability. In mammals, however, CBF1/RBPJ nuclear translocation has been shown to also depend on the binding of either co-repressors or NICD [48].
Based on their Rel homology region domains (RHR), CSL proteins are considered distant relatives of the rel class of transcription factors [49]. Our data now add new details to this relationship. The immunoglobulin-like folds, corresponding to RHR-c, build the centre of the CSL C-terminal domain (CTD) that contacts both NICD and MAM in the activator complex https://doi.org/10.1371/journal.pgen.1006774.g008 [50,51], as well as H in the repressor complex [8,11]. The homologous structure in NFκB for example binds to the inhibitory protein IκB resulting in cytoplasmic retention of NFκB (reviewed in [52,53]). Similar to NICD, IκB contains six or more Ank-repeats that contact RHR-c, or CTD in CSL (reviewed in [52]). In general, rel proteins become activated through their nuclear translocation, which is inhibited by IκB proteins. The signalling cascade eventually activating rel proteins results in a phosphorylation and degradation of IκB, and subsequent nuclear translocation of NFκB (reviewed in [53]). Clearly, CSL proteins differ in several respects. Firstly, they act as monomers unlike the rel type transcription factors [5]. Secondly, their nuclear import requires a co-factor rather than the release from an inhibitor. In both examples, however, regulation of nucleo-cytoplasmic shuttling and degradation are important steps in the process of signal transduction.

Generation of the Su(H) attP founder line
Genome engineering was performed as outlined before [16,17]. Genomic fragments were derived from two overlapping λ phages isolated from the Kr Sb10 /SM1 library [54] and cloned into pGX-attP [16]: into the Kpn I site of the 5'MSC a 3.4 kb Eco RI fragment and into the Bgl II / Xho I sites of the 3'MSC a 3.9 kb Bam HI / Xho I fragment. The construct was inserted randomly into the genome in the HR-starter line by P-element mediated germ line transformation [55]. Homologous recombination at the Su(H) locus was induced after crossing in y 1 w 1118 ; P {ry Ã , 70Flp}11P{v+,70I-SceI}2B Sco/CyO (BL6934) and subjecting the offspring to a heat shock regimen as described before [56]. One recombinant event was isolated from about 850 mosaic F1 virgins. The founder line Su(H) attP was derived by Cre-mediated deletion of the white + marker and vector-remains by crossing in y 1 w 67c23 ; sna Sco / CyO, P{w[+mC] = Crew}DH1 (BL1092) as described before [17,56]. Su(H) attP carries a deletion of 1.75 kb of genomic DNA and contains an attP site instead. The line was verified by PCR and sequence analysis.

Generation of Su(H) constructs and Su(H) mutant fly lines
The genomic insertion construct was derived from the already mentioned λ phage clone and contains a 2.3 kb Eco RV / Nde I fragment; it was cloned as Bam HI / Xho I fragment into pGE-attB GMR to be inserted into the attP site of the founder line as outlined before [16,17]. After floxing the white + and vector sequences, the resultant Su(H) gwt line contains a duplication of 590 bp of intronic sequences plus remaining attR, loxP, and vector sequences in the intron and the 3' UTR. This strategy was chosen to avoid possible splice defects. The three alanine substitution mutations were originally introduced in Su(H) cDNA [11]. They were shuttled into the genomic Su(H) DNA whenever possible, resulting in the loss of some of the three introns: Su(H) LL and Su(H) LLF lack intron 3, and Su(H) LLL lacks introns 2 and 3. The respective control lines Su (H) gwtΔi3 and Su(H) gwtΔi2i3 were generated as well. In addition, Su(H) gwt-mCh , Su(H) LLF-mCh and Su(H) LLL-mCh were produced by C-terminal in frame fusion of mCherry that was derived from pRRins [34]. Mutant and control constructs were inserted into the attP site of the founder line by site specific recombination and white + plus vector sequences were deleted as described above. The integration frequency varied between 2-7% of larvae surviving the injection. All lines were confirmed by PCR, diagnostic restriction digests and sequence analysis.
Notch activity was induced along the antero-posterior boundary by crossing UAS-Ser with ptc-Gal4 as described earlier [36], or by crossing UAS-RICN [32] with dpp-Gal4 [59] in a wild type or Su(H) mutant background. Mosaics were induced with the FLP/FRT technique [61,62]; to this end FRT40A was recombined with the Su(H) alleles of interest and crossed with y 1 w Ã ; P{w +mC = Ubi-GFP.D}33 P{w +mC = Ubi-GFP.D}38 P{ry +t7.2 neo-FRT)40A / CyO (BL5189). Offspring was heat shocked for 1h at 37˚C, at first to second instar larval stage and dissected as wandering third instar larvae. Scanning electron micrographs were taken from adult female flies using a table-top NeoScope (JCM-5000; Nikon, Tokyo, Japan).

Immunochemistry
For Western blots, about 10 homozygous third instar larvae were homogenized in 100μl binding buffer (20mM HEPES pH 7.6, 150mM KCl, 2.5mM MgCl 2 , 10% glycerol, 0.05% NP-40, 1mM DTT, ROCHE complete ULTRA protease inhibitor mini tablet) [63], and protein amounts were adjusted by larval weight and normalized by Bradford assay. Rabbit anti-Su(H) (1:1000; Santa Cruz Biotech, Dallas, USA) or mouse anti-beta-tubulin A7 (1:3000) (developed by M. Klymkowsky; obtained from DSHB, the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Dept. of Biology, Iowa City, IA 52242) and goat anti-rabbit or anti-mouse coupled to alkaline phophatase (from Jackson Immuno-Research, via Dianova, Hamburg, Germany) were used for detection. The blots were cut before separate detection of Su(H) and beta-tubulin proteins.
Wing imaginal discs from late third instar larvae were dissected and stained according to standard protocols. The following antibodies were used: rabbit anti-Su (

RT-PCR and RT-qPCR
Poly(A) + RNA was extracted from 100mg of female flies from either Oregon R or Su(H) gwt using the Poly ATract 1000 kit (Promega, Mannheim, Germany), and cDNA generated with ProtoScriptII First Strand cDNA Synthesis Kit (New England Biolabs, Frankfurt, Germany) according to the suppliers' protocols. The PCR spans the first intron that contains the manipulations: Genomic DNA from wild type and from Su(H) gwt should yield a 1012 bp and a 1745 bp amplificate, respectively. The amplificate from cDNA is expected to be 238 bp in size when spliced normally. The following primer pair was used: Pu , 5' CCG GCC ACA CAT CGA GGA  GAA G 3' and Pl, 5' CGC GCA TAG TTG TGC TCC CTG TTC G 3'. qPCR was done on three biological replicates of each genotype with 40 homozygous larvae at~72 hours after egg deposition at 25˚C. Poly(A) + RNA was extracted with Poly ATract 1 1000 kit (Promega, Mannheim, Germany) and the concentration determined in a μCuvette with the BioPhotometer Plus (Eppendorf, Hamburg, Germany). 1μg was treated with 0.4U DNase I (New England Biolabs, Frankfurt, Germany) and reverse transcribed in 0.3μg batches with the ProtoScriptII Kit using oligo-dT primers (New England Biolabs, Frankfurt, Germany). Real time qPCR was performed with Blue S'Green qPCR Kit (Biozym, Hessisch-Oldendorf, Germany) on 2μl of cDNA (0.012μg) in 10μl end volume using MIC magnetic induction cycler (bms, Pots Point Australia) always including target and no-template controls; a hot start (95˚C 2 min) and 40 cycles of 95˚C 5s / 68˚C 10s was followed by a melt curve analysis (72˚C to 95˚C at 0.3˚/s) to select for specific amplification. Absence of DNA was tested in a non-RT control for every sample; RNA integrity was confirmed by 5'-3' Cq analysis. CTCF (PP30808), cyp33 (PP14577), DNApol-α60 (PP9936), eRF1 (PP11596), hisRS (PP13550), and Tbp (PP1556) were assayed as internal references; primer pair sequences (in parentheses) are listed at DRSC FlyPrimer bank [65].  Fig. Su(H) protein expression in salivary glands. (A, B) Salivary glands from homozygous third instar larvae of the given genotype were doubly stained for Pzg as nuclear marker and for Su(H) and mCherry, respectively. Drastic reduction of nuclear Su(H) protein expression is apparent in the mutants. (C) Quantification of staining intensity was performed on stacks of pictures cutting through the entire gland using Image J. Intensity of all the nuclei from one gland was compared with the total cytoplasmic signal from that gland (n = 3-5). Loss of nuclear staining and no concurrent cytoplasmic enrichment is observed. (D) Signal intensity of Su(H) and Pzg staining of individual nuclei (25-50 per gland; 3-5 glands per genotype) were recorded using Image J. A four-to five-fold drop in Su(H) expression is seen in the Hbinding deficient Su(H) as well as in the H alleles indicated. Differences between mutants and control are highly significant by ANOVA two-tailed Tukey-Kramer approach for multiple comparisons ( ÃÃÃ , p<0.001). Size bar represents 100μm in all panels. (TIF)

S6 Fig. Reduction of Su(H) protein levels is not a result of altered transcriptional regulation. (A)
Su(H) mRNA expression was quantified by qPCR relative to cyp33 and Tbp as reference genes. Efficiencies for Su(H) (0.97), for cyp33 (0.94) and for Tbp (0.92) were taken into account for the relative quantification [65]. mRNA was derived from homozygous mutant larvae and compared to Su(H) gwt . Data are assembled from three biological and two to four technical replicates. Mini-max depicts 95% confidence, median corresponds to expression ratio. Values are close to 1, i.e. close to Su(H) gwt expression; differences of about 20% between Su (H) gwt Fig 5A-5C. Signals were recorded at 20 points each within homozygous wildtype, mutant, and heterozygous clones at identical positions for mCherry and for GFP within each disc. To allow for comparison of intensity between control and mutant, homozygous (hz) or heterozygous (htz) mCherry signals were set in relation to GFP signals from heterozygous (htz) cells. Su(H) gwt-mCh protein is about four-fold enriched compared to Su(H) LLF-mCh or Su(H) LLL-mCh protein. Signal intensity of mCherry correlates well with gene dose (heterozygous is about half of homozygous, htz/hz), in contrast to the GFP signals, where the heterozygous signal is more than two-fold weaker than the homozygous signal. Error bars indicate standard deviation (n = 5-7 biological samples). (E) Quantification of nuclear versus cytoplasmic Su(H) protein detected with rabbit anti-Su(H) antibodies in clones as shown in Fig 4. Signals were recorded within and just next to 20 nuclei each within homozygous wild type (dark columns) or homozygous mutant clone (light columns) at identical positions within each disc. GFP signals were used to identify cell clones and Pzg signals to identify the nuclei. Grey values were determined with Image J, and were related to the nuclear signal of the control. Red columns represent nuclear, grey columns cytoplasmic signals; dark columns represent homozygous wild type and light columns homozygous mutant. Error bars show standard deviation (n = 2-4 biological samples). In wild type cells Su