Zinc Finger Protein Zn72D Promotes Productive Splicing of the maleless Transcript

Abstract

In organisms with sex chromosomes, dosage compensation equalizes gene expression between the sexes. In Drosophila melanogaster males, the male-specific lethal (MSL) complex of proteins and two noncoding roX RNAs coat the X chromosome, resulting in a twofold transcriptional upregulation to equalize gene expression with that of females. How MSL complex enrichment on the X chromosome is regulated is not well understood. We performed an RNA interference screen to identify new factors required for dosage compensation. Using a Drosophila Schneider S2 cell line in which green fluorescent protein (GFP)-tagged MSL2 localizes to the X chromosome, we assayed ∼7,200 knockdowns for their effects on GFP-MSL2 distribution. One factor identified is the zinc finger protein Zn72D. In its absence, the MSL complex no longer coats the X chromosome. We demonstrate that Zn72D is required for productive splicing of the transcript for the MSL protein Maleless, explaining the dosage compensation defect. However, Zn72D is required for the viability of both sexes, indicating its functions are not sex specific. Consistent with this, Zn72D colocalizes with elongating RNA polymerase II, implicating it as a more general factor involved in RNA metabolism.

Transcription is a highly regulated process, ensuring the production of appropriate levels of gene products to direct cellular proliferation and differentiation programs. Regulation occurs at every step, from the initiation and elongation of transcription and splicing of pre-mRNAs to the export and translation of the mature transcripts. One specific example of regulation at the level of transcription elongation is dosage compensation. Organisms in which males and females are distinguished by sex chromosomes undergo dosage compensation to achieve equal gene expression between the sexes. Female mammals (XX) inactivate one of their two X chromosomes in order to equalize gene expression with males (XY), which have only one X chromosome (36). Recent studies indicate that there is also upregulation of gene expression from the single X in males and the single active X in females to provide proper diploid gene expression (22, 43). In Caenorhabditis elegans, hermaphrodites (XX) downregulate transcription from both X chromosomes by half to match the expression of that in males (XO) (41). In Drosophila melanogaster, X-linked gene expression is upregulated twofold on the single X chromosome in males (XY) to equalize expression with that of females (XX) (36). Dosage compensation in both mammals and Drosophila requires noncoding RNA-containing protein complexes (36). In Drosophila males, the male-specific lethal (MSL) complex (also known as the dosage compensation complex) is localized to the single X chromosome, where it directs the upregulation of gene expression (reviewed in references 36 and 52).

The MSL complex is composed of five proteins, MSL1, MSL2, MSL3, MLE (Maleless), and MOF (Males absent on the first), and two noncoding RNAs, roX1 and roX2. All MSL proteins and at least one roX RNA must be expressed for the localization of the entire MSL complex to the X chromosome and for proper X-linked gene expression. msl2 mRNA is translated exclusively in male cells, ensuring sex-specific deployment of the MSL complex (5, 19, 29). MSL2 stabilizes MSL1, and these two proteins, in the absence of any one of the other MSL proteins, bind to approximately 35 sites along the length of the X chromosome (28, 37, 40). When MSL3, MLE, MOF, and at least one of the roX RNAs are also expressed, the complete complex forms and localizes to additional sites on the X chromosome. This was originally proposed to be the result of spreading of the complex from the ∼35 sites to additional sites along the X chromosome (28, 37, 40) and was more recently proposed to be the result of the complex binding first to ∼35 high-affinity sites and then to additional lower-affinity sites (10, 12, 15, 20, 44). MOF, a histone acetyltransferase, hyperacetylates histone H4 at lysine 16, resulting in the twofold increase in expression of X-linked genes (1, 49). The localization and function of each component of the complex are highly regulated. For example, the acetyltransferase activity of MOF and the helicase and/or the ATPase function of MLE are both required for the complex to associate with the X chromosome in regions beyond the chromatin entry sites (21). MOF and MLE, as well as MSL3, require RNA for their localization to the X chromosome, and in turn, the roX RNAs are stabilized by the localization of the MSL complex to the X chromosome (2, 8, 40, 46). In addition, transcription of roX RNAs is controlled by the MSL proteins (4, 34, 45). Together, these data uncover a series of interactions between components of the MSL complex that are required to ensure the complex's correct localization and activity.

While the translational regulation of msl2 by sex-lethal proteins and the regulation of roX transcription by the MSL proteins are well documented, the factors that regulate expression of the remaining components of the MSL complex remain largely uncharacterized. We carried out an RNA interference (RNAi) screen to identify novel factors involved in the localization of the MSL complex to the X chromosome. Using this screen, we identified the zinc finger protein Zn72D as a new protein required for MSL complex localization and dosage compensation. Zn72D is essential for development in both males and females. It is not enriched on the X chromosome in males but rather colocalizes with elongating RNA polymerase II. We found that Zn72D is required to promote the production of the correctly spliced form of mle mRNA, thus elucidating its role in dosage compensation. Together, these data suggest that Zn72D may have a more general role in RNA metabolism and that it indirectly regulates dosage compensation.

MATERIALS AND METHODS

Cell culture and generation of stable cell lines.

S2 cells and Kc cells were grown in Schneider's medium plus 10% fetal bovine serum, penicillin, and streptomycin. Cells were maintained according to an Invitrogen Drosophila expression system protocol. The green fluorescent protein (GFP)-MSL2 line was created by cloning a PCR product of MSL2 into the NotI/AgeI sites in pAC5.1/V5His-B, which carries Emerald GFP as a EcoRI-NotI fragment (a gift from Renny Feldman and Pat O'Farrell). pAFH-Zn72D(HA-Zn72D), pAM-MLE(Myc-MLE), and pAM-MLE2intWT (a construct containing two introns) were created using an Invitrogen Gateway system. Twenty micrograms of GFP-MSL2 plus 1 μg pCoHygro (Invitrogen) was transfected into S2 cells, using the protocol described in an Invitrogen Drosophila expression system protocol, and selected with 300 μg/ml hygromycin-B. All other plasmids (20 μg each) were transfected with 1 μg pCoBlast (Invitrogen), and cells were selected with 15 μg/ml blasticidin-S HCl.

RNAi screen.

The RNAi library was constructed as described previously (16), and additional information about this library can be found at http://rnai.ucsf.edu. Double-stranded RNAs (dsRNAs) were added to S2 cells (∼15 μg/ml) in 50% conditioned-50% fresh Schneider's medium in a 96-well format. Four days later, cells were transferred to glass-bottom imaging plates (Greiner Bio-One) coated with concanavalin A (0.5 mg/ml) and allowed to settle for 1 h. Cells were then fixed in 1% formaldehyde in 1× phosphate-buffered saline (PBS) for 10 min, and washed three times with 1× PBS, and Fluoromount-G (Southern Biotech) was added. A visual screen was performed to assay for the loss of the GFP-MSL2-coated X chromosome, using a Nikon TE-200 inverted microscope. All dsRNAs produced for subsequent RNAi experiments were produced as described previously (16). The following primers were used to make the dsRNAs: msl1 was made with the primers RNAi1.1-msl1 (GGGCGGGTAATTACCTTTTGGAATTGGA) and RNAi1.2-msl1 (GGGCGGGTGGTGGACTGATGGTTGGCTA); mle (in the RNAi library) was made with the left primer GGGCGGGTTTATGGCTTCGTACTCTAGCACC and the right primer GGGCGGGTAAGTTAAGCCAGTTGTCAACGC; an alternative mle dsRNA was made with primers mleRNAi#2F (GGGCGGGTCCCAAAAATCGCCAGCGG) and mleRNAi#2R (GGGCGGGTCGCGAATGTTGTTCGTCTGC); and the Zn72D3′UTR was made using primers Zn72D3′UTR(s) (GGGCGGGTGCGGCGAGAATAGGTTATATAC) and Zn72D3′UTR(as) (GGGCGGGTCCGCTTCGTTCTAGTATTTGTG).

Immunofluorescence.

S2 cells were allowed to settle on concanavalin A-treated coverslips for 1 h and then fixed in 3.7% formaldehyde-0.1% Triton X-100 in 1× PBS for 5 min, washed twice in 0.1% Triton X-100, and blocked for 30 min in 0.1% Triton X-100-5% goat serum-1× PBS, and primary antibodies were added overnight at 4°C. Polytene chromosomes were prepared following the online protocol (http://www.epigenome-noe.net/researchtools/protocol.php?protid=1). Polytene spreads were stained for 1 h at 37°C for MSL1 and MLE staining and at 4°C overnight for polymerase II (Pol II)-stained chromosomes. The antibodies were used at the following dilutions: rabbit anti-MOF at 1:500, rabbit anti-MSL1 at 1:200, guinea pig anti-MLE at 1:200 (gifts from J. Lucchesi), rabbit anti-GFP (catalog no. ab290-50; Abcam) at 1:750, mouse anti-GFP (catalog no. JL-8; Clontech) at 1:500, mouse anti-HA (catalog no. HA.11; Covance) at 1:200, Pol II antibody H5 and H14 at 1:50 each, and mouse anti-c-Myc (catalog no. sc-40; Santa Cruz Biotechnology) at 1:500. Coverslips were washed twice in 0.1% Triton X-100 and blocked for 5 min, and secondary antibodies (all from Vector Laboratories; used at a dilution of 1:200, except for anti-immunoglobulin M, which was from Jackson Immunologicals and was used at a dilution of 1:1,000) were applied for 45 min at 37°C, followed by 4′,6′-diamidino-2-phenylindole (DAPI) staining for 5 min. All samples were visualized with an Olympus BX60 microscope, and images were collected with a Hamamatsu ORCA-ER digital camera using Openlab 4.0.1 software and assembled with Adobe Photoshop 7.0. Levels were adjusted to enhance contrast.

Western blotting.

S2 cells with or without dsRNA treatment for 6 days were counted, spun down, and lysed in 1× sample buffer (2× sample buffer is a solution of 8.3% glycerol, 1.25% sodium dodecyl sulfate, 0.1 M Tris-HCl, pH 6.7, 0.083 mg/ml bromophenol blue, 50 μl/ml 2-mercaptoethanol) at a concentration of 5 × 10⁴ cells/μl. Samples were boiled for 5 min and spun down at 14,000 rpm at 4°C for 20 min. Fifteen microliters of lysate per lane was loaded onto a 4 to 15% Tris-HCl gradient gel. For Western blotting of larvae, three third-instar larva equivalents in 2× sample buffer were loaded per lane (larvae were homogenized in an Eppendorf tube, boiled, and spun down to remove debris). Gels were transferred to nitrocellulose, blocked with 1% nonfat dry milk-0.05% Tween-1× PBS, and probed overnight at 4°C with guinea pig anti-MLE antibody at a 1:500 dilution, mouse anti-γ-tubulin (catalog no. GTU-88; Sigma) at a 1:1,000 dilution, mouse anti-HP1 (catalog no. C1A9; Developmental Studies Hybridoma Bank) at a 1:2,000 dilution, or chicken anti-Zn72D serum (1:200 dilution). The Zn72D antibody was created against a Zn72D peptide, TYREHLEGQKHKKREASL. Donkey anti-guinea pig Cy3 (1:1,000 dilution), donkey anti-mouse Cy3 (1:500 dilution), and donkey anti-chicken Cy3 (1:500 dilution) (Jackson ImmunoResearch) were each used as a secondary antibody and were detected using a Typhoon 9400 instrument and quantitated using Quantity One software (Bio-Rad). MLE levels were normalized to that of HP1 or γ-tubulin.

Flies.

The 43S2 fly line was a gift from Helena Richardson, and its generation is described in the report by Brumby et al. (7). The mutation was identified by sequencing PCR products along the length of the Zn72D mutant gene. The GFP-Zn72D transgene was generated by using an Invitrogen Gateway system to clone Zn72D into pTGW. This construct was used to create transgenic flies (BestGene, Inc.). pTGW-Zn72D flies were crossed to w118 hsGal4 (III) flies (a gift from Pat O'Farrell), and third-instar larvae were heat shocked for 1 h to induce GFP-Zn72D expression.

qRT-PCR.

Cells were treated with dsRNAs for 5 to 6 days (as described above for S2 cells; and Kc cells were incubated with dsRNA in serum-free medium for 1 h, and then twice the volume of complete Schneider's medium was added). Cells were Trizol extracted, treated with DNase I (Worthington), and phenol-chloroform extracted; RNA samples were quantitated; and 2.5-μg RNA samples were put into a 50-μl reverse transcription (RT) reaction. cDNA was diluted threefold, 2.5 μl was used in 20 μl quantitative (q)PCR assays using SYBR Master Mix (Applied Biosystems), and qPCR was done on an ABI 7300 instrument. Each sample was normalized to rp49 transcript levels and to the sample not treated with dsRNA.

qPCR primers.

The following are the primers used for qPCR. Primers used for msl1 were QF1 (GAACAGGGCACACAAACGA) and QR1 (CCCCTGGGAAGTGCATTC); primers for msl2 were QF1 (GCATCCTTTGGTGCTTGTTC) and QR2 (GCTGCCCTGGAAGATATTGAA); primers for msl3 were QF2 (TGGCAAGCGAAAGGAAAA) and QR2 (GCCCCGGTTTCCCTTTAA); primers for mle (exons 2 to 3; red primers in Fig. 6) were QF1 (CGGAACACGCTAGGAGCTTT) and QR1 (TGAGCGCCGGCACAT); primers for mle (exons 3 to 4; blue primers in Fig. 6) were F1 (GATGAGGTGATTAAGGGTTTGG) and R3 (GAGGAATCTATACGGCTTAAG); the primer for mle (black primer in Fig. 6) was QF1adj (TGGGCCCGGAACACGC); primers for mof were QF3 (CAGGGAGACGGTCATCACA) and QR3 (CGGGATTTTCGCTTATATCGA); primers for Zn72D were QF2 (CGATGATAATCTGGACGATTCG) and QR2 (CGCCTACTGGCTTAATGTTGTC); primers for roX2 were QF1 (TTGCGCCTATGACAATCCTAA) and QR1 (GGCCATCCGAGCTACCTAAA); primers for mRpL16 were QF2 (TCAACACAGCCGGTCTTAAGTAT) and QR2 (GGCTGCTCCACATTCTGGTA); primers for rp49 were QF3 (GCCGTAATTGTCGTTTTTGG) and QR3 (CGAACAGCGCACGGACTA); primers for arm were arm-F (GCTGCTGAACGATGAGGATCA) and arm-R (CCAAAGCGGCTACCATCTGA); primers for CG14804 were CG14804-F (CTGAGCACAAGACGGCAGAG) and CG14804-R (GAGGGTCACGTTCACCTTGC); primers for RPII140 were RPII140-F (CACAATGGCGGCGGTT) and RPII140-R (ACGCAGATGTTCAGGCAGAGT) (51).

FIG. 6.

Zn72D promotes productive splicing at the second intron of the mle transcript. (A, top) The mle gene, which contains five exons (gray boxes) and four introns (black or green horizontal lines), is shown. Poly(A)-1 and poly(A)-2 are red and black vertical lines, respectively. (Bottom) Four mle transcript isoforms differ in their retention of part of intron 2 (isoforms I1 and I2) and their usage of the upstream or downstream poly(A) sites (P1 and P2). The partially retained intron is in green. (B) qRT-PCR in two regions of the mle transcript was performed with and without Zn72D knockdown. Primers detect the mle transcript at the exon 2-exon 3 junction (red; these are the primers used previously, as described in the legend to Fig. 5) and the exon 3-exon 4 junction (blue). The location of these primer sets along the length of mle is depicted in panel A as red and blue arrows, respectively. The forward primer in the red primer set spans the exon 2-exon 3 junction, and this primer set specifically amplifies I1. (C) RT-PCR with wild-type and Zn72D^−/− third-instar larvae, using primers that flank intron 2 of the mle transcript, shown as a black arrow pointing to the right and a red arrow pointing to the left in panel A. Three transcripts were amplified: the unspliced RNA; isoform I1, which decreases in Zn72D mutants; and isoform I2, which increases in Zn72D mutants. (D) Northern analysis with RNA collected from S2 cells either treated or untreated with Zn72D dsRNA. For untreated S2 cells, the top band corresponds to isoform (I2, P2); the middle band corresponds to both isoforms (I1, P2) and (I2, P1); and the lower band corresponds to isoform (I1, P1). Upon Zn72D knockdown, only I2 isoforms remain.

Northern blotting.

Northern blotting was performed using a NorthernMax kit (Ambion) and BrightStar-Plus membranes (Ambion). Ten micrograms of RNA from S2 cells treated with or without Zn72D dsRNA for 6 days was loaded per lane. Northern blots were probed with a labeled DNA probe that was antisense to the full-length MLE cDNA and then exposed to a phosphor screen overnight.

RESULTS

An RNAi screen identifies the zinc finger protein Zn72D, as required for MSL complex localization.

To identify new factors involved in the regulation of dosage compensation in Drosophila, we performed an RNAi screen with a male Drosophila Schneider S2 cell line fly expressing GFP-tagged MSL2, which localizes to the X chromosome (Fig. 1A). Using an RNAi library that consists of 7,216 dsRNAs targeting genes conserved among Drosophila, C. elegans, and mammals (16), we assayed for the loss of GFP-MSL2 on the X chromosome. In addition to identifying the known components of the MSL complex that were represented in the library, we identified two new candidates for the regulation of dosage compensation, Zn72D and Fumble. Upon knockdown of Zn72D or fumble, GFP-MSL2 and MOF were no longer enriched on the X chromosome (Fig. 1A and data not shown), consistent with the disruption of the MSL complex rather than an effect on expression of the MSL-GFP transgene.

FIG. 1.

Zn72D is required for localization of the MSL complex to the X chromosome in Drosophila Schneider S2 cells. (A) S2 cells expressing GFP-MSL2 (green) were either left untreated (top) or treated with Zn72D dsRNA (bottom). Cells were stained with anti-MOF (red) and DAPI (blue) to delineate the nuclei. (B) Wild-type S2 cells are shown either untreated (top) or treated with Zn72D dsRNA (bottom) and stained with anti-MSL1 (green) and anti-MLE (red). (C) Diagram of the domains in the Zn72D protein, identified in a BLAST search. Zn72D has three C2H2 zinc fingers, a DZF domain, two putative nuclear export sequences (NES), and a nuclear localization sequence (NLS). The asterisk indicates the approximate position of the frame shift mutation present in the Zn72D mutant flies.

Fumble is homologous to the mammalian pantothenate kinase, an enzyme required for acetyl coenzyme A synthesis (47). When S2 cells were treated with the histone deacetylase inhibitor TSA, levels of acetylation on histone H4 lysine 16 increased. When fumble was knocked down before cell treatment with TSA, there was no increase in the level of acetylation on histone H4 lysine 16, consistent with the disruption of acetyl coenzyme A production (data not shown). This implied that Fumble indirectly regulates MSL complex formation, as the acetylation activity of MOF is required for MSL complex localization (21). We focused our studies on the second candidate, Zn72D. When Zn72D was knocked down in wild-type S2 cells, MLE, MSL1, and roX2 were no longer enriched on the X chromosome (Fig. 1B and data not shown), confirming that Zn72D is necessary for the correct localization of the MSL complex. Zn72D has three C2H2 zinc finger domains of the type that are found in the U1C snRNP and the DZF domain, which is a domain found in some proteins containing a C2H2 zinc finger or dsRNA binding domains (Fig. 1C).

Zn72D is required in vivo for MSL complex localization and dosage compensation.

Next, we determined whether Zn72D was required for proper MSL complex localization to the X chromosome in male flies. In the 43S2 line (7), a single base pair insertion causes a frame shift located before the first zinc finger domain of Zn72D (Fig. 1C). Polytene chromosomes isolated from third-instar larvae were stained with antibodies to MSL1 and MLE. Wild-type males showed the complete banding pattern of the MSL complex on the X chromosome, while Zn72D mutant males had significantly fewer bands (Fig. 2). These data demonstrate that Zn72D is required for the correct localization of the MSL complex to the X chromosome in vivo.

FIG. 2.

Drosophila males require Zn72D for the complete localization of the MSL complex across the X chromosome. Immunofluorescence was performed with polytene chromosomes isolated from male third-instar larvae to detect MSL1 (top row, green) and MLE (second row, red) on the X chromosome in a wild-type (left column) and a Zn72D^−/− strain of larvae (center and right columns). DAPI (third row, blue) stains all chromosomes. The bottom row contains the merged images of MSL1, MLE, and DAPI staining.

To determine if Zn72D is required for proper dosage compensation, we performed qRT-PCR to assess steady-state levels of mRNA from X-linked genes when either mle or Zn72D was knocked down in S2 cells or in female Kc167 cells. We assayed three X-linked genes, arm, CG14804, and mRpL16, which are regulated by the MSL complex in male cells (25, 51). Levels were normalized to the autosomal rp49 gene. When mle was knocked down in S2 cells, we observed the expected decrease in X-linked gene expression and no effect on expression of the autosomal gene RpII140 (Fig. 3A). When Zn72D was knocked down in S2 cells, there was a similar decrease in X-linked gene expression (Fig. 3A). qRT-PCR performed with male Zn72D^−/− larvae also showed a decrease in levels of arm and CG14804 compared to that of the wild-type larvae; levels of mRpL16 were not affected in Zn72D^−/− larvae (data not shown). Upon the knockdown of mle or Zn72D in Kc167 cells, there was no change in the expression levels of arm, CG14804, or mRpL16, indicating that, like MLE, Zn72D regulates X-linked gene expression solely in males (Fig. 3B). However, all Zn72D mutants pupated but failed to hatch (7; also data not shown), indicating that Zn72D is required in females as well as in males. In contrast, the mutations in MSL complex proteins result in lethality at the larval or pupal stage of development only for males (6, 17, 26). This suggests that Zn72D has a broader function than its role in dosage compensation.

FIG. 3.

Zn72D is necessary for proper X-linked gene expression in males but not in females. (A) qRT-PCR for the X-linked genes arm, CG14804, and mRpL16 and the autosomal gene RpII140 was done after S2 cells were either untreated (gray bars), treated with mle dsRNA (white bars), or treated with Zn72D dsRNA (black bars). Samples were normalized first to the expression of rp49 and then to that of the untreated sample, setting the untreated to 1. The bar represents the average of two independent experiments, with qPCR performed in triplicate, and the error was determined using standard error propagation methods. (B) Female Kc cells were treated with dsRNAs, and qRT-PCR was performed as described in the legend to panel A.

Zn72D is associated with transcriptional puffs.

In order to gain insight into the role of Zn72D, we determined its cellular distribution. We expressed hemagglutinin (HA)-tagged Zn72D in S2 cells and performed immunofluorescence assays with an anti-HA antibody. Zn72D was localized predominantly in the nucleus (Fig. 4A). Unlike the MSL complex, Zn72D was not enriched on the X chromosome, consistent with a function for Zn72D outside of its role in dosage compensation. To determine if Zn72D was associated with chromosomes, we generated transgenic flies that express Zn72D fused to GFP under the control of the Gal4 upstream activation sequence. The GFP-Zn72D transgene encodes a functional Zn72D protein, as it rescues the MSL complex localization in S2 cells when endogenous Zn72D is knocked down with dsRNA targeting the 3′ untranscribed region, not present in the transgene (data not shown). The GFP-Zn72D transgenic flies were crossed to flies expressing heat-shock-inducible Gal4. Polytene chromosomes from heat-shocked third-instar larvae were isolated and stained with an anti-GFP antibody. GFP-Zn72D was localized to all chromosomes in both sexes, with no notable enrichment on the X chromosome. Instead, Zn72D appeared to be enriched on transcriptional puffs (Fig. 4B), which was confirmed by the colocalization of GFP-Zn72D and the elongating RNA Pol II phosphorylated on serine 2 of the carboxy-terminal domain (Pol IIo-Ser2) (Fig. 4C). Endogenous Zn72D also localizes to chromatin under normal and heat shock conditions (data not shown). This pattern of distribution suggests that Zn72D may have a role in regulating transcription, RNA splicing, processing, or transport, as factors that regulate these processes are all present on transcriptional puffs (3, 11, 31, 48, 50, 53).

FIG. 4.

Zn72D is nuclear, is enriched at transcriptional puffs, and colocalizes with elongating RNA Pol II. (A) S2 cells expressing HA-tagged Zn72D stained with anti-HA (red) and DAPI (blue). (B) Drosophila larvae were heat shocked to induce Gal4 to activate GFP-Zn72D expression, and polytene chromosomes were isolated and stained with anti-GFP (green) and DAPI (blue). In contrast to the MSL proteins, which are found only on the X chromosome, GFP-Zn72D is found on all chromosomes. (C) Polytene chromosomes from GFP-Zn72D-induced larvae were stained with anti-GFP (green) and anti-Pol IIo-Ser2 (red) and DAPI (blue).

Zn72D is required for normal levels of mle RNA and protein.

To determine if Zn72D functions in dosage compensation through the regulation of transcription or the stability of one or more of the MSL complex mRNAs, we used qRT-PCR to analyze mRNA levels of the components of the MSL complex when either Zn72D or mle as a control was knocked down in S2 cells. Upon mle knockdown, the levels of mle mRNA and protein were decreased (Fig. 5A and C), levels of msl3 remained unchanged, and levels of msl1 and msl2 mRNAs increased slightly (Fig. 5A). When Zn72D was knocked down (Fig. 5A and C), the levels of mle mRNA were reduced about 16-fold, while levels of msl1, msl2, and msl3 mRNA were unaltered (Fig. 5A). Levels of mof and roX2 decreased slightly when Zn72D or mle was knocked down, which was expected since mof is X linked and roX2 is destabilized in the absence of a functional MSL complex (Fig. 5A) (40). Additionally, MLE regulates the transcription of roX2 (4, 34); therefore, the decrease in the levels of mle when Zn72D is knocked down likely contributes to reduced levels of roX2. Supplementing the result with S2 cells, mle and roX2 transcripts were lower in male Zn72D^−/− third-instar larvae than in wild-type larvae (data not shown).

FIG. 5.

Zn72D is required for proper mle mRNA and protein levels. (A) S2 cells were treated with either (left) mle or (right) Zn72D dsRNA and qRT-PCR was performed, as described in the legend to Fig. 3 and Materials and Methods, to assay the transcript levels of msl1 (orange), msl2 (mauve), msl3 (green), mof (yellow), mle (red), roX2 (light purple), and Zn72D (blue). Samples were normalized first to rp49 and then to the untreated sample, setting untreated to 1. (B) Kc167 cells were treated with (left) mle or (right) Zn72D dsRNA as above and the levels of mle (red) and Zn72D (blue) transcripts were assayed by qRT-PCR. (C) Western blots of (left and middle) extracts from S2 cells that had been treated with either no dsRNA, msl1, mle, or Zn72D dsRNAs or (right) extracts from wild-type and Zn72D^−/− third-instar larvae. Levels of MLE and Zn72D were normalized to the amount of γ-tubulin or HP1.

MLE is expressed in both males and females. If the role of Zn72D is to regulate mle RNA levels, we expected that the level of mle mRNA would decrease upon Zn72D knockdown in females as well as in males. To test this, the level of mle transcripts was assayed when Zn72D was knocked down in the female Kc167 cell line. The level of mle mRNA did decrease, indicating that Zn72D is required for mle expression in both sexes (Fig. 5B). Therefore, for males lacking Zn72D, the drop in mle mRNA levels might in turn affect MSL complex localization due to a decrease in the amount of MLE protein. This was indeed the case: the levels of MLE protein were reduced to less than one-fifth of the wild-type levels when Zn72D was knocked down in S2 cells and in Zn72D mutant larvae (Fig. 5C). MLE protein levels did not decrease when msl1 was knocked down, indicating that MLE protein is relatively stable even when the MSL complex does not form. Together, these results suggest that Zn72D is required for MLE expression by affecting the amount of mle mRNA that is produced. This function may be either direct or indirect, as we were unable to determine, by RNA immunoprecipitation, if Zn72D binds directly to mle RNA.

Zn72D promotes productive splicing of mle transcripts.

As we attempted to determine the fate of the mle mRNA, it became clear that the entire mle transcript was not being turned over to the degree indicated in our original qRT-PCR experiments. Other regions of the mle transcript did not show the same dramatic decrease as the region assayed in the original qRT-PCR experiment (Fig. 6B). mle is alternatively spliced at the exon 2-exon 3 splice junction (33). One splice isoform produces an mRNA that directs production of full-length MLE (isoform 1, Fig. 6A). The other isoform employs a downstream splice donor, resulting in the production of an mRNA that is 342 nucleotides longer and contains several in-frame stop codons (isoform 2, Fig. 6A). Isoform 2 would direct the translation of a truncated MLE protein of 226 amino acids, instead of the full-length protein of 1,293 amino acids. The primers we used to quantify mle mRNA levels detected only isoform 1, as the forward primer spans the exon 2-exon 3 junction (Fig. 6A). To determine if the loss of Zn72D affected the abundance of both isoforms, we examined the exon 2-exon 3 junction by using primers that flank the intron and therefore amplify both isoforms plus unspliced RNA. In Zn72D^−/− larvae and in Zn72D knockdown cells, the amount of isoform 1 decreased while the amount of isoform 2 increased (Fig. 6C and data not shown), suggesting that Zn72D regulates the relative amounts of the two mle isoforms rather than the absolute amount of mle mRNA. We performed Northern blotting as a second assay to measure the relative amounts of these two mle splice isoforms. Northern blotting indicated the presence of at least three mle transcripts (Fig. 6D), consistent with two poly(A) sites which are approximately 350 nucleotides apart (33) and the two splice sites that are also about 350 nucleotides apart. The identity of these transcripts was confirmed by using a probe that hybridizes to the region between the two poly(A) sites (data not shown). Upon Zn72D knockdown, the transcripts that could be attributed to splice isoform 1 decreased in abundance, while those that could be attributed to isoform 2 increased in abundance (Fig. 6D), indicating that Zn72D promotes the splicing of mle such that the full-length MLE protein is produced.

If the role of Zn72D in dosage compensation is to promote the productive splicing pattern of mle, then when an mle transcript that lacks introns is expressed, it should circumvent the requirement for Zn72D. To test this, we created a stable S2 cell line expressing the mle cDNA sequence tagged with Myc. Overexpressed Myc-MLE localized throughout the nucleus (Fig. 7A), consistent with the previous observation that overexpressed MLE localizes to all chromosomes in flies (46). MSL1 was still localized to the X chromosome in these Myc-MLE-overexpressing cells, indicating that the overexpressed protein did not interfere with MSL1 localization. When mle was knocked down in these cells, Myc-MLE expression was reduced, and MSL1 no longer appeared in a pattern consistent with enrichment on the X chromosome, indicating that even when MLE is overexpressed, the formation of the MSL complex is still sensitive to changes in the amounts of MLE. When Zn72D was knocked down, Myc-MLE levels did not decrease, and MSL1 localization was unaltered, indicating that the expression of the mle gene cDNA rescues the MSL localization defect observed when Zn72D is knocked down (Fig. 7A).

FIG. 7.

An mle cDNA transgene rescues Zn72D knockdown, but an intron-containing mle transgene does not. (A) S2 cells expressing Myc-tagged MLE cDNA were either untreated (top) or treated with mle (middle) or Zn72D dsRNAs (bottom) and stained with anti-MSL1 (green) and anti-Myc (red). Arrowheads (bottom) indicate cells that express Myc-MLE. (B) S2 cells expressing a Myc-tagged MLE transgene that contains the first two introns of mle were either untreated (top) or treated with mle (middle) or Zn72D dsRNAs (bottom) and stained with anti-MSL1 (green) and anti-Myc (red). (C) Western blots of S2 cells expressing transgenic Myc-MLE from the cDNA construct (left three lanes) or Myc-MLE containing two introns (right three lanes) and untreated or treated with mle or Zn72D dsRNAs, as indicated. Levels of full-length (indicated above the gel) and truncated (indicated below the gel) Myc-MLE were normalized to γ-tubulin levels and that of the sample not treated with dsRNA.

We next added the first two introns back into the Myc-mle gene cDNA, expressed the transgene in S2 cells, and knocked down either mle or Zn72D. Upon mle knockdown, Myc-MLE levels were decreased, and MSL1 was no longer localized on the X chromosome (Fig. 7B). In contrast to the Myc-mle cDNA, the Myc-mle transgene containing introns 1 and 2 was subject to Zn72D regulation: when Zn72D was knocked down, the intron-containing transgene did not rescue MSL1 localization (Fig. 7B). The level of protein expressed from the Myc-mle cDNA did not change significantly upon Zn72D knockdown, while the level of the full-length protein expressed from the intron-containing transgene decreased in the absence of Zn72D (Fig. 7B). In the intron-containing transgene line, an additional protein, of the approximate molecular weight predicted for a Myc-tagged protein that would be produced from isoform 2, was detected using Myc and MLE antibodies (Fig. 7C and data not shown). It is possible that this shorter product may account for the cytosolic Myc-tagged protein that was specific to the mle transgenic line that contained introns (Fig. 7B). While the increase observed in isoform 2 upon Zn72D knockdown led us to anticipate an increase in the abundance of the shorter protein product, none was observed. If the smaller Myc-MLE protein is unstable, then steady-state levels of the protein may not accurately reflect the amount of isoform 2. We were unable to detect a truncated MLE protein upon Zn72D knockdown in wild-type cells (data not shown), suggesting that the shorter protein produced from isoform 2 may be very unstable. In combination, our data are consistent with a role for Zn72D in promoting the usage of the correct splice site at intron 2 of the mle mRNA.

DISCUSSION

In an RNAi screen for proteins that regulate fly dosage compensation, we identified the zinc finger-containing protein Zn72D. Zn27D mediates its role in dosage compensation by promoting the proper splicing of the mle transcript, thus affecting MLE protein levels and the localization of the MSL complex to the X chromosome. Zn72D, either directly or indirectly, promotes productive splicing of mle in male and female cells and likely regulates additional targets, as both male and female Zn72D mutants pupate but do not hatch. Zn72D is enriched on transcriptional puffs and colocalizes with elongating RNA Pol IIo-Ser2 on polytene chromosomes, consistent with a general role in splicing regulation.

Zn72D is a zinc finger splicing factor.

Zn72D contains three zinc fingers that are similar to the zinc finger in the U1 small nuclear ribonucleoprotein C (U1C). U1C binds the 5′ splice site (5′ss) in a sequence-specific manner (13), and mutations in the zinc finger of U1C bypass the need for the DExH/D box helicase Prp28 to unwind the 5′ss and the U1 snRNA base pairing (9), suggesting that U1C stabilizes the 5′ss-U1 snRNA commitment complex. The similarity between the zinc fingers of Zn72D and those of U1C suggests the possibility that, like U1C, Zn72D might promote splicing through the regulation of the commitment complex formation between the 5′ss in pre-mRNAs and the U1 snRNA. Alternatively, Zn72D may recognize a 5′ss and inhibit its use in the splicing reaction, forcing the use of an alternative splice site. We were unable to detect an interaction between Zn72D and mle RNA, suggesting that Zn72D might instead play a more indirect role in 5′ss selection during mle splicing.

mle RNA is not subject to NMD.

The splice isoform that is upregulated upon knockdown of Zn72D includes several in-frame premature translation termination codons (PTCs). PTCs can signal nonsense-mediated mRNA decay (NMD). The PTC-containing mle transcript was easily detected, by both Northern blotting and RT-PCR. In addition, this transcript was readily detected in cytoplasmic extracts by RT-PCR (data not shown), indicating that it is exported from the nucleus. When we overexpressed the intron-containing MLE transgene, a smaller Myc-tagged MLE protein product was expressed in addition to the full-length tagged MLE, suggesting that a protein can be produced from the PTC-containing spliced transcript. The increase in the levels of the PTC-containing mle transcript upon Zn72D knockdown is unlikely to be a consequence of the disruption of NMD, since there was a compensatory decrease in the other mle splice isoform and no increase in overall mle levels. This suggests that mle transcripts are not significantly affected by NMD and that Zn72D is required for productive splicing of mle. It is not yet clear how premature termination codon-containing transcripts are recognized as aberrant and are targeted for NMD in Drosophila. Unlike mammals and yeast, in which these transcripts are recognized based on the presence of exon junction complexes on the transcript downstream from the premature termination codon, Drosophila mRNAs do not require the exon junction complexes for NMD (18). As the PTC-containing mle transcript apparently lacks the Drosophila NMD signal, mle may be a transcript that is useful for investigating the NMD pathway in flies, as one can ask why the PTC-containing mle transcript avoids the fate of other transcripts that contain PTCs.

Does Zn72D link splicing to localization and translation?

The closest mammalian homologue of Zn72D is ZFR; these proteins are 42% identical and 55% similar, and both proteins contain the same domain structure, with three zinc fingers followed by a DZF domain. ZFR was shown to be associated with chromosomes during meiosis, and it is required for normal development of the mouse embryo; homozygous Zfr mutants have gastrulation defects and die between 8 and 9 days of gestation (38, 39). Human ZFR was identified in a screen for Staufen2-interacting proteins and was implicated in nucleocytoplasmic shuttling of Staufen2 (14). Staufen1 and Staufen2 are dsRNA binding proteins that are implicated in the transport of mRNAs from the nucleus to cytoplasmic RNA granules (reviewed in references 30 and 42), which are clusters of ribosomes, translation factors, and mRNAs presumed to be incompetent for translation as they lack eIF4E, 4G, and tRNAs (32). Like its human counterpart, the Drosophila Staufen protein also plays a role in RNA localization: it is required for localization of the oskar mRNP complex to the posterior and of the bicoid mRNA to the anterior of the oocyte (reviewed in reference 27).

If Zn72D, like ZFR, is involved in localizing mRNP complexes within cells, this suggests it may be involved in linking splicing and mRNA localization. There is a precedent for the connection between splicing and RNA localization: splicing at a specific exon-exon junction of oskar is necessary for the proper localization of the oskar mRNA to the posterior (24), and members of the exon junction complex are also necessary for its localization (23, 35). Zn72D contains two putative nuclear export signals and a nuclear localization sequence, suggesting that Zn72D might participate in nucleocytoplasmic shuttling events. We did not observe any affect on localization for the dosage compensation complex upon RNAi-mediated knockdown of staufen (data not shown), so any potential homologous role of Zn72D in shuttling is likely mediated by other proteins.