The expression of Drosophila melanogaster Hox gene Ultrabithorax is not overtly regulated by the intronic long noncoding RNA lncRNA:PS4 in a wild-type genetic background

Abstract Long noncoding RNAs (lncRNAs) have been implicated in a variety of processes in development, differentiation, and disease. In Drosophila melanogaster, the bithorax Hox cluster contains three Hox genes [Ultrabithorax (Ubx), abdominal-A, and Abdominal-B], along with a number of lncRNAs, most with unknown functions. Here, we investigated the function of a lncRNA, lncRNA:PS4 that originates in the second intron of Ubx and is transcribed in the antisense orientation to Ubx. The expression pattern of lncRNA:PS4 is complementary to Ubx in the thoracic primordia, and the lncRNA:PS4 coding region overlaps the location of the large insertion that causes the dominant homeotic mutation Contrabithorax-1 (UbxCbx-1), which partially transforms Drosophila wings into halteres via ectopic activation of Ubx. This led us to investigate the potential role of this lncRNA in regulation of Ubx expression. The UbxCbx-1 mutation dramatically changes the pattern of lncRNA:PS4, eliminating the expression of most lncRNA:PS4 sequences from parasegment 4 (where Ubx protein is normally absent) and ectopically activating lncRNA:PS4 at high levels in the abdomen (where Ubx is normally expressed). These changes, however, did not lead to changes in the Ubx embryonic transcription pattern. Targeted deletion of the two promoters of lncRNA:PS4 did not result in the change of Ubx expression in the embryos. In the genetic background of a UbxCbx-1 mutation, the lncRNA:PS4 mutation does slightly enhance the ectopic activation of Ubx protein expression in wing discs and also slightly enhances the wing phenotype seen in UbxCbx-1 heterozygotes.


Introduction
Long noncoding RNAs (lncRNAs) are a ubiquitous group of RNA transcripts that are over 200 nucleotides long, do not code for potential proteins or peptides longer than 100 amino acids, and are transcribed from a variety of genomic locations, including enhancers, promoters, and intergenic regions (Rinn and Chang 2012;Yamamoto and Saitoh 2019). Such lncRNAs have diverse roles in development, differentiation, and disease (Yao et al. 2019). lncRNAs are primarily localized in the nucleus where they can regulate gene expression by a variety of mechanisms, including interactions with enzymatic complexes that modify and remodel chromatin, direct transcriptional regulation via interaction with Preinitiation Complex, and regulation of alternative splicing (Squillaro et al. 2020). One of the classic examples of regulatory lncRNAs is Xist which controls female X chromosome inactivation in mammals (Yao et al. 2019). Another example is the partial derepression of human and mouse HoxD genes by mutations in the lncRNA HOTAIR (Rinn and Chang 2012;Li et al. 2013), although other studies that ablated Hotair function in mice reported little or no changes in Hox gene embryonic patterning function (Schorderet and Duboule 2011;Lai et al. 2015;Amâ ndio et al. 2016;Yao et al. 2019).
Numerous lncRNAs have been detected within the clusters of Hox genes-evolutionarily conserved complexes of genes coding for homeobox-containing transcriptional factors that control axial patterning in bilateral animals (McGinnis and Krumlauf 1992;Kumar and Krumlauf 2016). These lncRNAs are transcribed from both coding and noncoding strands of the Hox gene clusters and can be found within Hox introns, in intergenic regions, or flanking the Hox gene clusters (Kumar and Krumlauf 2016). They can function both in cis, by regulating the adjacent Hox genes and in trans, by regulating the expression of Hox genes in other Hox clusters or by regulating or modulating the activity of non-Hox genes (Kumar and Krumlauf 2016).
The Drosophila bithorax Hox cluster (BX-C) contains three Hox genes: Ultrabithorax (Ubx), abdominal-A, and Abdominal-B, as well as many noncoding RNAs (Lipshitz et al. 1987;Sanchez-Herrero and Akam 1989;Cumberledge et al., 1990;Bae et al., 2002;Pease et al. 2013;Schor et al. 2018). Some of these noncoding RNAs are processed into microRNAs (miRNAs) that can regulate Hox expression (reviewed in Garaulet and Lai 2015), but most have no known function. An example of such a non-miRNA-containing lncRNA is bxd, which is encoded in upstream regulatory regions of the Ubx gene. Functionally, the bxd RNA has been reported to repress early transcription of Ubx in cis, perhaps by readthrough through the Ubx promoter (Petruk et al. 2006); however, the ablation of bxd RNA revealed that the early transcription repression of Ubx transcription in the bxd expressing region was very transient and no embryonic or adult Ubx mutant phenotypes were detected from bxd RNA elimination (Pease et al. 2013).
Contrabithorax is a dominant mutation that results in a transformation of some adult structures of the second thoracic segment (T2) into those of the third thoracic segment (T3), e.g., transforming wings into halteres (Lewis 1963). The Ubx Cbx-1 allele results from an insertion of 17 kb from a Ubx upstream regulatory element into the second intron of Ubx and is the only known Contrabithorax allele that maps to that region (Bender et al. 1983;Lewis 1955). This inverted insertion contains parts of two Ubx upstream regulatory regions: 16 kb of the pbx region and 1 kb of bxd.
In Contrabithorax mutants, Ubx transcript and protein expression are abnormally activated in the wing discs, primarily in the posterior compartment of the disc, which results in a partial transformation of wing toward haltere morphology (White and Akam 1985); however, the mechanisms by which Cbx mutations ectopi-

Materials and methods
Determination of the extent of lncRNA: PS4 transcription unit We used 5 0 and 3 0 RACE [5 0 /3 0 RACE kit, second generation (Roche)] to determine the start and the end of lncRNA:PS4 transcription. The extent of the lncRNA:PS4 transcription unit was determined as follows: the region between the lncRNA:PS4 promoter and Ubx transcription start site was virtually subdivided into regions of about 1 kb long. These regions were cloned by PCR amplification from the genomic DNA of w 1118 flies. Antisense RNA probes were generated using in vitro transcription and the presence of transcripts from each region was assessed using wholemount in situ hybridization. To determine if any of the transcribed regions were spliced, RT-PCR of 900-1100 bp amplicons spanning the entire predicted transcript was produced, followed by sequencing. For RT-PCR template generation, total RNA was extracted with RNAeasy mini kit (Qiagen) and treated with Turbo DNA-free kit (Ambion) to remove DNA.

Determination of lncRNA: PS4 expression pattern
Whole-mount in situ hybridization was performed as previously described (Kosman et al. 2004). The Ubx cDNA probe was generated from the BSUbx1a plasmid (Harding et al. 1985). Another probe to detect Ubx nascent transcripts was prepared with a pBluescript plasmid clone that included the 5 0 -most 1.5 kb of the first intron of Ubx. The wingless probe was generated from a 3 kb clone in pBluescript plasmid (a gift from B. Cohen). lncRNA:PS4 transcription was detected via the M probe (upstream of Ubx Cbx-1 insertion), or the X probe (downstream of Ubx Cbx-1 insertion, Figure 1B). Images were obtained using Leica SP4, Leica SP5, and Leica SP8 fluorescent confocal microscopes and processed in Adobe Photoshop software.
CRISPR/Cas9-generated mutations in lncRNA:PS4 promoters To generate a targeted deletion of the sequences encompassing both putative promoters of lncRNA:PS4, two gRNA sequences were designed using CRISPR Optimal Target Finder (Gratz et al. 2014) and cloned into pCFD4-U6:1_U6:3tandem-gRNAs (Addgene plasmid No. 49411), using the protocol described in Port et al. (2014). The first guide RNA sequence targeting the site 182 bp upstream of the "upstream" promoter was placed under the control of the U6:1 promoter in pCFD4 vector and its sequence was GGAGTAAATTTATCTGGCTCT (in the genome, this sequence would be followed by a 3 0 PAM: CGG). The guide RNA sequence targeting the "downstream" promoter was placed under the control of the U6:3 promoter and its sequence was GGTTCATTTCATTTGCCCAA (in the genome, this sequence would be followed by a 3 0 PAM: CGG). gRNA cloning was verified by sequencing.
Transgenic lines carrying the gRNA construct were generated by BestGene Inc (Chino Hills, CA, USA). Males of this gRNA line [genotype (y1sc*v1; PfCaryP-vþgattP40/CyO] were crossed to females carrying Vasa-Cas9 transgene either on the X chromosome (Bloomington stock No. 52669) or on the third chromosome (Bloomington stock No. 51324) and the progeny of these crosses was back-crossed. To screen for the presence of the desired deletions, we PCR-amplified the region surrounding the lncRNA:PS4 promoters from whole-fly genomic DNA, followed by sequencing of the PCR fragments. One line of transgenic flies was isolated from the cross to a female with a Vasa-Cas9 transgene on the third chromosome. This line has a 324 bp deletion that eliminates both the upstream promoter and the downstream promoter of lncRNA:PS4. Homozygotes for this deletion chromosome survive to adulthood at 5% the expected frequency, although this effect on viability does not map to the 324 bp deletion or to the Ubx region, as the chromosome with the small deletion is fully viable when crossed to a chromosome bearing the Ubx 109 mutation [Df(3R)Ubx 109] , which deletes the entire Ubx locus.

Results
Characterization of lncRNA:PS4 transcription unit lncRNA:PS4 was identified by Pease and colleagues in a systematic survey of noncoding RNAs originating from the Ubx gene and its upstream regulatory elements (Pease et al.. 2013). Pease and colleagues characterized the lncRNA:PS4 transcription unit as spanning $15 kb and including nonprotein coding sequences from the first and the second Ubx introns (Pease et al. 2013). Here, we characterized the exact boundaries of the lncRNA:PS4 transcription unit, using 5 0 and 3 0 RACE, followed by verification using fluorescent in situ hybridization with $1 kb probes that mapped throughout the lncRNA:PS4 region ( Figure 1).
Using 5 0 RACE, we determined that lncRNA:PS4 has two transcription start sites (P1 and P2), separated by 116 bp and located 14,299 and 14,182 bp downstream of the transcription start site of Ubx, respectively. Promoter site 2 is preceded by a good match, TCATTT, to the Drosophila InR promoter consensus sequence of TCAGTY (Ngoc et al. 2019, Supplementary Figure S1). Promoter 2 (P2, Figure 1) also includes a putative DPE promoter sequence (Supplementary Figure S1). Promoter 1 (P1) contains a weaker match, TCACTG, to the InR promoter consensus and no matches to a DPE consensus sequence (Supplementary Figure S1). In our 5 0 RACE reactions, the genomic coordinate for the first 5 0 nucleotide of the P1 transcripts is 16,720,127, and for P2 the initiation site maps to nucleotide 16,720,245 (Drosophila melanogaster genome release r6.37; Larkin et al. 2021). The 3 0 RACE results indicated that lncRNA:PS4 transcripts are polyadenylated and located the 3 0 end of lncRNA:PS4 transcripts to nucleotide 16,729,960 on Chromosome 3R (D. melanogaster genome release r6.37; Larkin et al. 2021). The predicted length of lncRNA:PS4 transcription unit was 9833 bp for transcripts starting from promoter P1 and 9715 bp for transcripts starting from promoter P2.
To determine if any of the sections of the predicted transcript were spliced, we performed RT-PCR of 900-1100 bp amplicons spanning the entire predicted transcript, followed by sequencing. All RNA-generated sequences were the same size and sequence as genomic DNA, indicating that the lncRNA:PS4 transcript contains no introns. The only potential protein coding region identified by PhyloCSF in the lncRNA:PS4 transcript could in theory encode a 39 amino acid peptide, MLQMPTVKTKVPCCHFMNCF CSVGKSTTLTLATHKVLPS (Lin et al. 2011). We were unable to detect the conservation of this protein coding region in other Drosophila species. No palindromic miRNA precursor sequences were identified in the lncRNA:PS4 transcription unit. In wild-type embryos lncRNA:PS4 transcripts (green) originate from the second intron of the Ubx gene, driven by two alternative promoters (P1 and P2). lncRNA:PS4 is transcribed in antisense direction relative to Ubx transcripts, contains no introns, and is terminated in the first intron of Ubx. The lncRNA:PS4 promoter deletion eliminates both P1 and P2 promoters. (B) In Ubx Cbx-1 mutants, a 17 kb insertion that contains pbx and bxd regulatory elements maps $1.1 kb downstream of lncRNA:PS4 transcription start sites, and produces an additional, new hybrid transcript (blue). Note that the Ubx Cbx-1 hybrid transcript terminates closer to the transcription start of Ubx than the wild-type lncRNA:PS4 transcript. Locations of sequences in the in situ hybridization probes M, X, and first intron are indicated. and posterior borders, respectively. Here, we show that lncRNA:PS4 RNA is first detected by in situ hybridization during stage 5 of Drosophila embryonic development (Figure 2A). The expression pattern of lncRNA:PS4 between stages 6 and 12 was previously characterized (Pease et al. 2013). During stages 6 and 7, the lncRNA:PS4 pattern of expression resolves into a broad and strong anterior domain and weaker posterior stripes ( Figure 2B and Supplementary Figure S2). Costaining with a wingless probe at stage 6 allowed us to map the broad anterior domain of expression to the posterior part of parasegment 4 and the entirety of parasegment 5 (segment T2 and the anterior compartment of T3; Supplementary Figure S3), thus abutting the anterior boundary of Ubx expression. lncRNA:PS4 is excluded from the ventral mesoderm primordia at this stage (Pease et al. 2013).
We were also interested in precisely mapping the lncRNA:PS4 transcription domain relative to that of Ubx, in whose intron lncRNA:PS4 resides. Throughout stages 6-10, Ubx and lncRNA:PS4 transcription domains were largely nonoverlapping; however, in a few nuclei, both lncRNA:PS4 and Ubx transcription were detected in parasegments 5 through 12 (Supplementary Figures  S2 and S3). High-resolution imaging at stage 7 (Figure 2, B-E) shows that in the domain of strong Ubx transcription, lncRNA:PS4 transcripts are absent from almost all Ubx transcribing nuclei (Figure 2, C-E).
To better understand the cellular localization of lncRNA:PS4 transcripts, we performed high-resolution imaging of stage 6 embryos after in situ hybridization with lncRNA:PS4 and Ubx probes and costaining with lamin antibodies that outline the nuclear envelope (Figure 2, F and G). lncRNA:PS4 signals are localized almost exclusively in the nuclear compartment ( Figure 2F). There is a prominent signal corresponding to lncRNA:PS4 nascent transcripts ( Figure 2F, arrowheads), as well as other small lncRNA:PS4 nuclear RNA signal "specks" that are not associated with the lncRNA:PS4 locus, presumably originating from polyadenylated transcripts that accumulate in that compartment ( Figure 2B, arrows). Ubx, which was detected with a probe against its first intron, was mostly observed as the nascent transcripts being made from Ubx loci on the two homologous chromosomes ( Figure 2G, arrowheads).
Using in situ hybridization, we made multiple attempts to detect spatially localized transcription from lncRNA:PS4 in wing, haltere and leg discs, using our most sensitive method of tyramide amplification. We never observed a localized signal within discs, or between different discs. The positive controls were in situ hybridizations with engrailed and Ubx cDNA probes, which had easily detectable localized signals. When using quantitative RT-PCR with gene-specific primers covering region X ( Figure 1B), we did detect low-level transcripts from this region in the discs we tested (wing, leg, and haltere; Supplementary Table S1). The abundance of these low-level transcripts was slightly higher in haltere and leg discs than in wing discs, the opposite of what we expected from the embryonic expression pattern of lncRNA:PS4. The positive control was Ubx expression levels using quantitative RT-PCR on RNA from the same discs.
lncRNA:PS4 location relative to Ubx Cbx-1 insertion In order to determine the exact location of Ubx Cbx-1 insertion in the lncRNA:PS4 region, we performed genomic PCR amplification followed by sequencing analysis of the approximate region of Ubx Cbx-1 insertion in Ubx Cbx-1 /T(2; 3)ap Xa , ap Xa flies (Bloomington). This allowed us to map the Ubx Cbx-1 insertion location to position 16,721,339 of Chromosome 3R (D. melanogaster genome release r6.37; Larkin et al. 2021). Thus, the Ubx Cbx-1 insertion is located within the lncRNA:PS4 transcribed region, 1.06 kb downstream of the lncRNA:PS4 P2 transcription start site. The effect of the Ubx Cbx-1 insertion on transcription of the lncRNA:S4 region The Ubx Cbx-1 insertion profoundly changes the extent and the pattern of lncRNA:PS4 embryonic transcription (Figure 3, A-C). The insertion splits the lncRNA:PS4 transcription unit and results in two separate transcripts, as shown schematically in Figure 1B: a transcript containing $1 kb of 5 0 lncRNA:PS4 sequences fused to an unknown amount of Cbx 1 insertion sequence, and another transcript initiated within Ubx Cbx-1 insertion that is fused to the 3 0 most 8.6 kb of lncRNA:PS4. The homozygous Ubx Cbx-1 mutant embryo shown in Figure 3, A-C demonstrates the different expression patterns of these two transcripts. In this in situ hybridization, lncRNA:PS4 transcription was imaged with two probes ( Figure 1B): probe M contains 1.4 kb of sequence upstream and 0.4 kb downstream of the Ubx Cbx-1 insertion ( Figure 3B) and probe X, located downstream of the insertion, thus detecting the transcription of the transcript that initiates within the Ubx Cbx-1 insertion and has 3 0 sequences of lncRNA:PS4 ( Figure 3A). To simplify, we will refer to this transcript as the Cbx hybrid transcript.
The strongest domain of the Cbx hybrid expression was posterior to parasegment 6 and its expression extended uniformly to the posterior boundary of parasegment 12 ( Figure 3A). This transcript was also observed in a few nuclei located in parasegments 4-6 ( Figure 3, A and C). In addition to the abundant posterior expression relative to wild-type lncRNA:PS4, the Cbx hybrid transcript was not excluded from embryonic mesoderm, unlike wild-type lncRNA:PS4 (Figure 3, A and C, arrow). The Cbx hybrid transcript pattern mimics the normal expression pattern of the bxd transcript (Lipshitz et al. 1987;Pease et al. 2013;Petruk et al. 2006). In Ubx Cbx-1 mutants, the bxd promoter and regulatory sequences lie within the 17 kb Ubx Cbx-1 insertional mutation, in the opposite orientation to their normal direction in the BX-C complex. The expression of the Cbx hybrid transcript, therefore, appears to be driven by bxd enhancers located in the insertion. In the hybrid transcript, transcription of the 3 0 region of lncRNA:PS4 extends beyond its normal termination site to at least the vicinity of the Ubx first exon, since Cbx hybrid transcripts can be detected with a probe corresponding to the 5 0 most 1.5 kb of the first intron of Ubx (Supplementary Figure S4, A and B).
The M probe primarily detected the expression of the lncRNA:PS4 that is encoded upstream of the Cbx 1 insertion ( Figure 3B). This 5 0 1 kb region of lncRNA:PS4 continues to be transcribed in parasegments 4 and 5 and continues to be largely excluded from the mesoderm, similar to the lncRNA:PS4 in wild- type embryos ( Figure 3B, arrow). Because M probe has a small amount of homology to sequences downstream of the Cbx 1 insertion ( Figure 1B), it also weakly detects the Cbx hybrid transcript (see the weak mesodermal signal, probably resulting from codetection of the Cbx hybrid transcript in Figure 3, B and C). Thus, in Ubx Cbx-1 mutants, the lncRNA:PS4 transcription unit is split into two transcripts: the 1 kb upstream of the insertion that is expressed in the largely wild-type lncRNA:PS4 pattern and another, containing over 8 kb of lncRNA:PS4 downstream of the insertion that is expressed in bxd pattern ( Figure 3C).
As the Ubx Cbx-1 insertion eliminates the parasegment 4 and 5 expression of the 3 0 -most 8.6 kb of lncRNA:PS4 transcribed region, we next investigated whether this change was associated with the expansion of Ubx transcription into that region. We detected no expansion of embryonic Ubx transcription into parasegments 4 and 5 of Ubx Cbx-1 mutants (Supplementary Figure S4, C and D), suggesting that PS4-PS5 expression of lncRNA:PS4 was not necessary for the repression of Ubx transcription anterior to its normal domain in embryos. Furthermore, the abundant and broadened transcription of the 3 0 portion of lncRNA:PS4 sequences in parasegments 6-12 of Ubx Cbx-1 mutants were not associated with suppression of Ubx transcription in those parasegments (Supplementary Figure S4, C and D).
Using high-resolution microscopy, we investigated whether the presence of the Cbx hybrid transcript interferes with transcription of the Ubx gene from the same locus. To that end, we detected Ubx transcription using a probe for the 5 0 -most region of the first, 7.5 kb-long Ubx intron, which should detect the first $10 min of Ubx transcription (Figure 3, D-G). In some nuclei we detected Ubx and Cbx hybrid transcripts that appeared to derive from the same DNA molecule (Figure 3, F and G), thus arguing against transcriptional interference by Cbx hybrid transcript. One caveat to this statement is that it is possible that some of the imaged chromosomes contained replicated sister chromatids and so the overlap of Cbx hybrid and Ubx transcript signals could be due to adjacent DNA molecules in cell cycle G2.
CRISPR-mediated knockout of lncRNA:PS4 has no effect on Ubx expression, but it interacts genetically with the Ubx Cbx-1 mutation in developing wings To further test the function of the lncRNA:PS4 transcript, we generated a CRISPR/Cas9-mediated deletion of 324 bp that removed both of its promoters. Whole-mount in situ hybridization demonstrated that lncRNA:PS4 transcription was largely eliminated in the homozygous mutants (Supplementary Figure S5). Next, we investigated the effects of the lncRNA:PS4 mutation on Ubx expression. In embryos homozygous for the promoter deletion, the abundance and pattern of Ubx transcripts were the same as in wild-type embryos. Only 5-6% of the lncRNA:PS4 homozygous mutant flies emerged as adults (with most homozygotes dying in late larva and pupa stage), apparently because of an additional mutation elsewhere on the third chromosome, since trans-heterozygotes of the lncRNA:PS4 promoter deletion over a deletion of the Ubx locus [Df(3R)Ubx 109 ], resulted in viable adults at approximately wild-type frequencies. We then examined if the deletion of lncRNA:PS4 promoters resulted in ectopic activation of Ubx protein in larval wing discs, or in wing abnormalities. The wing discs and the wings of the surviving homozygous adults were similar to those of flies heterozygous for the deletion and contained no wing deformations ( Figure 4A), or ectopic activation of Ubx protein in wing discs ( Figure 4B).
We also examined whether the lncRNA:PS4 promoter deletion mutation interacted genetically with the Ubx Cbx-1 mutation. In wing discs heterozygous for Ubx Cbx-1 , we observed ectopic Ubx protein in the posterior region of the disc ( Figure 4D), as previously described (White and Akam 1985). In our hands, adults that were heterozygous for the Ubx Cbx-1 insertion had only slight defects in wing morphology ( Figure 4C), likely because of accumulation of modifiers in this strain. As reported previously (Castelli-Gair et al. 1992), Polycomb (Pc 3 ) mutants enhance the amount and extent of Ubx expression in the wing disc of Ubx Cbx-1 mutants ( Figure 4H), and have a stronger transformation of wings toward halteres ( Figure 4G). The lncRNA:PS4 promoter deletion also enhances the adult wing phenotype of Ubx Cbx-1 (Figure 4E), albeit much more mildly than Pc 3 . The ectopic expression of Ubx protein in the posterior compartment of wing discs from Ubx Cbx-1 / lncRNA:PS4 larvae was also enhanced ( Figure 4F), compared with that seen in discs of larvae heterozygous for Ubx Cbx-1 .

Discussion
Although in embryos the antisense lncRNA:PS4 transcripts are expressed in cells that are largely nonoverlapping with Ubx, the function of lncRNA:PS4 seems to have little if any effect on the pattern of Ubx expression in otherwise wild-type animals. Two lines of evidence support this conclusion. First, a small deletion that eliminates almost all lncRNA:PS4 transcription has no detectable effect on Ubx expression in embryos and in imaginal discs. Second, the abundant transcription of 3 0 lncRNA:PS4 sequences driven by the Ubx Cbx-1 insertion in posterior regions of the embryo has no detectable repressive effect on Ubx transcription in such posterior regions.
In animals that are heterozygous for Ubx Cbx-1 and lncRNA:PS4 promoter mutations there is an enhancement of the Ubx Cbx-1 phenotype, and one can interpret this as noncomplementation between these two alleles, suggesting that lncRNA:PS4 has a very subtle repressive effect on Ubx transcription in wing discs which can only be observed in a Ubx Cbx-1 mutant background. One possible mechanism by which the lncRNA:PS4 region may provide this subtle repression is by helping to set a repressive chromatin state in the cells in which it is expressed, as was suggested for the bxd lncRNA (Pease et al. 2013).
One puzzle is how the enhancer/promoter for lncRNA:PS4 that resides in the Ubx locus is able to activate transcription within parasegment 4, whereas the rest of the Ubx locus is repressed in parasegment 4. This seems potentially inconsistent with the "open for business" model, in which the Ubx locus as a whole has been observed to be associated with H3K27 trimethylation repressive modifications in the bulk of embryonic parasegment 4 cells (Bowman et al. 2014;Maeda and Karch 2015). Those modifications recruit Polycomb group proteins that maintain the off state of Ubx in those cells. However the expression of lncRNA:PS4 is transient in most cells of parasegment 4, and may help recruit Polycomb group complexes at very early embryonic stages that then quantitatively assist in the maintenance of Ubx repression in the cells of that parasegment through later stages of embryogenesis and beyond.
The lncRNA:PS4 promoters are both contained within a predicted Polycomb Repression Element (Supplementary Figure S1; Né gre et al. 2011). It may be that either lncRNA:PS4 transcripts themselves contribute to the function of this PRE or that they are merely a signal associated with the PRE regulatory element, as noncoding RNAs are often produced in the vicinity of cis-regulatory elements (Natoli and Andrau 2012). If this is the case, it is possible that the Ubx Cbx-1 -dependent effect of lncRNA:PS4 promoter deletions on Ubx expression in the wing is due to the sequences removed from the predicted PRE, and not due to elimination of lncRNA:PS4 transcripts per se. It is also possible that lncRNA:PS4 transcripts have a repressive effect on Ubx in some cells that we did not test, but an important effect of lncRNA:PS4 on early embryonic patterning via regulation of Ubx is not supported by our results.
In some respects, the Hox-encoded lncRNA:PS4 transcripts resemble that of the Hox cluster-encoded HOTAIR transcripts of mice. Different studies have proposed that HOTAIR either has an important (Rinn et al. 2007;Li et al. 2013 2016), these inconsistent findings might be explained by the different genetic backgrounds in which HOTAIR mutant alleles were tested. As described here, the function of lncRNA:PS4 transcripts also seem sensitive to genetic background, as lncRNA:PS4 mutant animals only exhibit ectopic expression of Ubx and abnormal wing shapes in the background of a Ubx Cbx-1 mutant allele.

Data availability
Strains and plasmids are available upon request. Reagents and primer sequences not included in the Materials and Methods section are listed in the Reagents Table. The GenBank accession number of lncRNA:PS4 is OK501974.
Supplementary material is available at G3 online.

Funding
This research was supported by NIH grant R01HD28315 to W.M.