Reversible acetylation of Hpo regulates the Hippo pathway

Significance

Histone acetylation, a crucial epigenetic code, influences the accessibility of chromatin and controls gene expression. The histone deacetylase Hdac3 is able to remove acetyl groups from histones, resulting in chromatin condensation and gene repression. Despite being present in both the cytoplasm and nucleus, the function of cytoplasmic Hdac3 is yet to be determined. This study reveals that Hpo, a key component of the Hippo pathway, undergoes reversible acetylation by Nej and Hdac3 in the cytoplasm, which in turn regulates signaling output. Hdac3 interacts with Hpo to prevent its acetylation, thereby increasing its stability. These findings not only uncover the reversible acetylation of Hpo in controlling the Hippo pathway, but also identify Hpo as a cytoplasmic substrate of Hdac3.

Abstract

The Hippo pathway, which is highly conserved from Drosophila to mammals, plays a crucial role in regulating organ size and tissue homeostasis. Dysregulation of this pathway has been linked to various diseases, including tumors. The pathway controls the subcellular localization of the transcription coactivator Yki through core kinases Hpo and Wts, ultimately influencing the expression of target genes. Extensive studies have shown that most components in this pathway undergo posttranslational modifications, such as phosphorylation and ubiquitination. Nevertheless, the role of acetylation in Hippo signaling is still not fully understood. In this study, we find that Hpo is subject to reversible acetylation by Nej and Hdac3, thereby regulating its stability. Loss of Hdac3 leads to an increase in the expression of Hippo target genes, which is completely rescued by overexpressing Wts, positioning Hdac3 upstream of Wts. Although Hdac3 localizes in both the cell cytoplasm and nucleus, only the cytoplasmic Hdac3 is involved in regulating the Hippo pathway, where it interacts with Hpo to deacetylate and stabilize it. Additionally, knockdown of the acetyltransferase Nej decreases the expression of Hippo target genes, a phenotype that can be reversed by simultaneously silencing Hdac3. Taken together, these findings shed light on the role of reversible acetylation in controlling the Hippo pathway and provide insights into growth regulation and potential therapeutic approaches for related diseases.

The Hippo pathway plays a crucial role in coordinating cell proliferation and apoptosis to ensure proper development, tissue homeostasis, and organ size, with its dysregulation frequently contributing to tumorigenesis (1, 2). In Drosophila, the core of the pathway involves a kinase cascade: Hippo (Hpo) kinase binds to the adaptor Salvador (Sav) for autoactivation, leading to the phosphorylation and activation of the downstream kinase Warts (Wts) (3–5). Upon activation, Wts phosphorylates and inhibits the transcriptional coactivator Yorkie (Yki) in collaboration with its partner Mob as Tumor Suppressor (Mats) (6). Yki phosphorylation by Wts-Mats takes place at multiple serine residues (S111, S168, S250), resulting in the sequestration of Yki in the cytoplasm (7). In the absence of Yki, the transcription factor Scalloped (Sd) recruits Tondu-domain-containing Growth Inhibitor (Tgi) to activate a default repression mechanism that inhibits the expression of target genes (8). When the Hippo pathway is inactivated, unphosphorylated Yki enters the nucleus to relieve Sd-mediated default repression, ultimately leading to the expression of target genes (8, 9). Well-documented Yki-Sd target genes include cycE and diap1, which promote cell proliferation and suppress cell death respectively (5). The high conservation of Hippo pathway components and regulatory mechanisms in metazoans makes Drosophila an excellent model for this pathway and its related tumors (10–12).

Given the critical role of the Hippo pathway in cell proliferation and apoptosis, its activity should be strictly controlled at multiple layers, including posttranslational modifications (PTMs). Phosphorylation is the first PTM identified in this pathway, which modulates the activity of the core kinase cassette and Yki localization (5, 6). Ubiquitination is another important PTM in the Hippo pathway, as the majority of components undergo ubiquitination modification (13). Ubiquitin modification not only influences protein stability but also affects their interactions and localization, thereby impacting the overall Hippo signaling outcome (14–17). Several components of the Hippo pathway also experience methylation, which affects the signaling output (18, 19). Recently, it has been demonstrated that lactylation modification is involved in regulating the Hippo pathway (20, 21). Despite some studies revealing that the Hippo pathway is controlled by acetylation during oxidative stress or tumorigenesis, the physiological function of acetylation in this pathway remains unclear (22, 23). Protein acetylation occurs through the transfer of an acetyl group from acetyl-coenzyme A (acetyl-CoA) to lysine residues by acetyltransferases (24). This modification is reversible due to the action of deacetylases. Histone deacetylases (HDACs) belong to the largest branch of the deacetylase family (25). Despite histones being the most well-characterized substrates of HDACs, a growing number of nonhistone proteins have been found to be deacetylated by HDACs (26). To date, most of the reported nonhistone substrates of HDACs are nuclear proteins, such as the transcription factors P53 and STAT3, but their cytoplasmic substrates are still unclear (27, 28).

The human genome encodes eleven HDACs, which display a certain degree of functional redundancy, making it challenging to analyze their biological roles (29, 30). Drosophila offers a possibility to address this challenge because it possesses only five HDACs (31). These five HDACs present distinct subcellular localizations: HDAC1 and HDAC11 are predominantly located in the nucleus, HDAC3 and HDAC4 exist in both the nucleus and cytoplasm, and HDAC6 is mainly in the cytoplasm (32–36). The various localizations suggest that HDACs play diverse roles by deacetylating nuclear or cytoplasmic proteins. Moreover, certain HDACs, such as HDAC3, have the capability to perform their functions in a manner that does not rely on their deacetylase activities (37).

To systematically explore the role of HDACs in the Hippo pathway, we conducted an RNAi-mediated screening in Drosophila wing discs. The unbiased screening revealed that knockdown of hdac3 apparently enhanced the expression of Hippo target genes, leading to eye overgrowth. Somatic clones carrying hdac3 null alleles also exhibited increased expression of Hippo target genes, which was rescued by introducing wild-type Hdac3 rather than deacetylase-deficient Hdac3. Even though Hdac3 showed localization in the cytoplasm and nucleus, only cytoplasmic Hdac3 was capable of restoring the upregulation of Hippo target genes due to hdac3 depletion. Genetic epistasis analysis positioned Hdac3 upstream of Wts, parallel to Hpo. Biochemical studies showed that Hdac3 interacts with Hpo to suppress Hpo acetylation. Interestingly, loss of hdac3 led to a decrease in Hpo protein level, while overexpression of hdac3 increased Hpo stability. Finally, the acetyltransferase Nej bound to Hpo and promoted the expression of Hippo target genes, which was completely counteracted by expressing Hdac3. Overall, our study not only uncovered the reversible acetylation of Hpo in controlling the Hippo pathway, but also identified Hpo as a cytoplasmic substrate of Hdac3.

Results

Knockdown of hdac3 Upregulates the Expression of Yki Target Genes.

To examine the correlation between HDACs and the Hippo pathway, we performed an RNAi-mediated screen targeting all identified Drosophila HDACs. Knocking down HDACs in wing discs using the ptc-Gal4 driver and monitoring the diap1-lacZ reporter revealed that only knockdown of hdac3 (V20814 from VDRC) led to an apparent increase in diap1-lacZ expression (Fig. 1 A and B′ and SI Appendix, Fig. S1 A–E′). To rule out potential off-target effects, we utilized two additional RNAi lines (2128R-1 from NIG and BL31633 from BDSC) that target distinct regions of the hdac3 gene sequence (SI Appendix, Fig. S1J). Both RNAi lines increased the expression of diap1-lacZ (SI Appendix, Fig. S1 F–G′). The effectiveness of these three RNAi lines was verified by the RT-qPCR assay (SI Appendix, Fig. S1K). When hdac3 was knocked down in both the wing and eye discs via the Flp-out technique (38), it led to the upregulation of diap1-lacZ expression in GFP-positive clones (SI Appendix, Fig. S1 H–I′), suggesting a cell-autonomous manner. Furthermore, we also examined the expression of other well-known Yki target genes. Compared with control discs (Fig. 1 C and C′, E and E′, and G and G′), knockdown of hdac3 increased the expression of ban-lacZ (Fig. 1 D and D′), ex-lacZ (Fig. 1 F and F′), and fj-lacZ (Fig. 1 H and H′). The RT-qPCR assay revealed that knocking down hdac3 elevated the mRNA levels of ex and diap1, but did not affect the expression of core Hippo pathway components (SI Appendix, Fig. S1K), implying that the regulation of Hippo pathway activity by Hdac3 occurs at the protein level.

Fig. 1.

Knockdown of hdac3 activates Yki target genes (A–B′) Wing discs of the control (A and A′) and hdac3 knockdown by ptc-Gal4 (B and B′) were stained to show diap1-lacZ (white) and GFP (green). Of note, knockdown of hdac3 led to an increase in diap1-lacZ. The GFP marks the expression pattern of ptc-Gal4 in wing discs. (C–D′) Knockdown of hdac3 elevated the expression of ban-lacZ. (E–F′) Knockdown of hdac3 promoted ex-lacZ expression. GFP marks the expression region of ci-Gal4 in wing discs. (G–H′) Wing discs of the control (G and G′) and hdac3 knockdown by hh-Gal4 (H and H′) were stained with fj-lacZ (white) and Ci (red). hh-Gal4 drives UAS transgenes to express in the posterior compartment of the wing disc, where Ci is not expressed. (I–K) Scanning electron microscope images of adult eyes from the control fly (I), hdac3 knockdown with V20814 (J), hdac3 knockdown with 2128R-1 (K). Knocking down hdac3 resulted in enlarged eye sizes, without affecting ommatidial morphology. Quantitative analyses of eye sizes are shown on the Right. (Scale bar, 100 µm for all images.)

Due to the important function of the Hippo pathway in growth regulation, we aimed to assess the effect of hdac3 knockdown on organ size in Drosophila. Compared to the control eye, knockdown of hdac3 led to an increase in eye size by approximately 20% (Fig. 1 I–K and SI Appendix, Fig. S2 A–C). Upon examination with scanning electron microscopy, it was observed that knockdown of hdac3 did not cause histological abnormalities in the eyes (Fig. 1 I–K), indicating that Hdac3 mainly influences tissue growth rather than morphogenesis. To investigate the cellular basis of tissue enlargement, we examined cell proliferation using the mitotic marker phospho-histone H3 (PH3) (39). In comparison to the internal control (GFP-negative), the area with hdac3 knockdown (GFP-positive) exhibited denser PH3 signals (SI Appendix, Fig. S2 D–D″), indicating upregulated cell proliferation. Taken together, the RNAi-mediated screening identifies that knocking down hdac3 increases Yki activity and promotes organ overgrowth.

Loss of hdac3 Activates the Expression of Yki Target Genes.

To confirm the results obtained by RNAi, we utilized the hdac3^N allele, which contains a premature stop codon at position Q57, resulting in a truncated nonfunctional peptide (40). Since hdac3^N homozygote is embryonic lethal, we employed the FLP/FRT technique to create somatic hdac3^N homozygous mutant clones (41). In line with the RNAi findings, hdac3^N homozygous clones, marked by the absence of RFP signals, displayed increased fj-lacZ expression (Fig. 2 A–A″). In addition, we examined the expression of other Yki target genes in hdac3^N homozygous clones, and found that the depletion of hdac3 led to elevated levels of diap1-lacZ (Fig. 2 B–B″), ban-lacZ (Fig. 2 C–C″), and ex-lacZ (Fig. 2 D–D″). The finding that hdac3 loss results in the widespread activation of Yki target genes indicates that Hdac3 may be involved in the control of Hippo signaling transduction.

Fig. 2.

Loss of hdac3 leads to elevated expression of Yki target genes (A–A″) A wing disc carrying hdac3^N mutant clones was stained to show the expression of RFP (red) and fj-lacZ (white). Clones with homozygotic hdac3^N alleles are labeled by the lack of RFP signals. Depletion of hdac3 results in a notable increase in fj-lacZ. The genotype is ubx-flp; fj-lacZ/+; hdac3^N FRT82B/ubi-RFP FRT82B. (B–B″) Loss of hdac3 increased the expression of diap1-lacZ. The genotype is ubx-flp; +/+; hdac3^N diap1-lacZ FRT82B/ubi-RFP diap1-lacZ FRT82B. (C–C″) Knockout of hdac3 upregulated the level of ban-lacZ. The genotype is ubx-flp; +/+; hdac3^N ban-lacZ FRT82B/ubi-RFP ban-lacZ FRT82B. (D–D″) Depletion of hdac3 led to an elevation in ex-lacZ. The genotype is ubx-flp; ex-lacZ/+; hdac3^N FRT82B/ubi-RFP FRT82B. (Scale bar, 100 µm for all images.)

Having established that hdac3 deficiency results in the activation of Yki target genes, we then sought to explore the effects of hdac3 overexpression. Compared to the control disc (SI Appendix, Fig. S3 A and A′), overexpression of hdac3 did not result in a decrease in fj-lacZ level (SI Appendix, Fig. S3 B and B′). In addition, ectopic expression of hdac3 failed to alter the levels of ex-lacZ (compare SI Appendix, Fig. S3 D and D′ with SI Appendix, Fig. S3 C and C′) and diap1-lacZ (compare SI Appendix, Fig. S3 F and F′ with SI Appendix, Fig. S3 E and E′). These findings suggest that endogenous Hdac3 is sufficient to maintain the proper Hippo signaling output, and further increasing the amount of Hdac3 does not change its effectiveness. This idea is reinforced by the fact that the elevation of diap1-lacZ as a result of hdac3 RNAi was completely rescued through the expression of exogenous hdac3 (compare SI Appendix, Fig. S3 H and H′ with SI Appendix, Fig. S3 G and G′).

The Regulation of the Hippo Pathway By Hdac3 Relies on Its Deacetylase Activity.

To investigate whether Hdac3 regulates the Hippo pathway through its deacetylase activity, we employed a deacetylase-defective mutant of Hdac3 with His140 changed to Ala (Hdac3-H140A). Compared with the control disc (Fig. 3 A and A′), overexpression of hdac3-H140A substantially elevated the ban-lacZ level (Fig. 3 B and B′). Additionally, overexpression of hdac3-H140A also upregulated the levels of fj-lacZ (compare Fig. 3 D and D′ with Fig. 3 C and C′) and diap1-lacZ (compare Fig. 3 F and F′ with Fig. 3 E and E′). The similar phenotypes observed in Hdac3-H140A and hdac3 RNAi suggest that Hdac3-H140A plays a dominant-negative role. To verify this notion, we examined whether wild-type Hdac3 was able to counteract the elevation of Yki target genes induced by Hdac3-H140A. Indeed, coexpression of hdac3 fully restored the increase in diap1-lacZ (compare Fig. 3 G and G′ with Fig. 3 F and F′) and fj-lacZ (compare Fig. 3 H and H′ with Fig. 3 D and D′). Collectively, these findings reveal that the deacetylase activity of Hdac3 is crucial for its regulation of the Hippo pathway.

Fig. 3.

Hdac3 regulates the Hippo pathway through its deacetylase activity (A–B′) A control wing disc (A and A′) and a wing disc expressing hdac3-H140A by en-Gal4 (B and B′), were stained to show Ci (red) and ban-lacZ (white). Notably, ectopic expression of hdac3-H140A led to an increase in ban-lacZ. en-Gal4 drives UAS transgenes to express in the posterior compartment of the wing disc, where Ci is lacking. (C–D′) Overexpression of hdac3-H140A by hh-Gal4 upregulated fj-lacZ in the wing discs. (E–F′) Overexpressing hdac3-H140A with ptc-Gal4 increased diap1-lacZ levels. (G and G′) A wing disc simultaneously expressing wild-type hdac3 and hdac3-H140A was stained to show diap1-lacZ (white) and GFP (green). Hdac3 was able to rescue the upregulation of diap1-lacZ induced by Hdac3-H140A. (H and H′) The elevation of fj-lacZ caused by Hdac3-H140A was completely blocked by introducing Hdac3. (Scale bar, 100 µm for all images.)

Cytoplasmic Hdac3 Is Responsible for Regulating the Hippo Pathway.

Hdac3 primarily functions as a transcriptional silencer through forming a repressor complex with Smrter (Smr) in the nucleus (42). Smr serves as a scaffold that recruits Hdac3 to histones, facilitating histone deacetylation and the establishment of a condensed, transcriptionally repressive chromatin state (43, 44). In contrast to the clear upregulation of diap1-lacZ observed with hdac3 RNAi (compare SI Appendix, Fig. S4 B–B″ with SI Appendix, Fig. S4 A–A″), knockdown of smr under comparable conditions failed to induce its expression (SI Appendix, Fig. S4 C–C″). The same outcomes were also seen in ex-lacZ (SI Appendix, Fig. S4 D–F″), indicating that Hdac3 regulates the Hippo pathway independently of the Hdac3-Smr repressor complex. Since the previous study has demonstrated that Hdac3 localizes in both the cytoplasm and nucleus (34), we sought to determine where the regulation of the Hippo pathway by Hdac3 takes place. To this end, we developed two location-anchored Hdac3 mutants, one in the cytoplasm and the other in the nucleus. A nuclear localization sequence (NLS) was fused to the C-terminus of Hdac3 to create a Myc-Hdac3-NLS variant. On the other hand, we also generated a membrane-tethered form of Hdac3 by introducing a myristoylation signal at its N terminus (Myr-Hdac3-HA). In comparison to wild-type Hdac3, which could be found in both the cytoplasm and nucleus (SI Appendix, Fig. S5 A–A″), Myc-Hdac3-NLS was exclusively located in the nucleus (SI Appendix, Fig. S5 B–B″), while Myr-Hdac3-HA was only in the cytoplasm (SI Appendix, Fig. S5 C–C″), suggesting the effectiveness of these variants.

We produced random somatic clones of cells carrying GFP-marked hdac3^N alleles in wing discs using the Mosaic Analysis with a Repressible Cell Marker (MARCM) technique (45). As expected, loss of hdac3 led to a sharp increase in fj-lacZ (Fig. 4 A–A″), which was rescued by reintroducing exogenous hdac3 (Fig. 4 B–B″). The upregulation of fj-lacZ by hdac3 depletion could be restored by Myr-Hdac3-HA (Fig. 4 D–D″), rather than Myc-Hdac3-NLS (Fig. 4 C–C″), indicating the importance of cytoplasmic Hdac3 in the Hippo pathway regulation. We validated this result using diap1-lacZ as a reporter (SI Appendix, Fig. S5 D–G″). Furthermore, the increase in fj-lacZ levels caused by hdac3 deficiency was counteracted by wild-type Hdac3 (Fig. 4 B–B″), but not Hdac3-H140A (Fig. 4 E–E″), highlighting the necessity of the deacetylase activity of Hdac3 in modulating the Hippo pathway. Taken together, despite Hdac3 being present in both the cytoplasm and nucleus, its role in controlling the Hippo pathway is primarily carried out in the cytoplasm.

Fig. 4.

Cytoplasmic Hdac3 is responsible for controlling fj-lacZ expression (A–A″) A wing disc carrying hdac3^N mutant clones was stained to show the expression of GFP (green) and fj-lacZ (white). Clones with homozygotic hdac3^N alleles are labeled by the presence of GFP signals. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ; hdac3^N FRT82B/tub-Gal80 FRT82B. (B–B″) A wing disc containing hdac3^N mutant clones and hdac3 overexpression was stained to show the expression of GFP (green) and fj-lacZ (white). GFP-positive cells carry both endogenous hdac3 depletion and exogenous hdac3 overexpression simultaneously. Overexpression of wild-type hdac3 was able to rescue the upregulation of fj-lacZ induced by hdac3 knockout. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-hdac3; hdac3^N FRT82B/tub-Gal80 FRT82B. (C–C″) Overexpression of hdac3-NLS failed to counteract the elevation of fj-lacZ in the context of hdac3 depletion. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-Myc-hdac3-NLS; hdac3^N FRT82B/tub-Gal80 FRT82B. (D–D″) The increase of fj-lacZ by hdac3 depletion was restored by introducing Myr-Hdac3. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-Myr-hdac3-HA; hdac3^N FRT82B/tub-Gal80 FRT82B. (E–E″) Hdac3-H140A could not rescue fj-lacZ expression in the hdac3 knockout setting. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-hdac3-H140A; hdac3^N FRT82B/tub-Gal80 FRT82B. (Scale bar, 100 µm for all images.)

Hdac3 Acts Upstream of Wts, Parallel to Hpo.

In order to understand the mechanism of Hdac3 regulating the Hippo pathway, we performed genetic epistasis analysis. Compared to the control eye disc (Fig. 5 A and A′), knockdown of hdac3 resulted in a noticeable increase in diap1-lacZ (Fig. 5 B and B′). Due to the inhibitory effect of the Hippo pathway on Yki activity, upregulation of its components will result in the suppression of Yki target genes, including diap1-lacZ (39). Since the above findings have demonstrated that cytoplasmic Hdac3 is responsible for regulating the Hippo pathway, we speculated that Hdac3 possibly targets a cytoplasmic component of this pathway. To this end, we successively overexpressed these components in the context of hdac3 RNAi and assessed the diap1-lacZ levels to evaluate the rescue effect. Overexpression of fat (ft), dachsous (ds), or expanded (ex) did not counteract the increase of diap1-lacZ caused by hdac3 RNAi in both the eye and wing discs (Fig. 5 C–E′ and SI Appendix, Fig. S6 A–D′). To quantitatively assess the rescue effects, we conducted RT-qPCR using RNA isolated from adult fly eyes. Despite the effective reduction in fj and cycE mRNA levels due to the overexpression of ft, ds, or ex, they did not restore the upregulation of target genes resulting from hdac3 RNAi (SI Appendix, Fig. S6G), positioning Hdac3 downstream of these components. The upregulation of diap1-lacZ induced by hdac3 knockdown was only partially inhibited by hpo overexpression in both the eye discs (compare Fig. 5 F and F′ with Fig. 5 B and B′) and wing discs (compare SI Appendix, Fig. S6 E and E′ with SI Appendix, Fig. S6 A and A′). In contrast, overexpression of wts was able to fully reverse hdac3 RNAi-caused diap1-lacZ elevation (Fig. 5 G and G′ and SI Appendix, Fig. S6 F and F′).

Fig. 5.

Hdac3 functions upstream of Wts (A–B′) Eye discs of the control (A and A′) and hdac3 knockdown by GMR-Gal4 (B and B′) were stained to show diap1-lacZ (white) and GFP (green). GMR-Gal4 drives UAS transgenes to express in the posterior compartment of the eye disc, as shown by GFP staining. (C–E′) Eye discs expressing ft plus hdac3 RNAi (C and C′), ds plus hdac3 RNAi (D and D′), ex plus hdac3 RNAi (E and E′) were stained with diap1-lacZ (white) and GFP (green). (F-G′) Eye discs expressing hpo plus hdac3 RNAi (F-F′), wts plus hdac3 RNAi (G and G′) were stained with diap1-lacZ (white) and Ci (red). Of note, overexpression of wts fully rescued the upregulation of diap1-lacZ caused by hdac3 knockdown. (H and H′) Loss of hdac3 led to a notable increase in fj-lacZ expression. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ; hdac3^N FRT82B/tub-Gal80 FRT82B. (I and I′) Overexpression of hpo weakly rescued the upregulation of fj-lacZ induced by hdac3 knockout. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-hpo; hdac3^N FRT82B/tub-Gal80 FRT82B. (J and J′) Overexpression of wts completely counteracted the elevation of fj-lacZ caused by hdac3 depletion. The genotype is hs-flp; tub-Gal4 UAS-GFP/fj-lacZ UAS-wts; hdac3^N FRT82B/tub-Gal80 FRT82B. (Scale bar, 100 µm for all images.)

Since the previous study showed that increased hpo expression in the wing disc leads to apoptosis (8), we aimed to investigate if the proapoptotic effect could be mitigated by knocking down hdac3. Remarkable activation of apoptosis was seen in the wing disc when hpo was overexpressed using nub-Gal4, as evidenced by Caspase3 staining (SI Appendix, Fig. S6H). In addition, overexpression of hpo produced a slender wing disc (marked by the red dotted line), reducing its size (SI Appendix, Fig. S6H). Knockdown of hdac3 by 2128R-1 (SI Appendix, Fig. S6I) or V20814 (SI Appendix, Fig. S6J) not only decreased the Caspase3 signals, but also enlarged the sizes of wing discs, indicating that hdac3 knockdown suppresses the proapoptotic function of Hpo.

To confirm the genetic position of Hdac3 in controlling the Hippo pathway, we employed the MARCM method to overexpress hpo or wts in hdac3^N mutant clones and analyzed Yki target gene expression. Loss of hdac3 led to a remarkable increase in fj-lacZ level (Fig. 5 H and H′), which was blunted by introducing exogenous Hpo (Fig. 5 I and I′). Nevertheless, overexpression of wts could completely block the upregulation of fj-lacZ in hdac3^N mutant clones (Fig. 5 J and J′). To determine whether the activation of Yki target genes upon hdac3 loss depends on Hpo function, we employed the hpo^KS240 allele (46). This allele harbors a premature stop codon at W518, thereby abolishing kinase dimerization and autoactivation (47). Homozygous hdac3^N or hpo^KS240 alleles are larval lethal but can survive until embryonic stages 13 to 15 (12 h after egg laying at 25 °C). RT-qPCR analysis revealed a dramatic activation of Yki target genes in both hpo^KS240 and hdac3^N homozygous embryos (SI Appendix, Fig. S6K). However, the upregulation of Yki target genes following hpo loss remained unaffected upon hdac3 codepletion (SI Appendix, Fig. S6K), suggesting that Hdac3 regulates the Hippo pathway through Hpo. Overall, these findings argue that Hdac3 functions upstream of Wts, in parallel with Hpo, to modulate the Hippo pathway.

Hdac3 Interacts With Hpo to Maintain Its Stability.

After demonstrating that Hdac3 works in parallel with Hpo, we proceeded to investigate the possibility of Hpo being a substrate of Hdac3. We initially examined the protein–protein interaction between Hdac3 and Hpo. The coimmunoprecipitation (co-IP) assay showed that both Fg-Hdac3 and Fg-Hdac3-H140A were able to pull down HA-Hpo equally (Fig. 6A), suggesting that the loss of deacetylase activity did not impact its binding to Hpo. A known substrate, histone H4 (48), was used as a positive control (SI Appendix, Fig. S7A). Additionally, HA-tagged Hdac3 could co-IP endogenous Hpo protein (Fig. 6B). A recent study revealed that a commercially available antibody against human HDAC3 (A2139, ABclonal) also recognizes Drosophila Hdac3 (49). Using this antibody in co-IP assays, we detected the interaction between endogenous Hdac3 and Hpo in S2 cells (Fig. 6C). Moreover, this interaction was confirmed through in vitro pull-down experiments using GST-Hdac3 and His-Hpo expressed in E. coli BL21 (SI Appendix, Fig. S7B), indicating a direct interaction. To determine which domains are responsible for this interaction, we constructed various truncated mutants for co-IP assays (Fig. 6D). The C-terminus of Hdac3, rather than its N terminus, interacted with Hpo (Fig. 6E). On the other hand, Hpo bound to Hdac3 through its N terminus (Fig. 6F and SI Appendix, Fig. S7C).

Fig. 6.

Hdac3 interacts with Hpo to enhance its stability (A) Immunoblots of immunoprecipitates (IP) and whole-cell lysate (WCL) from HEK293T cells transfected with the indicated constructs. Both Hdac3 and Hdac3-H140A were able to pull down Hpo. (B) Hdac3-HA was able to pull down endogenous Hpo in S2 cells. (C) Endogenous Hdac3 pulled down endogenous Hpo in S2 cells. (D) Schematics showing the structures of Hdac3 and Hpo as well as the truncated constructs used in subsequent co-IP experiments. (E) Fg-Hdac3-C, rather than Fg-Hdac3-N, pulled down HA-Hpo. The arrowhead indicates heavy IgG. (F) Hpo-N could interact with Hdac3. (G) Hdac3 decreased the acetylation of HA-Hpo, but Hdac3-H140A did not show this effect. (H) Immunoblots of WCL from HEK293T cells transfected with the indicated constructs and subjected to CHX treatment for distinct intervals. Hdac3 prolonged the half-life of Hpo protein. Actin acts as a loading control. (I) Immunoblots of WCL from adult heads expressing hdac3 RNAi. Knockdown of hdac3 resulted in a decrease in endogenous Hpo protein. Actin acts as a loading control. (J–M′) Wing discs from the control (J and J′), V20814 (K and K′), 2128R-1 (L and L′), UAS-hdac3 (M and M′) under hs-HA-hpo background were stained to show HA (white) and GFP (green). Knocking down hdac3 resulted in a decrease in HA-Hpo protein, while the overexpression of hdac3 slightly increased HA-Hpo. (N and N′) Overexpression of hdac3-H140A downregulates HA-Hpo, similar to the effect of hdac3 knockdown. (O and O′) Clones with homozygotic hdac3^N alleles displayed a decrease in HA-Hpo protein. (Scale bar, 100 µm for all images.)

The interaction between Hdac3 and Hpo suggests that Hpo could be a potential substrate of Hdac3. To test this speculation, we utilized a pan-acetyl-lysine antibody to analyze the acetylation level of Hpo protein with or without Hdac3. As expected, Hdac3 decreased the acetylation of Hpo, whereas Hdac3-H140A did not display this effect (Fig. 6G). An in vitro deacetylation assay demonstrated that incubation with recombinant GST-Hdac3 suppressed the acetylation of His-Hpo (SI Appendix, Fig. S7D), establishing Hpo as a bona fide enzymatic substrate of Hdac3. The previous study has revealed that Hdac3 stabilizes cyclin A through promoting its deacetylation (50). To investigate if this situation is applicable to Hpo, we conducted a Cycloheximide (CHX) pulse–chase (en rule) assay and found that Hdac3 is able to prolong the half-life of Hpo protein (Fig. 6H). Furthermore, knocking down hdac3 led to a dramatic decrease in endogenous Hpo protein (Fig. 6I).

To explore whether Hdac3 modulates Hpo stability in vivo, we generated a pCaSpeR-hs-HA-hpo transgene, in which the hpo coding sequence (CDS) was inserted downstream of the hsp70 (hs) promoter (51). Hence, the production of HA-tagged Hpo protein was controlled by the hs promoter, ruling out the influence of Hdac3 on HA-hpo transcription. After a 2-h heat shock at 37 °C 12 h before dissection, the wing disc displayed widespread HA-Hpo expression, confirming the successful establishment of this transgene (Fig. 6 J and J′). Compared to the internal control (GFP-negative), knockdown of hdac3 using V20814 (Fig. 6 K and K′) or 2128R-1 (Fig. 6 L and L′) resulted in a noticeable reduction in HA-Hpo levels. The observed decrease in hs-HA-Hpo upon hdac3 knockdown was not due to its effect on the hs promoter, as hdac3 RNAi did not alter the expression of hs-Myc-Wts (SI Appendix, Fig. S7 E–H′). However, overexpression of hdac3 only led to a weak increase in HA-Hpo (Fig. 6 M and M′), supporting that the endogenous Hdac3 is enough to maintain HA-Hpo stability. In line with its dominant-negative role, Hdac3-H140A decreased HA-Hpo protein (Fig. 6 N and N′), similar to the effect of hdac3 RNAi. In addition, the level of HA-Hpo protein was reduced in hdac3^N mutant clones (Fig. 6 O and O′). We found that knockdown of hdac3 in S2 cells decreased Hpo protein levels, without affecting its phosphorylation status (as measured by the pHpo/total Hpo ratio) (SI Appendix, Fig. S7I). In sum, these findings reveal that Hdac3 physically interacts with Hpo to remove acetyl groups from it, ultimately leading to the stabilization of Hpo.

The Acetyltransferase Nej Is Involved in Regulating the Hippo Pathway.

The finding that Hdac3 deacetylates Hpo implies the existence of an acetyltransferase responsible for Hpo acetylation. To identify this acetyltransferase, we carried out two sets of experiments. First, the prediction via an online web server (GPS-PAIL) showed that lysine residues K657 and K668 in Hpo are potential acetylation sites for the conserved acetyltransferase CREBBP (also known as Nej in Drosophila). Additionally, we performed immunoprecipitation (IP) with Fg-tagged Hpo as the bait, followed by mass spectrometry (MS) assays to identify preys. Adult eyes from control flies (GMR-Gal4 > UAS-lacZ), flies overexpressing Fg-hpo (GMR-Gal4 > UAS-Fg-hpo) were collected for IP-MS experiments (the results of the IP-MS will be published elsewhere due to ongoing projects). We found that Nej was the only acetyltransferase present in the preys. Therefore, Nej was selected for further study. Nej is able to both “read” and “write” protein lysine acetylation marks through its BROMO and HAT domains, respectively (52, 53). The co-IP result confirmed the interaction between Hpo and Nej’s BROMO domain (Fig. 7A). We investigated whether Nej plays a role in Hippo pathway regulation by analyzing the expression of Yki target genes with manipulated Nej levels. Compared with the control wing disc (Fig. 7 B and B′ and D and D′), knockdown of nej using BL32577 from BDSC resulted in a noticeable reduction in fj-lacZ (Fig. 7 C and C′) and diap1-lacZ (Fig. 7 E and E′). Furthermore, we silenced randomly nej using the FLP-out method (54), and observed a pronounced decrease in ex-lacZ levels in nej RNAi cells (Fig. 7 F and F′). The downregulation of Yki target genes induced by nej knockdown is unlikely due to off-target effects, as three different RNAi lines (THU4858, BL32577, THU2811) targeting distinct regions of nej produced a similar decrease in ban-lacZ (SI Appendix, Fig. S8 A–D″). RT-qPCR assays showed that knocking down nej with various RNA lines led to a consistent reduction in the expression of Yki target genes (SI Appendix, Fig. S8E). Conversely, overexpression of nej led to an increase in ex-lacZ (Fig. 7 G and G′), highlighting the importance of appropriate Nej levels for the output of Hippo signaling.

Fig. 7.

The acetyltransferase Nej is involved in Hippo pathway regulation (A) HA-Hpo was able to pull down Myc-Nej-BROMO. (B–C′) Wing discs of the control (B and B′) and nej knockdown by hh-Gal4 (C and C′) were stained to show fj-lacZ (white) and GFP (green). Knockdown of nej reduced the expression of fj-lacZ. (D–E′) Knocking down nej resulted in a noticeable reduction in diap1-lacZ levels. (F and F′) Knockdown of nej using Ay-Gal4 downregulated ex-lacZ. GFP marks the region of nej knockdown. (G and G′) Ectopic expression of nej using Ay-Gal4 led to an increase in ex-lacZ. (H and H′) A wing disc simultaneously expressing nej RNAi (BL-32577) and hdac3 RNAi (2128R-1) was stained with diap1-lacZ (white) and GFP (green). The decrease in diap1-lacZ caused by nej RNAi was rescued by knocking down hdac3. (I–I″) A control eye disc was stained to show diap1-lacZ (white) and GFP (green). diap1-lacZ exhibited uniform expression throughout the eye disc. (J–N″) Eye discs expressing nej (J–J″), nej-F2161A (K–K″), nej plus hdac3 (L–L″), nej plus hdac3-NLS (M–M″), nej plus Myr-hdac3 (N–N″) were stained with diap1-lacZ (white) and GFP (green). Of note, Nej-induced upregulation of diap1-lacZ was restored by introducing Hdac3 or Myr-Hdac3, not Hdac3-NLS. (O) A proposed model for reversible acetylation of Hpo by Nej and Hdac3 in the cytoplasm to regulate the Hippo pathway. (Scale bar, 100 µm for all images.)

After demonstrating the involvement of Nej in the Hippo pathway, we sought to examine if its function was antagonized by Hdac3. Despite nej knockdown causing a notable decrease in diap1-lacZ expression (Fig. 7 E and E′), this effect was completely reversed when hdac3 was silenced simultaneously (Fig. 7 H and H′). Compared with the control eye disc (Fig. 7 I–I″), overexpression of wild-type nej using GMR-Gal4 resulted in a noticeable increase in diap1-lacZ levels (Fig. 7 J–J″), while the acetylase-dead mutant Nej-F2161A (55) had no effect on diap1-lacZ expression (Fig. 7 K–K″), suggesting that the enzymatic activity of Nej is necessary for its activation of Yki target genes. The upregulation of diap1-lacZ induced by nej overexpression was rescued by introducing wild-type Hdac3 (Fig. 7 L–L″) or Myr-Hdac3 (Fig. 7 N–N″), instead of Hdac3-NLS (Fig. 7 M–M″), demonstrating that cytoplasmic Hdac3 counteracts Nej’s function in activating Yki target genes. Despite numerous efforts, we failed to acquire a full-length Nej expression plasmid, likely due to its high-molecular-weight coding sequence. As an alternative strategy, we employed a commercial human CREBBP expression plasmid for subsequent biochemical assays, leveraging the high functional homology between human CREBBP and Drosophila Nej (56). CREBBP cotransfection, as expected, induced Hpo acetylation, a process that was counteracted by Hdac3 (SI Appendix, Fig. S8F). Consistent with the bioinformatic prediction, mutation of either K657 or K668 to arginine (R) reduced CREBBP-mediated Hpo acetylation (SI Appendix, Fig. S8F). Strikingly, simultaneous mutation of both residues to R (K657/668R) virtually abolished CREBBP-mediated acetyl group incorporation (SI Appendix, Fig. S8F), validating K657 and K668 as the primary acetylation sites on Hpo. Cotransfection of CREBBP accelerated the degradation of Myc-tagged Hpo (SI Appendix, Fig. S8G). In stark contrast, the Hpo-K657/668R variant exhibited increased stability and was insensitive to CREBBP-mediated regulation (SI Appendix, Fig. S8H), suggesting that acetylation at these sites promotes Hpo proteolysis.

To determine where Nej-mediated Hpo acetylation takes place, we manipulated Nej localization using the importin inhibitor Importazole (IPZ) and the exportin inhibitor Leptomycin B (LMB) (57, 58). Although Nej was enriched in the nucleus (SI Appendix, Fig. S9 A and B″), IPZ treatment triggered its cytoplasmic retention (SI Appendix, Fig. S9 C–C″). The effectiveness of IPZ was validated by its ability to prevent the nuclear entry of Myc-HDAC3-NLS (compare SI Appendix, Fig. S9 F–F″ with SI Appendix, Fig. S9 E–E″). In contrast, inhibition of nuclear export with LMB caused nuclear accumulation of Nej (SI Appendix, Fig. S9 D–D″). Under physiological conditions, treatment with either IPZ or LMB did not result in a noticeable alteration in diap1-lacZ (SI Appendix, Fig. S9 G–I′). Nevertheless, the upregulation of diap1-lacZ due to nej overexpression was strengthened by IPZ but weakened by LMB (SI Appendix, Fig. S9 J–L′), emphasizing the crucial role of cytoplasmic Nej in Hippo pathway regulation. In conclusion, these findings suggest that Nej and Hdac3 play antagonistic roles in controlling the Hippo pathway in the cytoplasm (Fig. 7O).

Discussion

The Hippo pathway was initially discovered in Drosophila due to its critical role in organ size control (5). It is an evolutionarily conserved mechanism for regulating cell proliferation, apoptosis, cell fate determination, and tissue homeostasis (59). Malfunctions in this pathway have been implicated in various human diseases, including cancers (60). Therefore, the transduction of Hippo signaling is tightly modulated through multiple mechanisms, including PTMs, to prevent aberrant activation (13, 61). Despite several components, such as MST1 (62), YAP (63), and MOB1 (22), undergoing acetylation modification in mammals, the role of acetylation in Drosophila Hippo signaling has yet to be determined. The N-acetyltransferase Mnat9 regulates the Hippo pathway independently of its enzymatic activity (64). The alterations in diap1-lacZ expression following hdac1 knockdown in Drosophila epithelial cells may result from the secondary effects of apoptosis, since the mechanism is unclear (65). In this study, we found that Hpo is subject to reversible acetylation by Nej and Hdac3, thereby regulating its stability. An RNAi screen identifies Hdac3 as a positive regulator of the Hippo pathway. Depletion of hdac3 results in elevated expression of Hippo target genes, which can be rescued by introducing cytoplasmic Hdac3, rather than nuclear Hdac3, suggesting that Hdac3-mediated regulation of the pathway occurs in the cytoplasm. Genetic epistasis studies place Hdac3 upstream of Wts, parallel to Hpo. Moreover, Hdac3 interacts with Hpo to deacetylate it, thereby increasing its stability. Finally, the acetyltransferase Nej binds to Hpo and promotes the expression of Hippo target genes, which is completely counteracted by expressing Hdac3. Overall, our study not only uncovered the reversible acetylation of Hpo in controlling the Hippo pathway, but also identified Hpo as a cytoplasmic substrate of Hdac3.

Despite the upregulation of Yki target genes upon hdac3 knockdown, there is no apparent organ overgrowth. Knockdown of hdac3 in the mid-third instar stage eye disc with GMR-Gal4 produces a slightly enlarged eye by increasing cell numbers. At this stage, the eye disc is undergoing photoreceptor specification, showing insensitivity to both proliferative and apoptotic cues (66, 67). It has been demonstrated in a previous study that the reduction of hdac3 leads to a decrease in cell size by antagonizing the PI3K-Akt pathway (48). Nevertheless, another study reveals that depletion of hdac3 triggers apoptosis, without affecting cell proliferation, therefore leading to a small size (40). The presence of phenotypic diversity following hdac3 knockdown implies that Hdac3 plays multiple roles in various biological processes. The finding that Hdac3 suppresses Yki activity is in line with Yki’s known function in determining organ size. Overexpression of yki by GMR-Gal4 during the mid-third instar leads to eye overgrowth (68), while its overexpression by ey-Gal4 at an early stage reduces eye size (69). Knocking down hdac3 with ey-Gal4 results in a small eye (48), whereas with GMR-Gal4, it enlarges eye size, suggesting that Hdac3 exerts distinct effects on growth regulation depending on the context. The similar impacts of hdac3 RNAi and UAS-yki on eye size hint that Hdac3 might function by suppressing Yki.

Hdac3 was originally discovered as a critical epigenetic regulator for its capacity to change the acetylation status of histones (70). By removing acetyl groups from histones, Hdac3 induces a compacted chromatin structure that hinders DNA’s accessibility to the transcription machinery, ultimately repressing gene transcription (70). Here, we provide strong evidence to rule out the possibility that depletion of hdac3 activates Yki target genes by deacetylating histones. First, knockdown of hdac3 exclusively promotes the expression of Yki target genes, without affecting the transcription of Hippo pathway components. Second, the upregulation of Yki target genes in hdac3 mutant clones can be rescued by introducing membrane-anchored Hdac3, rather than nuclear localized Hdac3, indicating that the regulation of Hdac3 upon the Hippo pathway takes place in the cytoplasm. Third, the effects of hdac3 RNAi are completely abolished by expressing wts, positioning Hdac3 upstream of Wts. In addition to histones, an increasing number of nonhistones have been identified as substrates of Hdac3. Most of these nonhistones are nuclear proteins, such as GATA2 (71), STAT1 (72), p65 (73), and Foxo1 (74). This study uncovers the cytoplasmic protein Hpo as a true substrate for Hdac3, deepening the understanding of the function of Hdac3.

Recently, Hdac3 has been increasingly recognized as an oncoprotein in cancer initiation and progression. Elevated levels of Hdac3 are observed in various tumors, including gastric cancer (75), pancreatic cancer (76), and breast cancer (77), and are inversely associated with patient prognosis. Hdac3 primarily facilitates tumor development by repressing the expression of genes involved in tumor suppression, apoptosis, and cell cycle arrest (78). Hence, Hdac3 is regarded as a promising therapeutic target for tumor patients. Several Hdac3 inhibitors, such as RGFP966, are undergoing preclinical trials for tumor treatment (79). Nevertheless, Hdac3 can also exert an anti-tumor role in some situations. Hdac3 suppresses de novo lipogenesis by decreasing the expression of fatty acid synthase, thereby inhibiting metastasis of prostate cancer (80). Hdac3 deficiency in female mice leads to constant liver injury and the spontaneous emergence of hepatocellular carcinoma in an estrogen-dependent manner (81). In the mouse liver, inhibiting Hdac3 promotes nuclear translocation of Yap through an unclear mechanism, leading to liver damage (82). Therefore, Hdac3 plays both promoting and inhibitory roles in tumorigenesis depending on the genetic contexts. The finding in this study that inhibiting hdac3 with hdac3 RNAi or the dominant-negative form Hdac3-H140A can activate Yki transcriptional activity implies that caution should be exercised when selecting Hdac3 inhibitors as drugs for the treatment of tumors, especially Hippo-related tumors.

Methods and Materials

DNA Constructs.

Fg-Hdac3, HA-Hpo, Myc-Hpo, and Myc-H4 constructs were generated by amplifying the corresponding CDS using Vazyme DNA polymerase (P505), and inserting them into the pcDNA3.1-Fg, pcDNA3.1-HA, or pcDNA3.1-Myc backbone vectors. Fg-Hdac3-H140A, Myc-Hpo-K657R, Myc-Hpo-K668R, and Myc-Hpo-K657/668R were created through PCR-based site-directed mutagenesis. Truncated constructs expressing Fg-Hdac3-N (aa1-219), Fg-Hdac3-C (aa220-438), HA-Hpo-N (aa1-310), HA-Hpo-C (aa311-669), and Myc-Nej-BROMO (aa1249-1806) were made by inserting the corresponding CDS into pcDNA3.1-HA, pcDNA3.1-Fg, or pcDNA3.1-Myc vectors. The plasmids for pull-down analyses were constructed by cloning Hdac3, Hdac3-N, Hdac3-C, Hpo, Hpo-N, and Hpo-C sequences into pGEX-4 T-3 or pET32a vectors. pcDNA3.1-CREBBP-HA was purchased from the Miaoling Plasmid platform. A nuclear localization signal (NLS) from SV40 (PPKKKRKV) was inserted at the C-terminus to generate the pUAST-Myc-Hdac3-NLS construct. A myristoylation signal (MGSSKSKPKDPSQRRRSLE) was introduced at the N terminus to create the pUAST-Myr-Hdac3-HA plasmid.

Drosophila Genetics.

ptc-Gal4, en-Gal4, hh-Gal4, ci-Gal4, nub-Gal4, Ay-Gal4, ap-Gal4, C765-Gal4, GMR-Gal4, UAS-lacZ, UAS-GFP, diap1-lacZ, ban-lacZ, ex-lacZ, fj-lacZ, UAS-ft, UAS-ds, UAS-ex, UAS-hpo, UAS-wts, mCherry-RNAi have been described in our previous studies (16, 38, 39, 83). hdac3-RNAi (BL31633), hdac3^N FRT82B (BL32246), UAS-hdac3-H140A (BL55070), hpo^KS240 FRT42D (BL25085), nej-RNAi (BL32577), UAS-nej (BL32573), UAS-nej-F2161A (BL32574), smr-RNAi (BL34087), CyO-GFP balancer (BL5194), TM3-GFP balancer (BL6663), wts-RNAi (V106174), hdac3-RNAi (V20814), hdac1-RNAi (V46930), hdac6-RNAi (V108831), nej-RNAi (THU4858), nej-RNAi (THU5193), nej-RNAi (THU1718), nej-RNAi (THU2811), hdac3-RNAi (2128R-1), hdac4-RNAi (1770R-3), and hdac11-RNAi (3119R-2) stocks were obtained from Bloomington Drosophila Stock Center (BDSC), Vienna Drosophila Resource Center (VDRC), TsingHua Fly Center (THFC) or National Institute of Genetics (NIG). The detailed information of all stocks has been described in FlyBase database (http://flybase.bio.indiana.edu/). ubx-flp, hs-flp, FRT82B, ubi-RFP, tub-Gal4, tub-Gal80 were gifted from Prof. Junzheng Zhang’s lab, China Agricultural University, China. UAS-Fg-hdac3, UAS-Myr-hdac3-HA, UAS-Myc-hdac3-NLS, pCaSpeR-hs-Myc-wts, and pCaSpeR-hs-HA-hpo were created by injecting the indicated constructs into w¹¹¹⁸ Drosophila embryos according to the previous method (84).

Generating Clones.

Clones were generated by FLP/FRT-mediated mitotic recombination. The genotype of hdac3 mutant clones in Fig. 2 was ubx-flp; hdac3^NFRT82B/ubi-RFP FRT82B. The genotype of hdac3 MARCM clones was hs-flp; tub-Gal4 UAS-GFP/+; hdac3^NFRT82B/tub-Gal80 FRT82B. The detailed genotype information is provided in the figure legends. To generate MARCM clones, the 1st-instar larvae underwent a 1.5 h heat shock at 37 °C, followed by incubation at 25 °C for 2 to 3 d. The 3rd-instar larvae were subjected to dissection according to standard protocols. The presence of GFP expression served as a marker for the mutant clones.

Immunostaining.

Immunostaining and image capture were carried out according to our previous studies (85). Briefly, third-instar larvae were dissected in phosphate buffered saline (PBS) and fixed with freshly made 4% formaldehyde in PBS for 20 min at room temperature. After fixation, the larvae were washed three times with PBS supplemented with 0.1% Triton X-100 (PBT). Subsequently, the larvae were incubated overnight with primary antibodies in PBT at 4 °C, followed by three washes with PBT and a 2 h incubation with fluorophore-conjugated secondary antibodies at room temperature. After being washed with PBT for three times, the imaginal discs were separated and soaked in 40% glycerol. Images were captured using the Zeiss confocal microscope. For inhibitor treatment, discs from mid-third instar larvae were cultured in S2 cell medium (HyClone) with 25 µM IPZ (MedChemExpress) or 60 ng/mL LMB (Beyotime) for 3 h. An equal amount of DMSO was used as a control. The primary antibodies used in this study were as follows: rat anti-Ci (1:10, DSHB); rabbit anti-β galactosidase (1:500, MBL); mouse anti-Myc (1:200, Santa Cruz); mouse anti-HA (1:200, Santa Cruz); mouse anti-Fg (1:500, Sigma); rabbit anti-cleaved Caspase-3 (1:200, CST); rabbit anti-PH3 (1:400; Abcam). All secondary antibodies (Jackson ImmunoResearch) were diluted 1:500 for use in this study.

RNA Extraction and RT-qPCR.

Total RNA was extracted from adult eyes with indicated genotypes using TRIzol (Invitrogen) according to standard protocols, and subsequently reverse transcribed with HiScript^® Q RT SuperMix with gDNA wiper (Vazyme) as instructed by the manufacturer. Real-time PCR was conducted on ZY/VQ-100A (Yuanzan) utilizing the ChamQ SYBR^® Color qPCR Master Mix (Q711, Vazyme). The primers were shown as follows: hdac3, 5’-CAG CAG GTG ACT CCA CAA A-3’ (forward) and 5’-CCG CCC GAC CAG TTT AT-3’ (reverse); hpo, 5’-CTC TTC GGC AGC ATC TCC-3’ (forward) and 5’-GCT TCT GGT CGT CGT TGT TT-3’ (reverse); wts, 5’-GCA TCA AGC AGG ACC TAA CC-3’ (forward) and 5’-TCG TTG CGT CCG TTGG-3’ (reverse); yki, 5’-TGC CTA ATC GCT AAG ATA ATTC-3’ (forward) and 5’-CAG GTT GTT GGA CTT GATC-3’ (reverse); ex, 5’-TCC TTG CTG AAA CAG ACTA-3’ (forward) and 5’-GGC TTA CGG TAG ATC CTT-3’ (reverse); diap1, 5’-CCC AAG TCC TCA AAT TCA AA-3’ (forward) and 5’-ACA ATG TAG GCT TAC GAT AAC-3’ (reverse); fj, 5’-CTT TGA GCC TGA GGT CTG GG-3’ (forward) and 5’-TGC GAA ATG GAG CTG GGA TT-3’ (reverse); cycE, 5’-AAC AAG AAT TAA CCG TAA CTGA-3’ (forward) and 5’-GCC TTA ATC GTA TGA ACA CTTA-3’ (reverse); actin, 5’-GAT CAT TGC TCC TCC TGA GC-3’ (forward) and 5’-ACT CCT GCT TGC TGA TCC AC-3’ (reverse). Relative quantification was processed through the comparative 2^−ΔΔCt method, with the results shown as the mean ± SD of values obtained from a minimum of three experiments.

Cell Culture, Transfection, and Immunoblotting.

S2 cells were cultured in serum-free insect cell medium (HyClone) with 1% penicillin/streptomycin (Sangon Biotech) at 25 °C. HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (Excell) and 1% penicillin/streptomycin (Sangon Biotech) at 37 °C with 5% CO₂. Transfection was carried out with PEI (Sigma) in accordance with the instructions provided by the manufacturer. Cells were collected for co-IP and IB assays 48 h post transfection as previously described (16). To knockdown hdac3 in S2 cells, the double-stranded RNA (dsRNA) was synthesized by the T7 RNAi Transcription Kit (Vazyme) according to the manufacturer’s instructions. DNA templates targeting hdac3 (aa1-173) and gfp (aa1-177) were generated by PCR and used for producing dsRNAs. The gfp-dsRNA was used as a negative control. The dsRNA was transfected into S2 cells using Lipo2000 (Vazyme) according to the manufacturer’s instructions. The antibodies used for IP and IB were as follows: mouse anti-Fg (1:500 for IP, 1:5,000 for IB, Sigma); mouse anti-HA (1:200 for IP, 1:2,000 for IB, Santa Cruz); rabbit anti-Hpo (1:2,000 for IB, from Prof. Shian Wu, Nankai University, China); rabbit anti-Wts (1:1000 for IB, from Prof. Shian Wu); rabbit anti-pWts (1:1000 for IB, from Prof. Shian Wu); mouse anti-Actin (1:5000 for IB, Proteintech); mouse anti-Myc (1:2,000 for IB, Santa Cruz); rabbit anti-Ace-Lys (1:1000 for IB, CST); rabbit anti-GST (1:2,000 for IB, Proteintech); mouse anti-His (1:1000 for IB, Proteintech); rabbit anti-HDAC3 (1:1000 for IB, 1:200 for IP, ABclonal); rabbit anti-pMST1/MST2 (1:1000 for IB, CST); goat anti-mouse HRP (1:10,000, Abmax); goat anti-rabbit HRP (1:10,000, Abmax).

Pull-Down Assay.

GST fusion proteins and His fusion proteins were expressed in E. coli BL21 upon IPTG induction, then purified by corresponding affinity beads (Beyotime). Subsequently, the beads were incubated with lysates from HEK-293T cells expressing the corresponding proteins at 4 °C for 3 h. Following three washes, the bound proteins were analyzed by western blot.

In Vitro Acetylation Assay.

GST-Hdac3 and His-Hpo purified from E. coli were incubated in HAT assay buffer (50 mM Tris-HCl pH 8.0, 10% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol) with acetyl-coenzyme A (60 μM) at 30 °C for 2 h, then the reaction products were analyzed by western blot.

Statistical Analysis.

The sizes of adult eyes, GFP⁺ areas of wing discs and the densities of IB bands were measured using Image J software. All data presented in the figures are representative of three or more independent replicates and are displayed as means ± SD. Statistical significance was assessed using a two-tailed unpaired Student’s t test in Prism software (GraphPad), with significance thresholds established at *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and ns (not significant, P > 0.05).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.