The Activity of Mblk-1, a Mushroom Body-selective Transcription Factor from the Honeybee, Is Modulated by the Ras/MAPK Pathway*

We previously identified a gene, termedMblk-1, that encodes a putative transcription factor with two DNA-binding motifs expressed preferentially in the mushroom body of the honeybee brain, and its preferred binding sequence, termed Mblk-1-binding element (MBE) (Takeuchi, H., Kage, E., Sawata, M., Kamikouchi, A., Ohashi, K., Ohara, M., Fujiyuki, T., Kunieda, T., Sekimizu, K., Natori, S., and Kubo, T. (2001) Insect Mol Biol 10, 487–494; Park, J.-M., Kunieda. T., Takeuchi, H., and Kubo, T. (2002) Biochem. Biophys. Res. Commun. 291, 23–28). In the present study, the effect of Mblk-1 on transcription of genes containing MBE in Drosophila Schneider's Line 2 cells was examined using a luciferase assay. Mblk-1 expression transactivated promoters containing MBEs ∼2–7-fold. Deletion experiments revealed that RHF2, the second DNA-binding domain of Mblk-1, was necessary for the transcriptional activity. Furthermore, mitogen-activated protein kinase (MAPK) phosphorylated Mblk-1 at Ser-444 in vitro, and the Mblk-1-induced transactivation was stimulated by phosphorylation of Ser-444 by the Ras/MAPK pathway in the luciferase assay. These results suggest that Mblk-1 is a transcription factor that might function in the mushroom body neuronal circuits downstream of the Ras/MAPK pathway in the honeybee brain.

The honeybee Apis mellifera L. is a social insect, and colony members perform various exquisite communications to maintain colony activities. Worker bees inform the other foragers of the direction and distance of a food source using dance language (1,2), which might require complex processing of sensory information in their brains. Little is known, however, regarding the molecular basis of their highly advanced behavior.
Mushroom bodies (MBs) 1 are believed to be involved in sensory integration, learning, and memory in insects (3,4). The honeybee MBs are well developed when compared with those of other insects. In the honeybee, the ratio of volume of MBs to that of whole brain is ϳ12%, whereas that of Drosophila is ϳ2% (5). Moreover, each MB of the honeybee has two calyces composed of two morphologically distinct types of interneurons, the large-and small-type Kenyon cells (5)(6)(7). On the other hand, in Drosophila, there is only one calyx, and the Kenyon cells are morphologically indistinct (8). These observations suggest that MB function is closely associated with the advanced honeybee behaviors.
To identify molecules involved in the highly advanced behaviors of the honeybees, we previously used the differential display method to identify a gene, termed Mblk-1, that is expressed preferentially in the large-type Kenyon cells of the honeybee brain (9). Mblk-1 encodes a novel protein consisting of 1598 amino acid residues with significant similarity to a nuclear factor encoded by the Drosophila melanogaster CG18389/E93 gene. The CG18389/E93 gene was identified previously as an ecdysone-inducible gene in the prepupal salivary gland (10) and was reported to encode a nuclear protein that is required for ecdysone-triggered programmed cell death during metamorphosis (11). The expression of CG18389/E93 in the adult and the biochemical characteristics of the protein, however, have not been examined.
Two putative DNA-binding motifs, termed RHF (region conserved between honeybee and fruit fly) 1 and RHF2, a nuclear localization signal, and Gln run were conserved between Mblk-1 and Drosophila E93 protein (9). RHF2 has significant sequence homology with proteins encoded by genes from nematoda (a polypeptide predicted by an open reading frame of the Caenorhabditis elegans cosmid T01C1), human (three polypeptides predicted by open reading frames of the chromosome 4 clone RP11-173B23 map 4, chromosome 11 clone RP11-162M10 map 11, and chromosome 10 clone RP11-175019, respectively), mouse (12), and sea urchin (a polypeptide predicted by an open reading frame of the Strongylocentrotus purpuratus EST253 coelomocyte cDNA 5Ј-end), suggesting that the intracellular functions of these proteins are conserved among the animal kingdom. The binding site selection method was used to identify the preferred binding sequence of Mblk-1 as 5Ј-CCCTATC-GATCGATCTCTACCT-3Ј, termed MBE (Mblk-1-binding element). Truncated Mblk-1 protein that contains either RHF1 or RHF2 can also bind MBE but with much lower affinity than intact Mblk-1. An in vitro pull-down assay indicated that RHF1 and RHF2 afford homodimeric bindings, suggesting that Mblk-1 functions as a dimer (13). These results suggest that Mblk-1 is a transcription factor that functions in the MB neural circuits in the honeybee brain. The molecular function of Mblk-1, however, has not been characterized previously.
In general, long term memory formation requires protein synthesis. This has been confirmed in animals ranging from insects and mollusks to mammals (14). In the honeybee, the formation of long term memory lasting 4 days requires both de novo transcription and translation (15). Mitogen-activated protein kinase (MAPK) has a role in long term memory in a number of different learning paradigms in invertebrates and vertebrates (16 -23). Therefore, the MAPK signaling pathway is a good candidate involved in long term changes in neuronal gene expression triggered by extracellular stimuli.
In the present study, we used a luciferase assay to determine whether Mblk-1 transactivates promoters containing MBEs and can be modulated by the Ras/MAPK pathway. The results indicated that Mblk-1 is a transcription factor that might function in MB neural circuits directly modulated by the Ras/ MAPK pathway.

EXPERIMENTAL PROCEDURES
Cell Culture-SL-2 cells (Schneider's Line 2 cells derived from D. melanogaster embryos) (24) were maintained in Schneider's Drosophila medium (Invitrogen) with the addition of heat-inactivated fetal bovine serum (Sigma), 5 mg/ml polypeptone, and antibiotics (100 units/ml penicillin G and 0.1 mg/ml streptomycin) (Invitrogen). The cells were grown in monolayers at 27°C Plasmid Construction and Mutagenesis-A luciferase reporter vector containing either the Hsp70Bb core promoter from pUAST (pGL3H) or the P-element core prompter and hsp70 leader from WTP-1 (pGL3PH) was prepared by amplifying the corresponding sequences by PCR and ligating each of these PCR products to the pGL3-basic vector (Promega Co., Madison, WI) containing the firefly luciferase gene. Two or six tandem copies of either MBE or UAS G (Gal4 upstream activating sequence) were subcloned into the upstream regions (at positions Ϫ11 to Ϫ28) of pGL3H or pGL3PH.
To create the RHF1 deletion mutant (⌬586 -636) of Mblk-1, we utilized a HindIII site, which was located just at the end of RHF1. The DNA fragment that corresponds to ϩ1691 to ϩ2030 with an extra nucleotide for the HindIII site at the 3Ј-end was amplified by PCR, and the resulting PCR product was used to replace the Asp718-HindIII fragment of the Mblk-1-(384 -808) cDNA. The insert of this plasmid was then used to replace the NotI-BlpI fragment of the full-length Mblk-1 cDNA, and the resulting insert was subcloned into pPac-PL (⌬586 -636/pPac-PL). To create the RHF2 deletion mutant (⌬1031-1088) of Mblk-1, DNA fragments that correspond to ϩ3188 to ϩ3364 and ϩ3539 to ϩ5071 with extra nucleotides for the Asp718 site at the 5Ј-and 3Ј-end, respectively, were amplified by PCR. The resulting PCR products were then digested with Asp718, ligated to each other, subcloned, and digested with BsgI and SpeI. The resulting fragment was used to replace the BsgI-SpeI of the Mblk-1-(776 -1207) plasmid. The resulting plasmid was again used to replace the BlpI-BstZ17I of the full-length Mblk-1 plasmid and subcloned into pPac-PL (⌬1031-1088/pPac-PL).
To create the Mblk-1S444A mutant, PCR was performed using the first sense primer, 5Ј-CACCTCTCGCACCGCAGAGCGACAGTAGCA-3Ј, where the underline indicates nucleotides corresponding to mutated Ala-444; the second sense primer, 5Ј-aaaTGATCAACCACCTCTCG-CACCGCAGAG-3Ј; the antisense primer, 5Ј-CTAGGTACCGGT-GAGAGCC-3Ј), which correspond to ϩ1588 to ϩ1714 of Mblk-1; and the full-length Mblk-1 cDNA as a template. The PCR product was subcloned, digested with BclI-Asp718, and used to replace the BclI-Asp718 of the Mblk-1-(384 -808) plasmid. This mutated Mblk-1-(384 -808) plasmid was used to replace the NotI-BlpI of the full-length Mblk-1 plasmid and subcloned into pPac-PL (Mblk-1S444A/pPac-PL). The pPac expression plasmids containing the actin 5C promoter and either Ras1 V12 (constitutively active Ras1) or MAPK Sem (constitutively active MAPK) cDNA were kind gifts from Dr. T. Hsu (Medical University of South Carolina, Charleston, SC) (26) Transfections and Reporter Assay-Transfection experiments were performed essentially as described previously (27). SL-2 cells (2-5 ϫ 10 5 cells/ml) were cultured in 1 ml of Schneider's Drosophila medium containing heat-inactivated fetal bovine serum and 5 mg/ml polypeptone in a 12-well plate for 24 h at 27°C to allow them to adhere to the dish, and the medium was discarded. A mixture of plasmid DNA (0.5 g) was incubated with 2 l of Cellfectin reagent (Invitrogen) in 0.1 ml of Drosophila serum-free medium (Invitrogen) for 30 min, and then 0.4 ml of Drosophila serum-free medium was added to increase the volume. The resulting total mixture was added to the adhered cells and incubated for 4 h to accomplish transfection. The medium was then replaced with fresh Schneider's Drosophila medium containing heat-inactivated fetal bovine serum and 5 mg/ml polypeptone, and incubation was continued. The reporter gene activities were assayed 42-44 h later. Cells were collected and lysed in the reporter lysis buffer (Promega Co.), and luciferase activity in the lysate was measured in a luminometer (Lumat LB 9507; Berthold) immediately after addition of the substrate luciferin (Promega Co.). ␤-Galactosidase activity in the lysate was measured using o-nitrophenyl-␤-D-galactopyranoside as a substrate, and the values were used to normalize the efficiency of transfection. The mixture of plasmid DNA (0.5 g) consisted of the luciferase reporter vector (50 ng), an actin 5C-␤-galactosidase reporter vector (50 ng), and 0.0 -0.4 g of mutant Mblk-1 expression vector with 0.0 -0.4 g of empty pPac-PL (total, 0.5 g).

Mblk-1 Is a Sequence-specific Transcriptional Activator-
The luciferase assay was used to examine whether Mblk-1 transactivates genes containing MBEs in their promoters. Drosophila SL-2 cells were cotransfected with a luciferase reporter vector containing MBEs and a minimal promoter and an Mblk-1 expression vector driven by the Drosophila actin 5C promoter. Two kinds of minimal promoters were used: the Hsp70Bb core promoter and the P-element core promoter with an hsp70 leader. With the Hsp70Bb core promoter, Mblk-1 expression increased luciferase activity ϳ2-fold depending on the MBE copy number (Fig. 1A). In contrast, neither the transfection of an empty expression vector instead of the Mblk-1 expression vector nor the reporter vector containing UAS G s instead of MBEs increased the activity.
Using the P-element core promoter with the hsp70 leader, Mblk-1 expression did not increase the activity for the reporter vector containing two copies of MBE (Fig. 1B). Luciferase activity increased ϳ7-fold for the reporter vector containing six MBE copies as compared with that containing UAS G s. This increase was not observed when an empty vector was used instead of the Mblk-1 expression vector. These results indicate that Mblk-1 can transactivate minimal promoters driven by MBE, but not those driven by UAS G . Furthermore, the level of transactivation correlated with the MBE copy number. Thus, Mblk-1 is a sequence-specific transcriptional activator.
Identification of the Functional Domains of Mblk-1-To identify the functional domains of Mblk-1, we created various deletion mutants of Mblk-1 and examined their effects on the expression of the reporter gene. Six series of N-terminal deletion mutants were first constructed because this region contains some characteristic domains such as 25-amino-acid residues that share high sequence homology (68%) with CG18389/ E93 (amino acid positions 29 -53), Thr runs (106 -130), and Gln runs (164 -177) (28). There were no significant differences between the transcriptional activities of this series of deletion mutants and wild-type Mblk-1 ( Fig. 2A), indicating that Mblk-1 function is independent of the specificity of the N-terminal 383 residues.
Another series of six deletion mutants was constructed to assess the importance of the RHF1 and RHF2 domains. When the C-terminal 390 amino acid residues were deleted (⌬1208 -1597), there was no appreciable effect (Fig. 2B). When the C-terminal 399 residues including the RHF2 domain (⌬809 -1597) were deleted, however, the transcriptional activity of Mblk-1 was almost completely lost. In contrast, the luciferase activity gradually decreased as the N-terminal regions were deleted (⌬1-383, ⌬1-775). In addition, deletion of the 430 residues including the RHF2 domain (⌬1205-1597) caused complete loss of transcriptional activity (Fig. 2B). These results indicate that the 399 residues including the RHF2 domain are necessary for Mblk-1 function.
To directly assess the significance of RHF1 and RHF2 for Mblk-1 transcriptional activity, RHF1 and RHF2 deletion mutants of Mblk-1 were created and examined using the luciferase assay (Fig. 2C). The transactivation activity of the RHF2 deletion mutant (⌬1031-1088) was almost completely lost. In contrast, RHF1 (⌬586 -636) deletion did not have an appreciable inhibitory effect, indicating that the RHF2 domain is necessary for Mblk-1 transcriptional activity (Fig. 2C). In contrast, RHF1 is dispensable, at least in this assay system.
Mblk-1-induced Transactivation Was Stimulated by the Ras/ MAPK Pathway-We next tested whether phosphorylation at Ser-444 by MAPK affected Mblk-1 transcriptional activity. The increase in luciferase activity was reduced to ϳ65% when the Mblk-1S444A protein was expressed instead of intact Mblk-1 (Fig. 5A). These results strongly suggest that the activity of Mblk-1 can be modulated, at least in part, by direct phospho- rylation by MAPK. To test this possibility further, we examined whether the Ras/MAPK pathway stimulated Mblk-1 transcriptional activity. For this, either the pPacMAPK Sem or the pPacRas1 V12 plasmid, which express an activated form of Drosophila MAPK (37) or Ras1 (38,39), respectively, was cotransfected, and luciferase activity was examined. Cotransfection of pPacMAPK Sem or pPacRas1 V12 increased the transcriptional activity of intact Mblk-1 ϳ2-fold (Fig. 5B). In contrast, neither MAPK Sem nor Ras1 V12 had any effect on basal activation. These results clearly indicated that Mblk-1-induced transactivation can be stimulated by the Ras/MAPK pathway. Mblk-1S444A transcriptional activity, however, was also increased to some extent by expression of activated Ras1 or MAPK, suggesting that Mblk-1 was also modulated by the Ras/MAPK pathway at a site other than Ser-444. Treatment with forskolin induces activation of endogenous PKA in SL-2 cells (40). Thus, we also examined the effect of PKA activation by treatment with forskolin on Mblk-1transcriptional activity. There was no detectable effect, however, on Mblk-1 transactivation (data not shown), suggesting that PKA has little effect on Mblk-1 activation, although PKA phosphorylated Mblk-1 in vitro. DISCUSSION We previously identified the preferred binding sequence of Mblk-1, termed MBE. It remained uncertain, however, whether Mblk-1 has transcriptional activity and whether MBE is important for Mblk-1-mediated transcriptional activity. We report the first direct evidence for the transcriptional activity of Mblk-1/E93 insect proteins. Thus, the honeybee Mblk-1 is the first transcription factor identified that is expressed preferentially in the MBs of the insect brain.
Deletion experiments revealed that Mblk-1 contains functional regions for activation. Among them, RHF2 was necessary for Mblk-1 activity. Although truncated Mblk-1 containing RHF1 can also bind to MBE in vitro (13), there was no appreciable decrease in the transactivation activity of RHF1 deletion mutants in the luciferase assay. At present, it is not clear whether RHF1 has some functions in Mblk-1. It is possible, however, that truncated RHF1 binds to MBE in vitro because the domain is readily exposed to bind to MBE, whereas RHF1 is usually hidden in the intact Mblk-1 molecule, to be exposed only in response to a particular signal(s). Our luciferase assay might have lacked such a particular signal(s) and thus failed to detect any effect of RHF1 deletion.
Covalent modification by phosphorylation is a potential route for Mblk-1 regulation. Previous studies established a crucial role of second messenger-dependent kinases in the modulation of neuronal activity, and their involvement in learning and memory (29 -32) and many different types of stimuli that affect gene expression also leads to the activation of protein kinases (41). Thus, it is likely that Mblk-1 function is also regulated by phosphorylation. We demonstrated that Mblk-1 activity could be modulated by direct phosphorylation by the Ras/MAPK pathway. Specifically, we identified Ser-444 as one of the important phosphorylation sites involved in determining the magnitude of the Mblk-1 transactivating capacity. It remains unknown, however, how the activity of Mblk-1 can be stimulated by the phosphorylation by MAPK. We previously reported that Mblk-1 functions as a dimer using an in vitro pull-down assay (13). Phosphorylation did not have a significant effect, however, on homophilic protein-protein interactions (data not shown). Some noteworthy possibilities are: 1) phosphorylation allows translocation of Mblk-1 into the nucleus (42), 2) DNA binding activity of Mblk-1 might be modulated by phosphorylation (43), and 3) phosphorylation might affect interaction of the transactivation domains with the transcriptional machinery (44,45). Long term memory formation is generally dependent on protein synthesis (14), and a role for MAPK in long term memory has been demonstrated in a number of different learning paradigms in invertebrates and vertebrates (17)(18)(19)(20)(21)(22)(23)32). Among the transcription factors involved in learning, memory, and neuronal plasticity, cAMP-response element-binding protein (CREB) is best characterized (46,47). CREB transcriptional activity is also stimulated by the Ras/MAPK pathway, and the Ras/MAPK-dependent phosphorylation of CREB is performed by several different kinases, including members of the ribosomal S6 kinase and mitogen-and stress-activated protein kinase families (48 -50). Furthermore, CREB is also activated via PKA and CaMK pathways (51,52). Similarly, the phosphorylation of Mblk-1 by PKA and CaMKII in vitro (Fig. 3B) and the partial increase in the transcriptional activity of Mblk-1S444A by the Ras/MAPK pathway (Fig. 5B) suggest that Mblk-1 is modulated via various signaling pathways other than the Ras/ MAPK pathway.
We previously demonstrated that gene expression for inositol 1,4,5-triphosphate (IP 3 ) receptor, CaMKII, and IP 3 phosphatase is concentrated in the large-type Kenyon cells of the honeybee brain (34,53,54). PKA is also expressed preferentially in the large-type Kenyon cells (33). To our knowledge, Mblk-1 is the first MB-selective transcription factor that might participate in transcriptional activation of some genes for proteins involved in synaptic plasticity like IP 3 receptor, CaMKII, PKA, and IP 3 phosphatase in the MB neural circuits and might therefore be responsible for the status of MBs as the main association and memory centers of the honeybee brain. The identification of possible target genes for Mblk-1 and its biologic function might provide important clues to the molecular basis that underlies the MB functions.