Reduction of Cellular Lipid Content by a Knockdown of Drosophila PDP1 γ and Mammalian Hepatic Leukemia Factor

In exploring the utility of double-stranded RNA (dsRNA) injections for silencing the PAR-domain protein 1 (Pdp1) gene in adult Drosophila, we noticed a dramatic loss of fat tissue lipids. To verify that our RNAi approach produced the expected Pdp1 knockdown, the abdominal fat tissues sections were stained with PDP1 antibodies. PDP1 protein immunostaining was absent in flies injected with dsRNA targeting a sequence common to all known Pdp1 isoforms. Subsequent experiments revealed that lipid staining is reduced in flies injected with dsRNA against Pdp1γ (fat body specific) and not against Pdp1 ε (predominantly involved in circadian mechanisms). Drosophila PDP1γ protein shows a high homology to mammalian thyrotroph embryonic factor (TEF), albumin D site-binding protein (DBP), and hepatic leukemia factor (HLF) transcription factors. In an in vitro model of drug- (olanzapine-) induced adiposity in mouse 3T3-L1 cells, the mRNA content of HLF but not TEF and DBP was increased by the drug treatment. A knockdown of the HLF mRNA by transfecting the cultures with HLF dsRNA significantly reduced their lipid content. Furthermore, the HLF RNAi prevented olanzapine from increasing the cell lipid content. These results suggest that the PDP1/HLF system may play a role in physiological and drug-influenced lipid regulation.


Introduction
Contrary to the previous belief that in adult Drosophila (fruit fly) RNA interference (RNAi) cannot occur by exogenous administration of double-stranded RNA (dsRNA), it was conclusively demonstrated that this type of systemic RNAi is operative in adult flies [1]. This mechanism explains the efficacy of dsRNA injections into adult Drosophila, applied as a tool for targeted gene knockdown in adult organisms [2][3][4]. The advantage of this RNAi method is that it avoids the unwanted developmental alterations and the possible side effects of genetic manipulations involved in alternative RNAi approaches. In the course of exploring the utility of dsRNA injections for silencing the PAR-domain protein 1 (Pdp1) gene in adult Drosophila, we noticed a peculiar RNAi-related phenotype, a dramatic loss of fat tissue lipids. Here, we report the follow-up study aimed at exploring this serendipitous discovery.
Drosophila Pdp1 encodes a transcription factor highly homologous to the proline-and acidic amino acid-rich (PAR) subfamily of mammalian bZIP transcription factors, albumin D site-binding protein (DBP), hepatic leukemia factor (HLF), and thyrotroph embryonic factor (TEF). PDP1 was originally identified as a regulator of the muscle activator region [5]. Subsequently, it was established that Pdp1 is a component of the Drosophila circadian network and that its expression is directly activated by dClock/Cycle genes [6,7]. The Pdp1 gene encodes multiple transcripts, which are differentially expressed during embryogenesis [8]. Of the six Pdp1 isoforms, Pdp1 is the one that shows a circadian expression and is involved in the regulation of the Drosophila circadian behavior [9]. Whereas Pdp1 is predominantly expressed in the nervous system, Pdp1 is predominantly expressed in the fat body [8]. The mammalian homologous factors DBP, TEF, and HLF also show a circadian rhythm in their expression/accumulation; their absence results in epilepsy [10]. Furthermore, these mammalian proteins have recently been linked to a mechanism of fatty acid regulation [11]. To address possible similarities between Drosophila Pdp1 and its mammalian homologues, in addition to the experiments in 2 Journal of Lipids

Target
Forward fruit flies, we employed a model of drug-induced adiposity in mammalian 3T3-L1 cells in vitro [12].  staining; that is, the effect was rather obvious. To verify this observation, we employed the ORO lipid staining. In these experiments, a total Pdp1 RNAi was achieved by injecting flies with the dsRNA targeted at a sequence common to all known Pdp1 isoforms. This resulted in the loss of ORO lipid staining as exemplified by the staining of Drosophila abdominal sections ( Figure 1). To verify that our RNAi approach produced the expected Pdp1 knockdown, we stained the whole fly sections including the abdominal fat tissue with PDP1 protein antibodies. PDP1 immunostaining revealed a robust nuclear PDP1 staining in the cells of flies injected with control dsRNA and the absence of PDP1 staining in Pdp1 dsRNA injected flies ( Figure 2). Also, in these sections, we confirmed the lipidreducing effect of the total Pdp1 dsRNA; the number and the size of lipid-containing droplets in fat tissue cells were reduced in the RNAi samples ( Figure 2). The previous results were obtained with the dsRNA targeted at a sequence common to all known Pdp1 isoforms.

Drosophila
In subsequent experiments, we investigated the Pdp1 isoform specificity of the lipid-reducing phenotype. Hence, flies were injected with dsRNA targeted specifically against Pdp1 (predominantly present in the nervous system and involved in circadian mechanisms) and Pdp1 (expressed in the fat body), respectively, and their body lipid content was quantified. Only the Pdp1 dsRNA significantly reduced the body lipid content (Figure 3).

Mammalian HLF dsRNA RNAi.
Drosophila PDP1 protein shows a high homology to TEF, DBP, and HLF members of the PAR subfamily of mammalian bZIP transcription factors ( Figure 4) [13]. To explore possible similarities between Pdp1 and these factors, we selected the mouse 3T3-L1 preadipocytes in vitro. These cells have been used as a model for drug-induced adipogenic effects; that is, treatment of 3T3-L1 cells during differentiation into adipocytes with the antipsychotic drug olanzapine increases their lipid content [12]. In this model, we found that the adipogenic olanzapine treatment increases the mRNA content of HLF but not TEF and DBP ( Figure 5). On the other hand, a knockdown of the endogenous HLF mRNA by transfecting the cultures with HLF dsRNA significantly reduced their lipid content ( Figure 6). In an experiment in which olanzapine and vehicle treatments were initiated 24 h after the initiation of transfection and conducted for the next 24 h, we found that olanzapine treatment increased lipid content in naïve and sham-dsRNA transfected cells but not in HLF dsRNAtransfected cells (Figure 7).

Discussion
In this work, we confirmed and expanded our serendipitous observation that a systemic PDP1 knockdown in adult flies, induced by injections of Pdp1 dsRNAs, leads to a significant lipid decrease, and we found that similar phenotype can be induced by HLF RNAi in mouse adipocytes in vitro. Collectively, our results demonstrated that a reduction of the PDP1 /HLF transcription factor leads to a decreased lipid content.
Both mammalian HLF and Drosophila PDP1 are known components, that is, output regulators, of circadian cycles and as such have been linked to metabolic regulation [15]. The expression of Drosophila circadian genes (i.e., peripheral clocks) in the fat body has been shown to play a role in the  regulation of fly metabolism [16]. Of the six Pdp1 isoforms, Pdp1 is the one that is characterized by a prominent circadian expression and is involved in the regulation of the Drosophila circadian behavior [9]. One of the circadian functions of Pdp1 is in regulating the circadian output gene takeout [17]. In male Drosophila, takeout is abundant in the fat body and plays a role in courtship behavior of these flies [18]. It was suggested that Pdp1 -mediated regulation of the fat body genes may influence this type of fly behavior [17]. In our Drosophila experiments, lipid staining was decreased by Pdp1 and not Pdp1 RNAi. Hence, it would be interesting to elucidate whether in addition to Pdp1 also Pdp1 regulates the expression of the output genes such as takeout and whether this mechanism mediates the observed lipiddecreasing effects of Pdp1 knockdown. Our experiments with adipocytes show that in addition to systemic PDP1/HLF alterations (e.g., systemic PDP1 knockdown in adult flies), also direct cellular HLF alterations (e.g., knockdown in 3T3-L1 cells) can reduce lipid content.
In our in vitro experiments, a drug-induced adiposity was accompanied by increased levels of HLF mRNA, whereas HLF RNAi was accompanied by decreased lipid content. Furthermore, it was previously reported that in a mouse model of severe reduction of lipid accumulation and severe loss of body weight, the liver HLF mRNA levels, along with the TEF and DBP mRNA levels, are significantly reduced [19]. Hence, HLF appears to be involved in the physiological regulation of cellular lipid levels.
In our experimental conditions, the RNAi-mediated HLF mRNA reduction significantly diminished a drug -induced adiposity (i.e., olanzapine). Olanzapine belongs to a class of drugs known as the second generation antipsychotic drugs (SGADs). All these compounds are capable of triggering significant weight gain associated with adverse metabolic alterations [20,21]. It has been proposed that these side effects may occur by a direct stimulatory action of SGADs on adipocytes [12,22,23]. Our in vitro experiments confirmed the direct adipogenic action of olanzapine and found that this action can be diminished by HLF reduction. In Figure 4: Homology between Drosophila PDP1 and mammalian bZIP transcription factors HLF, TEF, and DBP. Shown is the protein sequence alignment in the b-ZIP structural regions with the highest homology, analyzed using the online CLUSTALW multiple sequences alignment tool [13]. The NCBI database accession information is as follows. Drosophila PDP1 Q9TVQ4; HLF: mouse NP 766151, rat Q64709, and human NP 002117; TEF: mouse NP 059072, rat NP 062067, and human NP 003207; DBP: mouse NP 058670, rat NP 036675, and human NP 001343. Highlighted in yellow are the sequences of Drosophila PDP1 and mouse HLF. the therapy of psychiatric patients with SGADs, a better understanding of the mechanisms that lead to metabolic side effects is needed to identify the risk factors that facilitate and exacerbate this SGADs-associated clinical problem. Our results suggest for the first time that the HLF pathway could be such a mechanism. Furthermore, the observed direct susceptibility of the adipocyte HLF and lipids to regulation by drugs (e.g., olanzapine-increased HLF mRNA and lipid contents) suggests that future pharmacological tools could be tailored specifically to the adipocyte HLF pathway to interfere therapeutically with the mechanisms of adiposity.  Control dsRNA HLF dsRNA * Naïve Figure 6: Effect of mammalian HLF RNAi on lipid content in mouse 3T3-L1 cells. Cells in culture were transfected with control and HLF dsRNA and processed for lipid assay 48 h later (see text for details). Lipid content in 3T3-L1 cells was reduced by HLF dsRNA treatment ( * < 0.001 versus naïve; = 6; mean ± standard error mean), which also reduced HLF mRNA content (not shown).

Lipids (unit)
Vehicle Olanzapine Dimitrijevic for technical help, Robert V. Storti for PDP1 antibodies, and the late Erminio Costa, Director of the Psychiatric Institute, for support.