Identification of new genes involved in human adipogenesis and fat storage

Abstract

Since the worldwide increase in obesity represents a growing challenge for health care systems, new approaches are needed to effectively treat obesity and its associated diseases. One prerequisite for advances in this field is the identification of genes involved in adipogenesis and/or lipid storage. To provide a systematic analysis of genes that regulate adipose tissue biology and to establish a target-oriented compound screening, we performed a high throughput siRNA screen with primary (pre)adipocytes, using a druggable siRNA library targeting 7,784 human genes. The primary screen showed that 459 genes affected adipogenesis and/or lipid accumulation after knock-down. Out of these hits, 333 could be validated in a secondary screen using independent siRNAs and 110 genes were further regulated on the gene expression level during adipogenesis. Assuming that these genes are involved in neutral lipid storage and/or adipocyte differentiation, we performed InCell-Western analysis for the most striking hits to distinguish between the two phenotypes. Beside well known regulators of adipogenesis and neutral lipid storage (i.e. PPARγ, RXR, Perilipin A) the screening revealed a large number of genes which have not been previously described in the context of fatty tissue biology such as axonemal dyneins. Five out of ten axonemal dyneins were identified in our screen and quantitative RT-PCR-analysis revealed that these genes are expressed in preadipocytes and/or maturing adipocytes. Finally, to show that the genes identified in our screen are per se druggable we performed a proof of principle experiment using an antagonist for HTR2B. The results showed a very similar phenotype compared to knock-down experiments proofing the "druggability". Thus, we identified new adipogenesis-associated genes and those involved in neutral lipid storage. Moreover, by using a druggable siRNA library the screen data provides a very attractive starting point to identify anti-obesity compounds targeting the adipose tissue.

Figure 1. siRNA screening procedure.

Two read-out-assays were performed in 96-well-plates to determine accumulation of neutral lipids and DNA-content of untransfected and transfected (pre)adipocytes (N = negative control: control (scrambled) siRNA; P = positive control, PPARγ-siRNA; T = transfection reagent, no siRNA). Outer wells were filled with PBS to reduce edge effects.

Figure 2. Data correction and quality control of the primary screen.

(A) Correction of neutral lipid values with respect to changes in cell number. Z-scores of lipid values are illustrated before and after the amount of DNA was factored in. (B) Q-Q plot of normally distributed quantiles against screening result (Z-score) quantiles (red circles = positive control/PPARγ-siRNA; blue circles = negative control/scrambled siRNA). A perfect fit to a normal distribution is represented by red dotted line. (C) Experiment-wide quality plot focusing on controls. Signal from positive (red dots; PPARγ-siRNA) and negative (blue dots; control (scrambled) siRNA) controls plotted against plate number. The distance between the two distributions was quantified by the Z′-factor (0.42). For data normalization, the method ‘normalized percent inhibition’ (NPI) was applied.

Figure 3. Ingenuity Pathways Analysis of positive hits identified in the primary screen.

(A) Depicted are the three most significant canonical pathways with cAMP signaling being the most prominent. (B) Top five associated network functions of related genes. Two of the five top networks are believed to be directly involved in lipid metabolism. (C) The network displaying the highest score shows PPARγ and RXRA, known regulators of adipogenesis, as central genes.

Figure 4. Network of axonemal dyneins.

(A) Dynein network: The green coloring indicates 4 out of 5 hits regarding axonemal dyneins. DNAH7 (also identified in our screen) was not part of the network. Blue symbols represent dyneins which were part of the library but were not determined as hits. Dyneins colored in white could not be investigated using the druggable siRNA library. (B) IPA showed an accumulation of motor proteins including dyneins and kinesins. (C) Messenger RNA levels of DNAH7, DNAH8 and DNAH17 in the course of adipocyte differentiation. Expression data at day 0 (preadipocytes) were set as 100%. All CT-values analyzed were between 28 and 34.

Figure 5. Knock-down of axonemal dyneins and classification of corresponding phenotypes.

InCell-Western Analyses were performed using aP2- and Perilipin-specifc antibodies. (A) Depicted are immunofluorescence images showing aP2 and Perilipin stainings. (B) InCell-Western-Image using aP2 and Perilipin antibodies showing PPARγ knock-down (rimmed in red) and scrambled siRNA edged in blue. (C) Classify of ‘target phenotypes’ into the following categories: (1) target-knock-down that reduce differentiation, [aP2 and Perilipin A signals decreased ≥20% compared to controls, green arrow]; (2) target-knock-down that stimulate differentiation [aP2 and Perilipin A signals increased ≥20% compared to controls, red arrow] and (3) target-knock-down that caused no changes in differentiation [no significant changes in aP2 and Perilipin signals]. For targets highlighted in grey, signals of both markers decreased or increased. For those targets under laid in yellow, none marker changed. (D) Messenger RNA levels of adipocyte-relevant transcription factors as well as IL6 and IL1ß after DNAI2- and DNAH8 knock-down. Differentiation was initiated 3 days after transfection. The qRT-PCR was performed on day 5 post-transfection. Results are depicted as mean ± SD.

Figure 6. Genes identified and validated in our siRNA-screening showing the most significant changes in gene expression during adipogenesis were again transfected with specific siRNAs to classify the corresponding phenotype.

Knock-down of genes caused a decreased lipid accumulation (upper box) or increased lipid accumulation (Lower box). InCell-Western Analyses were performed using aP2- and Perilipin-specific antibodies to classify target phenotypes into the following categories: (1) target-phenotype that reduce differentiation, [aP2 and Perilipin A signals decreased ≥20% compared to controls, green arrow]; (2) target-knock-down that stimulate differentiation [aP2 and Perilipin A signals increased ≥20% compared to controls, red arrow] and (3) target-phenotype that caused no changes in differentiation [none alterations in aP2 and Perilipin signals]. For targets highlighted in grey, signals of both markers decreased or increased. For those targets under laid in yellow, none marker changed.

Figure 7. Effects of the HTR2B-antagonist RS127445 on neutral lipid accumulation during human primary (pre)adipocyte differentiation.

(A) Increase in lipid concentration in maturing adipocytes after treatment with 50 µM RS127445. Neutral lipid accumulation is shown relative to untreated control cells set as 100%. Results are depicted as mean ± SD (n = 10). Significant differences are marked with an asterisk (* for p<0.0001). (B) Cell viability of differentiating adipocytes after treatment with 50 µM RS127445 is shown relative to untreated control cells set as 100%. Results are depicted as mean ± SD. (C) Fluorescence microscopy after incubation with or without 50 µM RS127445 and lipid staining (yellow: neutral lipids; scale bar: 200 µm).