Cellular and Molecular Life Sciences https://doi.org/10.1007/s00018-018-2799-7 Cellular andMolecular Life Sciences

ORIGINAL ARTICLE

UDP‑glucose glucosyltransferase activates AKT, promoted proliferation, and doxorubicin resistance in breast cancer cells

Marthe‑Susanna Wegner1 · Nina Schömel1 · Lisa Gruber1 · Stephanie Beatrice Örtel1 · Matti Aleksi Kjellberg2 · Peter Mattjus2 · Jennifer Kurz3 · Sandra Trautmann1 · Bing Peng4 · Martin Wegner5 · Manuel Kaulich5 · Robert Ahrends4 · Gerd Geisslinger1,3 · Sabine Grösch1

Received: 5 October 2017 / Revised: 19 February 2018 / Accepted: 13 March 2018 © Springer International Publishing AG, part of Springer Nature 2018

Abstract The UDP-glucose ceramide glucosyltransferase (UGCG) is a key in the synthesis of glycosylated , since this enzyme generates the precursor for all complex (GSL), the GlcCer. The UGCG has been asso- ciated with several cancer-related processes such as maintaining cancer stem cell properties or multidrug resistance induc- tion. The precise mechanisms underlying these processes are unknown. Here, we investigated the molecular mechanisms occurring after UGCG overexpression in breast cancer cells. We observed alterations of several cellular properties such as morphological changes, which enhanced proliferation and doxorubicin resistance in UGCG overexpressing MCF-7 cells. These cellular efects seem to be mediated by an altered composition of -enriched microdomains (GEMs), especially an accumulation of globotriaosylceramide (Gb3) and glucosylceramide (GlcCer), which leads to an activation of Akt and ERK1/2. The induction of the Akt and ERK1/2 signaling pathway results in an increased expression of multidrug resistance protein 1 (MDR1) and anti-apoptotic and a decrease of pro-apoptotic . Inhibition of the protein kinase C (PKC) and phosphoinositide 3 kinase (PI3K) reduced MDR1 gene expression. This study discloses how changes in UGCG expression impact several cellular signaling pathways in breast cancer cells resulting in enhanced proliferation and multidrug resistance.

Keywords Glycosphingolipids · Glycosphingolipid-enriched microdomains · Multidrug resistance · MDR1 · Glucosylceramide · Apoptotic

Introduction the fundamental cellular processes in breast (cancer) cells, which are leading, for example, to promoted proliferation In the year 2015, breast cancer was declared as the second and multidrug resistance development. Multidrug resist- leading cause of cancer death in women in industrial coun- ance of cancer cells is the main cause of therapy failure. It tries [1]. This fact underlines the importance to investigate is accomplished by alteration of myriad cellular signaling cascades resulting, for example, in enhanced expression of multidrug resistance proteins, which transport toxic sub- Electronic supplementary material The online version of this stances out of cancer cells. article (https​://doi.org/10.1007/s0001​8-018-2799-7) contains supplementary material, which is available to authorized users.

* Marthe‑Susanna Wegner 3 Fraunhofer Institute for Molecular Biology and Applied [email protected]‑frankfurt.de Ecology IME, Project Group Translational Medicine and Pharmacology (TMP), Frankfurt am Main, Germany 1 pharmazentrum frankfurt/ZAFES, Institute of Clinical 4 Leibniz-Institut für Analytische Wissenschaften, ISAS e. V., Pharmacology, Johann Wolfgang Goethe University, Otto‑Hahn‑Straße 6b, 44227 Dortmund, Germany House 74, Theodor Stern‑Kai 7, 60590 Frankfurt am Main, Germany 5 Institute of Biochemistry II, Johann Wolfgang Goethe University, Theodor Stern‑Kai 7, 60590 Frankfurt am Main, 2 Biochemistry, Faculty of Science and Engineering, Åbo Germany Akademi University, Artillerigatan 6A, III, BioCity, 20520 Turku, Finland

Vol.:(0123456789)1 3 M.-S. Wegner et al.

The UDP-glucose ceramide glucosyltransferase (UGCG) the Golgi apparatus where they are used for the synthesis was frst cloned by Ichikawa et al. [2] and is connected to of (constituted of ceramide and phospho- processes of multidrug resistance in cancer cells. This in choline or a phosphoethanolamine group) or cerebrosides the cis-Golgi apparatus residing protein transfers a glucose (Fig. 1). Subsequently, cerebrosides can be used as precur- moiety in β-linkage to the position 1 hydroxyl group of cera- sors for synthesis of lactosylceramides (LacCer), which are mide, which results in glucosylceramide (GlcCer) formation also named . Globosides can be metabolized to (Fig. 1). GlcCer, also named cerebrosides, serve as precur- . sors for all complex glycosphingolipids (GSLs). , Knockout of the UGCG in mice leads to embryonic which are used for GlcCer synthesis, are produced in the lethality during the phase of gastrulation [4]. In addi- endoplasmic reticulum by six mammalian ceramide syn- tion, a constitutive disruption of this protein in mice epi- thase (CerS) isoforms, which have a substrate specifcity dermis results in loss of skin barrier function and death for acyl-CoenzymeAs (acyl-CoAs) of defned chain length due to dehydration [5]. Amen et al. showed that GlcCer [3]. Accordingly, each CerS isoform produces ceramide spe- is essential for proper formation of the lamellar body, cies of a specifc chain length. Ceramides are transported to regular , and composition of lipids in the

Fig. 1 Schematic overview of the potential mechanisms of UGCG- membrane protein activities and activation of signaling pathways like derived GSLs infuencing membrane lipid composition resulting Akt and ERK1/2. Activation of these kinases increases proliferation in cellular signaling pathway induction. Overexpression of UGCG and MDR1 expression. GSL glycosphingolipid, GEM glycosphin- results in increased GlcCer concentration leading to Gb3 accumu- golipid-enriched microdomain, P-gp P-glycoprotein, bis I bisindolyl- lation and augmented integration of GlcCer in plasma membrane maleimide I, PKC protein kinase C, PI3K phosphoinositide 3-kinase structures. This results in altered biophysical membrane properties of (PI3K) glycosphingolipid-enriched microdomains (GEMs), which may alter

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and… stratum corneum [6]. All these parameters are important Results for maintaining water permeability function. Deletion of the UGCG in nervous system-specifc cells leads to Establishing a stably UGCG overexpressing MCF‑7 disturbance of brain tissue by the loss of Purkinje cells cell line (reviewed in [7]). Moreover, long-term pharmacological inhibition of the UGCG with eliglustat in patients with MCF-7 cells were transfected with an UGCG expression Gaucher disease type 1 was well tolerated [8]. plasmid (MCF-7/UGCG OE) or a control vector (MCF-7/ The UGCG is overexpressed in several cancer types, for pTarget) (MCF-7/naiv = no transfection). After selec- example, in metastatic breast cancer tissue resulting in a tion of the cells with G418 over several weeks, UGCG poor patient prognosis [9] and colon cancer cells [10]. This mRNA and protein expression was analyzed. Figure 2a overexpression correlates with an enhanced expression of shows a signifcantly increased UGCG mRNA expression P-glycoprotein 1 (P-gp) (also ATP-binding cassette sub- in MCF-7/UGCG OE cells as compared to MCF-7/naiv family B member 1, ABCB1), which is encoded by the mul- and MCF-7/pTarget cells. This is verifed at protein level tidrug resistance protein 1 (MDR1) gene. The exact molecu- by Western blot analysis (Fig. 2b). In addition, the UGCG lar mechanisms by which UGCG and MDR1 are connected overexpression in MCF-7/UGCG OE cells was confrmed are unknown, but it could be shown that MDR1 regulates by immunocytochemistry (Fig. 2c). In all three MCF-7 UGCG promoter activity as well as UGCG regulates MDR1 cell types, UGCG co-localizes with GM130, a marker for expression (reviewed in [11–13]). Liu et al. showed that the cis-Golgi apparatus. In summary, the overexpression globo-series GSL produced by UGCG activity alter MDR1 of UGCG protein in MCF-7/UGCG cells has been shown. expression [14]. Overexpression of UGCG in combination with chemotherapeutic agents leads to increased Gb3 and Gb5 concentrations in GSL-enriched microdomains (GEM). Morphological and physiological changes This results in cSrc tyrosine kinase activation, decreased following UGCG overexpression β-catenin phosphorylation, and increased nuclear β-catenin. It is assumed that nuclear β-catenin may bind in a com- Overexpression of UGCG in MCF-7 cells leads to an plex with the T-cell factor 4 (Tcf4) to the Tcf4/lymphoid enlarged cytoplasm as compared to control cells (Fig. 3a). enhancer factor (LEF) binding motif at the MDR1 pro- In addition, MCF-7/UGCG OE cells exhibit an up to 70% moter and thus enhancing promoter activity. This leads to increased nucleus size as quantifed by calculating the enhanced P-gp expression and subsequently to efux of anti- nucleus-to-cytoplasm (N:C) ratio in the diferent MCF-7 cancer drugs from cells. P-gp is also postulated to function cells (Fig. 3b). This ratio is defned as the ratio of the as a fippase, transporting GlcCer from the outer to the inner nuclear area divided by the cytoplasmic area indicating leafet of the Golgi apparatus, where it can be metabolized to abnormal nuclear morphology. MCF-7/UGCG OE cells more complex sphingolipids like GSLs (reviewed in [15]). exhibit also a promoted proliferation indicated by a fve- The regulation of the UGCG is rarely investigated. Beside fold higher living cell number than the control cell number CpG island methylation of the UGCG promoter in ductal after 5 days of culturing (Fig. 3c). This efect can par- breast cancer cells, which seems to be important for drug tially be reversed by pretreatment with PPMP, an UGCG resistance [16], doxorubicin in combination with the estro- inhibitor, indicating that the UGCG is important for this gen receptor (ER) subtype α leads to an increased UGCG proliferation promoting efect. The results are confrmed promoter activity possibly mediated via a Sp1-binding site by UGCG knockdown (UGCG KD) experiments (supple- [17]. The UGCG protein activity is infuenced by the dimeri- mental 1A). MCF-7/UGCG KD cells are smaller as com- sation with c-Fos [18] and the neuroendocrine-specifc pro- pared to control and MCF-7/UGCG OE cells (supplemen- tein Reticulon-1C [19]. tal 1B). Staining of the cytoskeleton and cell membrane Here, we investigated the cellular mechanisms afected by by β-actin and β-catenin validates the diferent cell sizes UGCG overexpression in breast cancer cells. Our data indi- following UGCG overexpression and UGCG knockdown cate that an increased UGCG expression leads to transforma- (supplemental 2A). In addition, MCF-7/UGCG KD cells tion of the cell metabolism including an altered morphol- exhibit a signifcantly reduced living cell number follow- ogy and transcriptional upregulation of MDR1. In addition, ing 5 days of cultivation (supplemental 1C). MCF-7 cells anti-apoptotic gene expression is enhanced and pro-apop- were treated for 48 h with doxorubicin, which is frequently totic gene expression reduced. These cellular changes result used for chemotherapy in breast cancer patients. MCF-7/ in a doxorubicin resistance and promoted proliferation of UGCG OE cells are less sensitive to doxorubicin than MCF-7/UGCG cells. Therefore, the UGCG is a key enzyme MCF-7 control cells (Fig. 3d). The viability of MCF-7/ in breast (cancer) cells signaling leading to severe alterations of the cell metabolism.

1 3 M.-S. Wegner et al.

Fig. 2 Stable UGCG overexpression in MCF-7 cells. a Expression as a mean of n = 5 ± SEM. Unpaired t test with Welch’s correction. analysis of UGCG mRNA in MCF-7 cells by qRT-PCR. The expres- c Immunocytochemistry of MCF-7 cells. Cells were incubated with sion is related to the housekeeping gene RPL37A. Data are repre- an anti-UGCG and anti-GM130 (cis-Golgi apparatus) antibody and sented as a mean of n = 7–11 ± SEM (standard error of the mean). subsequently incubated with secondary antibodies. DAPI (4′,6-diami- Unpaired t test with Welch’s correction. b Western blot analysis of dino-2-phenylindole) was used to stain DNA. Images were recorded MCF-7 cells. Densitometrically evaluated UGCG protein concen- by Axio Observer. Z1 microscope (Carl Zeiss AG, Oberkochen, Ger- tration and a representative blot is displayed. Data are represented many). *p ≤ 0.05, **p ≤ 0.01 naiv cells is reduced to 50% at a doxorubicin concentra- UGCG overexpression infuences mRNA expression tion of 7.75 µM. MCF-7/pTarget cells are 50% reduced of metabolizing and MDR1 in their viability at a doxorubicin concentration of 7 µM, whereas MCF-7/UGCG OE cells are vital up to 50% at Overexpression of UGCG in MCF-7 cells has a severe a concentration of 60 µM doxorubicin. PPMP treatment impact on the mRNA expression of several sphingolipid increases the doxorubicin sensitivity of MCF-7/naiv cells metabolizing enzymes. Figure 4a shows that in MCF-7/ to 2.6 µM. The doxorubicin sensitivity of MCF-7/pTar- UGCG OE cells, the mRNA expression of CerS2, -5, get cells remains unaltered following PPMP treatment. CERK, and SMS1 is signifcantly increased as compared MCF-7/UGCG OE cells are resensitivated to doxorubicin to MCF-7/pTarget cells. In contrast, the mRNA expres- by PPMP treatment showed by 50% cell viability at a sion of CerS4, -6, aSMase, nSMase1, and -2 and aCDase doxorubicin concentration of 40 µM (Fig. 3d). These data are downregulated in these cells (Fig. 4b). In addition, indicate that the UGCG is involved in the process of drug MCF-7/UGCG OE cells exhibit an increased Gb3 syn- resistance. thase mRNA expression as compared to control cells, whereas the GM3 synthase mRNA expression is unaltered

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and…

Fig. 3 Morphological and physiological changes following UGCG resented as a mean of n = 5–11 ± SEM. Unpaired t test with Welch’s overexpression in MCF-7 cells. a Transmitted light image acquisition correction. d Cell viability analysis with and without 2 µM PPMP shows an enlarged cytoplasm of MCF-7/UGCG OE cells. b Nucleus- pretreatment over 5 days. Doxorubicin was added for 48 h. Data are to-cytoplasm ratio of MCF-7 cells. Data are represented as a mean represented as a mean of n = 4 ± SEM. Unpaired t test with Welch’s of n = 6 ± SEM. Unpaired t test with Welch’s correction. c Cell pro- correction. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 liferation with and without 2 µM PPMP over 5 days. Data are rep- following UGCG overexpression (Fig. 4d). To exclude the contacts in the 3D spheroid model or due to nutrition possibility that the changes in mRNA expression are a shortage. Supplemental 4 shows that MCF-7/UGCG OE monolayer-dependent efect, 3D MCF-7 spheroid assembly spheroids are smaller in size and more densely packed was performed, and subsequently, the qRT-PCR analysis than control cell spheroids. In addition, MCF-7/naiv and repeated (supplemental 3A). In spheroids from MCF-7/ MCF-7/pTarget spheroids exhibit cell-free spaces repre- pTarget and MCF-7/UGCG OE cells, nearly all enzymes sented by the DAPI staining, which indicates a nutrition of the sphingolipid pathway were comparably expressed as defcit in MCF-7/UGCG OE spheroids or a necrotic core under monolayer conditions, with respect to the expression of MCF-7/naiv and MCF-7/pTarget spheroids. Hence, of CerS2, CerS5 and SMS1 mRNA, which are decreased the diferences in the mRNA expression levels of vari- in MCF-7 spheroids compared to 2D cultured MCF-7 ous enzymes of the sphingolipid pathway occur due to cells. The diference regarding CerS2, CerS5, and SMS1 the overexpression of the UGCG gene in these cells and mRNA expression is possibly due to increased cell–cell are not a monolayer-dependent efect. In addition, Fig. 4a

1 3 M.-S. Wegner et al.

Fig. 4 Analysis of mRNA expression of sphingolipid metabolizing ing 48 h without stimulation, 2 µM PPMP, 2 µM bisindolylmaleimide enzymes and MDR1 in MCF-7 cells by qRT-PCR. a Anti-apoptotic/ I, or 25 µM Ly294002. d Gb3 and GM3 synthase mRNA expression. pro-proliferative enzymes: MDR1, CerS2, CerS5, CERK, and SMS1. The mRNA expression is related to the housekeeping gene RPL37A. b Pro-apoptotic/anti-proliferative enzymes: CerS4, CerS6, aSMase, Data are represented as a mean of n = 3–10 ± SEM. Unpaired t test nSMase1, nSMase2, and aCDase. c MDR1 mRNA expression follow- with Welch’s correction. *p ≤ 0.05, **p ≤ 0.01

shows an UGCG-dependent increase of MDR1 mRNA MDR1 mRNA expression is decreased in MCF-7/UGCG expression. Treatment of MCF-7/UGCG OE cells with KD cells as compared to MCF-7/UGCG OE cells and con- 2 µM PPMP, an UGCG inhibitor, 2 µM bisindolylma- trol cells (supplemental 2B). leimide I [protein kinase C (PKC) inhibitor], or 25 µM Ly294002 [phosphoinositide 3-kinase (PI3K) inhibitor] for The impact of an increased UGCG expression 48 h abolishes the UGCG-mediated increase of MDR1 on sphingolipid levels mRNA expression in MCF-7/UGCG OE cells (Fig. 4c). In contrast, MDR1 mRNA expression of MCF-7/pTarget Strong expression of the UGCG leads to altered levels of cells was not afected by treatment with PPMP, Bisindolyl- several sphingolipid metabolites. First, we analyzed com- maleimide I, or Ly294002. This is possibly based on the plex sphingolipids by HPTLC analysis, which revealed that fact that MCF-7/pTarget cells already exhibit a very low MCF-7/UGCG OE cells show lower concentrations of total basal MDR1 mRNA expression as compared to MCF-7/ ceramide and GlcCer, whereas Gb3, a globo-series gly- UGCG OE cells (Fig. 4a). Consistent with these data, cosphingolipid, is strongly increased in its concentration in

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and…

Fig. 5 Determination of sphingolipid concentrations in MCF-7 cells analytes were related to the respective ceramide spot. d Total Cer, by LC–MS/MS and HPTLC analysis. a Representative HPTLC plate Glc-, and LacCer levels determined by LC–MS/MS. e Concentrations showing the separation of ceramides, DAG, non-complex GSLs like of ­C14:0-, ­C16:0-, ­C18:1-, ­C18:0-, ­C20:0-, ­C24:0-, and C­ 24:1-Cer determined GlcCer, LacCer, and complex GLSs like Gb3 and gangliosides and by LC–MS/MS f. Concentrations of C­ 16:0-, ­C24:0-, and C­ 24:1-GlcCer, more complex GSLs. b Densitometrical analysis of the HPTLC plate and ­C16:0-, ­C24:0-, and C­24:1-LacCer determined by LC–MS/MS. lines 1 and 2 (control). c Densitometrical analysis of the HPTLC *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001 plate lines 3 and 4 (treatment with 2 µM PPMP for 48 h). All given comparison to MCF-7/pTarget cells (Fig. 5a, lane 1 and 2, levels in MCF-7/UGCG OE cells as compared to MCF-7/ b). This is in line with the qRT-PCR analysis showing an pTarget cells. This is also shown in Fig. 5e, f. Interestingly, increased Gb3 synthase mRNA expression in MCF-7/UGCG the LC–MS/MS analysis shows an increased LacCer level in OE cells as compared to control cells (Fig. 4d). Figure 5a, b MCF-7/UGCG OE cells, which is not detectable by HPTLC. shows also a slight increase in the concentration of ganglio- The HPTLC analysis indicates a shift of long-to-very long sides and more complex GSLs in MCF-7/UGCG OE cells chain LacCer, which could not be proven by LC–MS/MS (line 2). Interestingly, also a shift from LacCer with very due to the lack of standards for the determination of very long chain length to LacCer with long chain length is detect- long chain LacCer species. Figure 5e shows a reduction able in MCF-7/UGCG OE cells. PPMP treatment reduces all of C­ 14:0-, C­ 18:1-, C­ 18:0-, and mainly ­C24:0-ceramide level in glycosylated sphingolipids in their concentration in MCF-7/ UGCG overexpressing cells. This efect is partially reversed pTarget as well as in MCF-7/UGCG OE cells as shown in by PPMP pretreatment. PPMP treatment clearly reduced Fig. 5a line 3 and 4 and Fig. 5c. The determination of sphin- ­C16:0, ­C24:0-, and ­C24:1-GlcCer in both cell lines (Fig. 5f), golipid levels by HPTLC is verifed and analyzed in more which verifes the HPTLC analysis. Supplemental 5A shows detail by LC–MS/MS (Fig. 5d–f). Figure 5d supports the a decrease of the -1-phosphate (Sph-1p) con- HPTLC analysis by showing decreased total Cer and GlcCer centration following UGCG overexpression as compared to

1 3 M.-S. Wegner et al. control cells, which may be the outcome of an accelerated which is accompanied by a reduction of total Cer, GlcCer, ceramide metabolism to complex sphingolipids. Stimulation and Sph-1p concentration (Fig. 1). with PPMP leads in MCF-7/pTarget and MCF-7/UGCG OE cells to a light reduction of Sph-1p, which is not signif- UGCG overexpression alters the composition cant. To investigate whether or not sphingolipid levels fol- of glycosphingolipid‑enriched microdomains lowing UGCG overexpression exhibit the same pattern in (GEMs) MCF-7/UGCG OE monolayer cultured cells as in MCF-7/ UGCG OE spheroids, the LC–MS/MS analysis was repeated Next, we investigated whether or not the alterations in the for MCF-7/UGCG OE spheroids. Our data show the same sphingolipid composition have an impact on the formation alterations of the sphingolipid levels in MCF-7/UGCG of lipid domain structures in cellular membranes. Therefore, OE spheroids as in monolayer cultured cells (supplemen- we isolated GEMs by sucrose gradient centrifugation. The tal 3B–D). However, LacCer species are slightly reduced lipid content of the separated sucrose fractions was subse- in MCF-7/UGCG OE spheroids as compared to monolayer quently analyzed by LC–MS/MS. Fraction 2 of MCF-7/ conditions. In summary, our data indicate that an increased UGCG OE cell membranes exhibits a signifcant increase UGCG expression leads to an enhanced formation of Gb3 of GlcCer content in comparison to MCF-7/pTarget cell and gangliosides and more complex GSLs in MCF-7 cells, membrane fraction 2 (Fig. 6a). In contrast, if we sum up

Fig. 6 Determination of sphingolipids in glycosphingolipid-enriched n = 3 ± SEM. c Cholesterol-level determination by ELISA of frac- microdomains (GEMs) of MCF-7/cells. a GEMs were isolated and tion 1–5. Data are represented as a mean of n = 3 ± SEM. Caveolin-1 total GlcCer levels of fractions 1–10 were determined by LC–MS/ protein content determined by Western blot analysis. Data are repre- MS. Data are represented as a mean of n = 3 ± SEM. Unpaired t test sented as a mean of n = 3 ± SEM. One representative blot of three is with Welch’s correction. b GEMs were isolated and total GlcCer displayed. d Sphingomyelin concentration determination of whole- levels of fractions 1–10 were determined by LC–MS/MS. Results cell lipid extract by LC–MS/MS. SM = Sphingomyelin. Data are rep- of fraction 1–10 are summarized. Data are represented as a mean of resented as a mean of n = 3 ± SEM. *p ≤ 0.05, **p ≤ 0.01

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and… the GlcCer concentration of all fractions, the total GlcCer which might contribute to the observed cellular changes content is decreased slightly in MCF-7/UGCG OE cells as such as increased cell proliferation and doxorubicin drug compared to MCF-7/pTarget cells (Fig. 6b), which is in line resistance (Fig. 1). with the LC–MS/MS and HPTLC data showing a loss of GlcCer (Fig. 5). Supplemental 5B shows that also the total sphinganine and total ceramide levels are increased in frac- Discussion tion 2, whereas total LacCer is slightly decreased. Particu- larly, the concentrations of ­C14:0-, ­C16:0-Cer, and ­C16:0-, and We were able to show that an increased UGCG mRNA ­C18:1-GlcCer are increased in fraction 2 of MCF-7/UGCG expression in MCF-7 cells leads to cellular changes such as OE cells as compared to MCF-7/pTarget cells (data not an altered enlarged cytoplasm, which promoted prolifera- shown). Further analysis revealed out that fractions 2 and 3, tion and doxorubicin resistance. These efects are abolished which are enriched in GlcCer, contain the highest cholesterol by pretreatment of the cells with PPMP, an UGCG inhibi- and caveolin-1 content (Fig. 6c). Cholesterol and caveolin-1 tor, and UGCG knockdown, indicating that these efects are are markers for GEMs. These fractions could also be defned UGCG-mediated. as the lipid fractions, since the protein content is low, but the Our LC–MS/MS analysis shows a loss of Cer and Glc- sodium potassium ATPase (SPATPase), a plasma membrane Cer and a slight increase of LacCer. The accumulation of protein, is clearly detectable (supplemental 6). Figure 6d Gb3 and gangliosides in UGCG overexpressing MCF-7 shows the sphingomyelin (SM) concentration in whole-cell cells indicates that GlcCer are metabolized immediately lipid extracts from MCF-7/pTarget and MCF-7/UGCG OE to LacCer and subsequently to complex GSLs, especially cells. UGCG overexpression has no efect on SM, respec- Gb3. This is accomplished by an increased Gb3 synthase tively; further proceeding of ceramides to sphingomyelin mRNA expression in MCF-7/UGCG OE cells. The GM3 can be excluded. Our data indicate that MCF-7/UGCG OE synthase mRNA expression remains unaltered indicating a cells show an alteration of the GEM composition in their transcriptional-independent regulation. Possibly, the GM3 membranes as compared to control cells. GEMs are known synthase exhibits an increased enzyme activity due to altered to be signaling platforms for several membrane-associated membrane composition following UGCG overexpression proteins [20, 21]. Accordingly, UGCG overexpression might resulting in GM3 accumulation. The reduction of ­C24:0-Cer afect membrane-associated proteins and cellular signaling and ­C24:0-GlcCer in MCF-7/UGCG OE cells indicates that pathways, which is debated in the “Discussion”. preferably ceramides with a chain length of ­C24:0 are further metabolized to more complex GSL in UGCG overexpressing Efect of UGCG overexpression on signaling cells. This result is in line with studies from Yamaji et al., cascades in MCF‑7 cells who investigated the acyl chain length of sphingomyelin and GSLs in HeLa cells, which expressed variable amounts of To reassess our hypothesis that MCF-7/UGCG OE cells CerS2. By performance of metabolic labeling experiments, exhibit an altered composition of their signaling platforms they could show that preferentially very long chain cera- in the cellular membranes, we analyzed various signaling mides such as C­ 24:0-Cer are transported to the UGCG via a pathways using the Pathscan­ ® Intracellular Signaling Array thitherto unknown mechanism [22]. This study showed also Kit. The antibody-based analysis of the phosphorylation that ­C16:0-Cer is predominantly used for sphingomyelin syn- status of various key signaling proteins shows an increased thesis, which seems also to be true in our study showing only phosphorylation of Akt308, glycogen synthase kinase a slight decrease of C­ 14:0-, C­ 16:0-, C­ 18:1-, and C­ 18:0-Cer in 3β (GSK-3β), Bad, proline-rich Akt substrate of 40 kDa MCF-7/UGCG OE cells. However, although the GlcCer con- (PRAS40), ribosomal protein S6 (rpS6), and extracellular centrations decrease following UGCG overexpression our signal-regulated kinase 1/2 (ERK1/2) in MCF-7/UGCG OE data indicate that the remaining GlcCer are integrated in the cells compared to MCF-7/pTarget cells (Fig. 7). Activation cell membrane leading to an enrichment of membrane struc- of these signaling pathways is abolished by PPMP pretreat- tures with high GlcCer content in MCF-7/UGCG OE cells ment (supplemental 7). The antibody-based analysis was as compared to control cells. However, Sph-1p levels are repeated for 3D MCF-7/UGCG OE spheroids and shows decreased in MCF-7/UGCG OE cells as compared to control also a positive phosphorylation status of Akt308, PRAS40, cells, which are contradictory to the concept of sphingolipid rpS6, and ERK1/2 in the 3D cell culture (supplemental 8). rheostat, which claims the pro-proliferative character of In addition, data were confrmed with MCF-7/UGCG KD Sph-1p and an anti-proliferative character of Cer (reviewed cells (supplemental 9), which verify an UGCG-dependent in [23]). However, the role of Sph-1p in cellular processes is phosphorylation of Akt308, GSK-3β, Bad, PRAS40, rpS6, still under investigation and controversial (reviewed in [24]). and ERK1/2. In summary, our data indicate that in MCF-7/ The lowered Sph-1p level following UGCG overexpression UGCG OE cells, several signaling pathways are induced, may be the consequence of an accelerated sphingolipid

1 3 M.-S. Wegner et al.

Fig. 7 Analysis of phosphorylation status of key signaling molecules sis of the array. Data are represented as a mean of n = 5–7 ± SEM. b in MCF-7 cells by an antibody-based array. Array was performed Representative arrays were displayed. RFU relative fuorescence unit. according to the manufacturer’s protocol. a Densitometrical analy- *p ≤ 0.05 metabolism resulting in accumulation of complex GSL. microdomains (GEMs) localized in the plasma membrane However, further analysis of the fractionated cellular lysates of MCF-7/pTarget and MCF-7/UGCG OE cells. Whether by sucrose gradient centrifugation revealed that the frac- or not the accumulation of Gb3 or changes in another lipid tions 2 and 3, which contain increased GlcCer concentra- is responsible for altered membrane protein activity and, tions, also contain the highest cholesterol and caveolin-1 therefore, leading, for example, to promoted proliferation level. Caveolins are one of the main constitutes of GEM, needs to be investigated. It is known that GEMs are signal- respectively rafts and are used as raft markers [25, 26]. The ing platforms for several membrane-associated proteins [20, data identify fractions 2 and 3 as glycosphingolipid-enriched 21]. This leads to the assumption that UGCG overexpression

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and… might afect membrane proteins and subsequently cell sign- the mTOR signaling pathway. Activation of the mTOR sign- aling pathways. An example for this might be the epider- aling pathway is also indicated by a positive phosphoryla- mal growth factor receptor (EGFR), which is connected to tion status of rpS6, which is a downstream target of mTOR, promoted MCF-7 cell proliferation. Rogers et al. showed leading, for example, to cell cycle progression and subse- that EGFR signaling in cancer cells is reliant on the locali- quently to accelerated cell growth (reviewed in [28]). Beside zation of EGFR in intact lipid rafts [27]. Disruption of the Akt, the ERK1/2 signaling pathway is induced, which may receptor localization leads to decreased Ras activation and also contribute to promoted proliferation (reviewed in [29]) subsequent downregulation of ERK signaling resulting in (Fig. 1). Treatment of MCF-7/UGCG OE cells with PPMP, proliferation inhibition. Investigating the activation of key an UGCG inhibitor, and UGCG knockdown abolishes the signaling proteins in MCF-7/UGCG OE cells revealed a phosphorylation of Akt, GSK-3β, Bad, PRAS40, rpS6, and positive phosphorylation status of the key signaling pro- ERK1/2 signifcantly indicating an UGCG-dependent activa- teins Akt308, GSK-3β, Bad, PRAS40, rpS6, and ERK1/2, tion of these signaling pathways. Since the mTORC1 com- indicating that overexpression of the UGCG activates anti- plex regulates translation initiation and ribosome biogenesis, apoptotic and pro-proliferative signaling pathways. We perfect conditions for an increased gene expression of ana- assume that the altered composition of the GEMs leads to bolic genes are given when the mTOR pathway is activated the phosphorylation of Akt and ERK1/2 in MCF-7/UGCG like it is assumed for MCF-7/UGCG OE cells. Accordingly, OE cells. The serine/threonine kinase Akt is phosphorylated genes, which encode for cell metabolism regulator proteins, at threonine (T) 308, which is sufcient for its activation. are altered in its expression in the way that cell proliferation How Akt is phosphorylated exactly and, therefore, activated is intensifed (Table 1) (CerS2 [30, 31], CerS5 [32], CERK needs to be investigated in future studies. Akt308 is known [33], SMS1 [34–36], CerS4 and -6 [37], CerS6 (reviewed to phosphorylate GSK-3β and Bad leading to an inhibition of in [3], aSMase (reviewed in [38], nSMase1 [39], nSMase2 both proteins resulting in promoted cell survival and growth [40–42], and aCDase [43]). (reviewed in [28]) (Fig. 1). Akt308 phosphorylation also Expression analysis of CerS2, -5, and CERK in spheroids leads to PRAS40 phosphorylation, which negatively regu- shows a less expression in MCF-7/UGCG OE cells com- lates mTOR activity by binding directly to the mTORC1 pared to MCF-7/pTarget cells. We could show that MCF-7/ complex. Phosphorylated PRAS40 is inactive and dissoci- UGCG OE spheroids are packed more densely; accordingly, ates from the mTORC1 complex leading to an activation of the alterations in mRNA expression may be contributed to

Table 1 Summary of mRNA Protein MCF-7/UGCG cells compared Function References expression analysis data of to control cells MCF-7 cells MDR1 ↑ Multidrug resistance ↑ [9] CerS2 ↑ Colony formation ↑ [30] Anti-apoptotic [31] CerS5 ↑ Autophagy ↑ [32] CERK ↑ Tumor cell survival ↑ [33] Mammary tumor recurrence ↑ [33] SMS1 ↑ Proliferation ↑ [34] Anti-apoptotic [35] Regulation of raft structures (signaling [36] platforms) CerS4 ↓ Colony formation ↓ [37] Proliferation ↓ [37] CerS6 ↓ Colony formation ↓ [37] Proliferation ↓ [37] Pro-apoptotic [3] aSMase ↓ Pro-apoptotic [38] nSMase1 ↓ Pro-apoptotic [39] nSMase2 ↓ Pro-apoptotic ↓ [40] Cell viability ↓ [41] Growth rate ↓ [42] aCDase ↓ Tumor recurrence ↓ [43]

The protein name and the alterations regarding mRNA expression of MCF-7/UGCG OE cells compared to MCF-7/pTarget cells are given. In addition, the function and the respective reference are denoted

1 3 M.-S. Wegner et al. a defcit of energy supply in the core of MCF-7/UGCG OE conversion on the cytosolic surface of the Golgi apparatus. spheroids. In vivo these conditions would lead to angiogen- For LacCer synthesis, GlcCer must be fipped to the luminal esis. Cancer cells might stimulate angiogenesis by secretion leafet of the Golgi apparatus. Morad and Cabot postulate of Sph-1p [44], whereas the Sph-1p level in the spheroids in that P-gp fips GlcCer from the Golgi apparatus cytosol to our study was not detectable (data not shown). The cell-free the Golgi lumen, thereby contributing to multidrug resist- area in MCF-7/naiv and MCF-7/pTarget spheroids could ance (reviewed in [15]). This contribution is supported by also be an indicator for the development of a necrotic core the fact that beside ceramides, GlcCer induces toxic efects of the spheroid, which seems not to be induced in MCF-7/ in the cell [49], and by fipping GlcCer, it is used for non- UGCG OE spheroids. However, the cell-free core of MCF-7/ toxic complex GSL synthesis. Interestingly, the presence of naiv and MCF-7/pTarget spheroids might be a necrotic core, cholesterol, which infuences membrane packing and fuid- which is lacking in MCF-7/UGCG OE spheroids. Since 3D ity and which is increased in fraction 2 and 3 in MCF-7 MCF-7/UGCG OE spheroids show similar data, we assume cells, infuences P-gp activity (reviewed in [50]). Since the that the UGCG efect seems not to be monolayer-dependent. cholesterol levels in MCF-7/pTarget and MCF-7/UGCG OE However, MCF-7/UGCG OE cells exhibit high prolifera- cells are similar, it should be investigated whether or not tion rate despite the limited nutrient availability leading to the combination of cholesterol and GlcCer is more impor- the assumption that the UGCG has an infuence on energy tant for P-gp activity. Interestingly, treatment with a PKC metabolism or at least on the efcacy of energy uptake out (bisindoylmaleimide I) or PI3K (Ly294002) inhibitor or of the media. The data indicate that MCF-7/UGCG OE cells PPMP could abolish the efect, indicating that downstream alter their in the way that anabolic processes are of UGCG expression molecular mechanisms leads to an maintained and catabolic mechanisms are restricted resulting induction of MDR1 mRNA expression. A direct binding of in the described cellular impacts. MCF-7/UGCG OE cells the transcription factor nuclear factor (NF) ΚB to the MDR1 may achieve these efects by induction of autophagic pro- promoter results in decreased MDR1 gene and P-gp protein cesses. Shen et al. showed an UGCG inhibition-dependent expression [51]. Liu et al. showed that by UGCG activity autophagic fux in neurons [45]. Furthermore, CerS5 mRNA generated GSLs, MDR1 expression is mediated by signal- is increased in MCF-7/UGCG OE cells, which has also ing of the tyrosine kinase cSrc and β-catenin [14]. In this been associated with autophagy. Gosejacob et al. showed case, UGCG overexpression in combination with cytotoxic that adipose tissue of CerS5 knockdown mice exhibits less agents leads to increased globo-series GSLs in GEMs; and autophagy following high fat diet [46]. Because of the doxo- as a result, cSrc is activated and nuclear β-catenin increased. rubicin resistance of MCF-7/UGCG OE cells, we investi- A co-binding of β-catenin and the T-cell factor 4 (Tcf4) to gated the mRNA expression of the multidrug resistance pro- the Tcf4/lymphoid enhancer factor (LEF)-binding motif tein 1 (MDR1) in MCF-7 cells. The basal mRNA expression at the MDR1 promoter is assumed. The consequence is an of MDR1 is strongly increased in MCF-7/UGCG OE cells enhanced MDR1 promoter activity [14]. However, the exact indicating the mediation of the doxorubicin resistance by molecular mechanisms, which follow an UGCG overexpres- MDR1. It is already shown that an UGCG overexpression sion and lead to MDR1 gene expression induction, should be is accompanied by an increased MDR1 gene expression in investigated in the future. several cancer cell types [9] and that MDR1, which encodes In conclusion, UGCG overexpression in human breast the protein P-glycoprotein 1 (P-gp) (also ATP-binding cancer cells alters the lipid composition of the plasma cassette sub-family B member 1, ABCB1), which leads to membrane by increasing the integration of GlcCer resulting multidrug resistance, for example, against vinblastine. By in Akt and ERK1/2 activation. This leads to induction of blocking, the UGCG resensitization of multidrug-resistant MDR1 and enhanced anti-apoptotic gene transcription as breast cancer cells to anti-cancer drugs via downregulation well as reduced transcription of pro-apoptotic genes. Thus, of MDR1 has been achieved [10]. The induction of drug an enhancement of UGCG in tumor cells infuences several resistance to cytotoxic agents is accomplished by highly cellular signaling pathways leading to an altered phenotype, complex mechanisms and is the main cause of anti-cancer promoted proliferation, and chemotherapy resistance. therapy failure. In MCF-7 cells, inhibition of the sphingo- sine kinase-2 (SphK2), which produces Sph-1p, resensitizes the cells to standard chemotherapy [47], but, as mentioned Materials and methods above, MCF-7/UGCG OE cells exhibit a decreased Sph-1p level as compared to control. The P-gp protein expressed Cell culture at the plasma membrane transports cytotoxic substances out of the cell (reviewed in [48]) (Fig. 1). However, this The human breast adenocarcinoma cell line MCF-7 was pur- efux pump seems also to play a role in the Golgi apparatus. chased from the Health Protection Agency (European Col- The UGCG protein executes its task of ceramide to GlcCer lection of Cell Cultures, ECACC, Salisbury, UK). Cells were

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and… cultured in Dulbecco’s Modifed Eagle Medium (DMEM) diferent G418 concentrations. 200 μg/mL G418 was used containing high glucose, no phenol-red and no HEPES, for selection of stable transfected cells. 1% GlutaMAX, 1% sodium pyruvate, and 5% charcoaled fetal bovine serum (FBS) (Sigma-Aldrich, Deisenhofen, Germany). Cells were incubated at 37 °C in an atmosphere Determination of mRNA expression of MDR1 containing 5% CO­ 2. For selection of stably transfected cells, and sphingolipid metabolizing proteins G418 (Thermo Fisher Scientifc, Waltham, MA, USA) was by quantitative real‑time PCR (qRT‑PCR) added. Total RNA was isolated with RNeasy Mini Kit (Qiagen Cell treatment N.V., Venlo, The Netherlands) according to the manufac- turer’s protocol. RNA concentrations were determined MCF-7 cells were seeded at a proper density for the appro- photometrical by using the Infnite­ ® 200 NanoQuant mon- priate dish. Because of the high proliferation rate, MCF-7/ ochromator (Tecan Group, Männerdorf, Switzerland). The UGCG OE cells were seeded in a 50% lower density com- cDNA was synthesized from 300 ng total RNA using the pared to MCF-7/pTarget cells. Cells were stimulated with VERSO™ cDNA Kit (Thermo Fisher Inc., Waltham, MA, 2 μM dl-threo-1-phenyl-2-palmitoyl-amino-3-morpholino1- USA). Gene-specifc PCR products were assayed using propanol (PPMP) (Enzo Life Sciences, Farmingdale, NY, 5× QPCR Mix EvaGreen­ ® (ROX) (Bio&Sell, Feucht bei USA) over 6 days and media were renewed every 48 h. For Nürnberg, Germany) on a 7500fast quantitative PCR sys- MDR1 gene expression analysis, MCF-7 cells were treated tem ­(TaqMan®, Life Technologies, Darmstadt, Germany). with 0 and 2 µM bisindolylmaleimide I [Protein kinase C Relative gene expression was determined using the com- (PKC) inhibitor] (Cayman, Ann Arbor, MI, US) and 0 and parative ∆CT (cycle threshold) method, normalizing 25 µM Ly294002 [Phosphoinositide 3-kinase (PI3K) inhibi- relative values to the expression level of 60S ribosomal tor] (Cell Signaling, Cambridge, UK) for 48 h. protein L37a (RPL37A) as a housekeeping gene. Primers for CerS2, -4, -5, and -6 were synthesized by Biomers Stable transfection with a UDP‑glucose ceramide (Ulm, Germany) and primer for UGCG, acid sphingomy- glucosyltransferase (UGCG) expression plasmid elinase (aSMase), neutral sphingomyelinase 2 (nSMase2), acid (aCDase), MDR1, GM3 synthase, and UDP-glucose ceramide glucosyltransferase (UGCG) RPL37A were synthesized by Eurofns Genomics (Ebers- expression plasmid (pCMV6-ENTRY vector) was pur- berg, Germany) (Table 2). The forward and reverse primer chased from OriGene Technologies Inc. (Rockville, USA). sets for neutral sphingomyelinase 1 (nSMase1), ceramide The pTarget empty control vector was purchased from Pro- kinase (CERK), and Gb3 synthase were purchased From mega GmbH (Mannheim, Germany). Stable transfection RealTimePrimers (Elkins Park, Philadelphia, USA). From was performed with Lipofectamine­ ® 2000 Reagent (Inv- GeneCopoeia (Rockville, MD, USA), the primer set for itrogen by Life Technologies) according to the manufac- sphingomyelin synthase 1 (SMS1) mRNA detection was turer’s protocol. MCF-7 cells were transfected with 2 μg purchased. of the distinct plasmid and selected over 5 weeks with

Table 2 Oligonucleotides for Protein 5′-forward-3′ 5′-reverse-3′ Amplicon (bp) qRT-PCR CerS2 cca ggt aga gcg ttg gtt cca ggg ttt atc cac aat gac 142 CerS4 ctg gtg gta cct ctt gga gc cgt cgc aca ctt gct gat ac 248 CerS5 caa gta tca gcg gct ctg t att atc tcc caa ctc tca aag a 123 CerS6 aag caa ctg gac tgg gat gtt aat ctg act ccg tag gta aat aca 146 UGCG​ tgc tca gta cat tgc cga aga tgg aca ttg caa acc tcc aa 74 aSMase cac cca gga tga gaa tgg aaa gtc cgt cct cac cca cga t 59 nSMase2 caa caa gtg taa cga cga tgc c cga ttc ttt ggt cct gag gtg t 89 aCDase tgt gga tag ggt tcc tca cta ga ttg tgt ata cgg tca gct tgt tg 375 MDR1 ctc aga cag gat gtg agt tgg t aca gca agc ctg gaa cct at 116 Gb3 synthase Tac ctg gac acg gac ttc at Gga tgg aac acc act tct tg 226 GM3 synthase Gaa ctc ttg cca gag cac ga Ccc agt tct aat ccg tgc ag 104 RPL37A att gaa atc agc cag cac gc agg aac cac agt gcc aga tcc 94

The protein name, sequence, and amplicon size in bp are given

1 3 M.-S. Wegner et al.

Protein concentration determination by Western Scotland). The high-performance thin-layer chromatog- blot analysis raphy (HPTLC) silica plates were from Whatman, UK. 9 × 106 MCF-7 cells were scraped in 1 x PBS, centrifuged, For protein concentration analysis by Western blot, total pro- and frozen in liquid nitrogen. Cells were dried under the tein was isolated. Cells were resuspended in the following hood for 3 days and stored in a freezer (− 80 °C) until ana- bufer: 10 mM Tris (pH 8.0), 150 mM NaCl, 1% NP-40, lyzed. Each dried sample was solubilized in ice cold MQ- 0.5% deoxycholate, 0.1% sodium dodecyl sulfate, and 1% water (1 mL) by bath sonication (ice-water bath) for 5 min. 100X Halt™ Protease Inhibitor Cocktail (Thermo Fisher The solubilized samples were transferred to clean glass test Scientifc, Darmstadt, Germany). The lysate was sonicated tubes (with screw caps). Total lipids were extracted by a and centrifuged (14,000×g, 10 min, 4 °C). For determination modifed Blight and Dyer protocol: 1 mL of MQ-water was of total protein concentration, the Bradford method was per- added to each sample (2 mL of total water volume), 3 mL of formed [52]. 60 μg total protein extract was electrophoreti- chloroform:isopropanol (2:1, v:v) was added to each sample, cally separated by 12% sodium dodecyl sulfate (SDS)-PAGE and samples were vortexed thoroughly for 10 s, and rotated and electroblotted onto a nitrocellulose membrane (GE end-over-end for 20 min at RT. The samples were centri- Healthcare Life Sciences, Amersham, UK). After 1.5 h incu- fuged at 4000 rpm for 20 min at RT to separate the phases. bation of the membrane in Odyssey­ ® Blocking Bufer (LI- The organic phase (bottom) was carefully extracted with a COR Biotechnology, Lincoln, NE, USA) and 1× PBS (1:1), glass Pasteur pipette and placed in a clean glass tube, with- the membrane was incubated with the respective primary out disturbing the protein precipitate at the interphase. The antibodies: anti-UGCG antibody (Abcam, Cambridge, UK) chloroform:isopropanol (2:1, v:v) extractions were repeated and Hsp90 (BD Bioscience, Heidelberg, Germany). Anti- once, as described above, and combined with the previous UGCG antibody was incubated overnight, while the Hsp90 extracts. 3 mL of hexane was then added to each sample, and antibody was incubated for 60 min. The fractions, which samples were vortexed, rotated, and centrifuged as described are described in “Isolation of glycosphingolipid-enriched above. The hexane phases (top) were extracted and com- microdomains (GEMs)”, were incubated after SDS-PAGE bined with the previous chloroform:isopropanol extracts. with anti-caveolin-1 (Abcam, Cambridge, UK) primary anti- The organic solvent containing the extracted lipids was dried body to verify GEMs. For supplemental 9, membranes were under a stream of nitrogen. The dry samples were stored at incubated with anti-SPATPase, anti-Hsp90, anti-PDI, and − 20 °C. Each of the dried lipid sample was re-solubilized anti GAPDH primary antibodies (Abcam, Cambridge, UK). in an appropriate volume of chloroform:isopropanol (2:1, The membranes were analyzed on the ­Odyssey® infrared v:v). The lipid samples were normalized according to their scanner from LI-COR (Bad Homburg, Germany). corresponding protein concentrations and applied on an HPTLC plate using an autosampler. The HPTLC plate was Immunocytochemistry developed using the solvent system chloroform:methanol:a cetone:acetic acid:water (10:2:4:2:1). The sugar residues of MCF-7/naiv, MCF-7/pTarget, and MCF-7/UGCG OE the were visualized by spraying the plate with an cells were fxed in 4% paraformaldehyde, blocked with 5% orcinol solution (0.3% orcinol in 20% sulphuric acid) and by Odyssey­ ® Blocking Bufer (LI-COR Biotechnology, Lin- heating (< 5 min, 120 °C). Lipids were identifed with the coln, NE, USA), and incubated over night with anti-UGCG help of commercial standards that were run in parallel with and anti-GM130, anti-β-actin, or anti-β-catenin (Abcam, the samples on the HPTLC plate. Cambridge, UK) primary antibody. Subsequently, cells were The precipitated protein was isolated as follows: 1 mL incubated with fragment cy3- (Sigma-Aldrich, St. Louis, of chloroform and 1.5 mL of methanol were added to the Missouri, USA) and Alexa ­Fluor® 488- (Life Technologies, remaining aqueous phases of the samples. Samples were Carlsbad, CA, USA) conjugated secondary antibodies and briefy vortexed and centrifuged at 4000 rpm for 20 min. examined with an Axio Observer.Z1 microscope (Carl Zeiss The aqueous upper phase was carefully removed with a glass AG, Oberkochen, Germany). Pasteur pipette and discarded, leaving the interphase (with the precipitated proteins) untouched. 2 mL of methanol was Sphingolipid concentration determination added to each of the samples. Samples were again briefy by high‑performance thin‑layer chromatography vortexed and centrifuged at 4000 rpm for 20 min, to pellet (HPTLC) the precipitated proteins. The chloroform:methanol solution was carefully removed and discarded with a glass Pasteur All chemical reagents were of analytical grade or higher. pipette, without disturbing the pelleted proteins. The pellets Lipid standards were from Avanti Polar Lipids (Alabas- were carefully dried to completion, under a stream of nitro- ter, USA) or Matreya LLC (Pleasant Gap, USA). Organic gen. Each protein sample was re-solubilized in 500 µL of an solvents were from Rathburn Chemicals Ltd. (Walkerburn 8 M urea solution (8 M urea in PBS, pH 6.8, 0.5% SDS) by

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and… vigorous vortexing at RT. The protein samples were stored was done in Multiple Reaction Monitoring (MRM) mode at − 20 °C. The concentrations of the solubilized proteins with a dwell time of 20 ms for all analytes. Data acquisi- were determined by the method of Lowry. tion was done using Analyst Software V 1.6 and quantifca- tion was performed with MultiQuant Software V 3.0 (both Sphingolipid concentration determination by liquid Sciex, Darmstadt, Germany), employing the internal stand- chromatography–tandem mass spectrometry (LC– ard method (isotope dilution mass spectrometry). Variations MS/MS) in accuracy of the calibration standards were less than 15% over the whole range of calibration, except for the lower Cell pellets were resuspended in 150 µL water, while 50 limit of quantifcation, where a variation in accuracy of 20% or 200 µL cell culture supernatant was diluted with 100 µL was accepted. water. Samples were mixed with 150 µL extraction bufer (citric acid 30 mM, disodium hydrogen phosphate 40 mM), and 20 µL of the internal standard solution containing Determination of sphingomyelin concentrations sphingosine-d7, sphinganine-d7 (200 ng/mL each), sphin- by liquid chromatography–tandem mass gosine-1-phosphate-d7, C17:0 Cer, C16:0 Cer-d31, C18:0 spectrometry LC–MS/MS Cer-d3, C17:0 LacCer, C18:0 DHC-d3, C16:0 LacCer-d3, C18:0 GluCer-d5 (all avanti polar lipids, Alabaster, USA), For sphingomyelin concentration determination, MCF-7 cell and C24:0 Cer-d4 (Chiroblock GmbH, Bitterfeld-Wolfen) pellets were processed according to Peng et al. and deter- (400 ng/mL methanol each). The mixture was extracted mination of the analytes was established, respectively [53]. once with 1000 µL methanol/chloroform/hydrochloric acid (15:83:2, v/v/v). The lower organic phase was evaporated at 45 °C under a gentle stream of nitrogen and reconsti- Proliferation assay tuted in 100 µL of tetrahydrofuran/water (9:1, v/v) with 0.2 formic acid and 10 mM ammonium formate. Afterwards, MCF-7/naiv, MCF-7/pTarget, and MCF-7/UGCG OE cells amounts of sphingolipids were analyzed by liquid chroma- were seeded with 1 x ­106 cells/5 mm-dish and harvested on tography coupled to tandem mass spectrometry (LC–MS/ day 5 of stimulation, and the living cell number was deter- MS). An Agilent 1100 series binary pump (Agilent tech- mined with trypan blue and a Neubauer counting chamber. nologies, Waldbronn, Germany) equipped with a Luna C8 column (150 mm × 2 mm ID, 3 μm particle size, 100 Å pore size; Phenomenex, Aschafenburg, Germany) was used for Cell viability assay chromatographic separation. The column temperature was 35 °C. The HPLC mobile phases consisted of water with MCF-7/naiv, MCF-7/pTarget, and MCF-/UGCG OE cells 0.2% formic acid and 2 mM ammonium formate (mobile were seeded in 96-well plates and stimulated as mentioned phase A) and acetonitrile/isopropanol/acetone (50:30:20, in “Cell treatment”. Cells were treated with 0, 0.5, 4, 8, v/v/v) with 0.2% formic acid (mobile phase B). For separa- and 80 µM doxorubicin for 48 h. Subsequently, media were tion, a gradient program was used at a fow rate of 0.3 mL/ renewed with G418-free media. Water soluble tetrazolium min. The initial bufer composition 55% (A)/45% (B) was (WST)-1 reagent (Roche Diagnostics, Rotkreuz, Switzer- held for 0.7 min and then within 4.0 min linearly changed land) was added 1:10 and cells were incubated for 90 min to 0% (A)/100% (B) and held for 13.3 min. Subsequently, at 37 °C. Absorbance was measured by Infnite­ ® 200 PRO the composition was linearly changed within 1.0 min to reader (Tecan Group, Männerdorf, Switzerland) at a wave- 75% (A)/25% (B) and then held for another 2.0 min. The length of 450 nm and 620 nm as reference. total running time was 21 min and the injection volume was 15 μL. To improve ionization, acetonitrile with 0.1% for- mic acid was infused post-column using an isocratic pump Nuclear‑to‑cytoplasm (N:C) ratio measurement at a fow rate of 0.15 mL/min. After every sample, sample solvent was injected for washing the column with a 12 min The nuclear-to-cytoplasm (N:C) ratio is defned as the ratio run. The MS/MS analyses were performed using a triple of the nucleus area (AN) to the area of the cytoplasm (AC) quadrupole mass spectrometer API4000 (Sciex, Darmstadt, (A = π × r2). The radius (r) was calculated by measuring the Germany) equipped with a Turbo V Ion Source operating horizontal and vertical length of the nucleus and the cyto- in positive electrospray ionization mode. The MS param- plasm and the average of both values was divided by the eters were set as follows: Ionspray voltage 5500 V, source factor of 2. By the AxioVision software (Carl Zeiss AG, temperature 500 °C, curtain gas 30 psi, collision gas 12 psi, Oberkochen, Germany), the cytoplasm and the nucleus area nebulizer gas 40 psi, and heating gas 60 psi. The analysis were determined and, subsequently, the ratio calculated.

1 3 M.-S. Wegner et al.

Isolation of glycosphingolipid‑enriched Cambridge, UK) was used. This antibody array detects sev- microdomains (GEMs) eral phosphorylated signaling proteins. The assay was per- formed according to the manufacturer’s protocol. 12 × 106 MCF-7 cells were washed with DPBS twice and scraped. The cell pellet was processed in 566 μL MES bufer Statistics [0.15 M NaCl, 0.025 M MES, 1% Triton-X 100, 1× Roche Complete (7×)]. Samples were sonifed fve times for 20 s Data are presented as mean ± standard error of the mean (level 3, output 3%), mixed with 1.134 μL 67.5% saccharose (SEM). Statistical analysis was performed with GraphPad solution, and transferred to a saccharose density centrifuga- Prism 6 software. Signifcance of means was examined by tion tube. Samples were overlayed with 1.7 mL 37% saccha- the indicated statistical test in the legend of the appropriate rose solution and subsequently overlayed with 850 μL 5% fgure. saccharose solution. Centrifugation at 32.000 rpm at 4 °C for 20 h followed. Fractions each with 450 μL were removed. Acknowledgements This work was funded by the Deutsche Forschun- Fraction 10 was resuspended in 450 μL MES bufer and soni- gsgemeinschaft (WE 5825/1-1), SFB 1039 TP B05, the August Schei- fed (three times for 10 s at level 3 and output 3%) because of del-Stiftung, the Heinrich und Fritz Riese-Stiftung, and Minerva-Stif- its viscous nature. 50 μL were used to analyze sphingolipid tung. The supports by the Ministerium für Innovation, Wissenschaft und Forschung des Landes Nordrhein-Westfalen, the Senatsverwaltung concentration by LC–MS/MS [see “Sphingolipid concentra- für Wirtschaft, Technologie und Forschung des Landes Berlin, and tion determination by liquid chromatography–tandem mass the Bundesministerium für Bildung und Forschung and BMBF (Code spectrometry (LC–MS/MS)”]. 180 μL sample was purifed 031L0108A, 031A534B) are also gratefully acknowledged. In addi- by PD SpinTrap G-25 columns (GE Healthcare, Bucking- tion, funding by the Sigrid Jusélius Foundation and Magnus Ehrnrooth Foundation and Åbo Akademi University are acknowledged. hamshire, UK) according to the manufacturer’s protocol to separate saccharose from proteins for Western blot analysis. See “Protein concentration determination by Western blot analysis” for protein concentration analysis of the fractions. References

1. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo Cholesterol assay of glycosphingolipid‑enriched M, Parkin DM, Forman D, Bray F (2015) Cancer incidence and mortality worldwide: sources, methods and major patterns in microdomains (GEMs) GLOBOCAN 2012. Int J Cancer 136:E359–E386 2. Ichikawa S, Sakiyama H, Suzuki G, Hidari KI, Hirabayashi Y “Isolation of glycosphingolipid-enriched microdomains (1996) Expression cloning of a cDNA for human ceramide glu- cosyltransferase that catalyzes the frst glycosylation step of gly- (GEMs)” described samples were also used for the determi- cosphingolipid synthesis. Proc Natl Acad Sci USA 93:12654 nation of cholesterol in the diferent fractions. 10 µL stand- 3. Wegner MS, Schifmann S, Parnham MJ, Geisslinger G, Grosch S ards dissolved in 100% ethanol and 10 µL samples were (2016) The enigma of ceramide synthase regulation in mammalian added to 5 μL of emulsifer solution (15% Brij in 5 mM Tris/ cells. Prog Lipid Res 63:93–119 4. Yamashita T, Wada R, Sasaki T, Deng C, Bierfreund U, Sandhof HCl, pH 7.5) and incubated in a 96 well plate (Greiner BIO- K, Proia RL (1999) A vital role for glycosphingolipid synthesis ONE, Frickenhausen) at room temperature for 15 min. After- during development and diferentiation. Proc Natl Acad Sci USA wards 200 μL of the reaction solution was added and incu- 96:9142–9147 bated for 60 min at 37 °C. The reaction solution was freshly 5. Jennemann R, Sandhof R, Langbein L, Kaden S, Rothermel U, Gallala H, Sandhof K, Wiegandt H, Grone HJ (2007) Integrity prepared at a ratio of 1:1:0.01 of solution A (0.02 M phe- and barrier function of the epidermis critically depend on gluco- nol, 1.84 M MeOH, 1% methyl-β-cyclodextrin, ad 50 mM sylceramide synthesis. J Biol Chem 282:3083–3094 Tris/HCl pH 7.4), solution B (0.0019 M 4-aminoantipyrin, 6. Amen N, Mathow D, Rabionet M, Sandhof R, Langbein L, Gretz 1.84 M MeOH, 0.4% Brij (30% w/v), and 50 mM Tris/HCl, N, Jackel C, Grone HJ, Jennemann R (2013) Diferentiation of epidermal keratinocytes is dependent on glucosylceramide: cera- pH 7.4) and enzyme solution. The enzyme solution consisted mide processing. Hum Mol Genet 22:4164–4179 of cholesterol oxidase (12 U/mL) and peroxidase (8 U/mL). 7. Allende ML, Proia RL (2014) Simplifying complexity: geneti- Finally, the absorbance at 540 nm was measured by Tecan cally resculpting glycosphingolipid synthesis pathways in mice SpectraFluor Plus Reader (Tecan, Crailsheim, Germany). to reveal function. Glycoconj J 31:613–622 8. Mistry PK, Lukina E, Ben Turkia H, Shankar SP, Baris H, Ghosn M, Mehta A, Packman S, Pastores G, Petakov M, Assouline S, Balwani M, Danda S, Hadjiev E, Ortega A, Gaemers SJM, Tayag Detection of intracellular‑induced signaling R, Peterschmitt MJ (2017) Outcomes after 18 months of eliglustat pathways therapy in treatment-naive adults with gaucher disease type 1: the phase 3 ENGAGE trial. Am J Hematol 92:1170–1176 9. Gouaze V, Yu JY, Bleicher RJ, Han TY, Liu YY, Wang H, For analysis of signaling pathways in MCF-7 cells, the Gottesman MM, Bitterman A, Giuliano AE, Cabot MC ­Pathscan® Intracellular Signaling Array Kit (Cell Signaling, (2004) Overexpression of glucosylceramide synthase and

1 3 UDP‑glucose ceramide glucosyltransferase activates AKT, promoted proliferation, and…

P-glycoprotein in cancer cells selected for resistance to natural 28. Hers I, Vincent EE, Tavare JM (2011) Akt signalling in health product chemotherapy. Mol Cancer Ther 3:633–639 and disease. Cell Signal 23:1515–1527 10. Gouaze-Andersson V, Yu JY, Kreitenberg AJ, Bielawska A, 29. Wee P, Wang Z (2017) Epidermal growth factor receptor cell Giuliano AE, Cabot MC (2007) Ceramide and glucosylceramide proliferation signaling pathways. Cancers 9:52 upregulate expression of the multidrug resistance gene MDR1 30. Hartmann D, Wegner MS, Wanger RA, Ferreiros N, Schreiber in cancer cells. Biochim Biophys Acta 1771:1407–1417 Y, Lucks J, Schifmann S, Geisslinger G, Grosch S (2013) The 11. Jennemann R, Grone HJ (2013) Cell-specifc in vivo func- equilibrium between long and very long chain ceramides is tions of glycosphingolipids: lessons from genetic deletions of important for the fate of the cell and can be infuenced by co- enzymes involved in glycosphingolipid synthesis. Prog Lipid expression of CerS. Int J Biochem Cell Biol 45:1195–1203 Res 52:231–248 31. Mesicek J, Lee H, Feldman T, Jiang X, Skobeleva A, Berdy- 12. Ishibashi Y, Kohyama-Koganeya A, Hirabayashi Y (2013) New shev EV, Haimovitz-Friedman A, Fuks Z, Kolesnick R (2010) insights on glucosylated lipids: metabolism and functions. Bio- Ceramide synthases 2, 5, and 6 confer distinct roles in radiation- chim Biophys Acta 1831:1475–1485 induced apoptosis in HeLa cells. Cell Signal 22:1300–1307 13. Liu YY, Hill RA, Li YT (2013) Ceramide glycosylation cata- 32. Yamane M, Miyazawa K, Moriya S, Abe A, Yamane S (2011) D, lyzed by glucosylceramide synthase and cancer drug resistance. L-Threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol Adv Cancer Res 117:59–89 (DL-PDMP) increases endoplasmic reticulum stress, autophagy 14. Liu YY, Gupta V, Patwardhan GA, Bhinge K, Zhao Y, Bao J, and apoptosis accompanying ceramide accumulation via cera- Mehendale H, Cabot MC, Li YT, Jazwinski SM (2010) Gluco- mide synthase 5 protein expression in A549 cells. Biochimie sylceramide synthase upregulates MDR1 expression in the regu- 93:1446–1459 lation of cancer drug resistance through cSrc and beta-catenin 33. Payne AW, Pant DK, Pan TC, Chodosh LA (2014) Ceramide signaling. Mol Cancer 9:145 kinase promotes tumor cell survival and mammary tumor recur- 15. Morad SA, Cabot MC (2015) Tamoxifen regulation of sphin- rence. Cancer Res 74:6352–6363 golipid metabolism—therapeutic implications. Biochim Bio- 34. Burns TA, Subathra M, Signorelli P, Choi Y, Yang X, Wang Y, phys Acta 1851:1134–1145 Villani M, Bhalla K, Zhou D, Luberto C (2013) Sphingomyelin 16. Liu J, Zhang X, Liu A, Zhang D, Su Y, Liu Y, You D, Yuan L, synthase 1 activity is regulated by the BCR–ABL oncogene. J Kong X, Wang X, Sun P (2016) Altered methylation of glu- Lipid Res 54:794–805 cosylceramide synthase promoter regulates its expression and 35. Separovic D, Semaan L, Tarca AL, Awad Maitah MY, Hanada associates with acquired multidrug resistance in invasive ductal K, Bielawski J, Villani M, Luberto C (2008) Suppression of breast cancer. Oncotarget 7(24):36755–36766 sphingomyelin synthase 1 by small interference RNA is associ- 17. Zhang X, Wu X, Su P, Gao Y, Meng B, Sun Y, Li L, Zhou Z, ated with enhanced ceramide production and apoptosis after Zhou G (2012) Doxorubicin infuences the expression of glu- photodamage. Exp Cell Res 314:1860–1868 cosylceramide synthase in invasive ductal breast cancer. PLoS 36. Jin ZX, Huang CR, Dong L, Goda S, Kawanami T, Sawaki T, One 7:e48492 Sakai T, Tong XP, Masaki Y, Fukushima T, Tanaka M, Mimori 18. Crespo PM, Silvestre DC, Gil GA, Maccioni HJ, Daniotti JL, T, Tojo H, Bloom ET, Okazaki T, Umehara H (2008) Impaired Caputto BL (2008) c-Fos activates glucosylceramide syn- TCR signaling through dysfunction of lipid rafts in sphingo- thase and synthesis in PC12 cells. J Biol Chem myelin synthase 1 (SMS1)-knockdown T cells. Int Immunol 283:31163–31171 20:1427–1437 19. Di Sano F, Fazi B, Citro G, Lovat PE, Cesareni G, Piacentini M 37. Hartmann D, Lucks J, Fuchs S, Schifmann S, Schreiber Y, Fer- (2003) Glucosylceramide synthase and its functional interaction reiros N, Merkens J, Marschalek R, Geisslinger G, Grosch S with RTN-1C regulate chemotherapeutic-induced apoptosis in (2012) Long chain ceramides and very long chain ceramides neuroepithelioma cells. Cancer Res 63:3860–3865 have opposite efects on human breast and colon cancer cell 20. Mollinedo F, Gajate C (2015) Lipid rafts as major platforms for growth. Int J Biochem Cell Biol 44:620–628 signaling regulation in cancer. Adv Biol Regul 57:130–146 38. He X, Schuchman EH (2012) Potential role of acid sphingomy- 21. Staubach S, Hanisch FG (2011) Lipid rafts: signaling and sorting elinase in environmental health. Zhong nan da xue xue bao Yi platforms of cells and their roles in cancer. Expert Rev Proteom xue ban J Cent South Univ Med Sci 37:109–125 8:263–277 39. Yabu T, Shiba H, Shibasaki Y, Nakanishi T, Imamura S, Tou- 22. Yamaji T, Horie A, Tachida Y, Sakuma C, Suzuki Y, Kushi Y, hata K, Yamashita M (2015) Stress-induced ceramide genera- Hanada K (2016) Role of intracellular lipid logistics in the pref- tion and apoptosis via the phosphorylation and activation of erential usage of very long chain-ceramides in glucosylceramide. nSMase1 by JNK signaling. Cell Death Difer 22:258–273 Int J Mol Sci 17:1761 40. Park B, Lee YM, Kim JS, Her Y, Kang JH, Oh SH, Kim HM 23. Newton J, Lima S, Maceyka M, Spiegel S (2015) Revisiting the (2013) Neutral sphingomyelinase 2 modulates cytotoxic efects sphingolipid rheostat: evolving concepts in cancer therapy. Exp of protopanaxadiol on diferent human cancer cells. BMC Com- Cell Res 333:195–200 plement Altern Med 13:194 24. Tornquist K (2013) Sphingosine 1-phosphate and cancer: lessons 41. Kim WJ, Okimoto RA, Purton LE, Goodwin M, Haserlat SM, from thyroid cancer cells. Biomolecules 3:303–315 Dayyani F, Sweetser DA, McClatchey AI, Bernard OA, Look 25. Hakomori SI (2000) Cell adhesion/recognition and signal trans- AT, Bell DW, Scadden DT, Haber DA (2008) Mutations in the duction through glycosphingolipid microdomain. Glycoconj J neutral sphingomyelinase gene SMPD3 implicate the ceramide 17:143–151 pathway in human leukemias. Blood 111:4716–4722 26. Sabharanjak S, Sharma P, Parton RG, Mayor S (2002) GPI- 42. Marchesini N, Luberto C, Hannun YA (2003) Biochemical prop- anchored proteins are delivered to recycling endosomes via a dis- erties of mammalian neutral sphingomyelinase 2 and its role in tinct cdc42-regulated, clathrin-independent pinocytic pathway. sphingolipid metabolism. J Biol Chem 278:13775–13783 Dev Cell 2:411–423 43. Sanger N, Ruckhaberle E, Gyorfy B, Engels K, Heinrich T, 27. Rogers KR, Kikawa KD, Mouradian M, Hernandez K, McKinnon Fehm T, Graf A, Holtrich U, Becker S, Karn T (2015) Acid KM, Ahwah SM, Pardini RS (2010) Docosahexaenoic acid alters ceramidase is associated with an improved prognosis in both epidermal growth factor receptor-related signaling by disrupting DCIS and invasive breast cancer. Mol Oncol 9:58–67 its lipid raft association. Carcinogenesis 31:1523–1530

1 3 M.-S. Wegner et al.

44. Sukocheva O, Wadham C, Gamble J, Xia P (2015) Sphingosine- 49. Dahdouh F, Raane M, Thevenod F, Lee WK (2014) Nickel- 1-phosphate receptor 1 transmits estrogens’ efects in endothelial induced cell death and survival pathways in cultured renal proxi- cells. Steroids 104:237–245 mal tubule cells: roles of reactive oxygen species, ceramide and 45. Shen W, Henry AG, Paumier KL, Li L, Mou K, Dunlop J, Berger ABCB1. Arch Toxicol 88:881–892 Z, Hirst WD (2014) Inhibition of glucosylceramide synthase 50. Sharom FJ (2014) Complex interplay between the P-glycoprotein stimulates autophagy fux in neurons. J Neurochem 129:884–894 multidrug efux pump and the membrane: its role in modulating 46. Gosejacob D, Jager PS, Vom Dorp K, Frejno M, Carstensen AC, protein function. Front Oncol 4:41 Kohnke M, Degen J, Dormann P, Hoch M (2016) Ceramide syn- 51. Bentires-Alj M, Barbu V, Fillet M, Chariot A, Relic B, Jacobs thase 5 is essential to maintain C16:0-ceramide pools and con- N, Gielen J, Merville MP, Bours V (2003) NF-kappaB transcrip- tributes to the development of diet-induced obesity. J Biol Chem tion factor induces drug resistance through MDR1 expression in 291:6989–7003 cancer cells. Oncogene 22:90–97 47. White MD, Chan L, Antoon JW, Beckman BS (2013) Target- 52. Bradford MM (1976) A rapid and sensitive method for the quan- ing ovarian cancer and chemoresistance through selective inhibi- titation of microgram quantities of protein utilizing the principle tion of -2 with ABC294640. Anticancer Res of protein-dye binding. Anal Biochem 72:248–254 33:3573–3579 53. Peng B, Weintraub ST, Coman C, Ponnaiyan S, Sharma R, Tews 48. Katayama K, Noguchi K, Sugimoto Y (2014) Regulations of B, Winter D, Ahrends R (2017) A comprehensive high-resolution P-glycoprotein/ABCB1/MDR1 in human cancer cells. New J Sci targeted workfow for the deep profling of sphingolipids. Anal 2014:10 Chem 89:12480–12487

1 3