ER stress increases StarD5 expression by stabilizing its mRNA and leads to relocalization of its protein from the nucleus to the membranes.

StarD5 belongs to the StarD4 subfamily of steroidogenic acute regulatory lipid transfer (START) domain proteins. In macrophages, StarD5 is found in the cytosol and maintains a loose association with the Golgi. Like StarD1 and StarD4, StarD5 is known to bind cholesterol. However, its function and regulation remain poorly defined. Recently, it has been shown that its mRNA expression is induced in response to different inducers of endoplasmic reticulum (ER) stress. However, the molecular mechanism(s) involved in the induction of StarD5 expression during ER stress is not known. Here we show that in 3T3-L1 cells, the ER stressor thapsigargin increases intracellular free cholesterol due to an increase in HMG-CoA reductase expression. Activation of StarD5 expression is mediated by the transcriptional ER stress factor XBP-1. Additionally, the induction of ER stress stabilizes the StarD5 mRNA. Furthermore, StarD5 protein is mainly localized in the nucleus, and upon ER stress, it redistributes away from the nucleus, localizing prominently to the cytosol and membranes. These results reveal the increase in StarD5 expression and protein redistribution during the cell protective phase of the ER stress, suggesting a role for StarD5 in cholesterol metabolism during the ER stress response.

in response to sterol levels through the sterol regulatory element binding protein (SREBP) pathway ( 18,26 ) and in the early phase of ER stress ( 31 ), StarD5 expression is induced in response to ER stress, either in free cholesterolloaded mouse macrophages or in NIH-3T3 cells and HK-2 human proximal tubule cells after being treated with different ER stressors ( 20,26 ). However, it remains unknown which one of the three different classes of ER stress transducers (ATF6, IRE1, or PERK) mediates the regulation of the StarD5 gene.
The above-mentioned observations, i.e., that accumulation of free cholesterol leads to ER stress, that StarD5 binds cholesterol, and that StarD5 expression is induced in response to ER stress, led us to hypothesize that ER stress not only results in an increase in StarD5 expression but also in a redistribution of the StarD5 protein to subcellular locations where is particularly needed under conditions of ER stress (i.e., the cell membranes).
In this study, we show that induction of ER stress in 3T3-L1 cells increases levels of StarD5 mRNA and protein. ER stress also induces an accumulation of intracellular free cholesterol due to an increase in 3-hydroxy-3-methyl glutaryl CoA (HMG-CoA) reductase expression, and it leads to a redistribution of the StarD5 protein. The increase in StarD5 expression is driven by the IRE1 downstream effector XBP-1 (not by ATF6 or ATF4, the downstream effector of PERK), and it is also due to stabilization of its mRNA. We speculate that the increase in StarD5 expression and its subcellular redistribution plays a protective role by reducing the cholesterol levels in the ER during the cell protective phase of ER stress.

Materials
Antibodies and reagents. Alexa

Methods
Cell cultures and treatments. 3T3-L1 cells were obtained from ATCC and grown at 37°C in 5% CO 2. They were grown as fi broblast-like cells and were not differentiated to adipocytes. To induce ER stress, cells were treated in DMEM media with 10% LPDS with or without the addition of 2 M thapsigargin (Tg) or 5 g/ml tunicamycin (Tu) dissolved in dimethyl sulfoxide (DMSO). and binds to the unfolded or misfolded proteins, resulting in the activation of the three stress transducers ( 7 ). The activation of IRE1 generates a potent transcription activator called spliced X-box binding protein [XBP-1(s)] generated by the splicing of the XBP-1 precursor RNA. The activation of PERK promotes translation of the ATF4 mRNA. ATF6, XBP-1(s), and ATF4 are able to further induce the transcription of genes involved in the UPR ( 5,6,8 ).
Although the ER is the main site of cholesterol synthesis, cholesterol is rapidly transported to other organelles ( 2 ) because over-accumulation of free cholesterol in this organelle is one of the physiological stress conditions that induce UPR ( 9 ). In recent years, a number of specialized nonvesicular lipid transporters that are part of the steroidogenic acute regulatory related lipid transfer (START) domain superfamily of proteins have been shown to be involved in the traffi cking of cholesterol and other lipids between intracellular membranes (10)(11)(12). The START domain is composed of a 200-230 amino acid lipid-binding moiety, which is present in a wide range of proteins and has been implicated in several cellular functions, including lipid transport and metabolism, signal transduction, and transcriptional regulation ( 10,11,13 ). It has been proposed that all proteins with a START domain contain a similar binding pocket that binds different ligands, among them cholesterol derivatives ( 14,15 ).
StarD4, StarD5, and StarD6 are members of the StarD4 subfamily of proteins, whose functions remain poorly defi ned. These proteins have been shown to contain 205-233 amino acid residues, sharing 26-32% identity with each other ( 16 ). Therefore, they are predicted to be cytoplasmic proteins, like PCTP/StarD2 ( 16 ). StarD4 and StarD5 are expressed in a wide number of tissues. StarD4 is expressed at the highest levels in the liver ( 16 ). Recent studies have shown that StarD4 is also expressed in keratinocytes ( 17 ), in the mouse fi broblast cell line 3T3-L1, and in human THP-1 macrophages. StarD4 protein appeared to be localized in the cytoplasm and the ER ( 18 ). StarD5, however, is mainly localized in immune-related cells and kidney, localizing to the cytoplasm, Golgi, and ER membranes ( 19,20 ).
StarD1 and MLN64/StarD3 are able to bind cholesterol ( 10,21 ), although cholesterol is not the only ligand for StarD3, as the ability of StarD3 to bind lutein has recently been shown ( 22 ). StarD4 has been shown to bind cholesterol, 7 ␣ -hydroxycholesterol, and 7-hydroperoxycholesterol ( 23,24 ), whereas StarD5 binds cholesterol and 25-hydroxycholesterol in vitro ( 10 ). The functional signifi cance of cholesterol binding to some STAR proteins is illustrated by the observation that overexpression of StarD5 or truncated StarD3 increases free cholesterol in membranes ( 10,(25)(26)(27)(28), evidence of their ability to transfer cholesterol among different cellular compartments. Furthermore, overexpression of full-length StarD1 or MLN64/StarD3 is able to increase steroidogenesis after overexpression in cell cultures ( 25,29,30 ). Finally, overexpression of StarD4 increases cholesterol ester synthesis in the ER ( 23 ).
Expression of the StarD4 and StarD5 genes is regulated differently. Whereas StarD4 expression is highly regulated rinses with PBS. The samples were then incubated for 2 h at room temperature with 1% BSA and 0.05% Tween 20 in PBS (buffer A). The coverslips were then incubated for 16 h at 4°C in primary antibodies diluted in buffer A and then washed in PBS+0.05% Tween 20. Then the coverslips were incubated for 1 h at room temperature with fl uorescent conjugated secondary antibodies (Alexa Fluor 488 chicken anti-rabbit IgG for BiP and Alexa Fluor 568 donkey anti-goat IgG for StarD5) diluted in buffer A. DNA was stained with DAPI dye. Finally, the coverslips were washed again in PBS plus 0.05% Tween 20 and mounted on slides. The cells were viewed using a Carl Zeiss LSM 700 confocal microscope.

Subcellular protein fractionation
For subcellular localization of StarD5 protein, cellular fractions were isolated from 3T3-L1 cells by using the Subcellular Protein Fractionation Kit (Pierce). Each gel lane was loaded with an amount of protein equivalent to the same number of cells.

Transfection experiments
Cells were seeded onto 60 mm or 100 mm dishes one day before transfection. The number of cells seeded was adjusted so that the cells were 80% confl uent at the time of transfection. Transfections were performed using Lipofectamine LTX and PLUS reagents following manufacturer's recommendations.

Statistics
Data from real-time quantitative RT-PCR (qRT-PCR), immunoblots, fi lipin staining, cholesterol quantifi cation, and mRNA stability assays are reported as means ± SE of at least three separate experiments. Control and experimental groups were compared by Student t -test with the threshold for signifi cance at P < 0.05.

ER stress agents induce StarD5 mRNA expression in 3T3-L1 cells
It has been reported that StarD5 mRNA levels are induced by ER stress in macrophages, in NIH-3T3 cells ( 26 ), and in the HK-2 human proximal tubule cells ( 20 ). To further characterize the molecules and mechanisms involved in this process, we incubated 3T3-L1 cells with the ER stress inducers thapsigargin, a noncompetitive inhibitor of ER enzymes such as Ca 2+ -ATPase that blocks the ability of cells to pump Ca 2+ into the ER, and tunicamycin, a N -glycosylation inhibitor, for 20 h. We used qRT-PCR to quantify BiP and StarD5 mRNAs. As expected, both Tg and Tu induced ER stress in 3T3-L1 cells based on the expression of the ER stress marker BiP ( Fig. 1A ). StarD5 mRNA was also induced, consistent with previous reports ( 26 ). Furthermore, StarD5 protein is also increased upon treatment ( Fig. 1B ).

ER stress increases free cholesterol accumulation and HMG-CoA reductase mRNA levels in 3T3-L1 cells
In addition to the effect on StarD5 mRNA levels in ERstressed 3T3-L1 cells due to the Tg treatment, we noticed an increase in fi lipin staining ( Fig. 2A ), an indication of increased free cellular cholesterol. Free cellular cholesterol was quantifi ed by enzymatic methods ( Fig. 2B ) and correlated with an increase in HMG-CoA reductase expression ( Fig. 2C ). Free cholesterol and StarD5 mRNA RNA isolation and real-time quantitative RT-PCR Total RNA was isolated from cultured cells by using the SV Total RNA Isolation System (Promega). Two and one-half micrograms of RNA was reverse transcribed by using oligo-dT and M-MLV Reverse Transcriptase (Invitrogen), following manufacturer's instructions. BiP, CHOP, ATF4, XBP-1(s), StarD5, and HMG-CoA reductase mRNAs were quantifi ed using 20 ng of cDNA, forward and reverse primers at 125 nM each, and RT 2 Real-Time SYBR Green/ROX PCR Master Mix (SA Biosciences, Frederick, MD). The primers used are shown in Table 1 . A 7500 Fast Real-Time PCR System (Applied Biosystem, Foster City, CA) was used with the following thermal cycling profi le: 48°C for 30 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. The threshold was set in the linear range of normalized fl uorescence, and a threshold cycle (Ct) was measured in each well. Each sample was amplifi ed in triplicate for the genes of interest, and the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a control. Each Ct value for the mRNA of interest was then normalized to the corresponding value for GAPDH mRNA and expressed as a ratio.

Filipin staining and quantifi cation of cholesterol in 3T3-L1 cells
Cells were seeded on coverslips at 100% confl uency and were treated with thapsigargin (2 M) or DMSO in the presence or absence of 20 M mevinolin and 0.5 M mevalonate for the indicated lengths of time. Mevinolin is an HMG-CoA reductase inhibitor that stops all de novo cholesterol synthesis. The mevalonate concentration used is known to maintain synthesis of nonsterol metabolites of the mevalonate pathway needed for cell viability but it does not maintain cholesterol synthesis. Then the cells were washed three times with phosphate buffered saline solution (PBS) and fi xed in 3.7% formaldehyde in PBS for 10 min at 4°C followed by three rinses with PBS. The cells were permeabilized with 0.1% Triton X-100 in PBS for 30 min at 4°C followed again by three rinses with PBS. Then cells were stained with 1 mg/ml of fi lipin in PBS plus 0.05% BSA for 30 min at 37°C and washed three times, 7 min each, with PBS while rocking gently at room temperature in the dark. The coverslips containing the cells were mounted onto glass slides and viewed using a Nikon Ti-U microscope. Relative levels of free cholesterol from 3T3-L1 cell lysates were quantitated using the Free Cholesterol E kit from Wako.

Immunofl uorescence microscopic detection of StarD5 in 3T3-L1 cells
Cells were grown on coverslips and rinsed three times with PBS and then fi xed in 3.7% formaldehyde in PBS for 10 min at 4°C followed by three rinses with PBS. The cells were permeabilized with 0.1% Triton X-100 in PBS for 4 min at 4°C followed again by three was able to induce StarD5 mRNA expression ( Fig. 4B ). As expected, the ER-resident protein BiP was increased by ATF6 and XBP-1(s) factors.

ER stress stabilizes StarD5 mRNA
The expression studies shown in Fig. 3 indicate that the ER stress transcriptional factor XBP-1(s) is activated prior to StarD5 induction. The overexpression studies shown in Fig. 4 indicate that XBP-1(s) overexpression activates StarD5 expression. Together, these two studies suggest that StarD5 expression is activated under conditions of ER stress at the transcriptional level, perhaps by XBP-1(s). In an attempt to directly evaluate this process, we carried out promoter studies using a construct that contains 2,000 bp peaked at 6 h after the Tg treatment (see below). HMG-CoA reductase mRNA levels peaked 2 h after the addition of Tg, and after that, they steadily decreased, probably due to a general suppression of ER protein synthesis as a result of the ER stress. Sixteen hours after the addition of Tg, the cells were dead (data not shown).

XBP-1(s) overexpression induces StaD5 expression in 3T3-L1 cells
To further characterize the ER stress-mediated activation of StarD5 expression, we performed a time course study. 3T3-L1 cells were incubated with Tg for different times, and then StarD5 mRNA and other transcripts of key genes involved in the ER stress response were quantifi ed. Fig. 3 shows that expression of BiP (a protein involved in the folding and assembly of proteins in the ER) and CHOP (a transcription factor induced by ER stress and involved in apoptosis) increased over time in a similar manner as the ER stress transcription factors XBP-1(s) and ATF4, starting at 1 h after Tg addition. StarD5 mRNA, on the other hand, was induced 2 h after the Tg treatment, suggesting that the increase in StarD5 expression was either a consequence of the induction of XBP-1(s) and ATF4 expression by ER stress caused by Tg or a consequence of the induction of ATF6 (see below). All mRNA quantifi ed peaked at 4-6 h and then decreased after that, mainly due to cell death.
To test which ER stress transcription factor is involved in the regulation of StarD5, 3T3-L1 cells were transfected with expression plasmids encoding the transcription factors ATF6(1-373), XBP-1(s), and ATF4. As expected, immunofl uorescence revealed effi cient nuclear localization of the transcription factors ( Fig. 4A ). Then BiP and StarD5 mRNA levels were quantifi ed by qRT-PCR. Only XBP-1(s)  shown in Fig. 1 , StarD5 protein increased as a result of the ER stress. To confi rm the subcellular relocation of StarD5 as a result of the ER stress, we performed subcellular fractionation in 3T3-L1 cells treated as above, using I B-␣ as a cytosolic marker, Calnexin as an ER marker, and Lamin B1 as a of 5 ′ -fl anking region of the human StarD5 gene in front of the luciferase gene. After transfection of this construct into 3T3-L1 cells, they were incubated in the absence or presence of Tg. Luciferase activity was unchanged (data not shown), indicating that either StarD5 expression is not activated at the transcriptional level or that the elements responsible for that regulation are found outside the 2,000 bp of 5 ′ -fl anking region tested.
Given the lack of Tg-mediated regulation of the StarD5 proximal promoter region, we performed studies to characterize whether StarD5 mRNA is regulated by posttranscriptional mechanisms. Fig. 5 shows that induction of ER stress in the presence of the transcriptional inhibitor actinomycin D signifi cantly stabilized StarD5 mRNA.

StarD5 protein rearranges to perinuclear membranes in ER-stressed cells
We reported that StarD5 protein is localized in the cytosol; it is also loosely associated with the Golgi in macrophages ( 19 ). Because StarD5 binds cholesterol ( 10 ), we hypothesize that StarD5 might accumulate at cholesterol-loaded membranes. Immunofl uorence studies ( Fig. 6A ) showed that in 3T3-L1 cells StarD5 is located mainly in the nucleus, with some protein also in the cytosol. But when the cells were stressed with Tg, StarD5 rearranged to perinuclear membranes and to the cytosol ( Fig. 6A ). As controls, we used the ER marker BiP and DNA staining with DAPI. As   StarD5 is located in the cytosol and the Golgi ( 19 ). In these studies, we show that induction of ER stress by Tg treatment leads to an increase in free cholesterol accumulation as determined by fi lipin staining ( Fig. 2A ) and by enzymatic methods ( Fig. 2B ), which directly correlates with an increase in HMG-CoA reductase expression ( Fig. 2C ). We also show evidence that StarD5 mRNA is induced by a double mechanism: fi rst, the ER stress-induced transcriptional factor XBP-1(s) is capable of increasing StarD5 expression ( Fig. 4 ), and second, ER stress stabilizes the StarD5 mRNA ( Fig. 5 ). ER stress also relocalizes the StarD5 protein. In nuclear marker. Fig. 6B shows that, in agreement with the immunofl uorescence studies described above, StarD5 moved from a mainly nuclear location (and some minor localization in the cytosol) to a mostly membrane and cytosolic localization in Tg-stressed cells.

DISCUSSION
StarD5 is a cholesterol-binding protein ( 10 ) whose expression is induced by ER stress ( 26 ). In macrophages, plays a critical function in ER cholesterol homeostasis during ER stress. As excess ER free cholesterol causes ER stress ( 33 ), StarD5 expression increases under these conditions to transport excess cholesterol from the ER to the Golgi compartment, and from there to effl ux pathways to reduce ER free cholesterol levels. However, other delivery targets for this free cholesterol by StarD5 (for generation of oxysterols or for esterifi cation) cannot be ruled out. Therefore, StarD5 could play a role in the rapid and nonvesicular transport of cholesterol from the ER during the UPR as an early line of defense against excess accumulation of cholesterol in the ER. Although further studies will be necessary to determine the function of StarD5 in cholesterol homeostasis during the UPR, StarD5 could be a key target for prevention of ER stress-mediated diseases, including atherosclerosis.
3T3-L1 cells, StarD5 is located mainly in the nucleus, but under ER stress conditions, StarD5 concentrates to the membranes and cytosol ( Fig. 6 ).
It has been shown that StarD5 mRNA levels are induced by ER stress in NIH-3T3 cells ( 26 ). However, the mechanisms involved in this process remain to be determined. Promoter studies attempting to activate the StarD5 promoter under ER stress conditions using 2,000 bases upstream of the StarD5 5 ′ -fl anking region to drive the expression of a reporter gene failed, in agreement with earlier reports using 400 nucleotides of the promoter region ( 26 ). However, two observations strongly suggest that StarD5 transcription is activated by the ER stressor Tg. First, expression of the XBP-1(s) transcriptional factor precedes StarD5 mRNA induction ( Fig. 3 ). And second, XBP-1(s) overexpression alone activates StarD5 expression ( Fig. 4 ). If this is indeed the case, the ER stress regulatory elements in the StarD5 gene should be located somewhere outside the fi rst 2,000 nucleotides of the 5 ′fl anking region. These observations lead us to hypothesize that under normal (nonstress) conditions, StarD5 mRNA is rapidly degraded ( Fig. 5 ), but under ER stress conditions, StarD5 mRNA is stabilized by an unknown signaling, which is further increased by an increased transcription of the gene ( Fig. 4 ). An alternative explanation for these observations is that XBP-1(s) functions not only as a transcriptional activator but also as an mRNA stabilization factor. However, this function has not been described for XBP-1(s) yet. It should be noted that overexpression of two other transcriptional factors involved in the UPR, ATF6(1-373) and ATF4, failed to activate StarD5 expression. Therefore, only the branch of the UPR controlled by the signal transducer IRE1 seems to be involved in the activation of StarD5 expression.
ER stress not only dramatically increases StarD5 expression but also redistributes the StarD5 protein. In macrophages, StarD5 is found in the Golgi and cytosol ( 10 ). On the other hand, in 3T3-L1 cells under nonstress conditions, StarD5 is mainly nuclear with some protein found in the cytosol; however, when cells are ER stressed, it concentrates mostly at the membranes ( Fig. 6 ). Compared with macrophages, 3T3-L1 fi broblasts show very small, almost nonexistent Golgi when probed by immunofl uorescence with a Golgi marker (results not shown). Therefore, localization to the Golgi in 3T3-L1 fi broblasts is diffi cult to assess. One potential explanation for this relocation to membranes is that StarD5 gets to the membranes as a result of being bound to cholesterol, which is increased in the UPR ( Fig. 2 ), perhaps to protect the integrity of the ER membranes. What remains to be studied, however, is the role of StarD5 in the nucleus, which seems to be the major subcellular localization of StarD5 in 3T3-L1 cells under normal conditions.
In summary, in the present studies, we have identifi ed two mechanisms of action for the ER stress-mediated increase of StarD5 expression, namely, an increase in transcription mediated by the XBP-1(s) factor and stabilization of its mRNA. Furthermore, based on previous fi ndings ( 20,26 ) and those described here, we propose that StarD5