Coordinate Regulation of Phospholipid Biosynthesis and Secretory Pathway Gene Expression in XBP-1(S)-induced Endoplasmic Reticulum Biogenesis*

Development of the expansive endoplasmic reticulum (ER) present in specialized secretory cell types requires X-box-binding protein-1 (Xbp-1). Enforced expression of XBP-1(S), a transcriptional activator generated by unfolded protein response-mediated splicing of Xbp-1 mRNA, is sufficient to induce proliferation of rough ER. We previously showed that XBP-1(S)-induced ER biogenesis in fibroblasts correlates with increased production of phosphatidylcholine (PtdCho), the primary phospholipid of the ER membrane, and enhanced activities of the choline cytidylyltransferase (CCT) and cholinephosphotransferase enzymes in the cytidine diphosphocholine (CDP-choline) pathway of PtdCho biosynthesis. Here, we report that the level and synthesis of CCT, the rate-limiting enzyme in the CDP-choline pathway, is elevated in fibroblasts overexpressing XBP-1(S). Furthermore, overexpression experiments demonstrated that raising the activity of CCT, but not cholinephosphotransferase, is sufficient to augment PtdCho biosynthesis in fibroblasts, indicating that XBP-1(S) increases the output of the CDP-choline pathway primarily via its effects on CCT. Finally, fibroblasts overexpressing CCT up-regulated PtdCho synthesis to a level similar to that in XBP-1(S)-transduced cells but exhibited only a small increase in rough ER and no induction of secretory pathway genes. The more robust XBP-1(S)-induced ER expansion was accompanied by induction of a wide array of genes encoding proteins that function either in the ER or at other steps in the secretory pathway. We propose that XBP-1(S) regulates ER abundance by coordinately increasing the supply of membrane phospholipids and ER proteins, the key ingredients for ER biogenesis.

The endoplasmic reticulum (ER) 3 is a multifunctional organelle responsible for the folding and assembly of all proteins targeted to the secretory pathway (1). As such, the ER can adapt to accommodate an increased load of nascent polypeptides. For example, when B-lymphocytes differentiate into antibody-secreting plasma cells, an elaborate network of rough ER develops to facilitate immunoglobulin production (2)(3)(4). Likewise, the rough ER is highly developed in other specialized secretory cell types such as pancreatic acinar cells that secrete copious amounts of digestive enzymes (5). In contrast, the ER is sparse in non-secretory cells, such as reticulocytes (6). ER abundance, therefore, is regulated according to the demands on the secretory pathway. However, the mechanisms that regulate ER biogenesis are incompletely defined (7).
A key regulator of ER homeostasis is the unfolded protein response (UPR) pathway, a complex signaling system emanating from the ER membrane (8). When the protein folding capacity of the ER is challenged, the UPR relieves the resulting stress by repressing translation, increasing expression of ER chaperones and folding enzymes, and enhancing ER-associated degradation (8). In addition, recent studies have uncovered a connection between the UPR and ER abundance (9,10). The UPR-regulated transcription factor X-box-binding protein-1 (XBP-1) is required for proper development and function of plasma cells (11,12), pancreatic acinar cells, and salivary gland cells (13), all of which normally contain large quantities of rough ER necessary for high level synthesis of their respective secretory cargoes. In the absence of Xbp-1 expression, these cell types exhibit poorly developed ER, reduced expression of many ER proteins, and severely compromised secretory activity (11,13). Xbp-1 mRNA is modified by a novel splicing mechanism initiated by IRE1 (first identified in yeast, inositol requiring mutant), an ER transmembrane kinase/endoribonuclease that serves as a proximal transducer of the UPR (14,15). IRE1 executes site-specific cleavage of Xbp-1 mRNA and the resulting fragments are ligated to yield a transcript encoding a basic leucine zipper protein, termed XBP-1(S), that bears a strong transactivating domain (16 -18). Enforced expression of XBP-1(S) is sufficient to drive expansion of the rough ER (10), increase expression of a large number of ER proteins (9,19), and augment protein biosynthesis (9). Therefore, XBP-1(S) is necessary and sufficient for the biogenesis of functional ER, yet the mechanisms by which this transcription factor mediates these effects have not been fully delineated.
Synthesis of phophatidylcholine (PtdCho), the most abundant cellular phospholipid and a major component of ER membranes, is up-regulated in fibroblasts overexpressing XBP-1(S) (10). PtdCho is primarily produced by the cytidine diphosphocholine (CDP-choline), also known as the Kennedy pathway (20). In the rate-limiting step of the pathway, choline cytidylyltransferase (CCT) converts phosphocholine to CDP-choline in the presence of CTP (21). The phosphocholine moiety of CDPcholine is then transferred to diacylglycerol (DAG), yielding PtdCho (20). This final step is catalyzed by either cholinephosphotransferase (CPT1) (22) or choline/ethanolaminephosphotransferase (CEPT1), a bifunctional enzyme that can synthesize both PtdCho and phosphatidylethanolamine (PtdEtn) (23). We previously showed that fibroblasts overexpressing XBP-1(S) exhibit enhanced activities of CCT and CPT1/CEPT1 (10), but the relative role of these alterations in XBP-1(S)-induced Ptd-Cho biosynthesis and ER expansion has not been clarified. Neither is it clear whether an increased supply of PtdCho is sufficient for ER biogenesis.
Here, we report that the level and synthesis of CCT is up-regulated in fibroblasts overexpressing XBP-1(S). Furthermore, our data indicate that enhanced CCT activity in the CDP-choline pathway is the primary means by which XBP-1(S) up-regulates PtdCho production. We also demonstrate that increased synthesis of PtdCho alone in CCT-transduced cells is sufficient for only a meager expansion of rough ER. In contrast, XBP-1(S)transduced fibroblasts exhibit increased PtdCho synthesis, elevated expression of many ER proteins, and robust ER expansion. Thus, we propose that XBP-1(S) orchestrates ER biogenesis by coordinately regulating phospholipid biosynthesis and expression of ER proteins.
Immunoblotting-Cell lysates of pelleted cells were prepared using a 1% Nonidet P-40 lysis buffer (24). Microsomes were prepared from pelleted cells as described (23). Total protein concentrations of clarified Nonidet P-40 lysates and microsomal preparations were determined using the Bio-Rad protein assay. Proteins were resolved by SDS-PAGE and analyzed by chemiluminescent immunoblotting as previously described (24). The rabbit anti-CCT␣ antibody (27) was raised against full-length recombinant rodent CCT␣ (28). The rabbit anti-XBP-1 antibody was raised with the assistance of Rockland Immunochemicals against the N-terminal 81 amino acids of XBP-1 fused in-frame to glutathione S-transferase. Polyclonal rabbit antisera against BiP/GRP78, GRP94, ERdj3, and calnexin were generously provided by Dr. Linda Hendershot (St. Jude Children's Research Hospital, Memphis, TN). Rabbit anti-TRAP␣ was a gift from Dr. Chris Nicchitta (Duke University, Durham, NC). The rabbit anti-protein disulfide isomerase polyclonal antibody (Stressgen Bioreagents), mouse anti-␤-actin monoclonal antibody, clone AC-15 (Sigma), peroxidaseconjugated donkey anti-mouse IgG, and peroxidase-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories) were purchased.
Metabolic Labeling and Analysis of CCT␣ Synthesis-Cells were washed twice with warm phosphate-buffered saline and then cultured in warm media lacking methionine and cysteine (Invitrogen) for 20 min. Cells were then labeled for various intervals with [ 35 S]methionine and [ 35 S]cysteine using 100 Ci/ml of Tran 35 S-label (MB Biomedicals). Labeled cells were washed twice with cold phosphate-buffered saline and then solubilized in the dish on ice in cold lysing buffer (0.5% Nonidet P-40, 0.5% sodium deoxycholate, 150 mM NaCl, 50 mM Tris-HCL, pH 7.5, 20 g/ml leupeptin, 40 g/ml aprotinin, 100 g/ml phenylmethylsulfonyl fluoride). Post-nuclear supernatants were prepared, normalized to cell number, and precleared with protein A-Sepharose beads (Sigma) pre-coated with normal rabbit sera. Pre-cleared lysates were then incubated with protein A-Sepharose beads pre-coated with rabbit anti-CCT␣ antibodies. In each case, the beads were then washed 4 times with cold washing buffer (0.5% Nonidet P-40, 0.5% sodium deoxycholate, 400 mM NaCl, 50 mM Tris-HCl, pH 7.5), resuspended in reducing SDS-PAGE sample buffer, and boiled for 5 min. Samples were resolved by SDS-PAGE using 10% polyacrylamide gels. Gels were processed, and signals were visualized and quantified using a Typhoon PhosphorImager and ImageQuant software (Amersham Biosciences) as described (24). To measure the incorporation of radiolabeled methionine and cysteine into proteins, lysates from metabolically labeled cells were normalized to cell number, and equivalent amounts were spotted on Whatman filter paper in tripli-cate. Filters were boiled in 10% trichloroacetic acid and analyzed by scintillation spectroscopy.
Analysis of Phosphocholine, CDP-choline, PtdCho, and PtdEtn Synthesis-To assess production of phosphocholine, CDP-choline, and PtdCho, cells were cultured in choline chloride-free RPMI 1640 medium containing 2 Ci/ml [methyl- 3 H]choline chloride (81 Ci/mmol; Amersham Biosciences) for 2 h at 37°C. After labeling, cells were washed with cold phosphate-buffered saline and scraped in methanol/water (2:0.8, v/v). Lipids were extracted (29), radiolabeled PtdCho in the lipid fraction was assessed (10), and the aqueous phase was saved for analysis. After drying by vacuum, choline, phosphocholine, and CDP-choline in the aqueous phase were separated by thin layer chromatography in a solvent system of methanol, 0.78% NaCl, ammonium hydroxide (50:50:5, v/v). The spots were identified by co-migration with authentic standards, scraped, and assessed for radioactivity by scintillation spectroscopy. In some experiments, PtdCho synthesis was measured as previously described (10). To assess PtdEtn synthesis, cells were cultured for 2 h at 37°C in Dulbecco's modified Eagle's medium containing 2 Ci/ml [1-3 H]ethanolamine hydrochloride (60 Ci/mmol; American Radiolabeled Chemical). After labeling, radiolabel incorporated into chloroform-soluble metabolites was determined as in the analysis of PtdCho synthesis.
Analysis of CCT and CPT Enzymatic Activity-Pelleted frozen cells (2 ϫ 10 7 cells) were used for assays of CCT enzymatic activity as described previously (30,31). CPT activity was assessed as described previously (23) using microsomes prepared from pelleted frozen cells (2 ϫ 10 7 cells).
Electron Microscopy, Stereological Measurements of Rough Endoplasmic Reticulum, and Cell Size Determinations-Cells were processed and examined by electron microscopy as previously described (10). Two independent stereological sets containing electron micrographs of empty vector-, XBP-1(S)-, and CCT␣-transduced cells at magnifications of 8,000 and 20,000 were analyzed for ER surface area and ER volume as described previously (10). Average cell size was determined from light microscopy analysis of thick sections using the Scion image program (Scion). The mean cell area (m 2 ) was determined in a minimum of 10 fields.
Microarray Analysis-Total RNA was isolated using the RNeasy kit (Qiagen) and subjected to DNase I digestion to remove contaminating genomic DNA. Microarray analysis was performed for each sample using the Affymetrix GeneChip mouse genome arrays MOE430A or MOE430V2 at the Hartwell Center for Bioinformatics and Biotechnology, St. Jude Children's Research Hospital. For empty vector-and XBP-1(S)transduced cells, a total of four separate experiments was performed. For CCT␣-transduced cells, two separate experiments were performed. In each experiment, the integrity of the total RNA was determined using the Agilent Bioanalyzer Lab-on-a-Chip, and the RNA sample was converted to cDNA, labeled, and fragmented using procedures recommended by Affymetrix. Fragmented labeled cDNA (4.0 g) was hybridized for 16 h at 50°C to the array. After washing, staining and scanning of the arrays were performed according to the manufacturer's protocol (Affymetrix). The arrays were analyzed using the GeneChip Operating Software (GCOS), and global scaling was used to normalize the data from different arrays. Spotfire DecisionSite 8.2.1 and the NetAffx TM Analysis Center (32) were used to analyze array results. Genechips were hybridized in duplicate (St. Jude Children's Research Hospital). The statistical analyses were performed on data sets from all of the experiments that were independently processed and hybridized. The statistical significance of XBP-1(S) or CCT␣ overexpression was determined using two-tailed unpaired t test with the confidence intervals set at 95% and data with p values Յ0.05 were significant.
Phospholipid Analysis-Cells (2 ϫ 10 7 ) were pelleted and supernatant was removed. Cells were flash-frozen and stored at Ϫ80°C until analysis. Cell pellets were thawed on ice and resuspended in 1 ml of water or phosphate-buffered saline. The total volume was measured, and a 100-l aliquot was removed for protein determination (33). Lipids were extracted (3) from a 900-l aliquot using 2.4 ml of acetic acid in methanol (2%, v/v) and 1 ml of chloroform in the first step followed by 1.5 ml of chloroform and 1.2 ml of water in the second step to yield two phases, organic and aqueous. The organic phase was collected and dried. Lipids were resuspended in 100 l of chloroform/ methanol (2:1, v/v). A 1-l aliquot was loaded onto a thin-layer silica gel rod and developed first in ether, dried, and then developed in chloroform/methanol/acetic acid/water (50:25:8:2, v/v/ v/v). Lipids were detected by flame ionization using an Iatroscan instrument (Iatron Laboratories), and peaks were integrated with PEAKSIMPLE software (SRI Instruments). Peaks were identified by comigration with authentic standards, and the amount of each lipid species was calculated using standard curves for each.

Phospholipid Biosynthesis in XBP-1(S)-transduced Fibroblasts-
We previously demonstrated that enforced expression of the transcription factor XBP-1(S) is sufficient to up-regulate PtdCho biosynthesis in NIH-3T3 fibroblasts (10). Furthermore, analysis of the CDP-choline pathway for PtdCho biosynthesis in XBP-1(S)-transduced cells revealed basal activity for choline kinase (CK), enhanced activity for CCT, and markedly elevated activity for CPT (10). To further investigate the status of the CDP-choline pathway in XBP-1(S)-transduced cells, we metabolically labeled cells with [ 3 H]choline and monitored the output of each step in the pathway. Chromatographic analysis of soluble cellular material allowed for analysis of both phosphocholine and CDP-choline, the products of the CK-and CCTcatalyzed steps, respectively. The level of phosphocholine synthesis was similar in empty vector-and XBP-1(S)-transduced cells (Fig. 1A), in keeping with the similar CK activity present in both populations (10). In contrast, the level of CDP-choline synthesis was increased ϳ2-fold in XBP-1(S)-transduced cells (Fig. 1B), and as expected, this was accompanied by elevated production of PtdCho (Fig. 1C). These data indicate that the enhanced CCT activity present in XBP-1(S)-transduced cells increases the supply of CDP-choline for CPT-catalyzed production of PtdCho. Similar to PtdCho, the level of PtdEtn, the second most abundant phospholipid in cellular membranes, elevates in XBP-1(S)transduced fibroblasts (10). To further explore this observation, we performed metabolic labeling studies with [ 3 H]ethanolamine and found that the synthesis of PtdEtn was increased ϳ2-fold in XBP-1(S)-transduced cells (Fig. 1D). Therefore, enforced expression of XBP-1(S) augments cellular capacity for de novo synthesis of PtdCho and PtdEtn, two key phospholipid components of the ER membrane.
Effect of XBP-1(S) on the Level and Synthesis of CCT␣ in Fibroblasts-PtdCho is the most abundant phospholipid in cellular membranes; thus, we focused our studies on delineating the connection between XBP-1(S) and the regulation of PtdCho biosynthesis. Given that CCT catalyzes the rate-limiting step in the CDP-choline pathway, we investigated the mechanism by which its activity is augmented in XBP-1(S)-transduced cells. We reasoned that the elevation of CCT activity might involve an increase in the level of the CCT enzyme. In testing this hypothesis, we focused on the steady-state level of CCT␣, the predominant CCT isoform expressed in NIH-3T3 cells (10). Immunoblot analysis of Nonidet P-40 soluble material revealed a nearly 50% increase in CCT␣ in XBP-1(S)-transduced cells (Fig. 2). Thus, the enhanced CCT activity in XBP-1(S)-transduced fibroblasts correlates with an increased pool of CCT␣ enzyme.
Enforced expression of XBP-1(S) in fibroblasts does not induce increased levels of CCT␣ transcripts (10); therefore, the rise in CCT␣ enzyme in this system is mediated by a posttranscriptional or post-translational mechanism. A previous study showed that enforced expression of XBP-1(S) in Raji cells, a human B cell line, resulted in an overall increase in translation (9). Similarly, we observed a modest, but measurable, increase in total protein synthesis in XBP-1(S)-transduced NIH-3T3 cells (Fig. 3A). This led us to assess the effect of XBP-1(S) on the level of CCT␣ translation. First, we verified that radiolabeled CCT␣ could be specifically immunoprecipitated from NIH-3T3 cell lysates (Fig. 3B). We then metabolically labeled empty vector-and XBP-1(S)-transduced cells with [ 35 S]methionine and [ 35 S]cysteine for increasing intervals and assessed the incorporation of radiolabel into newly synthesized CCT␣ (Fig.  3C, upper panel) and into total proteins (Fig. 3C, middle panel). These experiments revealed that CCT␣ synthesis increased  ϳ60% in XBP-1(S)-transduced cells. Importantly, the effect of XBP-1(S) on translation was not specific for CCT␣, as revealed by the increase in total protein synthesis (Fig. 3A) and the profile of labeled proteins (Fig. 3C, middle panel) in XBP-1(S)transduced cells. Although the relative contributions of transcriptional and translational regulation to the overall increase in total protein synthesis cannot be distinguished from this study, the data indicate that CCT␣ is one of many proteins produced at a higher rate upon enforced expression of XBP-1(S). In addition, pulse-chase studies indicated that CCT␣ is a long-lived protein in NIH-3T3 cells (half-life Ն12-14 h) and is not further stabilized by enforced expression of XBP-1(S) (data not shown). These data suggest that elevated translation of CCT␣ is the primary mechanism by which the pool of CCT␣ enzyme increases in XBP-1(S)transduced fibroblasts.
Effect of High Level CCT and CPT Activity on PtdCho Biosynthesis in Fibroblasts-We next determined whether an increase in CCT activity is sufficient to up-regulate PtdCho synthesis in NIH-3T3 fibroblasts. As a comparison, we assessed the effect of elevated CPT activity on PtdCho production. To raise the level of these enzymatic activities, we used retroviral transduction to overexpress CCT␣, CPT1, and CEPT1 in NIH-3T3 cells. CCT␣transduced cells exhibited a large increase (ϳ10-fold) in CCT activity, much higher than that observed upon enforced expression of XBP-1(S), whereas neither CPT1 nor CEPT1 overexpression altered CCT activity (Fig. 4A). Total CPT activity was enhanced in both CPT1-and CEPT1-transduced cells to levels comparable with or higher than that present in cells overexpressing XBP-1(S) (Fig. 4B). CCT␣ overexpression did not modulate the CPT activity (Fig. 4B). These data confirmed that overexpressing CCT␣, CPT1, and CEPT1 effectively increased the respective enzymatic activities in NIH-3T3 cells. We then performed metabolic labeling experiments with [ 3 H]choline and found that PtdCho synthesis was induced in cells transduced with CCT␣ (Fig. 4C). However, neither CPT1 nor CEPT1 overexpression augmented PtdCho synthesis in NIH-3T3 cells (Fig. 4C). Therefore, elevated CCT activity is sufficient to increase Ptd-Cho synthesis in fibroblasts, in keeping with its role as the ratelimiting enzyme in the CDP-choline pathway (21). Interestingly, although overexpression of CCT␣ generated a much larger amount of CCT activity than did enforced expression of XBP-1(S) (Fig. 4A), PtdCho synthesis was up-regulated to a similar level under both conditions (Fig. 4C). These data sug- gest that the XBP-1(S)-mediated increase in CCT activity is sufficient for maximal output by the CDP-choline pathway.
Abundance of ER in CCT␣-Versus XBP-1(S)-transduced Fibroblasts-PtdCho is the most abundant phospholipid in cellular membranes, including ER membranes. Thus, the comparable increase in PtdCho synthesis in CCT␣and XBP-1(S)transduced fibroblasts prompted us to ask whether an increase in the supply of PtdCho is sufficient for ER expansion. Visual inspection of electron micrographs revealed that overexpression of CCT␣, like that of XBP-1(S), resulted in an apparent increase in cell size and in the abundance of intracellular mem-brane-bound organelles, many of which appeared to be ribosome-studded rough ER (Fig. 5A). Quantitative analysis of cell size confirmed that CCT␣-transduced cells were ϳ20% larger than control cells (empty vector cells, 145.6 Ϯ 8.9 m 2 ; CCT␣ cells, 174.2 Ϯ 11.4 m 2 ), whereas XBP-1(S) induced an ϳ50% increase in cell size (XBP-1(S) cells, 218.7 Ϯ 13.9 m 2 ) as previously reported (10). Stereological analysis of high power electron micrographs confirmed that both the surface area and volume of rough ER were enhanced in cells overexpressing CCT␣, 1.8-and 1.5-fold, respectively (Fig. 5B). Immunoblot analysis indicated that the modest increase in rough ER abundance in CCT␣-transduced cells could not be attributed to induction of XBP-1(S) synthesis (see Fig. 7). In contrast, as expected from our previous studies (10), XBP-1(S)-transduced cells exhibited a nearly 3-fold increase in both the surface area and volume of rough ER (Fig. 5B). These findings demonstrate that the magnitude of XBP-1(S)-mediated ER biogenesis was greater than that attained in cells overexpressing CCT␣ despite the equivalent level of PtdCho synthesis in both conditions.

Phospholipids and ER Proteins in CCT␣-Versus XBP-1(S)transduced Fibroblasts-The varying degrees of ER expansion
in CCT␣and XBP-1(S)-transduced fibroblasts provided an opportunity to assess how cellular lipid content and ER protein levels correspond to overall ER abundance. First, we compared the lipid content of CCT␣and XBP-1(S)-transduced cells and found that the total amount of PtdCho was increased similarly in both cell populations (Fig. 6). However, although the amount of PtdEtn was elevated in the XBP-1(S)-transduced cells, it was diminished in cells overexpressing CCT␣ (Fig. 6). There was a small increase in the level of cholesterol in CCT␣-transduced cells and little or no change in sphingolipid, cholesterol ester, and triglyceride levels upon enforced expression of either XBP-1(S) or CCT␣ (Fig. 6). Thus, CCT␣-and XBP-1(S)-transduced fibroblasts exhibited differences in lipid composition and content, including the ratio of PtdCho to PtdEtn, the two most abundant ER membrane phospholipids. Next, immunoblotting revealed that many ER resident proteins, including soluble components of the ER protein folding machinery (BiP/GRP78, GRP94, Erdj3, and protein disulfide isomerase) and a transmembrane subunit of ER translocons (TRAP␣), were up-regulated in XBP-1(S)-but not CCT␣-transduced cells (Fig. 7). These data demonstrate that elevated levels of PtdCho, PtdEtn, and resident ER proteins correlated with the greatest increase in rough ER.
Gene Expression Associated with ER Development-Our comparison of CCT␣and XBP-1(S)-transduced cells suggested that an increased supply of PtdCho was not sufficient for maximal ER expansion (Figs. 4 -6). Building upon these observations, we used microarray analysis to compare the profile of gene expression in fibroblasts 48 h after XBP-1(S) and CCT␣ transduction. Gene ontology analysis using the NetAffx TM Analysis Center (32) correlated the regulated probe sets with the most significantly perturbed cellular components in cells transduced by XBP-1(S). Similar to the findings of previous studies (9,19), our analysis indicated that XBP-1(S) exerts its greatest effects on genes involved with the ER network and ER to Golgi vesicle-mediated transport (supplemental Tables S1 and S2). Specifically, we found that enforced expression of XBP-1(S) in NIH-3T3 cells up-regulated transcripts for 122 identified genes (Ն2-fold, p Յ 0.05) that function either in the ER or at other steps in the secretory pathway ( Table 1). The cohort of ER genes up-regulated in the XBP-1(S)-transduced cells included factors involved in targeting and translocation of nascent polypeptides into the ER (such as Srp genes for components of the signal recognition particle and Sec61 genes for translocon subunits), folding and assembly of nascent polypeptides in the ER lumen (such as Hspa5, which encodes BiP/ GRP78, Dnaj genes for ER resident DnaJ proteins, and protein disulfide isomerase genes for protein disulfide isomerases), N-linked glycosylation (such as the Ddost and Dad1 genes for components of the oligosaccharyltransferase), and ER-associated degradation (such as Derl1 and Edem1, which encode Derlin1 and Edem). In addition, the XBP-1(S)-transduced cells exhibited increased expression of a large number of genes involved in vesicular trafficking and transport. These included genes implicated in anterograde transport (such as Sec genes for components of COPII vesicles), retrograde transport (such as Cop genes for components of COPI vesicles), and distal transport through the Golgi and beyond (such as Vamp genes for v-SNARE proteins). In sharp contrast to the gene expression profile in XBP-1(S)transduced fibroblasts, microarray analysis revealed that no secretory pathway genes were up-regulated (Ն2-fold, p Յ 0.05) in cells overexpressing CCT␣. This striking difference was consistent with the immunoblot analysis of ER proteins (Fig.  7). In fact, enforced expression of CCT␣ induced expression of only 13 genes and reduced expression of only 4 genes (supplemental Table S5).
XBP-1(S) also altered the expression of a small number of cytoskeletal and lipid metabolic genes in the transduced fibroblasts (supplemental Tables S3 and S4), and these were not up-regulated in the CCT␣transduced cells. Cytoskeletal genes that were up-regulated in response to XBP-1(S) expression included those encoding myosin VIIa and ankyrin 3. On the other hand, the expression of ankyrin 2, plakophilin 4, paralemmin, and protein band 4.1 was reduced at least 2-fold. Lipid metabolic genes whose expression increased included those involved in fatty acid modification (Pecr and Cyb5) and de novo phosphatidic acid formation (Agpat6), whereas the Elovl4 gene that mediates fatty acid elongation was down-regulated. The largest change in lipid metabolic gene expression involved the Lipin genes Lpin3 and Lpin1, which increased more than 8-and 3.4-fold respectively. Lipin, primarily studied in hepatocytes and adipocytes, has been implicated in regulating expression of lipid metabolic genes (34 -38). In addition, the yeast homolog of Lipin was recently shown to function as a phosphatidic acid phosphatase that catalyzes for- mation of DAG from phosphatidic acid (39). In contrast to the Lipin genes, expression of Ppap2b (40,41), which encodes a distinct phosphatidic acid phosphatase, was reduced after XBP-1(S) transduction.
In summary, the microarray analysis revealed that the marked expansion of the rough ER induced by XBP-1(S) correlated with enhanced expression of a wide array of resident ER proteins, a broad group of factors that function at other steps in the secretory pathway, a small group of lipid metabolic genes, and a small set of cytoskeletal genes. This profile of gene expression did not accompany the more modest increase in rough ER that occurred in the CCT␣-transduced cells.

DISCUSSION
Our data provide the first mechanistic insight into how the UPR, via XBP-1(S), can modulate the activity of an enzyme that plays a pivotal role in phospholipid production. The level and synthesis of CCT␣ protein, the rate-limiting enzyme in the CDP-choline pathway of PtdCho biosynthesis, is up-regulated in XBP-1(S)-transduced fibroblasts (Figs. 2 and 3), providing an explanation for the increased CCT activity (10) and elevated production of CDP-choline (Fig. 1B). Given that the CCT␣ gene is not a target of XBP-1(S) (10), we speculate that XBP-1(S) regulates expression of another gene(s) that influences the translation of CCT␣ and many other proteins in this experimental system. In the Raji human B cell line, enforced expression of XBP-1(S) led to an increase in assembled (80 S) ribosomes (9). It follows that such a mechanism could enhance total protein synthesis as was observed both in Raji cells (9) and NIH-3T3 fibroblasts overexpressing XBP-1(S) (Fig. 3A). However, there is evidence that XBP-1(S) might mediate events that target specific mRNAs for translation. Specifically, XBP-1(S) is required for optimal synthesis of immunoglobulin heavy chains but not of light chains in activated B cells (42). Delineation of the mechanisms by which XBP-1(S) regulates translation will require further study, and a reasonable next step would be to examine the status of ribosome assembly and the association of CCT␣ transcripts with polysomes in our experimental system.
Although translational regulation appears to be the primary means by which the level of CCT␣ increases in XBP-1(S)-transduced fibroblasts, we emphasize that CCT␣ abundance in different cell types and/or during various developmental processes might be controlled by other mechanisms (43). In this regard, we have found evidence for stabilization of CCT␣ protein turnover in lipopolysaccharide-stimulated CH12 B cells, a murine B cell lymphoma. 4 LPS triggers murine B cells to proliferate and to differentiate into antibody-secreting cells, a process that includes induction of PtdCho biosynthesis and expansion of rough ER (2)(3)(4). Thus, multiple mechanisms can influence the level of CCT␣, and its activity can also be regulated post-translationally by phosphorylation/dephosphorylation (43) and subcellular localization (44). Whether XBP-1(S) 4 S. Jackowski, personal communication.  and equivalent amounts of cell lysate protein were resolved by standard SDS-PAGE under reducing conditions. Chemiluminescent immunoblotting was performed for the indicated ER proteins (BiP, immunoglobulin-binding protein/GRP78; ERdj3; GRP94; PDI, protein disulfide isomerase; TRAP␣, translocon associated protein ␣) and, as controls, XBP-1(S), CCT␣, and ␤-actin. MARCH 9, 2007 • VOLUME 282 • NUMBER 10 mediates any post-translational control of CCT␣ activity and the relative roles of the various CCT regulatory mechanisms in modulating PtdCho biosynthesis under different conditions awaits further investigation.

XBP-1 and ER Biogenesis
Our studies reveal that although XBP-1(S)-transduced cells exhibit enhanced activities for both CCT and CPT (10), it is the increase in CCT activity that augments production of PtdCho in this system. Elevation of CCT activity by overexpression of CCT␣ was sufficient to up-regulate PtdCho biosynthesis to a level equivalent to that observed in XBP-1(S)-transduced cells, whereas elevation of CPT activity by overexpression of either CPT1 or CEPT1 had no effect on PtdCho biosynthesis (Fig. 4). These data are in agreement with previous studies of PtdCho biosynthesis in cells overexpressing CCT␣ (45)(46)(47)(48) or CEPT1 (49). Furthermore, the fact that PtdCho biosynthesis was upregulated in CCT␣-transduced cells (Fig. 4C), whereas CPT activity remained at a basal level (Fig. 4B), argues that increased CPT activity is not required for enhanced output of the CDPcholine pathway in NIH-3T3 fibroblasts. However, it is noteworthy that up-regulation of CPT activity correlates with induction of PtdCho biosynthesis in PC12 neuronal cells undergoing neurite outgrowth in response to nerve growth factor (50), lipopolysaccharide (LPS)-stimulated splenic B cells (4), and in LPS-stimulated CH12 B cells. 4 Therefore, although basal CPT activity supports maximal PtdCho biosynthesis in NIH-3T3 fibroblasts, it is possible that CPT activity must be enhanced for optimal output of the CDP-choline pathway in other cell types.
Finally, our comparative analyses of XBP-1(S)-and CCT␣transduced fibroblasts provide insight into the basic "ingredi-ents" required to build and equip a larger ER. It was fortuitous that the level of PtdCho increased in the CCT␣-transduced NIH-3T3 fibroblasts as cells typically balance enforced induction of PtdCho biosynthesis with PtdCho degradation (27,45,47). At least some of the accumulated PtdCho in CCT␣-transduced cells was utilized for ER membrane assembly as these cells exhibited a measurable increase in rough ER (Fig. 5B). The increase in rough ER in cells overexpressing CCT␣ was not accompanied by induction of XBP-1(S) (Fig. 7) or genes encoding ER proteins. Thus, although inhibition of PtdCho biosynthesis has been linked to UPR activation (51), the increased supply of PtdCho in CCT␣-transduced NIH-3T3 cells did not trigger the UPR or induction of ER protein expression. These data underscore that PtdCho, the most abundant phospholipid in ER membranes, can be a major determinant of ER abundance and provide further evidence that CCT␣ plays a key regulatory role in ER biogenesis. On the other hand, it is striking that the level of PtdEtn, but not other major lipids, rises along with PtdCho in XBP-1(S)-transduced fibroblasts (Fig. 6). It would be interesting to investigate the role of PtdEtn in ER biogenesis and the possibility that the PtdEtn supply might influence the UPR and expression of secretory pathway genes.
Only a small number of genes implicated in lipid metabolism were up-regulated in the XBP-1(S)-transduced fibroblasts. The potential roles of these genes in regulating lipid biosynthesis and ER membrane biogenesis, particularly the Lipin genes that have recently been implicated in DAG synthesis (39), certainly warrant further study. However, we note that elevated expres- sion of this set of lipid metabolic genes is apparently not essential for induction of PtdCho biosynthesis in NIH-3T3 fibroblasts as their expression was not modulated in CCT␣transduced cells. In addition, quantitative analysis of DAG abundance and metabolic labeling with [ 3 H]acetate revealed that both the level and synthesis of DAG were comparable in empty vector-and XBP-1(S)-transduced cells (data not shown). Although these data indicate that the supply of DAG in NIH-3T3 cells is sufficient to support increased synthesis of PtdCho and PtdEtn, they do not rule out a role for Lipin in DAG production. Thus, the potential connection between XBP-1(S), the Lipin genes, and DAG synthesis will require further investigation, particularly in other cell types and developmental processes involving ER biogenesis.
Proteins, like lipids, are key components of the ER. Indeed, expression of a large number of genes encoding ER proteins was up-regulated in XBP-1(S)-transduced cells (Table 1), and this correlated with a proliferation of rough ER that was nearly twice the level observed in cells overexpressing CCT␣ (Fig. 5B). This group of ER protein components includes both soluble and transmembrane proteins, many of which play well defined roles in the synthesis, maturation, transport, and degradation of secretory pathway proteins. It follows that increased expression of such factors would be necessary to equip an enlarged ER compartment for proper function. We further speculate that one or more ER proteins might play an integral role in the process of ER biogenesis, perhaps as a scaffold for membrane assem-bly or as a regulator of ER structure. Such functions might be performed by the Rrbp1-encoded p180 protein that was originally characterized as a ribosome receptor and shown to elicit proliferation of rough ER upon enforced expression in yeast (52,53). Notably, Rrbp1 transcripts are highly expressed in secretory tissues (54) and up-regulated in our XBP-1(S)-transduced cells ( Table 1).
Elucidation of the mechanisms that orchestrate organelle biogenesis remains fundamental to the understanding of basic cell biology. Remarkably, enforced expression of a single transcription factor, XBP-1(S), directs cells to construct more ER. We propose that XBP-1(S) mediates downstream events that intersect with regulation of both ER gene expression and lipid biosynthesis, both of which are essential for optimal ER biogenesis (Fig. 8). This model is entirely consistent with the profound defects in ER development observed in XBP-1-deficient-specialized secretory cell types (13). Thus, although many mechanistic details remain unclear, we suggest that a basic blueprint for XBP-1(S)-regulated ER biogenesis has now emerged.