Supplemental Materials and Methods


Gene-targeted knockout mice and LPS culture of SPL B cells.

     Cells were harvested from the spleens of mice lacking Blimp-1(prdm1flox/floxCD19Cre/+ and prdm1flox/-CD19Cre/+) and littermate control mice (Shapiro-Shelef et al., 2003).  After counting, cells were incubated in purification buffer (PBS, 1% BSA, 2mM EDTA, and 2% FCS) with a-mouse B220 conjugated to either APC or FITC (BD-Pharmingen) on ice for 20 minutes at a density of 2x108 cells/ml.  After washing, cells were incubated with goat a-rat IgG microbeads (Miltenyi Biotec) on ice for 20 min according to manufacturer instructions.  After washing, B220+ cells were purified by positive selection using magnetic separation columns (Miltenyi Biotec).  After purification, B220+ cells were plated at a density of 106 cells/ml and treated with LPS at empirically determined concentrations to induce maximal differentiation with minimal cell death (2-10ug/ml depending on lot, Sigma). On days 2 and 4 after LPS treatment, cells were harvested and resuspended in Trizol reagent (Invitrogen) for RNA purification. 

     For analysis of XBP1 in plasma cell differentiation, wildtype and xbp1-/-, rag2-/- chimeric mice were generated as described (Reimhold et al 2001).  Mice were maintained in pathogen-free facilities in accordance with the guidelines of the Committee on Animals of Harvard Medical School. Purified B cells from xbp1-/-, rag2-/- and wild-type chimeric mice were isolated from spleen and lymph nodes by magnetic CD43 depletion or CD19 magnetic bead selection, (Miltenyi Biotech), were plated at 1 x 106 cells/ml in RPMI 1640 supplemented with 10% fetal calf serum (FCS, Hyclone), glutamine (2mM), penicillin (50units/ml), streptomycin (50mg/ml), Hepes (100mM), nonessential amino acids (1X) sodium pyruvate (1mM) and b-ME (50mM) and stimulated with optimal concentrations of LPS (20 mg/ml, Sigma).  Cells treated for 2 and 4 days were harvested for RNA isolation and gene expression analysis. 

     In experiments using both genetically-deficient mice, the efficacy of the differentiation system was validated by staining for the plasma cell marker, syndecan-1, and measurement of immunoglobulin secretion (M.S.S. and N.I., unpublished data).


Cell Culture

     The following cells lines were grown in media (Invitrogen) with L-glutamine, penicillin/streptomycin and 10% fetal bovine serum (Hyclone): RAJI (human EBV+ GC B cell), WEHI231 (mouse mature B cell), and H929 (human multiple myeloma) in RPMI1640, and WS1 (human primary fibroblast, passages 8 through 11) and 293 (human kidney) in IMDM.


Cloning of XBP1 forms into retroviruses. 

     A  cDNA clone containing the full length unprocessed human XBP1 (ResGen, Invitrogen, Genbank Acc ID BC012841) was cut  with NcoI/XhoI and this fragment of XBP-1 was ligated to BS-Acmaf (Hurt et al., 2004) cut with the same enzymes. This construct was linearized with NcoI, to serve as the vector backbone. A fragment spanning the ATG of XBP-1 to the internal NcoI site of processed XBP-1 was generated by PCR with primers: 5’CAT GCC ATG GTG GTG GTG GCA GCC GC, and 5’CATGCCATGGGGAGATGTTCTGGAGGGGTGACAACTGGGCCTGCACCTGCTGCGGACTCAGCAGACCCGGCCACTG. This fragment was sequence verified before cloning into the vector backbone by NcoI ligation. This generated the full-length processed XBP1, XBP1-s. An Sfi/XhoI fragment of processed XBP1 is replaced with the Sfi/XhoI fragment isolated from the original cDNA clone to generate the full length unprocessed XBP1 which was then mutangenized by the Qiaquick mutagenesis method with the primer set:



to generate the splice-mutant form of XBP1, XBP1-u.  Inserts were then isolated by Sal/BglII, blunt cloned into the EcoRI site of Vxy-puro (Shaffer et al., 2002; Shaffer et al., 2000), creating the VXY-puro XBP1-s and VXY-puro XBP1-u vectors.


shRNA loop targeting XBP1




Mouse Lymphochip clone selection and analysis.

     Clones for this array were obtained from the I.M.A.G.E. consortium based on the following criteria:  Using BLAST and Homologene tools, mouse homologs of genes found on the human lymphochip were identified.  In addition, clones that were highly represented in mouse lymphoid cDNA libraries were also included. In as many cases as possible, more than one clone per cDNA was picked.  In total, the murine lymphochip contains over 12,000 cDNAs enriched for genes expressed in normal and malignant lymphocytes.  A listing of these elements is available in the Supplemental Table “Mouse Array Elements”.

     To be able to compare data across the time courses in Figure 1, the data for each time course were ‘time-zero transformed’.   In short, for a given time course, the value for an element at time zero (uninduced, no LPS) was subtracted from the value for that element at day 2 and also at day 4, giving the fold difference between the starting samples and each time point.  This reduces the variation among individual experiments, and allows data from all experiments to be analyzed together for determination of the genes of interest. 

     Our analysis focused solely on genes whose induction was impaired by loss of prdm1 or xbp1, as compared to WT cells.  We chose a 1.8fold difference cutoff, which translates to an approximate difference of 5fold when the number of differentiating cells in the culture is taken into account (30%).  Empirically, this provides the most consistent and reliable data on our platforms.  Potential target genes then had to meet the following conservative criteria for inclusion:

1-     Pass quality control standards for spot size, quality, and signal-to-noise ratio

2-     Pass sequence re-verification

3-     Behave consistently across the majority of array time course experiments (2 of 3 for prdm1 vs. WT, 2 of 2 for xbp1 vs. WT)

4-     Be more highly expressed in plasma cells than in B cells (see supplemental figures).


Mouse Lymphochip RNA reference pool

      The reference pool is a mixture of total RNA from 20 cell lines (many a kind gift of Dr. H. Morse):  6312 (proB), 18-81 (preB), 220-8 (preB), PD31 (preB), ED2 (preB), 70Z/3 (immature B), BCL1 (mature B), M1.2.4.1 (mature B), S107 (plasmacytoma), APBC4 (plasmacytoma),  2017 (preT), 3DO (mature T), B6206-3 (follicular lymphoma), NFS201 (centroblastic lymphoma), NFS204 (marginal zone lymphoma), NFS205 (diffuse large B cell lymphoma), NFS210 (T cell lymphoma), BSJ003-2 (Burkitt-like lymphoma), MAPS-1 (T cell lymphoma), and JZ74A1 (macrophage).  This pool provides a common denominator against which all samples may be compared, ultimately allowing the comparison of any experimental sample to every array in the dataset.


Mouse plasma cells vs. B cells

     Mature B cell samples included: five purified (CD19+ or B220+) splenic B cells samples, one mature B cell line (WEHI231), and 6 mature B cell c-myc-induced lymphomas (donated by S. Janz).  Plasma cell samples included: two plasmacytoma cell lines (B9, S107), four cmyc-induced plasmacytomas (donated by S.Janz), and six pristine-induced plasmacytomas (donated by M.Potter). See SupFig1. data.


Human Lymphochip elements

     A list of the elements on the Human Lymphochip can be found in the supplemental table “Human Array Elements”.


Human plasma cells vs. B cells

     Mature B cell populations and plasma cells were obtained from donated tonsils and bone marrow and were magnetically separated (MACS, Miltenyi Biotech) into CD19+ and CD19- fractions before FACS.  Mature B cells included naïve and memory B cells from tonsil and mature B cells from bone marrow.  Plasma cells were obtained from bone marrow aspirates.  Antibodies used in FACS were from Pharmingen and Caltag.

Mature B- CD19+, sorted for CD10-, IgM+, IgD+

Plasma cells- CD19-, sorted for CD138 hi, CD38+, CD20-

Naïve B- negatively selected with antibodies to CD11b, CD14, CD16, CD3, CD4, CD38, IgG, IgA, and glycophorin A, followed by FACS for CD20+, CD27- cells

Memory B- negatively selected with antibodies to CD11b, CD14, CD16, CD3, CD4, CD38, IgD, and glycophorin A, followed by FACS for CD20+, CD27+ cells.

Total RNA was isolated from between 60,000 and 20 million primary cells using the TRIzol reagent (Invitrogen).  The total RNA was amplified in duplicate using the standard T7 MessageAMPTM (Ambion) two-round amplification protocol to obtain between 25 and 50 micrograms of cRNA. For samples containing more than one million cells, only 1 mg of total RNA was used to initiate amplification. Six micrograms of amplified cRNA was labeled and hybridized to human lymphochips along with a standard pool of RNA as reference. The average gene expression was calculated for each population (PC vs mature B cells (naïve+memory+mature B).  See Figure 3B and Sup Fig2/3. data.


Assignment of gene function.

     Genes were organized into functional categories based on analysis of the OMIM ( and GO databases (


Flow cytometry and microscopy. 

     For flow cytometry using antibodies, cells were pelleted, medium was aspirated, and antibodies (Pharmingen) were added to pellets at empirically determined concentrations. Cells were incubated on ice for 15 mins., washed with 1xPBS+5%FBS, and analyzed on a FACSort (Becton Dickinson) using CellQuest software.  Staining with the ER marker BFA-BODIPY was performed by incubating live cells with the dye at 0.2ug/ml for 1hr. at 37 degrees C.  Cells were then pelleted, washed, and analyzed by flow cytometry.  Staining of cells with the ER/mitochondrial marker DiOC6 (Molecular Probes) was performed by incubating cells with the dye (0.5ug/ml) at 37 degrees C for 1 hr.  Cells were then cytospun onto glass slides after 15 mins. of fixation in paraformaldehyde with DAPI counterstaining.  Cells for microscopy were cytospun, fixed, and stained (DIFFQUIK, Fisher) as directed by the manufacturer. 

     Transient transfection of cells with ER-tagging vectors was performed on cells spun onto the bottom of 6 well plates using lipofectamine (Invitrogen) and 5ug of the vector.  A vector expressing gfp fused with a signal sequence and a KDEL ER-retention sequence (lys-gfp-KDEL) was obtained from the laboratory of Drs. E. Snapp and J. Lippincott-Schwartz. Lys(ozyme)-GFP-KDEL was constructed by fusing the complete coding sequence of hen's egg lysozyme with a COOH-terminal enhanced GFP (BD Biosciences Clontech) followed by a KDEL ER retention signal and stop signal (aag gat gaa ctg tag).   This was cloned into the BglII and BamH1 sites of the pSX vector. The HEL sequence was PCRed with BglII and BamH1 at the 5' and 3' ends, respectively.  EGFP was PCRed with a 5'BamH1 and 3'Xba1 after the stop codon. The two constructs were fused at the BamH1 site and then inserted into the pSX vector. 

     Staining with Mitotracker Red and Lysotracker Red (Molecular Probes) was performed by incubating cells with 50ng/ml of each dye at 37C for 45 mins.  Cells were then washed with 1xPBS and either analyzed by flow cytometry or cytospun onto slides for microscopy.


Determination of protein synthesis, degradation, and ribosome content Protein concentration

      The amount of total protein per cell was measured by pelleting equal numbers of cells in triplicate, washing 3 times with 1xPBS, resuspending in 100ul of 1xPBS, followed by sonication to lyse cells.  Samples were then centrifuged to remove debris, and the protein content of cellular extracts was determined by absorbance (595nm) using the Bradford Assay (BioRad).


Pulse Labeling

     Cells were radiolabelled at 37°C with 250 mCi [35S]L-Methionine or [3H]-Leucine (Amersham Pharmacia Biotech, Sweden). Labeling was terminated by adding ice-cold media containing 1 mg/ml of cold L-methionine or L-leucine. After washing twice, cells were chased at 37°C in the same media. Cell pellets were lysed in the extraction buffer [50 mM Tris-HCl (pH7.4), 150 mM NaCl, 1 mM EDTA and 1% Triton X-100] containing proteinase inhibitor cocktail (Boehringer Mannheim, Indianapolis, IN). Whole cell lysates were prepared by adding 10U DNase (Boehringer Mannheim, Indianapolis, IN) and 2% SDS, followed by incubation at 95°C for 5 min. SDS-PAGE confirmed that acid insoluble radioactivity accurately reflected incorporation into proteins and not other acid insoluble forms (S.Q., J.W.Y., data not shown).    For depletion of glyoproteins, Triton X-100 soluble lysates were incubated with agarose bound ConA (Vector Laboratories, Burlingame, CA) overnight at 4°C.  Superupernatants were collected after microfuging beads.  To quantitate radioactivity, 5 ml of samples were spotted onto  a filtermat (Wallac, Turku, Finland), and precipitated with 10% TCA.  TCA insoluble radioactivity was measured by Microbeta counter (Wallac, Turku, Finland). These results were also confirmed by independent transductions of RAJI cells and analysis of protein synthesis using [35S]-Met (A.L.S., S.Q., J.W.Y., data not shown).


Ribosome Content

    Raji cells with control vector, or expressing XBP1-s or XBP1-u were treated for treated for 20 mins. With 100mg/ml of cycloheximide (CHX) at 37C.  Cells were counted and equal cell numbers were transferred to triplicate tubes for washing, 3 times in 1xPBS with 100ug/ml CHX.  Cells were pellets were frozen in a dry ice/methanol bath.  Frozen cells were resuspended in 1ml of lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 10 mM MgCl2, 0.5% IGEPAL-630, 2 mM DTT, 100ug/ml CHX, 200U/ml RNasin (Promega), incubated on ice for 10 minutes, centrifuged at 10,000 x g for 10 minutes to clear the extract and 950ul of the supernatant was layered on a 10.5 ml continuous sucrose gradient (12%-45%)containing 50mM Tris-HCl (pH7.4), 50mM KCl, 10 mM MgCl2, 3mM DTT, 15U/ml RNasin.  Sucrose gradients were subjected to centrifugation at 38, 000 rpm in a SW40Ti rotor (Beckman) for 2 hours.  Gradients were fractionated from the bottom and the absorbance (260 nm) was monitored to detect ribosome fractions (Taha et al., 1999).


Array Data


Table 1.          Mouse Array Elements (data file)

Figure 1A.      mouse Blimp-1 target genes. (data file)

Figure 1B.       mouse Xbp-1 target genes. (data file)

Figure 2.         Target gene dependence on Blimp-1 and Xbp1. (data file)

Table 2.          Human Array Elements (data file)

Figure 3A.      Raji XBP1-s targets. (data file)

Figure 3B.       293 XBP1-s targets. (data file)

Figure 3C.      WEHI231 XBP1-s targets. (data file)


Supplemental Figures


Sfig1.  Blimp1 and XBP1 target expression in mouse B cells and PCs (Figure and data file)

Legend Sfig1.

Blimp-1 and XBP1 target gene expression was assessed by averaging the gene expression data for a series of primary B cell samples and B cell lines and comparing that to averaged gene expression across a set of primary plasmacytomas and plasma cell lines (PC).  All targets were expressed more highly on average in PCs as compared to B cells.


Sfig2.  XBP1-s target gene expression (from Raji and H929 experiments) in human B cells and PCs. (Figure and data file)

Legend Sfig2.

XBP1-s target gene expression (from Figure3A and Supplemental Fig. 3) was assessed by averaging the gene expression data for a series of primary B cell and comparing that to averaged gene expression across a set of primary plasma cell samples (PC).  All targets were expressed more highly on average in PCs as compared to B cells.




Sfig3.  Confirmation of XBP1-s target genes (Figure and data file)

Legend Sfig3. 

A. Quantitative RTPCR for XBP1-s target genes.  RNA from RAJI transduced with control or XBP1-s-expressing retroviruses was used for quantitative RTPCR (Taqman, ABI) with commercially available probe/primers sets.  The fold change in expression mediated by XBP1-s compared to control is plotted.  Data are representative of at least two independent transductions, tested in triplicate. 

B.  RNA interference demonstrates the dependence of gene expression on XBP1. The human multiple myeloma cell line H929 was transduced with control retrovirus (pRETROsuper) or the same retrovirus expressing an shRNA targeting human XBP1. Cells were harvested after selection and RNA was prepared for gene expression analysis (control Cy5, XBP1-shRNA Cy3). Array elements whose expression was decreased at least 1.5fold relative to control cells are shown.  Also shown is average expression difference between primary human plasma cells (PC) vs. mature B cells.