Supporting Information

Green et al. 10.1073/pnas.1019537108 SI Materials and Methods Sample Processing and RNA Isolation. Sorted B-cell populations. Total Detailed Sorting Protocols. Normal mouse B cells were isolated RNA was isolated from −80 °C frozen samples in Tripure re- from 8- to 12-wk-old mice housed in a pathogen-free environment agent (Roche), using the manufacturer’s protocol for single at animal resources at Children’s Hospital Boston, for a mini- cell suspensions. Total RNA was amplified to cRNA using the fi mum of 2 wk. For purification of B-cell subpopulations, single- Ovation RNA Ampli cation system V2 (NuGen Technologies), ’ cell suspensions were obtained from each lymphoid organ by conducted according to the manufacturer s protocol. Total physical dissociation using a cell strainer, stained with antibodies cRNA was labeled using the Affymetrix Gene Chip Labeling Kit to surface markers, and purified by FACS using either a MoFlow (Affymetrix), hybridized overnight on Mouse 430A 2.0 micro- arrays (Affymetrix), washed, and imaged as per the manufac- (Dako Cytomation) or a FACSAria (BD Biosciences). All cells ’ were kept on ice during the pre- and postsorting procedures, and turer s protocol. Murine lymphoma samples. Each murine lymphoma model sample dead cells were excluded using a live lymphocyte gate. Detailed was obtained directly from the laboratory of each investigator descriptions of protocols used for each subset follow. generating the models (Lig4/p53, F.W.A.; Bcl6-Tg and Bcl6/Myc- Pro-B and Pre-B cells. For each Pro-B cell or Pre-B-cell sample an- Tg, R.D.F.), and tissues were flash frozen in liquid nitrogen or dry alyzed, two femurs each from five mice were pooled, ground with ice when removed from mice. Representative sections were de- a mortar and pestle in PBS with 2% BSA, and passed through fined minimally as >80% effacement with lymphoma. Frozen a cell strainer. Whole marrow was stained with anti-B220, anti- fi lymphoma samples were homogenized using a mini-BeadBeater IgM, and anti-CD43 antibodies. Pro-B cells were de ned as B220- (Cole Parmer) in Tripure reagent (Roche). Total RNA was positive/IgM-negative/CD43–high-expressing cells, and Pre-B cells fi – isolated, ampli ed, and labeled prior to analysis using Affymetrix were B220-positive/IgM-negative/CD43 low-expressing cells (Fig. Mouse 430A 2.0 microarrays as described above. S5A). In all, 5 × 10e4 Pro-B cells and 25 × 10e4 Pre-B cells were collected, spun to concentrate, and immediately resuspended in T-Cell Signature Genes. The T-cell signature was derived using TriPure reagent (Roche). transcriptional nearest neighbors of probes for the T-cell markers Transitional, follicular, and marginal zone B cells. One whole spleen was Cd4 (1419696_at), Cd8a (1451673_at), and Cd8b1 (1426170_a_at). processed for each transitional cell sample obtained; spleens were Highly stringent criteria were used for selection of nearest neigh- harvested, placed on ice, physically dissociated using a cell strainer, bors: Pearson correlation coefficient <0.75 and an FDR < 0.05. The and subjected to red cell lysis (Ack lysing buffer, Invitrogen). union of the nearest neighbors consisted of 132 probes, including Whole-spleen single-cell suspensions were stained with anti-B220, the subjects of analysis, and corresponding to 64 discrete genes, anti-IgM, and anti-CD93, and transitional cells were defined as as follows; 1110013L07Rik, 2610019F03Rik, A130082M07Rik, B220 positive, IgM positive, and CD93 positive (Fig. S5B). One A430107P09Rik, Abcc3, Axl BC021614, Bcl11b, Cd247, Cd27, whole spleen was used to collect each sample of both follicular Cd28, Cd3d, Cd3e, Cd3g, Cd4, Cd6, Cd8a, Cd8b1, Cd96, Ckb, and marginal zone B cells. Spleens were processed before surface Cmbl, Crtam, Cxcr6, Edg8, Eg436235, F2rl1, Folr4, Gata3, Gbp6, staining as above for transitional B cells. Follicular B cells were Hfe, Hmox1, Hpgd, Icos, Id2, Igfbp4, Itgb7, Itk, Klk8, Lat, Leprotl1, defined as CD-19 positive, CD21 intermediate, and CD23 high Loc665446, Loc665500, Loc665506, Lrig1, Mboat1, Ms4a4b, Nsg2, cells, and marginal zone B cells as CD-19 positive, CD21 high, Pdlim4, Prkch, Prkcq, Qprt, Sh2d1a, Sidt1, Tcf7, Tcrb-J, Tcrb-V13, and CD23 low (Fig. S5C). A minimum of 1 × 10e5 cells were Tcrb-V8.2, Tgm2, Tgtp, Thy1, Tnfrst18, Trat1, Txk, and Zap70. acquired per each sample. Germinal center and plasmablast B cells. Mice (10 wk old) were ChIP-qPCR Protocols. GC B cells from 10 immunized mice were immunized with 100 μg NP-CGG (Biosearch Technologies) sus- obtained using the methods described above, with the following modification: Total splenic cells were magnetically sorted using the pended in alum, injected intraperitoneally. Spleens were har- fi vested 10 d after immunization and placed on ice, physically CD43 B-cell puri cation kit (Miltenyi Biotec) after red cell lysis, with the purified B cells subsequently stained to obtain B220- dissociated using a cell strainer, and subjected to red cell lysis. positive, PNA-positive B cells. Normal resting B cells were Three immunized mice were pooled for each GC B cells pop- obtained from three 10-wk-old 129/SvEv/Tac WT spleens that ulation, stained with anti-B220 and PNA, and selected based on fi fi – D were puri ed for total B cells using the CD43 B-cell puri cation B220-positive/PNA high-expressing cells (Fig. S5 ). Plasma- kit. FACS analysis validated that >98% of B cells were IgM blasts were isolated using two spleens for each sample and pro- surface positive and PNA surface negative. Sorted cells were cessed as for GC cells except that surface staining was performed washed in PBS on ice, and cross-linked using 1% formaldehyde, to select B220-positive, CD138-positive, CD93-negative cells (Fig. incubated at room temperature for 10 min. The cells were further E × S5 ). A minimum of 1 10e5 cells were acquired per each sample. processed using the ChIP assay kit recommended protocol (Up- Plasma cells. Plasma cells were isolated from mice that were im- state Biotechnologies) with the following modification: 125 nM μ munized with 100 g NP-CGG in alum, with a second immuni- glycine was added to reverse cross-linking for 5 min, whereupon zation performed after 21 d, and two femurs from each mouse cells were washed with PBS and resuspended in ChIP Assay Kit were harvested on day 35. Five mice were pooled for each plasma lysis buffer. Cells were sonicated to shear DNA (Misonix Soni- cell sample (10 femurs). Femurs were ground with a mortar and cators) to obtain chromatin in an average length between 400 bp pestle in PBS with 2% BSA and passed through a cell strainer. and 1,500 bp as verified by agarose gel electrophoresis on replicate Plasma cells were isolated using the CD138 Positive Plasma Cell samples of lymphocytes. Sonicated cell supernatants were diluted Isolation Kit (Miltenyi Biotec) as per the manufacturer’s protocol. 10-fold in ChIP Dilution Buffer with an aliquot of the supernatant To obtain maximal purity, bead-sorted cells were subjected to reserved for input control. A 5-μg quantity of anti-YY1 (rabbit subsequent sorting on a FACSAria to obtain a B220-low, CD93- polyclonal antibody, Abcam) or rabbit antimouse IgG (Pierce) negative, CD138-high B cells population that was >95% pure antibodies was added to the chromatin solution, and incubated (Fig. S5F). A total of 5 × 10e4 cells were obtained for each sample. overnight at 4 °C; next it was subjected to protein A agarose

Green et al. www.pnas.org/cgi/content/short/1019537108 1of6 separation and then concentrated by centrifugation, washed, and used to assess total genomic DNA and to normalize amounts be- detached from the beads. Eluted DNA was purified by phenol/ tween samples. Quantitative PCR primer sets were obtained from chloroform extraction and ethanol precipitation. DNA was sub- Applied Biosystems and used for: Abce1, Eef1B2, Gtf2A1, Syncrip, jected to real-time PCR using input DNA diluted in three in- Eif4G2 (Nat1), Mat2A, SrsF1, Btf3, and Aicda. The ΔΔCq creasing 10-fold dilutions to immunoprecipitated DNA. Primers to method was used to generate relative expression values. One mouse β-globin (Fwd-gacgagtcacagaggacttcctgg Rev- ctgtgaggag- sample each of resting splenic B cells and GC B cells was ana- cagagagggcaaag Fam-Probe-atggattgcagagccagagggagccat) were lyzed, in triplicate.

Fig. S1. Expression of known B-cell differentiation markers. (A) Heat map of 62 known markers of B-cell differentiation, clustered by correlation (rows), across each individual sample (columns) corresponding to the annotated B-cell differentiation categories. High expression is indicated in red and low expression in blue, and genes with multiple probe sets are represented by the mean of all probes (*). (B) Box plots of representative known B-cell differentiation markers.

Green et al. www.pnas.org/cgi/content/short/1019537108 2of6 Fig. S2. Enrichment of NFκB gene sets. (A) Enrichment of NFκB target gene sets, each represented by a single point, in up-regulated (red) and down-regulated (blue) signatures of each B-cell differentiation category. NFκB target gene sets were significantly enriched within the up-regulated gene signature in Tran- sitional B cells, and in the down-regulated gene signatures in Pro-B and Pre-B cells. (B) Heat map of genes driving enrichment of NFκB gene sets in Pro-B and Pre-B down-regulated signatures and the Transitional B-cell up-regulated signature.

Green et al. www.pnas.org/cgi/content/short/1019537108 3of6 Fig. S3. Coenrichment of Creb and E2f target gene sets. Because of the enrichment of Creb target gene sets in both Transitional and GCB up-regulated signatures, and E2f in both Pro-B and GCB up-regulated signatures, we performed a joint analysis to investigate the extent to which these Creb and E2f target genes are coenriched in these respective subsets. One-versus-all comparative gene expression analysis was performed to highlight shared signatures between Transitional and GCB (A) and between Pro-B and GCB (D). HG-GSEA of these shared signatures revealed significant enrichment of Creb (B) and E2f (E) target genes, respectively. This enrichment was driven by a number of genes with high expression in both Transitional and GCB (C) or Pro-B and GCB (F), respectively. However, it was noted that GCB cells showed markedly higher expression of these shared subsets.

Green et al. www.pnas.org/cgi/content/short/1019537108 4of6 Fig. S4. ChIP-seq validation of Yy1 target genes. Raw ChIP-seq signal data were obtained for the promoter regions, 2 kb upstream and 2 kb downstream of the transcription start site, for each of the Yy1 target genes that was present in the GCB-specific transcriptional signature. This revealed binding peaks for 30 of 33 genes, with only PFN2, SLC45A4, and RAB8B not showing significant peaks.

Green et al. www.pnas.org/cgi/content/short/1019537108 5of6 Fig. S5. Flow cytometry gates of sorted B-cell populations. B cells were sorted as described in SI Materials and Methods and gated in red. Representative FACS plots for each of the populations is shown as follows: Pro-B cells (A, Right), Pre-B cells (A, Left), Transitional B cells (B), Marginal Zone (C, upper gate), Follicular (C, lower gate), GCB (D), Plasmablast (E), and Plasma cell (F).

Other Supporting Information Files

Dataset S1 (XLSX) Dataset S2 (XLSX) Dataset S3 (XLSX) Dataset S4 (XLSX)

Green et al. www.pnas.org/cgi/content/short/1019537108 6of6