The AKT kinase signaling network is rewired by PTEN to control proximal BCR signaling in germinal center B cells

Abstract

B cell antigen receptor (BCR) and CD40 signaling are rewired in germinal center (GC) B cells (GCBCs) to optimize selection for high-affinity B cells. In GCBC, BCR signals are constrained, but the mechanisms are not well understood. Here we describe a GC-specific, AKT-kinase-driven negative feedback loop that attenuates BCR signaling. Mass spectrometry revealed that AKT target activity was altered in GCBCs compared with naive B cells. Retargeting was linked to differential AKT T308 and S473 phosphorylation, in turn controlled by GC-specific upregulation of phosphoinositide-dependent protein kinase PDK1 and the phosphatase PTEN. In GCBCs, AKT preferentially targeted CSK, SHP-1 and HPK1, which are negative regulators of BCR signaling. We found that phosphorylation enhances enzymatic activity of these proteins, creating a negative feedback loop that dampens upstream BCR signaling. AKT inhibition relieved this negative feedback and enhanced activation of BCR-proximal kinase LYN, as well as downstream BCR signaling molecules in GCBCs.

Fig.1 |. Phosphorylation of AKT is altered in GCBCs compared to NBCs.

a, AKT protein expression and phosphorylation was analyzed by immunoblot with freshly purified B1–8i NBCs and GCBCs (n=2); lower graph shows the quantitation of bands, with bars showing means of replicates. The ratios were normalized to those of NBCs, given a value of 1. b, Splenocytes from immunized MEG mice (n=3) were instantly fixed in 1.5% PFA and then analyzed by flow cytometry; lower graph shows fold change of median fluorescence intensity (MFI) and was normalized to NBCs as 1. Graph shows the quantitation as in (b). c, Western blot for p-PDK1(S241) and total PKD1 in purified MEG NBCs and GCBCs (n=2); lower graphs show the quantitation of bands as in (b). The ratios were normalized to those of NBCs, given a value of 1. d, Total splenocytes from immunized MEG mice (n=3) were stimulated with anti-IgM for indicated time points and p-AKT in GCBCs and NBCs was analyzed by flow cytometry. e, Purified MEG NBCs and GCBCs (n=2) were rested for 40 minutes and stimulated with anti-IgM for indicated time points and cell lysates were analyzed by immunoblot. Data represent two independent experiments with cells pooled from two mice per group in each experiment (a,c,e) or three independent experiments with one mouse tested in each experiment (b,d). P values, two-tailed unpaired Student’s t-test (a,b,c). (p-AKT(T308)/Actin: t=6.676, d.f.=2; p-AKT(T308)/total AKT: t=18.67, d.f.=2; total AKT/Actin: t=11.50, df=2; p-AKT(T308) normalized MFI: t=8.758, d.f.=4; p-PDK1/Actin: t=5.830, df=2; total PDK1/Actin: t=4.609, df=2).

Fig. 2 |. PTEN is highly expressed in GCBCs, controlling phosphatidylinositol phosphate generation and restraining AKT S473 phosphorylation.

a, Protein expression was analyzed by immunoblot with freshly purified B1–8i NBCs and GCBCs; lower graph shows the quantitation of immunoblot data with the mean value. The ratio of PTEN/Actin was normalized to NBCs as 1. (n=3 from three independent experiments, each sample contain cells pooled from two mice). P values, two-tailed unpaired Student’s t-test (t=4.161, d.f.=4). b, Purified MEG NBCs and GCBCs were stimulated with anti-IgM. Inositol lipids from cell lysates were measured by ELISA. Symbols are means and error bars are SEM from two independent experiments with cells pooled from five mice per group in each experiment; P values: two-way ANOVA with Sidak’s multiple comparisons test (PtdIns(3,4,5)P3: F= 44.76, d.f.=4; PtdIns(3,4)P2: F=33.68, d.f.=4; PtdIns(4,5)P2: F=11.31, d.f.=4). c, NBCs and GCBCs from MEG mice (n=2 except “no inhibitor” condition at 0, 1, and 5 minutes, where n=3)) were treated with the PTEN inhibitor (SF1670) for 10 minutes and then stimulated with anti-IgM. Inositol lipid species were analyzed by ELISA. Cells were pooled from five mice in each experiment. Symbols are means and error bars are SEM. P values compare DMSO and SF1670 treated samples for NBCs (red) and GCBCs (blue) by two-way ANOVA with Sidak’s multiple comparisons test (PtdIns(3,4,5)P3: F=76.71, d.f.=3; PtdIns(3,4)P2: F=214.4, d.f.=3; PtdIns(4,5)P2: F=88.16, d.f.=3). d, B1–8i splenocytes were treated with DMSO or SF1670 for 30 minutes before BCR stimulation with NP-Ficoll. Cells were fixed and analyzed by flow cytometry 5 min (p-AKT) or 20 min (p-S6) post stimulation. Right: fold change of MFI, normalized to DMSO-alone treated samples. Data are mean ± SEM from five mice from two independent experiments; P values, one-way ANOVA followed by Turkey’s multiple comparisons test (p-AKT NBCs: F=71.93, d.f.=16; p-AKT GCBCs: F=262.2, d.f.=16; p-S6 NBCs: F=64.86, d.f.=16; p-S6 GCBCs: F=98.72, d.f.=16).

Fig. 3 |. AKT targets different pathways in GCBCs compared to NBCs.

a, B1–8i NBCs, GCBCs from different time points (d7, d12 and d20), in vitro CpG stimulated B cells, and in vivo NP-Ficoll activated B cells were sorted by FACS. Immunoblot was performed on cell lysates using the AKT phospho-substrate antibody. One representative from two independent experiments is shown; cells were pooled from 4 to 9 mice in each experiment. b, A Venn diagram was constructed to depict the number of AKT phospho-substrates enriched in NBCs, activated NBCs, GCBCs and activated GCBCs using the criteria described in Methods from two independent experiments with cells pooled from three to five mice in each experiment. c, AKT substrates that were specific to NBCs, activated NBCs, GCBCs or activated GCBCs were compared to the whole mouse proteome using a binomial test for each gene ontology and pathway term in the PANTHER software package to calculate the probability (P-value) that the number of genes observed in each category occurred by chance.

Fig. 4 |. AKT targets proximal BCR signaling regulators in GCBCs.

a, The relative quantitation of proteins of interest immunoprecipitated by bead-conjugated AKT phospho-substrate antibody in Fig. 3 (n=2 from two independent experiments). b, MEG NBCs and GCBCs were pretreated with AKT inhibitor for 10 min and then stimulated with anti-IgM for 5 min or left unstimulated. Cell lysates were immunoprecipitated by AKT phospho-substrate antibody, and then probed by immunoblot for each of the indicated proteins. Data represent one of two independent experiments. c, In vitro kinase reactions were performed for 30 min by incubating recombinant pre-activated AKT with CSK, SHP-1, HPK1 or LYN as substrates. Immunoblot using the AKT phospho-substrate antibody was utilized to monitor phosphorylation of the target proteins. A representative of two independent experiments is shown.

Fig. 5 |. AKT-mediated phosphorylation of CSK, SHP-1 and HPK1 enhances their activity.

a, Mass spectrometry was performed on the in vitro kinase reaction of AKT and CSK to identify the AKT phosphorylation site on CSK (S284). S284 phosphorylation was modeled onto the crystal structure of CSK bound to C-SRC (PDB ID 3D7T). b, unphosphorylated CSK or AKT-phosphorylated CSK was reacted with LYN and immunoblot was utilized to monitor LYN Y507 phosphorylation. Densitometry was utilized to quantitate LYN Y507 phosphorylation (n=3 from three independent experiments). The resulting kinetic traces, shown are means ± SD, were fit with a linear model to determine the relative reaction rate for LYN phosphorylation. P values were derived with a two-tailed Student’s t-test (t=21.55, d.f.=4). c, Mass spectrometry was utilized to map the AKT phosphorylation site on SHP-1 to T394, which is in the catalytic domain. d, A phosphatase assay was performed to measure the activity of unphosphorylated SHP-1 and AKT-phosphorylated SHP-1. The resulting kinetic traces were fit with a linear model to determine the relative reaction rates (n=3 from three independent experiments). Shown are means ± SD, P values were calculated with a two-tailed Student’s t-test (t=4.963, d.f.=4). e, Unphosphorylated HPK1 and AKT-phosphorylated HPK1 were incubated with recombinant BLNK and immunoblot was performed with an antibody against phospho-threonine residue. Densitometry was utilized to quantitate BLNK phosphorylation (n=2 from two independent experiments). Shown are means ± SD. P values, two-way ANOVA with Sidak’s multiple comparisons test (F=15.95, d.f.=5).

Fig. 6 |. PTEN inhibition alters AKT signaling networks downstream of BCR signaling in GCBCs.

Purified MEG NBCs and GCBCs were incubated with PTEN inhibitor (SF1670) or DMSO for 30 min and then stimulated with anti-IgM for 5 min. The proteomic study workflow described in Supplementary Fig. 3 was utilized to identify AKT substrates (n=2 from two independent experiments). a, The PANTHER software package was utilized to identify pathways targeted by AKT in GCBCs. Heat map shows the relative protein abundance by normalizing the peak area for each experimental group to the maximum peak area observed for each protein. b, The peak area from the mass spectrometric assay was calculated for CSK, SHP1, HPK1 and ARP2. Bar graphs show the mean of the results with dots showing results of each independent experiment.

Fig. 7 |. AKT inhibition enhances proximal BCR signaling in GCBCs.

a, Purified NBCs and GCBCs from MEG mice were treated with DMSO or AKT inhibitor for 40 min before anti-IgM stimulation for the times indicated. Cell lysates were analyzed by immunoblot. Left panel, representative immunoblot of three independent experiments for the p-Tyr species indicated, with cells pooled from 3 to 5 mice in each experiment; right panel, quantitation of immunoblots from all three experiments, the ratios of pY396/pY507 of LYN and p-SYK/Actin were normalized to DMSO treated NBCs (time 0) as 1. Data are mean ± SEM; P values are comparing treatments (DMSO vs AKT inhibitor) by two-way ANOVA (two factors: treatment and time). (pY396/pY507 NBCs: F=0.005502, d.f.=20; pY396/pY507 GCBCs F=24.32, d.f.=20; p-SYK/Actin NBCs: F=0.3387, d.f.=20; p-SYK/Actin GCBCs: F=9.807, d.f.=20) b, Splenocytes from immunized MEG mice were treated with DMSO or AKT inhibitor for 40 min followed by anti-IgM stimulation for indicated time points. Cells were then analyzed by flow cytometry (n=4 from two independent experiments). Representative histogram (left panel) and statistical analysis for MFI of all samples (right panel) are shown for GCBCs. Data are mean ± SEM; P values are comparing treatments (DMSO vs AKT inhibitor) by two-way ANOVA (two factors: treatment and time). (p-BTK: F=9.034, d.f.=30; p-PLCγ2: F=13.15, d.f.=30).