Phosphoproteomic profiling of mouse primary HSPCs reveals new regulators of HSPC mobilization

Abstract

Protein phosphorylation is a central mechanism of signal transduction that both positively and negatively regulates protein function. Large-scale studies of the dynamic phosphorylation states of cell signaling systems have been applied extensively in cell lines and whole tissues to reveal critical regulatory networks, and candidate-based evaluations of phosphorylation in rare cell populations have also been informative. However, application of comprehensive profiling technologies to adult stem cell and progenitor populations has been challenging, due in large part to the scarcity of such cells in adult tissues. Here, we combine multicolor flow cytometry with highly efficient 3-dimensional high performance liquid chromatography/mass spectrometry to enable quantitative phosphoproteomic analysis from 200 000 highly purified primary mouse hematopoietic stem and progenitor cells (HSPCs). Using this platform, we identify ARHGAP25 as a novel regulator of HSPC mobilization and demonstrate that ARHGAP25 phosphorylation at serine 363 is an important modulator of its function. Our approach provides a robust platform for large-scale phosphoproteomic analyses performed with limited numbers of rare progenitor cells. Data from our study comprises a new resource for understanding the molecular signaling networks that underlie hematopoietic stem cell mobilization.

Figure 1

Bioanalytical platform for quantitative interrogation of signaling pathways in HSPCs. Two × 10⁵ highly purified primary murine HSPCs were sorted by flow cytometry, lysed, and trypsin-digested, followed by Fe³⁺NTA-IMAC phosphopeptide enrichment and isotope labeling with TMT reagents. Phosphopeptides were quantified by fully automated 3-D RP-SAX-RP chromatography, coupled to a ThermoFisher Orbitrap mass spectrometer. Data analysis and visualization was performed using a combination of multiplierz and R scripts. LC-MS, liquid chromatography–MS.

Figure 2

Phosphoproteomic interrogation identifies novel activated protein pathways in primary murine HSPCs. Resting BM (rest) or mobilized (mob) LSK HSPCs were harvested and sorted as described in “Methods.” (A) Comprehensive phosphoproteomic analysis of 2 × 10⁵ primary murine resting or mobilized HSPCs was done in triplicate, and this experiment was repeated 3 times. In total, this analysis identified 15 230 unique phosphopeptides and 4993 phosphoproteins. The phosphopeptide ratios followed a normal distribution and there was not a marked change in total phosphopeptide quantity with mobilization. Overall, 1018 phosphopeptides differed in relative amount by >2 SD between mobilized and resting cells. Of these, 572 phosphopeptide species were >2 SD more abundant in mobilized HSPCs (red), and 446 were >2 SD less abundant (green). Red trace shows Gaussian overlay. (B) Unsupervised hierarchical clustering analysis of all 18 samples demonstrates that mobilization results in durable phosphoproteomic changes in primary murine HSPCs. Mob1/rest1, mob2/rest2, and mob3/rest3 represent individual experiments; a, b, and c denote biological replicates within experiments. Sample phosphoprofiles consistently clustered with other samples in their biologic subgroup (mobilized or resting) and away from samples in the other subgroup, confirming the existence of a durable phosphoproteomic signature of mobilization. (C) NMF of data from all 3 experiments identifies two phosphoproteomic signatures that can be used to segregate mobilized from resting HSPCs. Using the nonsmooth NMF method of Pascual-Montano et al, 1250 iterations were run with a factorization rank of 2 to identify consensus clusters capable of segregating mobilized from resting phosphoprofiles. Heatmap shows the degree of concordance between sample groups (top bars) and consensus profiles (left bars). (D) Signature phosphoprotein residues were extracted and then filtered by featurescore to identify phosphoprotein residues most specific to each signature. A mixture expression profile heatmap, which summarizes the relative contribution of each signature to each sample was used to assign signatures, or metagroups (left), to the sample classes (top). A complete list of phosphoprotein residues in each metagroup is provided in supplemental Table 3.

Figure 3

ARHGAP25 deficiency in HSPCs leads to defects in mobilization. Arhgap25^−/− mice, obtained from KOMP, were confirmed at the DNA and protein level to be deficient in ARHGAP25 protein (not shown). (A) Mice lacking ARHGAP25 had significantly higher percentages of LSK HSPCs in their BM than controls (left), and lower proportions of HSPCs in their peripheral blood (center), suggesting increased central compartmentalization. This continued to be true after mobilization with Cy/G (right), indicating that ARHGAP25 is required for optimal mobilization. Each dot represents an individual mouse. (B) To confirm that this finding is due to intrinsic activity of ARHGAP25 in hematopoietic cells, Arhgap25^−/− BM was transplanted into lethally irradiated recipient mice. After 16 weeks of hematopoietic reconstitution, recipients of Arhgap25^−/− marrow were found to have increased percentages of LSK cells in the BM (left) and decreased percentages in the peripheral blood (center), demonstrating that ARHGAP25 function is intrinsic to the hematopoietic system in this context. After Cy/G treatment, diminished mobilization was again seen in recipients of Arhgap25^−/− marrow as compared with controls (right). These differences did not reach statistical significance, however, in part because recipients of control marrow reproducibly mobilized far less well than unmanipulated WT mice. Each dot represents an individual mouse. ctrl, control.

Figure 4

ARHGAP25 moderates CXCL12 signaling and is functionally affected by phosphorylation. (A) ARHGAP25 deficiency strengthens HSPC response to a CXCL12 gradient. HSPCs from Arhgap25^−/− or control mice were placed in the top wells of a transwell apparatus with CXCL12 in the bottom well (left), both wells (center), or neither well (right). After 1 hour, transmembrane migration was assessed by flow cytometry as described in “Methods.” Arhgap25^−/− HSPCs showed markedly increased migration across a membrane in response to a CXCL12 gradient (left), but not in the absence of a gradient (center, right). Results shown are representative of 3 independent experiments. (B) Arhgap25^−/− HSPCs were stained with antibodies to CXCL12 receptors to confirm that the augmented response to CXCL12 in Arhgap25^−/− HSPCs was not due to increased cell surface expression of CXCR4 or ROBO4. Shown are representative overlay histograms of CXCR4 (left) or ROBO4 (right) expression on Arhgap25^−/− LSKs (black traces) as compared with control LSKs (gray traces). MFIs and SDs are shown, as well as P value. N = 4 for each condition in this experiment, which was performed in triplicate. (C) Phosphorylation of GST-fused full-length (GST-WT) ARHGAP25 and its truncated fragments. Phosphorylation was performed using radiolabeled ATP and neutrophil cytosol as a kinase source, as described in “Methods.” Marked phosphorylation was observed in the full-length protein as well as in the PH and ID regions, whereas phosphorylation of GAP and CC domains was undetectable. (D) Mutation of S363 affects the ability of phosphorylated ARHGAP25 to inactivate Rac. GST-WT ARHGAP25 and GST-mutant (S363A) ARHGAP25 protein were phosphorylated with neutrophil cytosol and nonradiolabeled ATP. GTPase activation effect was measured 5 minutes after co-incubation with GST-Rac by nitrocellulose filter binding assay, as described in “Methods.” ctrl, control; MFI, mean fluorescence intensity.

Figure 5

Model of ARHGAP25 function. ARHGAP25 opposes CXCL12-CXCR4 signaling, promoting mobilization by converting GTP(active)-Rac to GDP(inactive)-Rac. Phosphorylation of ARHGAP25 on S363 inhibits its ability to inactivate Rac. GDP, guanosine diphosphate; GEF, guanine nucleotide exchange factor; PI3K, phosphatidylinositol 3-kinase; SFTK, Src family tyrosine kinase.