A respiratory chain controlled signal transduction cascade in the mitochondrial intermembrane space mediates hydrogen peroxide signaling

Significance

Both the mitochondrial respiratory chain and reactive oxygen species (ROS) control numerous physiological and pathological cellular responses. ROS such as hydrogen peroxide (H₂O₂) are thought to initiate signaling by broadly and nonspecifically redox-modifying signaling molecules, suggesting that H₂O₂ signaling may be distinct from other signal transduction pathways. Here, we provide evidence suggesting that H₂O₂ signaling is under control of what appears to be a typical signal transduction cascade that connects the respiratory chain to the mitochondrial intermembrane space-localized conserved Syk pathway and results in a focused signaling response in diverse cell types. The results thus reveal a mechanism that allows the respiratory chain to communicate with the remainder of the cell in response to ROS.

Abstract

Reactive oxygen species (ROS) such as hydrogen peroxide (H₂O₂) govern cellular homeostasis by inducing signaling. H₂O₂ modulates the activity of phosphatases and many other signaling molecules through oxidation of critical cysteine residues, which led to the notion that initiation of ROS signaling is broad and nonspecific, and thus fundamentally distinct from other signaling pathways. Here, we report that H₂O₂ signaling bears hallmarks of a regular signal transduction cascade. It is controlled by hierarchical signaling events resulting in a focused response as the results place the mitochondrial respiratory chain upstream of tyrosine-protein kinase Lyn, Lyn upstream of tyrosine-protein kinase SYK (Syk), and Syk upstream of numerous targets involved in signaling, transcription, translation, metabolism, and cell cycle regulation. The active mediators of H₂O₂ signaling colocalize as H₂O₂ induces mitochondria-associated Lyn and Syk phosphorylation, and a pool of Lyn and Syk reside in the mitochondrial intermembrane space. Finally, the same intermediaries control the signaling response in tissues and species responsive to H₂O₂ as the respiratory chain, Lyn, and Syk were similarly required for H₂O₂ signaling in mouse B cells, fibroblasts, and chicken DT40 B cells. Consistent with a broad role, the Syk pathway is coexpressed across tissues, is of early metazoan origin, and displays evidence of evolutionary constraint in the human. These results suggest that H₂O₂ signaling is under control of a signal transduction pathway that links the respiratory chain to the mitochondrial intermembrane space-localized, ubiquitous, and ancient Syk pathway in hematopoietic and nonhematopoietic cells.

The accumulation of oxygen on earth not only enabled aerobic respiration but also forced life to adapt to its toxic effects. Molecular oxygen (O₂) inevitably forms reactive oxygen species (ROS) such as superoxide (O₂⁻) and the more stable hydrogen peroxide (H₂O₂) due to its proclivity to react with univalent electron donors, such as flavin enzymes of the respiratory chain or NADPH oxidase that releases H₂O₂ into the extracellular space. Relevant other sources of extracellular and intracellular H₂O₂ include environmental exposure to ozone, UV light, and ionizing radiation (1, 2). Among the most conserved defense mechanisms against the oxidizing effects of H₂O₂ are detoxifying enzymes, such as catalase, as well as aquaporins that control the influx of H₂O₂ into the cell (3–7). These mechanisms are highly efficient and adaptable, and provide an explanation for the great variation in susceptibility to H₂O₂ among cell types despite the generally low intracellular concentrations across bacteria, plants, and mammals (3, 8, 9). Indeed, it requires up to 10 mM exogenous H₂O₂ to induce a measurable signaling response and more than 30 mM H₂O₂ to induce necrotic blebbing in some mammalian cells, while plant tissues can contain up to 100 mM H₂O₂ (10–13).

Life further evolved to communicate and convert the presence of H₂O₂ into diverse cellular responses appropriate to the amount of intracellular and extracellular H₂O₂ the organism is faced with (9, 14, 15). In the metazoan lineage, it appears that such mechanisms were usurped to amplify receptor-mediated signaling: Engagement of a large number of plasma membrane resident receptors, as well as extracellular H₂O₂, results in increased intracellular ROS production by the respiratory chain, and much of such induced cellular responses, including inflammasome signaling, NF-κB activation, and B-cell receptor (BCR)–induced proliferation, is ROS-dependent (9, 16–22). It is thus clear that extracellular and intracellular H₂O₂ is an ancient and essential mediator of cellular homeostasis linked to a wide range of physiological and pathological responses and numerous diseases (23, 24). However, the underlying mechanisms and their evolutionary purpose remain largely elusive.

A critical question is how the cell translates an encounter with H₂O₂ into a distinct cellular response. Early work demonstrated that H₂O₂ reversibly oxidizes deprotonated cysteine residues, and thereby inactivates protein tyrosine phosphatases. Thus inactivated phosphatases were hypothesized to shift the equilibrium of inactive to active kinases resulting in enhanced kinase activity (18, 25, 26). By now, a large number of redox modifications in phosphatases, kinases, adapters, receptors, and transcription factors have been proposed to modulate signaling, among them multiple components of the BCR signaling pathway such as tyrosine-protein kinase Lyn, tyrosine-protein kinase SYK (Syk), tyrosine-protein phosphatase nonreceptor types 6 and 11 (SHP1/PTPN6; SHP2/PTPN11), phosphatase and tensin homolog, and a mitogen activated protein (MAP) kinase serine/threonine phosphatase (27–36). However, it has been challenging to prove or disprove the physiological relevance of such a mechanism, and results have been difficult to reconcile as a whole (3, 26, 35). Nevertheless, these findings have resulted in the view that ROS signaling fundamentally differs from classical signal transduction because it broadly targets signaling molecules and thus assumes a relative independence of each redox-modified component in promoting signaling.

Although not excluding a contribution of cysteine modifications to ROS signaling, it is conceivable that H₂O₂ signaling, in fact, resembles other signaling pathways characterized by a cascade of phosphorylation events that originate locally and control a distinct signaling response in all tissues and species responsive to the ligand. Compatible with the existence of a few upstream mediators of H₂O₂ signaling, limited evidence indeed exists supporting that kinases, such as MAP kinases and Syk, are critical intermediaries of H₂O₂-induced signal transduction in yeast, plants, and chicken cells, respectively (13, 15, 37). However, the concept remains currently unexplored, in part, because it implies the existence of a “ROS receptor” that initiates signaling, a notion that has been dismissed previously (13, 18). Here, we provide evidence in support of such a model, demonstrating that H₂O₂ signaling is under control of the respiratory chain and the mitochondrial intermembrane space-localized Syk pathway leading to a circumscribed signaling response in B lymphocytes and fibroblasts.

Results

Syk Is Critical for H₂O₂-Induced Activation of Bkt, PLCγ2, JNK, and Akt.

Receptor-activated signal transduction cascades typically use the same intermediaries to induce a cellular response in all tissues responsive to the ligand and are often conserved in different species (38). Because H₂O₂ induces signaling across many tissues and species, we speculated that Syk is a critical mediator of H₂O₂ signaling in both hematopoietic and nonhematopoietic cells, as well as in different vertebrate species. To test this idea, we H₂O₂-stimulated ex vivo harvested mouse splenic B cells and freshly derived mouse embryonic fibroblasts (MEFs) briefly pretreated with the Syk inhibitor R406 (39, 40), as well as Syk-deficient DT40 B cells, which tolerate genetic Syk deficiency well, unlike primary B cells, and are derived from the genetically stable chicken DT40 B-cell line (41–44).

Both Syk inhibition with R406 in B cells and MEFs, as well as genetic Syk deficiency in DT40 B cells, resulted in strongly decreased H₂O₂, as well as anti-IgM–induced phosphorylation of tyrosine-protein kinase BTK (Btk), phospholipase Cγ2 (PLCγ2), and c-Jun N-terminal kinase (JNK) and some reduction in extracellular-signal regulated kinase (ERK) and p38 mitogen-activated protein kinase (p38) phosphorylation (Fig. 1 A and B). Consistent with decreased PLCγ2 phosphorylation, H₂O₂-induced calcium flux was greatly reduced in Syk-deficient DT40 B cells (Fig. 1C). However, H₂O₂ resulted in no or minor loss of phosphorylation at inhibitory Lyn Tyr507 and intermediate reduction of phosphorylation at activating Lyn Tyr396 and of SHP1 in Syk-inhibited and Syk-deficient cells (Fig. 1A). These findings suggest that Syk mainly acts downstream of Lyn in H₂O₂ signaling similar to signal transduction by the BCR but also has a role in feedback regulation of Lyn in response to H₂O₂. Stimulation with 0.1–0.6 mM H₂O₂, but not higher concentrations, revealed decreased Akt phosphorylation in Syk-inhibited B cells and MEFs and in Syk-deficient DT40 cells (Fig. 1D), highlighting a role for Syk also at low H₂O₂ concentrations and expanding on similar observations in DT40 B cells (45). Syk is thus critical for Btk, PLCγ2, JNK, and Akt but not, or less so, for Lyn and SHP1 activation in response to high and low extracellular H₂O₂ concentrations across tissues and species.

Fig. 1.

Syk is critical for H₂O₂-induced activation of Btk, PLCγ2, JNK, and Akt. (A and B) Immunoblots of mouse splenic B cells and primary MEFs pretreated with 2 μM R406 and Syk-deficient DT40 B cells stimulated with 1 mM H₂O₂ for 5 min or 50 μg/mL anti-mouse IgM for 3 min (B cells), 5 mM H₂O₂ for 10 min (MEFs), and 5 mM H₂O₂ for 5 min or 10 μg/mL anti-chicken IgM for 3 min (DT40 B cells). (C) Calcium flux of Syk-deficient DT40 B cells stimulated with 5 mM H₂O₂ or 10 μg/mL anti-IgM as determined by the emission ratio of Indo-1. (D) Immunoblots of mouse splenic B cells and primary MEFs pretreated with 2 μM R406 and Syk-deficient DT40 B cells stimulated as indicated for 10 min.

Syk Controls Tyrosine Phosphorylation of Major Pathways Involved in Basic Cellular Processes.

Cellular signal transduction induced by an external stimulus generally results in a circumscribed signaling response mediated by a few upstream kinases that reversibly phosphorylate downstream effectors (38). Syk inhibition with R406 in B cells and MEFs, as well as genetic Syk deficiency in DT40 B cells, resulted in strongly decreased H₂O₂-induced tyrosine phosphorylation of numerous protein species, suggesting that these proteins are direct or indirect tyrosine phosphorylation targets of Syk (Fig. 2A). To determine their identity, we performed label-free quantitative proteomics of phospho-Tyr–enriched lysates of H₂O₂-stimulated Syk-deficient DT40 B cells and H₂O₂-treated controls. The abundance of one-third of all phosphopeptides mapping to 455 homologous human genes was more than 2.5-fold decreased in H₂O₂-stimulated Syk-deficient DT40 B cells compared with controls (Fig. 2B and Dataset S1), suggesting that Syk is a major regulator of protein tyrosine phosphorylation in the presence of H₂O₂. These phosphopeptides included multiple peptides mapping to Btk and PLCγ2, consistent with decreased H₂O₂-induced phosphorylation of these proteins as judged by Western blotting (Fig. 1A and Dataset S1). Another 57 unique human homologs were identified that displayed an exclusive increase in phosphorylation in Syk-deficient cells, consistent with differential regulation by Syk (Fig. 2B and Dataset S1). Eighty-two percent of all Syk-regulated genes were found to be part of a network of proteins with known interactions and associations, suggesting a functional relationship (Fig. 2C). Closer examination revealed that the network components clustering around Syk contained numerous members of major signaling pathways related to the Syk, NF-κB, MAPK, PI3 kinase, JAK/STAT, and rho/ras/rac signaling pathways (Fig. 2C and Dataset S1), some of which are known Syk targets in response to immune receptor engagement (46). Further, the identified Syk targets were greatly enriched for basic cellular processes. They broadly fell into categories such as transcription, translation, protein folding, metabolism, cell cycle regulation, and tumor suppression, and they contained numerous functionally important and well-studied proteins, many of which have been implicated in ROS signaling (Table 1 and Dataset S1). In summary, these findings suggest that Syk is a critical mediator of a distinct signaling response to extracellular H₂O₂ focused on the regulation of basic cellular processes.

Fig. 2.

Syk is a major regulator of protein Tyr phosphorylation in the presence of H₂O₂. (A) Immunoblots of mouse splenic B cells and primary MEFs pretreated with 2 μM R406 and Syk-deficient DT40 B cells stimulated with 1 mM H₂O₂ for 5 min or 50 μg/mL anti-mouse IgM for 3 min (B cells), 5 mM H₂O₂ for 10 min (MEFs), and 5 mM H₂O₂ for 5 min or 10 μg/mL anti-chicken IgM (DT40 B cells). (B) Phosphorylated peptide abundance determined by label-free quantitative MS following enrichment for protein Tyr phosphorylation in H₂O₂-treated Syk-deficient DT40 B cells compared with H₂O₂-treated DT40 controls (5 mM H₂O₂ for 5 min). Red and blue dotted lines denote 2.5-fold increase, and decrease in pY peptide abundance, respectively. (C) Known network interactions of phosphorylated proteins regulated by Syk identified in the experiment described in B as determined by algorithms of the string database (87) (Left) and an extract (boxed) from this network focusing on signaling pathways clustering around Syk (Right). Key pathway members are highlighted in black.

Table 1.

Examples of tyrosine-phosphorylated proteins regulated by Syk in the presence of H₂O₂ grouped by biological process as identified by label-free quantitative proteomics of phospho-Tyr–enriched lysates of H₂O₂-stimulated Syk-deficient DT40 B cells and H₂O₂-treated controls

Biological process	No. of Syk-regulated pY proteins	Examples
Epigenetic regulation	14	MLL/KMT2A, DNMT1, KDM4A, PBRM1
Transcription	24	POLR1A, POLR2B, POLR3E, MED1, HMGB1, YY1, BACH1
Posttranscriptional regulation	35	DICER1, HNRNPK, SF3A1, SF3B, SMG1, SYNCRIP, SNRPC
Nuclear import/export	8	NUP98, NUP210, IPO7, DDX3X
Translation	34	EIF3A, EIF5, EEF1D, EEF2K, AARS, RPL26, RPL35A, RPS2, STAU1, DHX29, FXR1, UPF1
Protein folding	10	HSPA8, HSP90AA1, HSPD1, DNAJA1 CDC37
Proteolysis	34	PSMA3, PSMB5, ADAM17, ANAPC1, BTRC, CYLD, USP9X, USP16, ANXA2
Metabolic pathways	34	IGF2R, GAPDH, PDH1A, FDPS, ADSL, MDH2, SREBF2, FASN, LSS, CYP51A1, HGS, TOMM34, IRS1, GSK3B, LDHB
Cell cycle/tumor suppression	64	POLA1, TK1, KIF11, KIF15, DCTN2, CDC7, CDK1, CDK2, CDK5, LATS1, MLH1, RAD51, MGMT, VCP, BUB1, WEE1, STK4/MST1, DYRK2, NEDD9, PAK2, BCL11A, PDCD4, AXIN1
Redox regulation	4	SESN1, PRDX1, HVCN1, NCF4

Italicized names denote an increase in tyrosine phosphorylation of these proteins in H₂O₂-treated Syk-deficient DT40 cells, whereas tyrosine phosphorylation of all other proteins was decreased in the Syk-deficient cells.

Of note, the cellular models used in this study were remarkably resistant to the effects of H₂O₂ in our hands, consistent with some previous results (47, 48). Stimulation with a minimum of 1 mM H₂O₂ for 5 min or 5 mM for 10 min was required in primary B cells and MEFs, respectively, to detect robust H₂O₂-induced tyrosine phosphorylation of Syk pathway members Lyn, Syk, SHP1, Btk, and PLCγ2, as well as of many other proteins (Fig. 1D and Fig. S1A). These concentrations were well below saturation in primary B cells (Fig. S1A) and did not result in signs of disintegration after the stimulation period as suggested by intact cellular and mitochondrial ultrastructure in B cells and MEFs (Fig. S1B). In MEFs, H₂O₂ stimulation resulted in a dose-dependent decrease in cell recovery that was partially rescued by pretreatment with R406 and less so by cyclosporin A, an inhibitor of mitochondrial transition pore opening and H₂O₂-induced apoptosis (49) (Fig. S1 C and D). DT40 B cells required stimulation with 5 mM H₂O₂ for 5 min to detect Syk phosphorylation (Fig. 1D). Similar to MEFs treated with H₂O₂ and R406, these treatment conditions resulted in decreased cell recovery of DT40 B cells that were partially rescued by Syk deficiency (Fig. S1E). Further, we observed that the signaling response to H₂O₂ and susceptibility to H₂O₂-induced cell death were greatly affected by culture conditions. For example, H₂O₂ but not BCR signaling in B cells almost completely disappeared in response to prior serum deprivation for 2 h (Fig. S1F), suggesting that metabolic health is a critical determinant of H₂O₂ signaling. Overall, these results suggest that the high H₂O₂ concentrations used in parts of this study reflect physiological doses of extracellular H₂O₂ to B cells and MEFs because they were either greatly below saturation of the signaling response or elicited regulated “programmed” cellular responses.

Fig. S1.

Cellular responses to H₂O₂ in B cells and fibroblasts. (A) Immunoblots of mouse splenic B cells treated with different concentrations of H₂O₂ for 5 min or anti-IgM for 3 min and of primary MEFs treated with H₂O₂ for 10 min. (B) Transmission electron micrographs of primary B cells treated with 1 mM H₂O₂ for 5 min or 50 μg of anti-IgM for 3 min and of primary MEFs treated with 5 mM H₂O₂ for 10 min. (C and D) Phase contrast microscopy and cell numbers determined by trypan blue exclusion of MEFs that were cultured for an additional 4 d in complete medium after treatment with H₂O₂ in serum-free medium for 18 h. (Magnification: C, 20×.) (E) Cell numbers of DT40 B cells after culture for 18 h in complete medium following treatment with 5 mM H₂O₂ in serum-free medium for 10 min. (F) Immunoblots of B cells cultured in serum-free medium for the indicated times before stimulation, followed by treatment with H₂O₂ for 5 min and anti-IgM for 3 min. *P < 0.05; **P < 0.005; ***P < 0.0005.

Lyn but Not Protein Tyrosine Phosphatases Are Required for H₂O₂-Induced Syk Activation.

Signal transduction cascades are characterized by hierarchical signaling events, in which upstream mediators diversify and amplify the signaling input (38). Protein tyrosine phosphatases were previously proposed to initiate and promote H₂O₂ signaling as a result of redox-mediated inactivation (18, 25, 26). We therefore hypothesized that protein tyrosine phosphatases might be upstream activators of Syk, and that inhibition or loss of relevant phosphatases should therefore diminish H₂O₂ signaling in a cellular context. To address this question as it relates to the Syk pathway, we pretreated primary B cells and MEFs with the general protein tyrosine phosphatase inhibitor sodium orthovanadate (Na₃VO₄) (50), followed by stimulation with H₂O₂. Na₃VO₄ had little effect on protein tyrosine phosphorylation in the absence of H₂O₂ in B cells and MEFs (Fig. 3A), suggesting that decreasing protein tyrosine phosphatase activity is not sufficient to induce signaling. Consistent with these findings, cell recovery was normal after overnight culture of MEFs with Na₃VO₄ (Fig. S2 A and B). However, H₂O₂ combined with prior Na₃VO₄ treatment led to strongly enhanced phosphorylation of Syk pathway members Lyn, Syk, Btk, PLCγ2, and many other proteins, as well as increased phosphorylation of ERK, JNK, and p38. Enhanced cellular signaling in H₂O₂-stimulated Na₃VO₄-pretreated MEFs correlated with reduced cell recovery after overnight culture similar to treatments with higher concentrations of H₂O₂ that induce increased Syk pathway activation (Figs. S1 A and C and S2 A and B). Genetic deficiency of SHP1 and SHP2, the main phosphatases dephosphorylating Syk (51), resulted in normal or slightly enhanced phosphorylation of the Syk pathway following H₂O₂ stimulation in DT40 B cells (Fig. 3A). These results are thus consistent with a nonessential role of SHP1, SHP2, and other protein tyrosine phosphatases in H₂O₂-induced Syk pathway activation.

Fig. 3.

Lyn and the respiratory chain, but not protein tyrosine phosphatases, are required for H₂O₂-induced activation of the Syk pathway. (A) Immunoblots of H₂O₂-stimulated mouse splenic B cells (0.5 mM H₂O₂ for 5 min) or primary MEFs (2 mM H₂O₂ for 10 min) pretreated with the phosphatase inhibitor Na₃VO₄ (100 μM) and H₂O₂ stimulated SHP1- and/or SHP2-deficient DT40 cells (2 mM for 5 min). Stimulation of mouse splenic B cells and primary MEFs pretreated with 30 nM dasatinib (B) or 50 nM rotenone (C) and Lyn- and Lyn/Syk-deficient DT40 B cells treated with 1 mM H₂O₂ for 5 min and 50 μg/mL anti-mouse IgM for 3 min (B cells), 5 mM H₂O₂ for 10 min (MEFs), and 5 mM H₂O₂ for 5 min and 10 μg/mL anti-chicken IgM for 3 min (DT40 B cells).

Fig. S2.

Role of Na₃VO₄, mitochondrial ROS, and mitochondrial dysfunction in H₂O₂ signaling. (A and B) Phase contrast microscopy and cell numbers determined by trypan blue exclusion of MEFs that were cultured overnight in complete medium after treatment with H₂O₂ for 10 min and pretreatment with 100 μM Na₃VO₄ for 30 min in serum-free medium. (C) Phase contrast microscopy of MEFs cultured with 50 nM rotenone in serum-free medium for 16 h. (D) Immunoblots of primary B cells treated with different concentrations of complex I inhibitor rotenone for 30 min, followed by 5 min of H₂O₂ (1 mM) or 3 min of anti-IgM (50 μg/mL) stimulation. (E) Flow cytometry of MEFs treated with rotenone for 30 min and labeled with MitoSOX Red Mitochondrial Superoxide indicator. (F) Immunoblots of primary B cells and MEFs treated with complex I inhibitor rotenone (ROT, 500 nM), ATP synthase inhibitor oligomycin (OM, 5 μM), or respiratory chain uncoupler carbonyl cyanide m-chlorophenyl hydrazine (CCCP; 10 μM) for 30 min, followed by H₂O₂ (B cells, 1 mM; MEFs, 5 mM) or anti-IgM (50 μg/mL) stimulation. (Magnification: A and C, 20×.)

Lyn is a membrane-bound Src family kinase critical for Syk activation downstream of the BCR (52, 53). We therefore reasoned that Lyn might be critical for H₂O₂-induced activation of the Syk pathway in hematopoietic and nonhematopoietic cells as well. Treatment of primary B cells and MEFs with the Src family kinase inhibitor dasatinib (39) resulted in a reduction of both H₂O₂ and anti-IgM–induced tyrosine phosphorylation of Lyn, consistent with Lyn inhibition, as well as reduced Syk, SHP1, Btk, and PLCγ2 phosphorylation and reduced general protein tyrosine phosphorylation. Similarly, Lyn-deficient and Lyn/Syk-doubly deficient DT40 B cells treated with H₂O₂ exhibited reduced phosphorylation of Syk and its downstream target proteins, thus adding to earlier results (48) (Fig. 3B). Dasatinib treatment and genetic deficiency of Lyn also led to a reduction in H₂O₂- and anti-IgM–induced JNK and ERK but not p38 phosphorylation (Fig. 3B). In summary, these results suggest that Src family kinases, and Lyn in particular, are upstream regulators of Syk and SHP1 activation in response to H₂O₂ as well as BCR engagement.

Interference with the Respiratory Chain Selectively Diminishes H₂O₂-Induced Activation of Lyn and the Syk Pathway.

In addition to producing ATP through aerobic respiration, the mitochondrial respiratory chain is involved in signal transduction in the absence of apparent stimuli as well as in response to ROS-inducing stressors and extracellular H₂O₂ (54–57). To test whether the respiratory chain might also play a role in H₂O₂-induced activation of Lyn and the Syk pathway, primary B cells, MEFs, and DT40 B cells were treated with 50 nM rotenone, a complex I inhibitor (58). Consistent with results in other cell types (59), culture of MEFs with this concentration of rotenone and high-glucose culture medium for 16 h did not impair viability (Fig. S2C).

Treatment with rotenone alone for 30 min did not induce activation of the Syk pathway at a concentration of 50 nM and higher despite induction of mitochondrial ROS (Fig. 3C and Fig. S3 D and E). It thus appears that ROS is not sufficient to induce Syk signaling. However, addition of H₂O₂ to rotenone-pretreated cells resulted in almost complete loss of phosphorylation at activating Lyn Tyr396 but normal phosphorylation at inhibitory Lyn Tyr507 and only partial or no loss of p38 phosphorylation (Fig. 3C and Fig. S3D). These findings suggest that the mitochondrial respiratory chain has a selective role in H₂O₂-induced activation but not inhibition of Lyn nor H₂O₂-induced activation of p38 and that rotenone combined with short-term H₂O₂ treatment does not reduce cellular ATP to levels prohibiting ATP-dependent kinase signaling. Consistent with decreased Lyn activity, rotenone also led to a strong reduction of H₂O₂-induced tyrosine phosphorylation of the Syk pathway members Syk, SHP1, Btk, PLCγ2, JNK, ERK, and many other proteins (Fig. 3C). In contrast, rotenone treatment did not impair BCR-mediated activation of the Syk pathway in primary B cells and DT40 B cells, suggesting that BCR-induced activation of this pathway is independent of the respiratory chain (Fig. 3C). Similar results were obtained when mitochondrial respiratory chain function was perturbed directly or indirectly with the ATP synthase inhibitor oligomycin and electron transport chain uncoupler carbonyl cyanide m-chlorophenyl hydrazine, thus further consistent with a critical role of the respiratory chain in H₂O₂ signaling (Fig. S2F). Taken together, the mitochondrial respiratory chain thus has a selective role in H₂O₂-induced activation of the Syk pathway but not in activation of this pathway in response to extracellular ligand-mediated receptor engagement. Further supporting the notion that H₂O₂-induced activation of Syk and BCR-induced activation of Syk are distinct processes, BCR ligation but not H₂O₂ induced the appearance of slower migrating Syk-positive bands consistent with ubiquitinated Syk (60) (e.g., Fig. 3C). Overall, H₂O₂-mediated signal transduction is thus characterized by a series of hierarchical signaling events placing the respiratory chain upstream of Lyn, Lyn upstream of Syk and SHP1, and Syk upstream of a distinct signaling and cellular response in hematopoietic and nonhematopoietic cells.

Fig. S3.

Syk pathway localizes to liver mitochondria. (A) Cellular subfractionation of mouse liver mitochondria by Percoll gradient centrifugation to remove endoplasmic reticulum membranes. P100, microsomal fractions; S100, cytosolic fraction. (B) Confocal images of 4-μm tissue sections of normal mouse liver stained as indicated. White squares (Left) represent area of magnification shown on Right.

H₂O₂ Induces Phosphorylation of Mitochondria-Associated Lyn, Syk, and Many Other Proteins.

Signaling induced by the BCR and other plasma membrane-bound receptors initially clusters at the site of the engaged receptors (38, 61). Given our results that H₂O₂-mediated Syk signaling is controlled by the respiratory chain, we reasoned that Lyn and Syk activation occurs, at least in part, in association with the mitochondria. Immunopurification of mitochondria and associated membranes showed that H₂O₂ induced robust phosphorylation of Lyn, Syk, and many other distinct proteins in the mitochondrial fraction of both primary B cells and MEFs (Fig. 4A). Consistent with these findings, H₂O₂ treatment of B cells induced Lyn and Syk phosphorylation overlapping with complex I staining as judged by confocal imaging (Fig. 4 B and C). In contrast, Syk phosphorylation induced by BCR cross-linking on the B-cell surface was found exclusively at the outer rim of the cell, consistent with cap formation and localized signaling from the plasma membrane (61, 62) (Fig. 4 B and C). Furthermore, 13 Syk targets are known mitochondrial proteins among them essential metabolic enzymes, in support of a role for Syk in H₂O₂-mediated mitochondrial regulation (Dataset S1). H₂O₂ induced activation of the Syk pathway thus appears to take place in part though not exclusively associated with the mitochondria in line with the view that the respiratory chain is an upstream component of an H₂O₂-induced signal transduction cascade mediated by Lyn and Syk.

Fig. 4.

H₂O₂ induces phosphorylation of mitochondria-associated Lyn and Syk, which localize to the differential interference contrast mitochondrial intermembrane space and are transiently and stably associated with the mitochondrial membrane compartment. (A) Tom22-mediated mitochondrial immunopurification of B cells and MEFs stimulated with H₂O₂ (B cells: 1 mM H₂O₂ for 5 min, MEFs: 5 mM H₂O₂ for 10 min). Mito, mitochondrial fraction; WCL, whole-cell lysate. (B and C) Immunofluorescence staining and confocal images of mouse splenic B cells stimulated with 1 mM H₂O₂ for 5 min and 50 μg/mL anti-IgM for 3 min. DIC, differential interference contrast; Max, maximum. Mitochondrial subfractionation of mouse spleen mitochondria treated with hypoosmotic (swelling) buffer and PK (D) and quantitation of signal intensity of immunoblots (E). IMM, inner mitochondrial membrane; IMS, intermembrane space; OMM, outer mitochondrial membrane. (F) Carbonate extraction of mouse spleen mitochondria separating membrane integral (P) and soluble (SN) proteins. sol, soluble; Tm, transmembrane. (G) Confocal images of resting mouse splenic B cells stained as indicated.

A Pool of Cellular Lyn and Syk Localizes to the Mitochondrial Intermembrane Space.

To determine the precise spatial relationship of Syk pathway members with the mitochondria, we performed submitochondrial fractionation of mouse spleen mitochondria. A large portion of Lyn, Syk, and SHP1 remained associated with the mitochondria after digestion with proteinase K (PK), which resulted in degradation of outer mitochondrial membrane proteins as indicated by loss of Tom22 (Fig. 4 D and E). Lyn, Syk, and SHP1 almost disappeared after incubation in hypotonic buffer, rupturing the outer mitochondrial membrane and resulting in PK-mediated degradation of the intermembrane space-facing domain of inner mitochondrial membrane protein Tim23 (Fig. 4 D and E), overall suggesting localization of a sizable fraction of Lyn, Syk, and SHP1 in the mitochondrial intermembrane space. Separation of soluble and membrane-integrated mitochondrial proteins showed that Syk and SHP1 were partially associated with and Lyn was exclusively associated with the mitochondrial membrane fraction (Fig. 4F), thus paralleling the transient activating association of Syk and SHP1 with plasma membrane-bound phosphotyrosine motifs and the stable plasma membrane association of Lyn (53, 63). Consistent with mitochondrial localization, Lyn, Syk, and SHP1, but not the plasma membrane-bound phosphatase receptor-type tyrosine-protein phosphatase C (B220/CD45/Ptprc), overlapped with complex I in splenic B cells by confocal imaging (Fig. 4G). Lyn, Syk, and SHP1 were also detected in liver mitochondria from which endoplasmic reticulum remnants tethered to mitochondria were removed by Percoll gradient centrifugation (Fig. S3A). In line with these findings, Syk and phosphorylated Syk also overlapped with liver mitochondria by tissue immunofluorescence staining and confocal imaging, overall supporting mitochondrial localization also in nonhematopoietic cells (Fig. S3B). Taken together, these results suggest that a portion of cellular Lyn, Syk, and SHP1 localizes to the mitochondrial intermembrane space and appears to be transiently (Syk and SHP1) or stably (Lyn) associated with the mitochondrial membrane compartment.

The Syk Pathway Is of Early Metazoan Origin, Is Coexpressed Across Tissues, and Shows Evidence of Evolutionary Constraint in the Human.

Given the critical role of ROS signaling across biology, we finally reasoned that evidence may exist compatible with a function of Syk beyond linking immunoreceptor tyrosine–based activation motif (ITAM)-bearing receptors of the immune system to downstream pathways (46, 64). Indeed, database mining of large microarray and mRNA sequencing datasets, and our confirmatory quantitative PCR assay and immunohistochemistry of normal human and mouse tissues, showed that Syk transcript and protein were detectable in every tissue examined, although smaller amounts relative to total RNA and protein were found in most nonhematopoietic cells (Fig. 5A and Fig. S4 A–D). Analysis and categorization of the available 3,078 indexed research articles mentioning Syk suggested that Syk is functional in cell types derived from every organ system, although conclusive genetic evidence for a critical in vivo role exists only for hematopoietic tissues and mammary and vascular endothelial cells (Fig. 5B and Table S1). Further, expression of LYN, PTPN6, BTK, and PLCγ2, was correlated with SYK expression in a wide range of human tissues, whereas there were minor, no, or negative correlations with expression of the BCR-associated adapter CD79A (Igα), related family members, and other Syk targets as judged by both mRNA sequencing and microarray data (Fig. 5C and Table S2). These results suggest a constant stoichiometry of Syk with Syk pathway members, consistent with the idea that these proteins interact and form functional units or “signalosomes” in many different tissues.

Fig. 5.

The Syk pathway is coexpressed, is evolutionary ancient, and displays low missense variation in the human. (A) Microarray analysis of normal human tissues showing SYK transcript expression plotted as a box plot with Tukey whiskers (n = 688). The dotted line represents the median of all samples across tissues. (B) Categorization of evidence for a critical function of Syk in different cell types: established, abundant evidence using different cellular models and a combination of pharmacological and genetic approaches or pharmacological in vivo evidence demonstrating a critical role; likely, abundant evidence using different cellular models and a combination of pharmacological and genetic approaches or pharmacological in vivo evidence demonstrating a critical role; possible, at least one study in a cellular model using a combination of pharmacological and genetic approaches demonstrating a critical role. (C) Correlation of mRNA expression across different normal human tissues derived from mRNA sequencing datasets (n = 48). r, Pearson correlation coefficient. (D) Rooted phylogenetic tree of Syk orthologs in the animal kingdom. (E) Estimated evolutionary age of the Syk pathway in billion years (by). (F) Ratios of rare missense to rare synonymous variants for individual genes (dots) extracted from exomes of the Exome Aggregation Consortium. MAF, minor allele frequency. (G) Number of stimuli known to induce Syk-dependent signaling per category. Colors denote the number of stimuli per category with evidence that the stimulus induces cellular ROS (light blue), that the stimulus induces ROS and signals in a ROS-dependent manner (blue), and that ROS is required for stimulus-induced Syk phosphorylation (black). *P < 0.05; **P < 0.005; ***P < 0.0005.

Fig. S4.

Syk is ubiquitously expressed, and Syk orthologs are found across the animal kingdom. (A) SYK expression normalized to 18S in normal human tissues as determined by qPCR (mean ± SEM). (B) SYK expression in normal human tissues derived from human body map mRNA sequencing datasets. Bars represent average number of fragments per kilobase of transcript per million mapped reads (FPKM) from two biological replicates (dots). (C) SYK expression normalized to 18S of normal mouse tissues as determined by qPCR (shown as mean + SEM). (D) Immunohistochemistry against N-terminal Syk of formalin-fixed paraffin-embedded sections of normal tissues from 12-wk-old C57BL/6 mice. Similar results were obtained using a Syk antibody recognizing the C terminus of Syk. (E) Protein sequence alignment of orthologs of family members Syk, ZAP70, and SHARK characterized by conserved tandem Src homology 2 (SH2) domains and a kinase domain (shaded regions). Black regions denote aligned amino acid sequences.

Table S1.

Syk is critical for cellular signaling in cell types derived from all organ systems

Tissue	Evidence level
Hematopoietic system
B cells	3
T cells	3
Natural killer cells	3
Macrophages	3
Monocytes	3
Neutrophils	3
Dendritic cells	3
Eosinophils	2
Platelets	3
Erythroid cells	1
Basophils	3
Mast cells	3
Reproductive system
Ovaries	2
Sperm	1
Mammary epithelium	3
Prostate epithelium	1
Cardiovascular system
Vascular endothelial cells	3
Vascular smooth muscle cells	2
Gastrointestinal system
Intestinal epithelium	1
Pancreatic acinar cells	1
Hepatocytes	2
Gastric epithelium	1
Esophageal epithelium	1
Epidermis
Melanocytes	2
Keratinocytes	1
Skin fibroblasts	1
Connective tissue
Osteoblasts	1
Osteoclasts	2
Adipose tissue	1
Nervous system
Microglia	1
Retinal neurons	2
Pulmonary system
Pneumocytes	1
Mesothelial cells	1
Airway epithelial cells	1
Airway smooth muscle cells	1
Airway fibroblasts	1
Squamous epithelial cells	1
Renal system
Urothelial cells	1
Glomerular cells	1
Tubular cells	1
Endocrine system
Adrenal medulla	1

Annotation and categorization of evidence for a critical function of Syk in different cell types. Criteria used to determine the level of evidence were as follows: (1) possible function of Syk (at least one study in a cellular model using a combination of pharmacological and genetic approaches demonstrating a critical role), (2) likely function of Syk (abundant evidence using different cellular models and a combination of pharmacological and genetic approaches or pharmacological in vivo evidence demonstrating a critical role), (3) established function of Syk (mouse models with genetic loss of function studies and/or extensive genetic evidence in the human demonstrating critical in vivo role).

Table S2.

Syk and Syk pathway members are expressed at similar stoichiometry across tissues

SYK coexpression with gene*	Pearson correlation coefficient, r
	HBM (n = 48)	HBI (n = 504)
Src family kinase
LYN	0.96	0.74
FYN	0.29	−0.06 (ns)
SH2-PTP
PTPN6	0.86	0.81
PTPN11	−0.37	−0.35
Tec family kinase
BTK	0.84	0.53
TEC	0.06 (ns)	−0.02 (ns)
PLCγ
PLCG2	0.55	0.61
PLCG1	0.12 (ns)	0.03
Other
CD79A	0.45	0.55
AKT1	0.11 (ns)	0.28
RPS6KB1	0.14 (ns)	−0.18

Correlation between expression of SYK and indicated genes across tissues and samples as quantitated by mRNA sequencing and microarray analysis. HBI, human body index normal human tissues (microarray); HBM, human body map normal human tissues (mRNA sequencing); ns, P value for r not significant (P > 0.05).

^*Human Gene Organization (HUGO) Gene Nomenclature Committee (HGNC) nomenclature.

We identified known and predicted Syk orthologs in every vertebrate examined, as well as in evolutionarily distant groups of extant metazoans, including a member of the earliest group of metazoans, the sponge Amphimedon queenslandica (65), but not in yeast, plants, and bacteria (Fig. 5D, Fig. S4E, and Table S3). These findings add to earlier observations identifying Syk orthologs in the tunicate Hydra vulgaris and highlight a distribution of Syk orthologs throughout the animal kingdom (66). Similarly, orthologs of the Syk pathway members Lyn, SHP1, Btk, and PLCγ2 were found in the sponge A. queenslandica but not in premetazoan species. In contrast, all known ITAM-containing immune receptor-associated adapters were detected only in evolutionarily recent vertebrates. These findings thus suggest an evolutionary origin of the Syk pathway ∼1.2 billion y ago, closer to the evolutionary origins of members of the MAPK and mammalian target of rapamycin (mTOR) pathways than to the evolutionary origins of the ITAMs of the immune system (Fig. 5E and Table S4).

Table S3.

Syk orthologs are found across the animal kingdom

Species	RefSeq identification no.
	SYK	ZAP70
Vertebrata
Mammalia
Homo sapiens	NP_003168.2	NP_001070.2
Mus musculus	NP_035648.2	NP_033565.2
Rattus norvegicus	NP_036890.1	NP_001012002.1
Pan troglodytes	XP_520123.4	XP_515637.2
Oryctolagus cuniculus	XP_002708308.1	XP_002710048.1
Bos taurus	NP_001032542.1	NP_001179946.1
Ailuropoda melanoleuca	XP_002919153.1	XP_002912473.1
Loxodonta africana	XP_003420489.1	XP_003422076.1
Cavia porcellus	XP_003467605.1	XP_003471744.1
Sus scrofa	NP_001098422.1	XP_003481200.1
Cricetulus griseus	XP_007642341.1	XP_003498337.1
Callithrix jacchus	XP_002742746.1	XP_002757439.2
Canis lupus familiaris	XP_541312.3	XP_005626103.1
Aves
Taeniopygia guttata	XP_002192421.1	XP_002191935.1
Gallus gallus	NP_001026601.1	XP_418206.4
Amphibia
Xenopus tropicalis	XP_002937073.2	NP_001006825.1
Xenopus laevis	NP_001086665.1	NP_001087005.1
Actinopterygii
Danio rerio	NP_998008.1	NP_001018425.1
Oreochromis niloticus	XP_005449474.1	XP_003445176.1
Chondrichthyes
Callorhinchus milii	XP_007898221.1	XP_007899176.1
		SHARK
Tunicata
Ciona intestinalis	XP_002123000.1	XP_002121112.2
Echinodermata
Strongylocentrotus purpuratus	XP_003724543.1	XP_001180232.2
Hemichordata
Saccoglossus kowalevskii	XP_006822388.1	XP_006819141.1
Mollusca
Aplysia californica	XP_005099019.1	XP_005102068.1
Cnidaria
H. vulgaris	AAC28629.1	P53356.1
Porifera
A. queenslandica	XP_003385896.1	XP_003385679.1
Arthropoda
Drosophila melanogaster	nd	NP_524743.2
Tribolium castaneum	nd	XP_969153.1
Nasonia vitripennis	nd	XP_001602126.1
Aedes aegypti	nd	XP_001653430.1
Anopheles gambiae	nd	XP_311423.4
Culex quinquefasciatus	nd	XP_001862264.1
Drosophila ananassae	nd	XP_001959194.1
Drosophila erecta	nd	XP_001975400.1
Drosophila grimshawi	nd	XP_001995160.1
Drosophila mojavensis	nd	XP_002005639.1
Drosophila persimilis	nd	XP_002016089.1
Drosophila sechellia	nd	XP_002034084.1
Drosophila virilis	nd	XP_002049415.1
Drosophila willistoni	nd	XP_002068686.1
Drosophila yakuba	nd	XP_002092257.1
Drosophila simulans	nd	XP_002081734.1
Drosophila pseudoobscura ps.	nd	XP_001360545.2
Pediculus humanus corporis	nd	XP_002430981.1
Acyrthosiphon pisum	nd	XP_003248633.1
Ixodes scapularis	nd	nd
Nematoda
Caenorhabditis elegans	nd	nd
Caenorhabditis briggsae	nd	nd
Platyhelmintha
Schistosoma mansoni	nd	nd

Reference Sequencing (RefSeq) identification numbers of predicted and known metazoan Syk orthologs, including family members ZAP70 and SHARK, are given. Boldfaced words indicate a phylum or subphylum; italicized words indicate species. nd, not detected.

Table S4.

The Syk pathway arose early in metazoan evolution

Gene*	Species†	Time,‡ My
ITAM adapter
CD79A	Ornithorhynchus anatinus	220
CD79B	D. rerio	455
FCER1G	Monodelphis domestica	176
CD247	D. rerio	455
CD3D	Anolis carolinensis	325
CD3G	A. carolinensis	325
CD3E	A. carolinensis	325
NFAM1	A. carolinensis	325
TYROBP	D. rerio	455
Src family kinase
FGR	A. queenslandica	1,237
FYN	A. queenslandica	1,237
SRC	A. queenslandica	1,237
YES1	A. queenslandica	1,237
BLK	A. queenslandica	1,237
HCK	A. queenslandica	1,237
LCK	A. queenslandica	1,237
LYN	A. queenslandica	1,237
Syk family kinase
SYK	A. queenslandica	1,237
ZAP-70	A. queenslandica	1,237
SH2-PTP
PTPN6	A. queenslandica	1,237
PTPN11	A. queenslandica	1,237
Tec family kinase
BMX	A. queenslandica	1,237
BTK	A. queenslandica	1,237
ITK	A. queenslandica	1,237
TEC	A. queenslandica	1,237
TXK	A. queenslandica	1,237
PLCγ
PLCG2	A. queenslandica	1,237
PLCG1	A. queenslandica	1,237
Akt
AKT1	A. queenslandica	1,237
AKT2	A. queenslandica	1,237
AKT3	A. queenslandica	1,237
MAPK
MAPK14	Oryza sativa	1,628
MAPK1	O. sativa	1,628
MAPK3	O. sativa	1,628
MAPK8	Saccharomyces cerevisiae	1,368
MAPK9	Triticum aestivum	1,628
MAPK10	S. cerevisiae	1,368
mTOR
MTOR	T. aestivum	1,628
RPTOR	O. sativa	1,628
RICTOR	S. cerevisiae	1,368
RPS6KB1	O. sativa	1,628

Evolutionarily most distant species with an ortholog of a given gene and its estimated time of divergence from human species.

^*HGNC nomenclature.

^†Evolutionarily most distant species with ortholog.

^‡Estimated time (My) since species diverged from human species.

A low ratio of nonsynonymous to synonymous rare variants in humans and other species suggests purifying selection, thus allowing an estimate of the effects of missense variation in a given gene on reproductive fitness. Similar to genes of the MAPK and mTOR pathways, LYN, SYK, PTPN6, BTK, and PLCG2 displayed low ratios of rare missense variants to synonymous variants compared with the known ITAM-bearing immune adapters and many other immune-related genes as judged by mining exomes of 60,706 individuals assembled by the Exome Aggregation Consortium (Fig. 5F). Syk and the Syk pathway may thus also have a critical function in normal human physiology.

Literature curation revealed that 45 diverse stimuli ranging from hormones and growth factors to endogenous stressors such as high glucose induce signaling in a Syk-dependent manner (Fig. 5G and Table S5). Thirty-eight of these diverse stimuli are also known to induce signaling in a ROS-dependent manner, raising the possibility that a unifying mechanism of Syk activation by many stressors might be its activation by endogenous ROS (Fig. 5G and Table S5)_. In support of such a notion, osmotic stress and TNF induce Syk phosphorylation in a ROS-dependent manner (67, 68), suggesting that Syk critically mediates signaling not only in response to extracellular ROS but possibly also in response to intracellular ROS. Taken together, the ubiquitous expression of Syk, coexpression of Syk interaction partners in different tissues, occurrence of Syk across the animal kingdom, origin of the Syk pathway early in metazoan evolution, evidence for Syk signaling in numerous ROS-mediated processes, and signs of evolutionary constraint on the pathway in the human suggest a much broader role for Syk than currently appreciated and are compatible with a role in ROS signaling.

Table S5.

Stimuli using the Syk pathway depend on ROS

Stimulus	Evidence level
Cell matrix/cell cell
Collagen	3
Fibronectin	2
Thrombin	2
Fibrinogen	1
C-reactive protein	2
Von Willebrand factor	0
E-selectin	0
P-selectin	0
Cytokines
IFN-α	2
IL-1	2
IL-2	1
IL-4	2
IL-5	2
IL-6	2
CD40 ligand	2
B-cell activating factor	0
Receptor activator of NF-κB ligand	2
TNF	3
Endogenous stressors
Amyloid-β	2
Heme	2
Monosodium urate	2
Chondroitin sulfate	0
Oxidized low-density lipoprotein	2
Glucose	2
Arachidonic acid	2
Environmental stressors
LPS	2
CpG oligodeoxynucleotide	2
Hypoxia	2
Osmotic stress	3
UV irradiation	2
Growth factors
EGF	2
VEGF	2
PDGF	3
Granulocyte CSF	2
GM-macrophage CSF	2
Macrophage CSF	2
Stem cell factor	2
Thrombopoietin	2
Neurotransmitters
Epinephrine	2
Glutamate	2
Peptide hormones
Adiponectin	2
Angiotensin II	2
Endothelin	2
Insulin	2
Melanocyte-stimulating hormone	2
ITAM receptor ligands
Ig	2
BCR ligands	2
T-cell receptor ligands	2
Viral oncogenes with ITAMs	2
C-type lectin ligands	2

Stimuli known to induce Syk dependent signaling and/or cellular responses and their level of evidence linking them to ROS production, ROS-dependent signaling, and ROS-dependent Syk phosphorylation. Criteria to determine level of evidence were as follows: evidence for Syk-dependent signaling plus (0) none, (1) evidence that that stimulus induces cellular ROS, (2) evidence that ROS is required for stimulus-induced signal transduction and/or cellular outcome and (1), and (3) evidence that ROS is required for stimulus-induced Syk phosphorylation and (1, 2).

Discussion

Here, we provide evidence suggesting that H₂O₂ signaling has multiple distinguishing features of a signal transduction cascade. It is characterized by a sequence of events culminating in a distinct signaling response: The upstream respiratory chain selectively activates Lyn, resulting in activation of downstream Syk, which, in turn, controls tyrosine phosphorylation of pathways critically involved in signaling, transcription, translation, metabolism, and cell cycle regulation. Its upstream components reside and are active in physical proximity: H₂O₂-mediated Lyn and Syk activation occurs, at least in part, in proximity to the respiratory chain, and a pool of cellular Lyn and Syk localizes to the mitochondrial intermembrane space. Finally, it is controlled by the same mediators in different species and tissues responsive to H₂O₂: The respiratory chain and the conserved and ubiquitous Syk pathway mediate H₂O₂ signaling in diverse cell types that include mouse and chicken B cells as well as fibroblasts. The results thus provide a framework to conceptualize ROS signaling and offer a rationale for numerous avenues of investigation.

ROS and mitochondrial dysfunction have been linked to a large number of biological processes and diseases, including adipogenesis, neurodegeneration, cardiovascular disease, inflammation, and the aging process itself (69–74). Although only extracellular H₂O₂ was used in this study, it seems likely that intracellular H₂O₂ induced by receptors and other stressors also uses this pathway, overall suggesting that the immune kinase Syk might be critical for many more cellular responses and disorders than currently appreciated. Exploring how modulation of Syk activity and gene dosage affects different disease states will be particularly relevant for ongoing drug development efforts currently focused only on hematological malignancies and autoimmune disease (64, 75).

The finding that the respiratory chain is required for H₂O₂-induced Syk activation raises the intriguing possibility that an ITAM in one of the more than 100 largely uncharacterized mammalian respiratory chain subunits binds and activates mitochondrial intermembrane space-localized Syk. Although the present results implicate the respiratory chain in signal transduction, identification of a subunit with a functional ITAM would establish the respiratory chain as a bona fide signal transducer. Such a subunit might also mediate some of the many functions of the respiratory chain described that are independent of its ability to produce ATP (59, 76–78). The present data suggesting mitochondrial intermembrane space localization of the Syk pathway also raise the question of whether Syk and Lyn might directly tyrosine-phosphorylate and modulate respiratory chain function. Indeed, there is precedence for a Src family kinase to modulate complex II activity, consistent with the observation that the respiratory chain is extensively posttranslationally modified (79, 80).

Our proposed model that H₂O₂ signaling resembles canonical signal transduction implies the existence of an upstream ROS receptor that recognizes H₂O₂ with much higher sensitivity than its surroundings. The most intuitive location of such a sensor might be in the respiratory chain itself, which may have evolved to sense H₂O₂ at the site of its production and transmit a signal to the cell via mitochondrial intermembrane space localized signaling pathways. Although the reaction constant for oxidation of the abundantly present cysteine residues is generally low, iron and iron clusters display much higher reactivity with H₂O₂, thus offering a limited number of candidates as exquisitely sensitive receptor modules (1, 81, 82). In support of such a possibility, multiple iron cluster-containing proteins induce transcriptional changes in bacteria, and thus might represent ancestral ROS-sensing signal transducers (83).

Further, one might speculate that this pathway represents a mechanism of mitochondrial control over ROS-induced cellular processes such as differentiation and proliferation, or senescence and programmed cell death. Indeed, communicating mitochondrial health to the cell might be a critical prerequisite to the successful implementation of ROS-stimulated energetically demanding cellular processes. Further, the mitochondria might also activate the Syk pathway to induce growth arrest and/or programmed cell death as the present results suggest. Consistent with utilization of this pathway for a spectrum of cellular responses, Syk and mitochondrial dysfunction have both been implicated in cellular differentiation and proliferation, as well as in tumor suppression (15, 64, 75, 84).

Finally, it is striking that a kinase of early metazoan origin such as Syk is so critical for H₂O₂ signaling in vertebrate cells, given that cellular responses to ROS first evolved in bacteria (1). Perhaps the occurrence of the Syk pathway along with multicellularity reflects an adaptation specific to metazoan life that allows the integration of metabolic signals from the mitochondria with other extracellular and intracellular cues transmitted and amplified by ITAMs (85). Indeed, the existence of several hundred ITAM-containing proteins across biological processes has been suggested (86), which might fulfill this function linking tissue- and context-specific inputs to the basic and ubiquitous Syk pathway.

Materials and Methods

Cell Culture.

Primary B cells were isolated by depletion of CD43-positive cells from mouse spleen. Primary MEFs derived from embryonic day 14.5 embryos were cultured for 2–4 d before use. DT40 cell lines were imported from RIKEN and cultured in RPMI medium at 39.5 °C. All procedures were performed according to protocols approved by the Committee on Animal Care at the Massachusetts Institute of Technology and the University of Freiburg.

Phosphoproteomics.

A label-free quantitative liquid chromatography-tandem MS analysis was performed using an LTQ-Orbitrap-ELITE mass spectrometer (Thermoscientific), electrospray ionization–collision-induced dissociation, and SEQUEST search results following immunoprecipitation with Tyr phosphorylation motif antibody pY-1000 (Cell Signaling Technology).

Mitochondrial Subfractionation.

Crude mitochondria were resuspended in normosmotic buffer or hypotonic buffer [20 mM Hepes/KOH (pH 7.6)] and incubated for 15 min, followed by PK (Roche) digestion for 15 min.

Statistical Methods.

All statistical analyses were performed using Prism 6 (GraphPad Software, Inc.). Statistical significance was indicated as follows: *P < 0.05; **P < 0.005; ***P < 0.0005.

Supplementary materials and methods, including antibodies used in this study and listed in Table S6, are included in SI Materials and Methods.

Table S6.

Antibodies used in immunoblotting (WB), immunofluorescence (IF, IF-P), and immunohistochemistry (IHC) experiments in this study

Target antigen (human)	Function	Clone/catalog no.	Manufacturer	Application	Specificity/comments
pY396 Lyn	Activation loop tyrosine, required for activation (109)	EP503Y (rabbit monoclonal)	Abcam	WB (1:1,000)	Band absent or reduced in H2O2- and dasatinib-treated B cells and MEFs and Lyn−/− DT40 cells
pY507 Lyn	Negative regulatory function (109)	EP504Y (rabbit monoclonal)	Abcam	WB (1:1,000), IF (1:1,000)	Bands absent in H2O2- and dasatinib-treated B cells by WB and no signal in H2O2-treated Lyn−/− DT40 cells by IF
Lyn		44 (sc-15) (rabbit polyclonal)	Santa Cruz Biotechnology	WB (1:500), IF (1:100)	AB recognizes both Lyn isoforms in mouse but not chicken
Lyn		C13F9 (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)	AB recognizes smaller Lyn isoform in chicken and mouse, band absent in Lyn−/− DT40 cells
pY323 Syk	Positive regulatory function, PI3K, c-cbl binding site (53)	sc-293118 (rabbit polyclonal)	Santa Cruz Biotechnology	WB (1:200)	Band absent or reduced in
				IF (1:200)	Syk−/− DT40 cells and R406-treated B cells and MEFs, no signal in H2O2-treated Syk−/− DT40 cells by IF
				IF-P (1:50)
Syk	N-terminal Syk	N-19 (sc-1077) (rabbit polyclonal)	Santa Cruz Biotechnology	WB (1:1,000),	Recognizes both isoforms of Syk, absent band in Syk−/− DT40 cells, specific for Syk by IHC based on using peptide controls
				IHC (1:500)
Syk		D3Z1E (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000),	Recognizes both Syk isoforms
				IF (1:200),
				IF-P (1:50)
Syk	C-terminal Syk	C-20 (sc-929) (rabbit polyclonal)	Santa Cruz Biotechnology	IHC (1:500)	Specific for Syk by IHC based on using peptide controls
pY564 SHP1	Required for phosphatase activity, induced by cellular activation (110)	D11G5 (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)	Absent band in SHP1−/− DT40 cells but not WT DT40 cells treated with H2O2
SHP1		Y476 (rabbit monoclonal)	Abcam	WB (1:1,000)	AB does not recognize chicken SHP1
pY1217 PLCγ2	Induced by BCR cross-linking (111)	No. 3871 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)	AB recognizes phosphorylated mouse and chicken PLCγ2 and unphosphorylated PLCγ2 when used at a ratio <1:500
PLCγ2		Q-20 (sc-407)	Santa Cruz Biotechnology	WB (1:1,000)	AB recognizes mouse and chicken PLCγ2
pY223 Btk	Btk Y223 promotes BCR-induced calcium flux (90)	No. 5082 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)	AB recognizes phosphorylated mouse and chicken Btk
Btk		D3H5 (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)	AB recognizes mouse and chicken Btk
pTyr		4G10 (mouse monoclonal)	Upstate Biotechnology	WB (1:10,000)	Recognizes subset of tyrosine-phosphorylated proteins (112)
pS473 Akt	Activating residue	D9E (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)
pT308 Akt	Activating residue	C31E5E (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)
Akt		No. 9272 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)
pT202/Y204 ERK	Activating residues in activation loop	No. 9101 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)
ERK		No. 9102 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)
pT183/Y185 JNK	Activating residues in activation loop	81E11 (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)
JNK		No. 9252 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)
pT180/Y182p38	Activating residues in activation loop	D3F9 (rabbit monoclonal)	Cell Signaling Technology	WB (1:1,000)
p38		No. 9212 (rabbit polyclonal)	Cell Signaling Technology	WB (1:1,000)
VDAC1	OMM	20B12AF2 (mouse monoclonal)	Abcam	WB (1:1,000)
VDAC3	OMM	In-house (rabbit polyclonal)	Immunogen: H-CGKNF SAGGH KVGLG FELEA-OH	WB
Tom22	OMM	J-31 (mouse monoclonal)/sc-101286	Santa Cruz Biotechnology	WB (1:100)
Tom22	OMM	ab57523 (mouse monoclonal)	Abcam	WB (1:1,000)
Cytochrome C	IMS	37BA11/ab110325 (mouse monoclonal)	Abcam	WB (1:10,000)
Cytochrome C	IMS	7H8.2C12 (mouse monoclonal)/556433	BD Biosciences	WB (1:1,000)
ATP5A	IMM	15H4C4 (mouse monoclonal)	Abcam	WB (1:1,000)
				IF (1:500)
Complex I	IMM	18G12BC2 (mouse monoclonal)	Abcam	IF (1:500)
Ndufs3	IMM	3F9DD2 (mouse monoclonal)	Abcam	IF-P (1:200)
ATP5A	IMM	Ab14748 (mouse monoclonal)	Abcam	WB (1:1,000)
Cox IV	IMM	Ab33985 (mouse monoclonal)	Abcam	WB (1:1,000)
Cox6a	IMM	In-house (rabbit polyclonal)	Immunogen: H-CHNPH VNPLP TGYED E-OH	WB
Tim23	IMM	32/Tim23 (mouse monoclonal)/611222	BD Biosciences, immunogen: rat Tim23 amino acids 5–126	WB (1:1,000)
Tim44	Matrix	H-257 (rabbit polyclonal)/sc-366692	Santa Cruz Biotechnology	WB (1:200)
Lamin B	Nuclear membrane protein	M-20 (goat polyclonal)	Santa Cruz Biotechnology	WB (1:1,000)
PCNA	Protein with soluble cytosolic and nuclear isoform	PC10 (mouse monoclonal)	Santa Cruz Biotechnology	WB (1:1,000)
ACC	Catalyzes the biogenesis of long-chain fatty acids	Ab109368 (rabbit monoclonal)	Abcam	WB (1:1,000)
Calnexin	ER	Peptide antibody (rabbit polyclonal)/ab22595	Abcam	WB (1:100)
Actin		ACTN05 (C4) (mouse monoclonal)/ab3280	Abcam	WB (1:500)
GAPDH		6C5 (mouse monoclonal)/ab8245	Abcam	WB (1:5,000)
PDI	ER	RL90 (mouse monoclonal)/ab2792	Abcam	WB (1:1,000)
Calreticulin	ER	Peptide antibody (rabbit polyclonal) 2891S	Cell Signaling Technology	WB (1:750)
Anti-rabbit IgG	Secondary antibody	Anti-rabbit IgG, HRP linked (no. 7074)	Cell Signaling Technology	WB (1:5,000)
Anti-mouse IgG	Secondary antibody	Anti-rabbit IgG, HRP linked (#7076)	Cell Signaling Technology	WB (1:5,000)
Anti-goat IgG	Secondary antibody	Anti-goat IgG, HRP linked (no. 705-035-147)	Jackson Immunoresearch	WB (1:5,000)
Anti-rabbit IgG	Secondary antibody	Alexa Fluor 488 donkey anti-rabbit IgG (A21206)	Life Technologies	IF (1:1,000), IF-P (1:100)
Anti-mouse IgG	Secondary antibody	Alexa Fluor 488 goat anti-mouse IgG (A-11001)	Life Technologies	IF (1:1,000), IF-P (1:100)
Anti-mouse IgG	Secondary antibody	Alexa Fluor 633 goat anti-mouse IgG (A21050)	Life Technologies	IF (1:1,000), IF-P (1:100)

ER, endoplasmic reticulum; IMM, inner mitochondrial membrane; IMS, intermembrane space; OMM, outer mitochondrial membrane.

SI Materials and Methods

Mice.

C57BL/6J mice were bred in-house or purchased from The Jackson Laboratory (stock no. 000664). All procedures were performed according to protocols approved by the Committee on Animal Care at the Massachusetts Institute of Technology and the University of Freiburg.

Isolation of Primary Cells and Tissues.

B-cell isolation.

Mouse splenic B cells were harvested from adult male and female C57BL/6J mice. Splenic cells were isolated by grinding spleens through a 70-μm cell strainer (no. 352350; Corning) with a plunger (3-mL syringe), followed by red cell lysis with Gey’s solution for 1 min (88). Splenic CD43-negative B cells were isolated by removal of cells labeled with biotinylated CD43 antibody (clone R2/60, no. 13-0431-81; eBioscience) and magnetic beads coupled to streptavidin (BD Biosciences) using the BD IMagnet system and following the manufacturer’s instructions (BD Biosciences). This protocol resulted in recovery of, on average, 50 × 10⁶ cells per spleen, and 85% of lymphocytes were B cells as judged by CD19 staining (clone 1D3, no. 12-0193-81; eBioscience) and flow cytometry. The cells were kept on ice throughout the procedure and were used for no longer than 2–3 h postmortem.

MEF derivation and culture.

Primary murine embryonic fibroblasts were derived by first removing the heads and fetal livers from 20 to 30 freshly isolated embryonic day 14.5 embryos, followed by briefly mincing the bodies with fine scissors. Minced body parts were incubated in PBS and 0.025% trypsin EDTA (no. 25300-054; Life Technologies/Gibco) for 10 min in a 15-cm dish at 37 °C following trypsin inactivation with cold culture medium (DMEM, no.56499C; Sigma) with 10% (vol/vol) FBS (no. F2442; Sigma–Aldrich), 2 mM l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin (all from Life Technologies), and further processing on ice. Cells were dissociated by vigorously pipetting the suspension up and down with a 25-mL serological pipet and a Drummond Pipet-Aid for 2 min, after which the supernatant was filtered through 70-μm mesh (no. 352350; Corning) into 50-mL conical tubes. After adding cold PBS to the remaining embryo parts, the process was repeated. A 10-mL serological pipet was used as tissue pieces were broken up, and the process was continued until only fibrous material and a clear suspension remained. Suspensions were centrifuged at 500 × g for 5 min. Cells derived from two to three embryos were plated on a 15-cm dish (no. 430599; Corning) or cells from one embryo were plated on a 10-cm dish (no. 430293; Corning) or a six-well dish no later than 2 h postmortem. The cells were cultured in culture medium at 37 °C and in room air (20% oxygen) with 5% carbon dioxide. Thus, processed and plated MEFs displayed little cell death as judged by the minimal number of floating cells 1 d after derivation and 80–90% confluency after 2 d.

DT40 Cell Culture.

WT (RCB 1464) (89), Syk-deficient (RCB 1470) (42), Lyn-deficient (RCB 1471) (42), Syk/Lyn double-deficient (RCB1500) (90), SHP1-deficient (RCB1500) (91), SHP2-deficient (RCB 1502) (91), and SHP1/2 double-deficient (RCB 1501) (91) DT40 B-cell lines were directly imported from the RIKEN BioResource Center through the National BioResource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan (www2.brc.riken.jp/lab/cell/s5_list.cgi?key=LIB9). The cells were cultured in RPMI (no. 31800-089; Life Technologies/Gibco) containing 10% (vol/vol) heat-inactivated FBS (no. F2442; Sigma–Aldrich), 50 μM β-mercaptoethanol (no. M3148; Sigma–Aldrich), 1% chicken serum (no. 16110-082; Life Technologies), 4 mM l-glutamine (Life Technologies), 100 units/mL penicillin (Life Technologies), and 100 μg/mL streptomycin (Life Technologies) in 5% CO₂ at 39.5 °C as recommended by RIKEN. All cell lines were of early passage (lot 7 for DT40 cells, lot 2 for all genetically modified DT40 cells) and were kept in the laboratory for no more than five passages before use in experiments.

Cell Stimulation.

B-cell, DT40 B-cell, and MEF stimulation with exogenous ROS and anti-IgM.

For stimulation, B cells were resuspended in 1 mL of plain DMEM at 20 × 10⁶ cells per milliliter and warmed to 37 °C in a water bath. DT40 B cells were resuspended in 1 mL of plain RPMI at 5 × 10⁶ cells per milliliter and warmed to 39.5 C° in a water bath. H₂O₂ (no. H3410, diluted in sterile double-distilled water; Sigma–Aldrich), anti-mouse IgM Fab₂ fragments (no. 115-005-020; Jackson Immunoresearch) and goat anti-chicken IgM μ-chain (no. 603-101-007; Rockland Immunochemicals) were directly added to the cells. The conditions used were as follows in all experiments: for B cells, 1 mM H₂O₂ for 5 min and 50 μg/mL anti-mouse IgM for 3 min; for DT40 B cells, 5 mM H₂O₂ for 5 min and 10 μg/mL anti-chicken IgM for 3 min; with the exception of experiments with phosphatase inhibitors and SHP1/SHP2 deletion, in which B cells were stimulated with 0.5 mM H₂O₂ and DT40 B cells were stimulated with 2 mM H₂O₂ unless otherwise indicated. Immediately after addition of H₂O₂ or anti-IgM, B cells were mixed by inverting the tubes once and incubated with gentle agitation without further inversion until the end of the stimulation period. The stimulation was stopped by centrifuging cells at 500 × g for 1 min, followed by careful aspiration of all supernatant with a fine-needle aspirator, and lysis for 1 h in 100 μL of radioimmunoprecipitation assay (RIPA) buffer [PBS with 1% IGEPAL, 0.1% SDS, 0.5% sodium deoxycholate, and freshly added 1 mM Na₃VO₄, 1.5 mM sodium pyrophosphate, 15 mM sodium fluoride, and 50 mM β-glycerophosphate (all from Sigma–Aldrich), and cOmplete EDTA-free Protease Inhibitor Mixture Tablets (no. 11873580001; Roche)]. To determine cell recovery following H₂O₂ treatment, DT40 B cells were resuspended in complete RPMI medium and cultured overnight.

MEF cell layers with 80–90% confluency were washed once in warm PBS, and warm plain DMEM was added. Diluted H₂O₂ was directly added to the medium, and MEFs were placed back into the incubator for the incubation period. Unless otherwise indicated, all experiments were conducted stimulating MEFs with 5 mM H₂O₂ for 10 min and with 2 mM H₂O₂ in experiments when MEFs were pretreated with phosphatase inhibitors. The stimulation was stopped by washing the cell layer with ice-cold PBS, followed by careful aspiration and addition of 100 μL of RIPA buffer. After 30 min of gently tilting and rotating the plates at 4 °C, the cell layer was scraped off the plates with a cell lifter (no. 3008; Costar), collected, and incubated on ice for an additional 30 min. To determine cell recovery of freshly isolated MEFs following H₂O₂ treatment in plain DMEM, the medium was replaced with complete medium after overnight culture and the cell layers were cultured for the indicated times.

Inhibitors.

All chemical and pharmacological inhibitors were added to B cells, DT40 cells, MEFs, and brown and beige adipocytes after resuspension in or addition of plain DMEM and 30 min before stimulation with H₂O₂ or isoproterenol unless otherwise indicated. Na₃VO₄ (no. P0758; New England Biolabs) was directly added to the cells at a final concentration of 100 μM. Rotenone (no. R8875; Sigma–Aldrich) was freshly reconstituted in DMSO before use and added to the cells at a final concentration of 50 nM unless otherwise indicated. Oligomycin A (no. 75351; Sigma–Aldrich) and carbonyl cyanide m-chlorophenyl hydrazine (no. C2759; Sigma–Aldrich) were reconstituted in DMSO. Dasatinib (no. D-3307; LC Laboratories) was reconstituted in DMSO within 10 min before addition to the cells (B cells: 30 nM, MEFs: 100 nM final concentration) due to observed rapid loss of activity once dissolved in DMSO. R406 (free base aliquots stored at −80 °C, no. S1533; Selleck Chemicals) was reconstituted in DMSO, and B cells were treated with 2 μM R406. MEFs were treated with 5 μM R406, a dose required to lower H₂O₂-induced Syk phosphorylation to levels comparable to B cells treated with 2 μM R406.

Calcium flux.

A total of 5 × 10⁶ DT40 B cells per milliliter were loaded with 1 μM cell-permeant indo-1, AM (acetoxymethyl) (no. I-1223; Invitrogen/Molecular Probes) in RPMI 1640 medium without phenol red containing 5% FBS, 0.01% Pluronic F-127 (no. P6866; Molecular Probes/Life Technologies) in a mildly agitating thermomixer at 30 °C in the dark for 25 min following the manufacturer’s instructions. The cells were washed twice with and then resuspended in Krebs–Ringer solution containing 1 mM CaCl₂ and kept on ice in the dark until measurement and for no longer than 2 h. Following prewarming of the cells to 37 °C for 5 min, Ca²⁺ mobilization was immediately assessed by measuring the 405/485-nm emission ratio of indo-1 and cells stimulated with H₂O₂, anti-IgM, and ionomycin (1 μg/mL, no. I0634; Sigma–Aldrich) as a positive control using an LSRII flow cytometer (BD Biosciences). The data were analyzed with FlowJo software, version 8.7 (FlowJo, LLC).

Mitochondrial ROS production.

MEFs were treated with the indicated concentrations of rotenone in plain DMEM. MitoSOX Red Mitochondrial Superoxide indicator (no. M36008; Molecular Probes/Life Technologies) reconstituted in DMSO was added to the cell layer at a final concentration of 2.5 μM 10 min after rotenone was added. After rotenone treatment for a total of 30 min, the cells were washed with PBS, trypsinized, stained with TO-PRO-3 (no. T3605; Molecular Probes/Life Technologies) for dead cell exclusion, acquired with a Becton Dickinson FACSCalibur flow cytometer, and analyzed with FlowJo software.

Mitochondrial Isolation and Subfractionation.

Isolation of mitochondria from mouse tissue.

Six-week-old mice were starved overnight and killed, and the spleen and liver were removed. Tissues were cut with a scalpel into small pieces and suspended in mitochondrial isolation buffer B [20 mM Hepes/KOH (pH 7.6), 220 mM mannitol, 70 mM sucrose, 1 mM EDTA, 0.5 mM PMSF; 10 mL/g tissue]. Tissue suspensions were homogenized in a glass potter with 40 strokes on ice, and an aliquot was snap-frozen (total cell lysates). Samples were centrifuged at 800 × g for 15 min at 4 °C to remove whole cells and cell debris. Mitochondria were isolated by centrifugation at 7,000 × g for 15 min at 4 °C. The postmitochondrial supernatant [containing endoplasmic reticulum (ER) and cytosol, for example] was snap-frozen, whereas the mitochondrial pellet was resuspended in buffer B, and the centrifugation step was repeated once more, followed by resuspension of the mitochondria in buffer B. The protein concentration was determined by Bradford protein assay. Mitochondria were directly subjected to further analysis.

Subcellular fractionation.

To isolate the microsome (P100) and cytosol (S100), the postmitochondrial supernatant from the 7,000 × g centrifugation step during the mitochondrial isolation outlined above was centrifuged twice at 7,000 × g for 15 min at 4 °C to remove mitochondria. Subsequently, the S100 and P100 fractions were obtained by ultracentrifugation of the postmitochondrial supernatant at 125,000 × g for 1 h at 4 °C. The resulting supernatant (S100) was removed, and the microsome-containing pellet (P100) was resuspended in cold EDTA/MOPS (3-(N-morpholino)propanesulfonic acid) buffer [10 mM MOPS/KOH (pH 7.2), 1 mM EDTA]. Samples were snap-frozen in liquid nitrogen and stored at −80 °C.

Mitochondrial fractionation and protein localization.

To determine the submitochondrial localization of proteins, freshly isolated mitochondria were either resuspended in osmotic buffer (mitochondrial isolation buffer B), hypoosmotic buffer [swelling buffer: 20 mM Hepes/KOH (pH 7.6)] or 0.5% (vol/vol) Triton X-100 buffer B. Samples were incubated for 15 min on ice, followed by addition of PK (recombinant, PCR grade; no. 03115879001; Roche) at a final concentration of 10–25 μg/mL After additional incubation for 15 min on ice, PMSF was added at a final concentration of 2 mM to inactivate PK and the samples were stored on ice for another 10 min. Samples were then precipitated with trichloroacetic acid (TCA), or mitochondria were reisolated by centrifugation at 7,000 × g for 10 min at 4 °C, washed once with buffer, and solubilized with Laemmli buffer.

Carbonate extraction.

For separation of membrane-integrated and soluble proteins, mitochondria were treated with freshly prepared sodium carbonate solution [100 mM (pH 11.5 or 10.8)] for 30 min on ice (92, 93). Samples were subjected to ultracentrifugation at 125,000 × g at 4 °C for 1 h. The obtained supernatant was precipitated with TCA, lysed, and subjected to immunoblotting.

Percoll gradient.

The fractionation of crude mitochondria was performed as described (94). Isolated mitochondria were resuspended in mitochondria resuspension buffer [MRB; 250 mM mannitol, 5 mM Hepes/KOH (pH 7.4), 0.5 mM EGTA] and layered on top of Percoll medium [225 mM mannitol, 25 mM Hepes/KOH (pH 7.4), 1 mM EGTA, 30% (vol/vol) Percoll] in an ultracentrifugation tube. After centrifugation at 90,000 × g at 4 °C for 30 min, purified mitochondria were collected and diluted 10-fold with MRB buffer. Mitochondria were reisolated by centrifugation at 7,000 × g for 10 min at 4 °C and washed with MRB. After a second centrifugation step, the obtained mitochondrial pellet was resuspended in a small volume of MRB and stored at −80 °C.

Immunopurification of B-cell and MEF mitochondria and stimulation.

Stimulation.

For mitochondrial fractionation, 400 × 10⁶ CD43-depleted splenic cells resuspended at 20 × 10⁶ cells per milliliter of DMEM were stimulated with H₂O₂ in a gently agitating 37 °C water bath, followed by immediate centrifugation. Ten subconfluent 15-cm dishes with primary MEFs cultured for 2 d were treated with H₂O₂, followed by aspiration and addition of 4 mL of cold PBS to each plate. The plates were transferred to the cold room, cell layers were carefully scraped off the plates with a cell lifter (no. 3008; Costar), and cell suspensions were collected in 50-mL conical tubes, followed by immediate centrifugation. All subsequent steps were performed at 4 °C in the cold room.

Homogenization.

B-cell and MEF cell suspensions were centrifuged at 500 × g for 5 min at 4 °C and resuspended in 1 mL of ice-cold separation buffer [PBS with 0.5% fatty acid-free BSA, 2 mM EDTA (pH 7.2); 10× stock provided as part of the mouse mitochondrial isolation kit from Miltenyi Biotec (no. 130-096-946)] with protease inhibitors (cOmplete EDTA-free Protease Inhibitor Mixture Tablets, no.11873580001; Roche) and freshly added phosphatase inhibitors (1 mM Na₃VO₄, 1.5 mM sodium pyrophosphate, 15 mM sodium fluoride, 50 mM β-glycerophosphate; all from Sigma–Aldrich). For preparation of whole-cell lysates, 50 μL of B-cell or MEF cell suspension was transferred to a 1.5-mL tube, 1.4 mL of PBS was added, and the cell suspension was centrifuged at 500 × g for 5 min. The cell pellets were resuspended in 100 μL of RIPA buffer.

B-cell suspensions were transferred to a glass vessel and homogenized with 40 slow strokes of a fitting glass Dounce homogenizer (2-mL vessel with pestle B, no. ab110171; Abcam) exerting moderate force against the vessel wall. The cell suspension was transferred to a 1.5-mL tube and centrifuged at 500 × g for 5 min. The slightly turbid supernatant was transferred to a fresh 1.5-mL tube and centrifuged again at 500 × g for 5 min. The supernatant was then transferred to a 15-mL tube. The pellet remaining after the first centrifugation was resuspended in 1 mL of separation buffer, and the homogenization procedure was repeated. Due to their larger size, MEF cell suspensions were transferred to a glass vessel and homogenized with 40 slow strokes of a Teflon-coated pestle with slightly greater clearance (nos. 89026-386 and 89026-398; VWR), again exerting moderate force against the vessel wall. Cell suspensions were centrifuged once at 300 × g for 5 min. The turbid supernatant was separated into two 15-mL tubes, and the process was repeated with the remaining pellet.

Tom22 labeling and magnetic activated cell sorting (MACS) purification.

Separation buffer was added to the tubes with B-cell and MEF mitochondria to result in a total volume of 10 mL, and 50 μL of anti-mouse Tom22-labeled magnetic beads (no. 130-096-946; Miltenyi Biotec) was added to the suspension and slowly rotated at 4 °C for 1 h. Tom 22-labeled mitochondria were positively selected following the instructions of the manufacturer (no. 130-096-946; Miltenyi Biotec) using separation buffer with protease and phosphatase inhibitors. However, the columns were washed with PBS instead of separation buffer in the third wash step and eluted with 1 mL (B cells) and 0.5 mL (MEFs) of PBS to remove the protein-containing separation buffer. The eluate was transferred into a 1.5-mL tube and centrifuged for 30 min (B cells) and 15 min (MEFs) at 13,000 × g. The supernatant was carefully aspirated with a fine-needle aspirator, and the pellets were resuspended in 80 μL of RIPA buffer. After 30 min of lysis on ice, lysates were centrifuged twice for 10 min at 13,000 × g to remove debris and magnetic beads. Measurement of the protein concentration (DC Protein Assay Kit II; Bio-Rad) showed that the protein yield with this procedure approximated 30–50 μg for B-cell mitochondria and 80–100 μg for MEF mitochondria.

RNA Analysis.

Microarrays.

Normal tissue RNA expression arrays (HG-U133 Plus 2; Affymetrix) were collected from the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/geo/) (95), focusing on the human body index (HBI) dataset (GSE7307) and other normal samples. Low-quality arrays were removed using the affyPLM package from BioConductor (www.bioconductor.org). The collected data were preprocessed using the robust multiarray average algorithm of the affy package from Bioconductor. A custom probe-set definition was used for processing the arrays (96), such that each Entrez Gene ID was represented by one probe set. For analysis of coexpression, only HBI datasets (GSE7307) categorized as “normal” were used. Pearson correlation coefficients and their P values were calculated using Prism 6 (GraphPad Software, Inc.). The processed data sets can be queried at merav.wi.mit.edu/SearchByGenes.html.

mRNA sequencing.

Illumina human body map mRNA reads aligned to the human reference genome (Ensembl annotation GRCh37) were downloaded from Array Express (E-MTAB-513 at www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-513/, accessed September 2012) and quantified using Cufflinks (v2.0.2 at cole-trapnell-lab.github.io/cufflinks/ with Ensembl gene annotation GRCh37.67). Fragments per kilobase of transcript per million mapped reads were plotted, and Pearson correlation coefficients and their P values were calculated using Prism 6.

Quantitative PCR.

Total RNA was isolated from tissues or cells using TRizol reagent (Life Technologies/Ambion) and an RNeasy kit (Qiagen). A total of 300 ng was reverse-transcribed using SuperScript II reverse transcriptase (Life Technologies/Invitrogen) utilizing random hexamers (Life Technologies/Invitrogen). The cDNA was diluted 1:10, and 2.5 μL for a 96-well plate or 1 μL for a 384-well was used for real-time quantitative PCR (qPCR) analysis. qPCR was carried out on an ABI7900HT Fast real-time PCR system (Applied Biosystems) and analyzed using the delta delta cycle threshold (ΔΔCt) method normalized to 18S. RNA from various human and mouse tissues was purchased from Clontech Laboratories (nos. 636644 and 636643).

Protein Analysis and Microscopy.

Immunoblotting.

Lysates were centrifuged for 10 min at 13,000 × g to remove debris, and NuPAGE sample buffer and reducing buffer (Life Technologies) were added after measuring and adjusting the samples for protein concentration (DC Protein Assay Kit II). A total of 2–20 μg of lysate per sample was separated for 2–4 h at 60–100 V using 8% 26-well NuPAGE Bis-Tris Midi gels in MOPS or MES buffer (Life Technologies) and in Criterion cells (Bio-Rad) using respective adapters (Life Technologies). Protein was wet-transferred on PVDF membranes (Immobilon P; Millipore) in Criterion blotter cells (Bio-Rad) using 2× NuPAGE Transfer buffer (Life Technologies) with 10% methanol for 25 min at 1 A. After blocking the membrane in filtered (no. 595-4520; Nalgene) TBS-T (50 mM Trisbase, 150 mM NaCl, 0.1% Tween 20; no. P1379; Sigma–Aldrich) with 3% BSA (no. A7906; Sigma–Aldrich) for 1 h, blots were incubated in primary antibody diluted in TBS-T with 3% BSA sealed in hybridization bags and gently shaken at 4 °C overnight; they were then washed three times for 10 min in TBS-T, incubated in secondary antibody (Cell Signaling Technologies) diluted in TBS-T with 3% BSA with gentle shaking for 1 h at room temperature, and washed again three times for 10 min in TBS-T. Antibody binding was visualized using ECL Plus Western Lightning reagent (NEL 102; PerkinElmer), and blots were exposed to film (Kodak BioMax MR Film, no. 8701302; Carestream Health, Inc.). Films were scanned without adjustments using an Epson scanner. For immunoblotting of mitochondrial subfractions, samples were solubilized with Laemmli buffer and subjected to SDS/PAGE following established protocols.

Phosphoproteomics.

A label-free quantitative liquid chromatography (LC)-tandem MS (MS/MS) analysis of phosphorylated Tyr-enriched tryptic peptides from lysates of H₂O₂-treated Syk-deficient DT40 cells and controls was performed using an LTQ-Orbitrap-ELITE mass spectrometer (Thermoscientific), electrospray ionization (ESI)–collision-induced dissociation (CID), and SEQUEST search results similar to previously established protocols (97, 98), with the following modifications.

Sample preparation.

Four aliquots of 500 × 10⁶ cells per sample at a concentration of 5 × 10⁶ cells per milliliter of serum-free DMEM were treated with 5 mM H₂O₂ for 5 min and centrifuged at 500 × g for 10 min, followed by flash-freezing of the pellets in liquid nitrogen.

Phosphorylated tyrosine enrichment.

Pellets were lysed in urea lysis buffer, and tryptic peptides were generated as previously described (98). Total protein for each sample was normalized before trypsin digestion, and 15 mg of protein from each sample was used. The peptides were immunoprecipitated with immobilized phospho-Tyr motif antibody pY-1000 (no. 8954; Cell Signaling Technology).

MS and analysis.

The samples were analyzed in duplicate by LTQ-Orbitrap ELITE LC-MS/MS ESI-CID. Peptides were loaded directly onto a 10-cm × 75-μm PicoFrit capillary column (New Objective) packed with Magic C18 AQ reversed-phase resin (New Objective). The column was developed with a 90-min linear gradient of acetonitrile in 0.125% formic acid delivered at 280 nL⋅min⁻¹. The following MS parameter settings were chosen: MS run time =120 min, MS1 scan range (300.0–1,500.00), top 20 MS/MS (minimum signal = 500, isolation width = 2.0, normalized collision energy = 35.0, activation-Q = 0.250, activation time = 20.0, lock mass = 371.101237, charge state rejection enabled, charge state 1⁺ rejected, dynamic exclusion enabled, repeat count = 1, repeat duration = 35.0, exclusion list size = 500, exclusion duration = 40.0, exclusion mass width relative to mass, exclusion mass width = 10 ppm).

MS/MS spectra were evaluated using SEQUEST (99) and the Core platform from Harvard University (100). Peptide searches were performed against the most recent update of the National Center for Biotechnology Information (NCBI) Gallus gallus database with a mass accuracy of ±50 ppm for precursor ions and 1 Da for product ions. Results were filtered with a mass accuracy of ±5 ppm on precursor ions and presence of the intended motif. A 5% default false-positive rate was used to filter the SEQUEST results.

Label-free quantitation of phosphorylated peptide intensities was performed based on the peptide intensities measured by MS. Based on previous validation experiments (97, 101), 2.5-fold increases or decreases in phosphorylated peptide intensity were interpreted as Syk-regulated phosphorylation sites. Aliquots of the phosphoproteomics experiments were analyzed by immunoblotting for selected Syk targets as internal controls.

The chicken genes were mapped to their human homologs as judged by their chicken gene and protein names (www.genenames.org/cgi-bin/symbol_checker). Network and enrichment analysis of Syk-regulated genes was performed using algorithms displaying network interactions with medium confidence and databases integrated with the string database (www.stringdb.org, accessed January 2015) (87). Genes were manually grouped into pathways based on their string annotation.

Immunofluorescence staining.

Five hundred microliters of CD43-depleted splenic cells resuspended at 1 × 10⁶ cells per milliliter of PBS and 2% FBS was distributed on coverslips (no. 12-545-80; Fisher Scientific) previously coated with 10% poly-l-Lys (no. P8920; Sigma–Aldrich) PBS for 10 min in a 24-well plate (Corning) and centrifuged at 700 × g for 5 min. After washing with warm PBS and the addition of prewarmed plain DMEM, B cells were treated with H₂O₂ and anti-IgM at 37 °C and 5% CO₂. The stimulation was stopped by aspirating the medium, adding 4% paraformaldehyde (no. 15710; Electron Microscopy Sciences) in PBS and incubation for 20 min. Fixation and all subsequent steps were performed at room temperature, as well as in the dark as much as possible. For permeabilization, the slides were incubated in 0.25% Triton X-100 (no. T8787; Sigma–Aldrich) in PBS for 10 min, followed by three careful wash steps for 5 min with PBS. The slides were incubated in primary antibody diluted in filtered (no. 595-4520; Nalgene) PBS with 0.5% BSA for 1 h, washed three times for 5 min with PBS, and subsequently incubated with secondary antibody for 1 h and counterstained with DAPI (Life Technologies), followed by three washes with PBS for 5 min. Coverslips with B cells were mounted on slides (no. 48311-703; VWR) using small amounts of mounting medium (Prolong Gold antifade reagent, no. P36930; Life Technologies) and without exerting pressure on the coverslips. The slides were stored flat in the dark at 4 °C until acquisition.

Tissue immunofluorescence and immunohistochemistry.

Immunofluorescence of formalin-fixed, paraffin-embedded sections was performed following baking at 60 °C for 1 h, deparaffinization, and rehydration (100% xylene four times for 3 min each time, 100% ethanol four times for 3 min each time), as well as blocking of endogenous peroxidase activity with 3% H₂O₂ in ethanol for 15 min, followed by washing of the slides for 5 min with tap water. Subsequently, heat-mediated antigen retrieval was achieved by treatment in 1 mM EDTA buffer (pH 8.0) in a digital Pressure Cooker (Pacific Southwest Lab Equipment, Inc.).

For tissue immunofluorescence, the sections were incubated in primary antibody in PBS for 1 h, washed in PBS, and incubated in secondary antibody for 45 min, all performed in a humid chamber at room temperature, and then counterstained with DAPI and coverslipped.

For Syk immunohistochemistry, nonspecific binding of biotin and other antigens was blocked using a biotin blocking system (Dako), followed by blocking with 1.5% goat serum in PBS for 15 min. The sections were incubated in primary antibody in 1% BSA/PBS for 1 h in a humid chamber at room temperature and visualized with a Vectastain Elite ABC kit (Vector Laboratories, Inc.) with DAB (Dako) as a substrate and counterstained with Gill’s hematoxylin (Fisher Scientific).

Antibodies.

All antibodies used in immunoblotting, immunofluorescence, and immunohistochemistry experiments of this study are listed in Table S6.

Light microscopy.

Cell cultures were imaged using a phase contrast microscope (Eclipse TS100; Nikon) connected to a camera (model 2.1.1; SPOT Imaging Solutions, Diagnostic Instruments) and operated by SPOT 4.7 software. A bright-field Zeiss AxioPlan2 upright microscope connected to a CCD camera with QCapture software was used to image all tissue sections stained with immunohistochemical stains.

Transmission EM.

B cells and MEFs were fixed in 2.5% glutaraldehyde and 3% paraformaldehyde with 5% sucrose in 0.1 M sodium cacodylate buffer (pH 7.4), centrifuged at 500 × g for 5 min, and postfixed in 1% osmium tetroxide in veronal-acetate buffer. The cell pellet was stained en bloc overnight with 0.5% uranyl acetate in veronal-acetate buffer (pH 6.0) and then dehydrated and embedded in Embed-812 resin (Electron Microscopy Services). Sections were cut on a Reichert Ultracut E microtome with a Diatome diamond knife at a thickness setting of 50 nm and stained with uranyl acetate and lead citrate. The sections were examined using an FEI Tecnai Spirit at 80 kV and photographed with an Advanced Microscopy Techniques CCD camera.

Confocal imaging.

Confocal images were acquired using a Zeiss 710 NLO laser scanning confocal microscope on a Zeiss Axio Observer inverted microscope stand and a 63× magnification/1.4 N.A. Plan Apochromat oil immersion objective. All images were acquired in sequential acquisition mode, with 4× averaging, 2,048 × 2,048 pixels, and 8 bits per channel using Zen 2010 software (Zeiss). Alexa Fluor 488 fluorophores were excited with a 488-nm laser (emission: 498–577 nm), Alexa Fluor 568 fluorophores were excited with a 561-nm laser (emission: 568–645 nm), Alexa Fluor 633 fluorophores were excited with a 633-nm laser (emission: 638–755 nm), and DAPI was excited with a 405-nm laser (emission: 410–585 nm). Using a pinhole of 1 Airy unit, a single 488/561/633 dichroic beam splitter was used for both channels. Images of Alexa Fluor 488 fluorophores were pseudocolored green, images of Alexa Fluor 568 fluorophores were pseudocolored red, images of Alexa 633 fluorophores were pseudocolored magenta, and DAPI was pseudocolored blue. All images were processed with ImageJ 1.48S software (NIH) using the smoothening algorithm and then displayed with a linear range from the minimum to maximum signal intensity within each experiment for each channel. Adjusted green and magenta images were merged, and areas of similar signal intensity in each channel appear white. Differential interference contrast images were adjusted individually due to the variable intensity of the white light source across the slide.

Evolutionary Conservation.

Phylogenetic tree.

Syk orthologs were identified by searching for proteins and predicted proteins similar to human Syk (NP_003168.2) and the Drososphila melanogaster Syk ortholog SHARK (NP_524743.2) using the protein protein BLAST algorithm focused on the NCBI Reference Sequence protein database [BLASTP 2.2.26 (102, 103) at blast.ncbi.nlm.nih.gov/Blast.cgi, accessed March 2012 and revised June 2014]. Proteins annotated by the NCBI protein database (www.ncbi.nlm.nih.gov/protein) to possess tandem SH2 domains with (SHARK) or without (SYK, ZAP-70) interspaced ankyrin repeats and a protein tyrosine kinase domain were included. The results were further expanded by searching for Syk-like molecules in representative species using BLASTP and compared with other public databases such as GeneCards (www.genecards.org), HomoloGene (www.ncbi.nlm.nih.gov/homologene), Pfam (pfam.xfam.org), and the UCSC Genome Browser (genome.ucsc.edu), as well as with the published literature (reviewed in 46). The largest predicted isoform per protein (Syk, ZAP-70, or SHARK) and species were included. The protein sequences were aligned using MUSCLE (www.drive5.com/muscle/) (104) with default parameters and viewed in Jalview (105). A majority rule consensus tree based on maximum likelihood analysis was constructed by bootstrapping 100 replicates to assess the support of a given branch using PHYLIP (Phylogeny Inference Package, version 3.69, at evolution.genetics.washington.edu/phylip.html) and viewed with FigTree version 1.4.1 software (tree.bio.ed.ac.uk/software/figtree/).

Gene age estimate.

To estimate the evolutionary origin of the Syk pathway, we approximated the age of a gene to be the time of divergence between humans and the species in which the gene likely first occurred (106). First, all species with orthologs of Syk pathway members were downloaded from the GeneCards website (www.genecards.org, accessed November 2013), a curated database retrieving information from multiple public databases, and parsed using a custom script. To address the possibility that the Syk pathway is present in the sponge A. queenslandica, which was not included in the GeneCards database, we determined whether orthologs of Syk pathway members are predicted to exist in this sponge using the protein protein BLAST algorithm (BLASTP 2.2.26). The time of divergence of a species from the human species was retrieved from TimeTree2 (www.timetree.org), a database of reported molecular clock estimates (107, 108). Gene age estimates were plotted using Prism 6 and Adobe Illustrator (Adobe Systems, Inc.).

Literature Annotation.

A total of 3,078 PubMed entries containing the search term “Syk” and peer-reviewed reviews on Syk and ROS signaling (www.ncbi.nlm.nih.gov/pubmed, accessed May 2015) were screened for evidence of Syk function in primary cells, cell lines, and cancer tissue of a given cell type, and to identify stimuli that induce signaling and cellular responses in a Syk-dependent manner. Search results were validated and expanded by accessing the relevant full article versions through the Countway Library of Medicine at Harvard Medical School, as well as searching open access full-text articles using Google search engines.

Human Genetic Variation.

Variants with minor allele frequencies of less than −0.1% were extracted from the release 0.3 sites VCF (variant call format) from the Exome Aggregation Consortium (ftp://ftp.broadinstitute.org/pub/ExAC_release/release0.3/, accessed February 2015; n = 60,706 exomes). The canonical transcript, as defined by Ensembl version 75 and GENCODE version 19, of a gene was used to define the functional consequence (synonymous, missense, nonsense) for each variant, as well as the coding length of the gene. Immune genes were selected based on their inclusion in the Kegg pathways “IgA” and “Antigen processing” after removal of genes with known ubiquitous function as judged by their UniProt annotation. Ratios of missense variants to synonymous variants of groups of genes were compared by a two-tailed Mann–Whitney test using Prism 6.

Statistical Methods.

All statistical analyses were performed using Prism 6. Statistical significance was indicated as follows: *P < 0.05; **P < 0.005; ***P < 0.0005.