Bradykinin promotes neuron-generating division of neural progenitor cells through ERK activation

During nervous system development, neural stem and progenitor cells (NPCs) perform the task of proliferating, migrating, and differentiating in highly accurate patterns. Genetic programs and specific factors in the microniches control such accuracy. These factors influence proliferative and neuron-generating cell divisions, and differentiation programs (Calegari et al., 2005; Lako et al., 2009; Lukaszewicz et al., 2002). Proliferative division is rapidly completed, given that it has an abbreviated G₁-phase, and generates two identical precursor cells. By contrast, neuron-generating division generates two daughter cells, where one or both leave the cell cycle and undergo differentiation. Recent in vitro and in vivo studies have demonstrated that NPCs with a longer G₁-phase of the cell cycle perform a neuron-generating division (Calegari et al., 2005; Calegari and Huttner, 2003; Lako et al., 2009; Lukaszewicz et al., 2002; Neganova et al., 2009; Orford and Scadden, 2008). Similarly, expression of the gene TIS21 (also known as PC3 and BTG2) at the onset of neurogenesis identifies cells switching from proliferative to neuron-generating division. Expression of this gene, in the same way as elongation of the G₁-phase, serves as a marker for neuron-generating division (Calegari and Huttner, 2003; Iacopetti et al., 1999).

NPCs can be obtained from the telencephalon of mice embryos, and used to reproduce in vitro events of cortical development in vivo, such as proliferation, migration, neurogenesis and gliogenesis. As observed in vivo (Tropepe et al., 1999), fibroblast growth factor-2 (FGF-2) and epidermal growth factor (EGF) also induce proliferative division of NPCs in vitro, maintaining cells as multipotent and leading to the formation of neurospheres (Lukaszewicz et al., 2002; Reynolds and Weiss, 1996). After growth factor removal and cell adhesion, NPCs differentiate spontaneously (reviewed by Trujillo et al., 2009).

Distinct dynamics of ERK1 and ERK2 (ERK1/2; also known as MAPK3 and MAPK1, respectively; hereafter referred to as ERK) activation are believed to be the underlying cause of differences in cellular responses, such as proliferation or differentiation, as described for PC12 cells treated with growth factors (Marshall, 1995; Santos et al., 2007; Sasagawa et al., 2005; Traverse et al., 1992; Vaudry et al., 2002; von Kriegsheim et al., 2009). For example, whereas EGF stimulation causes transient ERK activation in the cytoplasm and results in proliferation, nerve growth factor (NGF) induces sustained ERK activation and nuclear translocation, resulting in neuronal differentiation. To achieve the latter sustained ERK activation, several signaling pathways including those mediated by phosphoinositide 3-kinase (PI3K), Ras, Ca²⁺, protein kinase C (PKC) and cAMP might be required. Once attained, sustained ERK activation followed by nuclear translocation is sufficient for induction of neurogenesis.

Recent studies have demonstrated a new role of bradykinin in promoting cell migration and neurogenesis during in vitro differentiation of pluripotent cells and embryonic NPCs (Martins et al., 2005; Pillat et al., 2015; Trujillo et al., 2012). The opposite occurs in NPCs treated with the kinin-B2 receptor (B2BkR, also known as BDKRB2) antagonist HOE-140 and also in NPCs from embryos of B2BkR^−/− mice, i.e. decreased cell migration and neurogenesis. B2BkR^−/− mice have decreased neuronal marker expression in several stages of development, suggesting the involvement of bradykinin in neuronal phenotype determination in vivo (Trujillo et al., 2012). However, the mechanisms triggered by bradykinin to coordinate migration, proliferation and neuronal differentiation remain unknown. Here, we provide mechanisms for neuron-generating divisions in NPCs and in hippocampus from newborn mice, and delineate the intracellular signaling pathways that might serve as key determinants of bradykinin-induced effects in the cell cycle followed by neuronal differentiation.

EGF and FGF-2 induce proliferative division of NPCs, keeping them in their undifferentiated state (Lukaszewicz et al., 2002; Reynolds and Weiss, 1992, 1996). In this context, we evaluated the effect of bradykinin on proliferation of undifferentiated cells stimulated by growth factors. EGF and FGF-2 (both at 20 ng/ml final concentrations) were added to the culture medium 1 h prior to bradykinin (1 μM) treatment and cells were analyzed after 24 h in this medium (Fig. 1A,B). The presence of bradykinin resulted in a significant reduction in proliferation in comparison with samples treated with growth factors only, as observed by their significantly lower BrdU incorporation after 2 h of treatment (P=0.0126). We sought to identify in which phase of the cell cycle NPCs stimulated with bradykinin in the presence of growth factors are primarily found. For this purpose, cells were double-labeled for BrdU (after 30 min of treatment) plus propidium iodide (Fig. 1C) and Ki67 plus propidium iodide (Fig. 1D). The addition of bradykinin to cells pretreated with growth factors resulted in an accumulation of cells in G₁-phase, compared to samples treated only with growth factors (Fig. 1E; P=0.0217).

It is important to stress that NPCs undergoing neuron-generating division have a longer G₁-phase than those performing proliferative division (which have an abbreviated G₁ phase) (Calegari et al., 2005; Orford and Scadden, 2008). Thus, given that the cell cycle length predicts cell fate, we assessed whether this neuron-generating ability in response to bradykinin signaling was accompanied by an increase in the length of the G₁-phase of the cell cycle. For this purpose two strategies were used for analysis, which are briefly described in the Materials and Methods section (Lukaszewicz et al., 2002; Nowakowski et al., 1989; Storey, 1989; Wilcock et al., 2007). Cell cycle and G₁-phase lengths of NPCs exposed to the growth factors EGF and FGF-2 and bradykinin were markedly increased compared to cell cycle and G₁-phase length of NPCs exposed to growth factors alone (Table 1), as revealed by flow cytometry experiments. These data demonstrate that proliferating NPCs in the presence of bradykinin adopted a slower cell cycle progression (longer G₁-phase), which is characteristic for cells that generate neurons.

For delineating intracellular signaling pathways that might serve as key determinants of bradykinin-induced effects in cell cycle followed by neuronal differentiation in NPCs, we examined the activities of various intracellular signaling pathways. A significant increase in intracellular cAMP concentration ([cAMP]_i) was observed after 1 and 2 min of exposure to bradykinin with a subsequent decay to baseline (Fig. 2A; P=0.012). The B2BkR can also catalyze the exchange of GTP for GDP on the α subunits of G_i and G_s proteins (Blaukat et al., 2000; Hall et al., 1993; Hanke et al., 2006; Liebmann et al., 1996, 1990), even though it has been often described as a prototypical G_q-protein-coupled receptor. Thus, we assessed if the B2BkR could also be coupled to any member of the G_s or G_i protein family in NPCs or if the G_q protein and its downstream pathway could affect bradykinin-stimulated cAMP production. Cells were treated with cholera toxin (0.5 μg/ml; 16 h) to activate all G_s proteins, with the G_q protein inhibitor YM254890 (10 μM; 15 min), with the PLC-β inhibitor U73122 (10 μM; 30 min), or with the G_i protein inhibitor pertussis toxin (200 ng/ml; 16 h). Samples were then stimulated or not for 2 min with bradykinin (1 μM), and changes in [cAMP]_i levels were determined. Pre-treatment with only pertussis toxin prevented the bradykinin-stimulated [cAMP]_i increase (Fig. 2B). In this context, it is known that βγ subunits of the G_i protein can activate adenylate cyclase II, IV and VII, and that βγ subunits are sensitive to pertussis toxin (Chen et al., 1995). Therefore, our data suggest that the B2BkR is also coupled to G_i proteins in NPCs, and that their βγ subunits might induce activation of adenylate cyclase and cAMP production.

Previous studies have shown that bradykinin induces an elevation in the concentration of intracellular free Ca²⁺ ([Ca²⁺]_i) by activating B2BkR in four models of neural differentiation (Appell and Barefoot, 1989; Martins et al., 2008, 2005; Nascimento et al., 2015; Trujillo et al., 2012). Thus, we aimed to find the intracellular signaling pathways that mediate the bradykinin-induced [Ca²⁺]_i increase. For this purpose, NPCs were pre-treated with YM254890, pertussis toxin, or YM254890 plus pertussis toxin. Surprisingly, treatments with YM254890 or pertussis toxin only partially inhibited the response of NPCs to bradykinin from 5.27±0.49 to 3.28±0.39 (37.7% of suppression) or 3.73±0.12 (29.2% of suppression), respectively (mean±s.e.m.; Fig. 2C,D). By contrast, simultaneous inhibition of both G_q and G_i proteins (YM254890 plus pertussis toxin) blocked the bradykinin-induced [Ca²⁺]_i increase (0.33±0.19; 93.7% of suppression), suggesting that both types of G proteins are involved in B2BkR-mediated responses, and that co-activation of G_q and G_i proteins exerts synergistic effects on bradykinin-induced [Ca²⁺]_i increases (P=0.0005), such as observed in experiments conducted with platelets (Shah et al., 1999). Subsequently, we provided evidence that the bradykinin-induced [Ca²⁺]_i increase occurs by mobilization of intracellular Ca²⁺ stores. Pre-treatment of NPCs for 1 h with thapsigargin, an inhibitor of the sarco- and endoplasmic reticulum Ca²⁺ ATPase (SERCA), completely blocked bradykinin-induced [Ca²⁺]_i increases (Fig. 2C,D; P=0.0002).

To determine which other signaling pathways might serve as key determinant of bradykinin effects in neural differentiation of NPCs, intracellular signal-transducing kinase activities were examined. For this, cells initially starved for 6 h were stimulated or not for 10 min with bradykinin (1 μM), followed by detection of phosphorylated (p)-STAT3, JNK family proteins (p-JNK), Akt family proteins (p-Akt) and p38 MAPK family proteins (p-p38) by flow cytometry (Fig. 2E,F). Bradykinin-treated samples did not show any significant differences in levels of phosphorylated JNK and STAT3, compared to untreated samples. In contrast, higher levels of p-p38 (P=0.0446) and p-Akt (P=0.0071) were observed after bradykinin stimulation. It is known that several signaling pathways, such as the PI3K–Akt, PKC, Ca²⁺, p38 MAPK and cAMP, can regulate ERK activation or integrate their signals in this signaling cascade (Blaukat et al., 2000; Bouschet et al., 2003; Murray, 1998; Westermarck et al., 2001). Based on this fact and known effects of ERK in controlling processes such as proliferation and neuronal differentiation, the influence of bradykinin on this MAPK was further studied (Fig. 2G,H). Phorbol 12-myristate 13-acetate (PMA) and inhibitors of MEK–ERK pathway U0126 and PD98059 were used as positive and negative controls, respectively, for ERK phosphorylation. Bradykinin increased the level of p-ERK expression when compared to control samples in neurospheres from wild-type (WT) mice (P=0.0001). PMA induced ERK phosphorylation, whereas bradykinin did not provoke any effect on ERK phosphorylation in neurospheres from B2BkR-knockout mice, confirming that bradykinin exerts its effects specifically through B2BkR stimulation.

The activation and regulation of ERK by one or more pathways result in distinct kinetics and cell localization, which in turn cause different cell responses. Namely, in PC12 cells, ERK induces either proliferation when activated transiently, or induces neuronal differentiation and suppression of proliferation when it remains activated for a prolonged time and translocates to the nucleus (Impey et al., 1999; Marshall, 1995; Santos et al., 2007; Sasagawa et al., 2005; Traverse et al., 1992; von Kriegsheim et al., 2009). In order to assess how long ERK activity is sustained, untreated NPCs (control) and NPCs treated with bradykinin (1 μM) for 2, 5, 10, 15, 30 and 60 min were analyzed by flow cytometry using an anti-p-ERK antibody. Bradykinin activated ERK for at least 60 min (Fig. 3A,B). NPCs starved for 6 h were kept for 15 min in the absence or presence of bradykinin (1 μM), fixed, and stained for p-ERK and the cytoplasmatic protein nestin. As observed in the representative images and bar plots (Fig. 3C,D), the proportion of cells expressing detectable levels of p-ERK in the nucleus increased significantly in bradykinin-treated samples compared to controls (P=0.0187). In agreement with results obtained in NPCs, bradykinin treatment induced a sustained ERK activation in PC12 cells, as revealed by flow cytometry experiments (Fig. 3A,B). Sublocalizations of ERK after bradykinin stimulation in PC12 cells starved for 6 h were further studied by using a protocol to separate nuclear and cytoplasmic protein extracts followed by western blotting assays. Increased p-ERK amounts in the nuclear extract fraction after bradykinin stimulation were observed, when compared to untreated cells (Fig. 3E,F; P=0.0336). No significant differences were observed in the ratio between p-ERK and total ERK concentrations in the cytoplasmic extract. NGF stimulation (positive control), as expected, resulted in an increase of p-ERK levels in the nucleus. In Fig. 3E, controls of separation efficiency confirmed the nuclear and cytoplasmic localization of histone H1 and β-actin protein, respectively.

As already published in earlier studies of our group, bradykinin treatment favors neural migration after differentiation of mouse and rat NPCs (Pillat et al., 2015; Trujillo et al., 2012). Neural cells present a radial migration pattern from the edge of neurospheres and the region limited by dotted lines comprised ∼95% of migrating cells. Here, we evaluated the dose–response curve of bradykinin (1 nM, 10 nM, 0.1 μM, 1 μM and 10 μM), and the involvement of PI3K–Akt, ERK and p38 MAPK pathways in bradykinin-induced neural migration. None of the pharmacological inhibitors of ERK (PD98059 20 μM; 10 μM U0126), p38 (10 μM SB203580) and PI3K–Akt (20 μM Ly294002) had any effect on cell viability (Fig. S1) and Akt activation by bradykinin occurred through PI3K, but ERK activation did not involve PI3K stimulation (Fig. S1). Although a dose of 0.1 µM of bradykinin promoted migration, because 1 µM had the maximum effect on neural migration, we used this concentration in the subsequent assay (Fig. 4A; Fig. S2). Pre-treatments with PD98059, U0126 and Ly294002 did not prevent the increase in bradykinin-induced migration. Nevertheless, inhibition of p38 MAPK by SB203580 resulted in a substantially lower number of migrating cells, and bradykinin had no effect on migration when p38 MAPK was inhibited. Cells were also treated with SB203580 plus EGF (20 ng/ml) as a positive control of the experiment. Taken together, these data suggest that bradykinin-induced cell migration during differentiation depends on p38 MAPK activation, but does not depend on the ERK and PI3K–Akt pathways.

Signaling pathways triggered by bradykinin that involved suppression of proliferative division stimulated by growth factors (FGF2 plus EGF) were investigated, by using the strategy of Dixon et al. (2002). Briefly, neurospheres were pre-treated for 30 min with EGF and FGF2 (20 ng/ml each) and then treated for 30 min with ERK (PD98059 and U0126), p38 (SB203580) and PI3K (Ly294002) inhibitors. Finally, bradykinin (1 μM) was maintained for 24 h in the culture medium. The pharmacological inhibition of ERK and p38 routes completely blocked the effect of bradykinin on suppressing growth-factor-stimulated proliferation, as shown with propidum iodide staining and quantified by flow cytometry (Fig. 4B). These results suggest that bradykinin suppressed the proliferative division of NPCs through activation of MAPKs (ERK and p38) cascades, but did not depend on the PI3K–Akt pathway (P=0.0487).

It is known that bradykinin treatment results in an increase of β3-tubulin and a decrease of GFAP-positive cells (Martins et al., 2008; Pillat et al., 2015; Trujillo et al., 2012). Here, we determined expression levels of further neural markers using flow cytometry techniques (Fig. 5A,B) and qualitative immunofluorescence studies (Fig. 5C,D), and investigated bradykinin-activated signaling pathways that function in fostering NPC neurogenesis. Chronic treatment with 1 µM bradykinin throughout the course of differentiation decreased the frequency of S100β-positive cells (a glial-cell-specific protein, in agreement with a previous work of our group, Trujillo et al., 2012), whereas the frequency of MAP2-positive cells (neuron-specific protein) was significantly increased (Fig. 5A,B). These effects were confirmed by microscopic analysis of immunostained cells (Fig. 5C,D). Now, we show, by flow cytometry analysis, that bradykinin completely lost its effect to promote neurogenesis when ERK signaling was blocked by the presence of 20 μM PD98059 or 10 μM U0126 (Fig. 5E). Furthermore, under these conditions gliogenesis was promoted and neurogenesis reduced compared to the untreated control differentiation (Fig. 5E), such as observed in experiments conducted in the presence of B2BkR antagonists (Trujillo et al., 2012). The importance of ERK in the mechanisms of neurogenesis imposed by bradykinin was confirmed by microscopic analysis of immunostained cells (Fig. 5F,G).

In this context, it is well known that Hes1 is a repressor-type basic helix-loop-helix (bHLH) gene in neurogenesis, given that it inhibits neurogenic bHLH transcription factors, such as MASH1 (also known as Ascl1) and Ngn2 (neurogenin 2, also known as Neurog2) (Kageyama et al., 2008). Thus, it is important to stress that Hes1 inactivation leads to acceleration of neurogenesis (Imayoshi et al., 2008). Expression of Hes1 mRNA was downregulated in NPCs in response to bradykinin stimulation (Fig. 5I). Interestingly, we found that bradykinin completely lost its effect on Hes1 expression under conditions of ERK inhibition. Moreover, expression of the neurogenic factor Ngn2 was significantly upregulated in response to bradykinin treatment, whereas pre-treatment with U0126 prevented this bradykinin-promoted effect (Fig. 5I). Taken together, these results confirm the importance of ERK in the mechanisms of neurogenesis imposed by bradykinin.

A further question was whether bradykinin exerts its effects in modulating differentiation through the PI3K–Akt or p38 MAPK pathways. Immunofluorescence and flow cytometry data shown in Fig. 5E and H, respectively, revealed that, even when PI3K–Akt is inhibited, bradykinin induced an increase in the percentage of cells expressing β3-tubulin and a decrease in the percentage of cells expressing GFAP at the end of differentiation (day 7). These data suggest that bradykinin acts on NPC differentiation independently from the PI3K–Akt pathway. Inhibition of p38 MAPK in the presence of bradykinin increased the population of cells expressing β3-tubulin and reduced the percentage expressing GFAP in comparison to control experiments (Fig. 5E,H). Thus, SB203580 plus bradykinin treatment resulted in similar effects to those observed in bradykinin-treated samples during differentiation. These data suggest that bradykinin acts in modulating NPC differentiation independently from the p38 pathway.

Given that in molecular mechanistic investigations of neurogenesis, PC12 cells provide a broad basis for comparison with previous studies, we also made use of this model in this study. We initially investigated the differentiation of non-treated (control, CTL), NGF-treated (50 ng/ml) or bradykinin-treated (1 μM) PC12 cells on an adherent matrix of poly-L-lysine and laminin for 48 h. NGF induced differentiation as expected, and bradykinin also enhanced differentiation because we observed a significant increase in the number of cells with more than three neurites (P=0.0334) and a reduction in the percentage of cells without neurites (P=0.0280) (Fig. 6A,B). As already mentioned, transient and sustained ERK activation regulates cell fates, such as growth and differentiation, in PC12 cells (Marshall, 1995; Santos et al., 2007; Traverse et al., 1992; Vaudry et al., 2002). Here, we performed the same kinetic experiments of ERK stimulation by EGF and NGF, which induce transient and sustained ERK activation, respectively (Fig. 6C,D). Additionally, it has been demonstrated that sustained ERK activation occurs by stimulation of two or more pathways that integrate activating this MAPK. Thus, the initial transient and the subsequent sustained phase can be disrupted (York et al., 1998). In view of that, we modulated the kinetics of ERK activation by bradykinin in order to evaluate the importance of sustained ERK activation in neurogenesis induction. For this, classical PKCs (cPKCs; activated by Ca²⁺, diacylglycerol and phosphatidylserine) were inactivated by using the inhibitor Gö6973 (Gö; 10 μM; 1 h). Fig. 6E,F shows that pre-treatment with Gö blocked sustained ERK activation stimulated by bradykinin. In other words, bradykinin started to activate ERK only transiently, similar to effects of EGF stimulation. Additionally, pre-treatment with Gö (Fig. 6G,H) and thapsigargin (1 μM for1 h; Fig. S1F,G) significantly decreased sustained ERK activation stimulated by bradykinin in NPCs. We can conclude, in advance, that ERK is activated by the integration of two or more pathways (including Ca²⁺ and cPKCs) upon stimulation with bradykinin in NPCs and PC12 cells. Thus, it is possible that the activation of ERK, regardless of Ca²⁺ and/or cPKC, comes from the cAMP (Bouschet et al., 2003) and/or Ras (Blaukat et al., 2000) and/or β-arrestin pathway (Khoury et al., 2014). In this context, it is known that some these pathways synergistically activate ERK. Moreover, each of them is individually necessary for attaining the profile of sustained ERK activation in PC12 cells (Bouschet et al., 2003; von Kriegsheim et al., 2009). Therefore, given that cPKC was necessary to attain the sustained ERK activation, i.e. the pre-treatment with Gö removed sustained ERK activation by bradykinin, we evaluated its effect on neuronal differentiation. The pre-treatment with this inhibitor reversed the bradykinin effect on neuritogenesis (Fig. 6A,B; cells with more than three neurites, P=0.0396, versus bradykinin and cells without neurites, P=0.0306 versus bradykinin), suggesting that sustained ERK activation might be a mechanism of inducing neurogenesis by bradykinin.

Finally, we investigated whether bradykinin and the ERK pathway affect cell division and neurogenesis in vivo. To assess this question, we injected (1 µl) vehicle, bradykinin (1 µM), HOE-140 (1 µM) or U0126 (10 µM) into the ventricles of newborn mice [postnatal day (P)0–P1]. After 1 week, mice (P7–P8) were deeply anesthetized and brains were dissected carefully. Sections were labeled for MAP2 (green), Ki67 (red), DCX (doublecortin; red), and NeuN (lilac) (Fig. 7A; Fig. S3A). It is important to stress that analysis of the proliferation rate in the neurogenic zone of the hippocampal dentate gyrus [region of interest (ROI) shown in Fig. 7A] is relevant, because this zone is composed of quiescent stem cells, and the proliferation of these cells is the first step in the process of neurogenesis (Orford and Scadden, 2008). For better characterization of the influence of bradykinin signaling on neurogenesis in vivo, proliferating cells (Ki67⁺ cells) within ROIs were quantified by two methods, (1) standard immunohistochemistry analysis counting Ki67⁺ cells (Fig. 7B) and (2) automatic analysis of these cells by StrataQuest software (Fig. 7C–E). Fig. 7C shows merged channels and ROI analyzed by this software, whereas Fig. 7D presents representative dot-plots of nuclear size versus relative fluorescence intensity of Ki67. Both types of analysis revealed that the presence of Ki67⁺ cells in the ROI was markedly increased in bradykinin-treated mice compared with those treated with vehicle or HOE-140 (Fig. 7B,E). Furthermore, under these conditions, bradykinin significantly lost its effect when ERK signaling was blocked by the presence of 10 μM U0126, as determined by StrataQuest software analysis (Fig. 7E).

The neurogenic effect of bradykinin and the importance of ERK in the mechanisms imposed by this kinin in vivo were also analyzed by microscopic analysis of DCX (a neurogenic marker) expression levels in neurogenic zone of the hippocampal dentate gyrus, estimated by mean fluorescence intensity (MFI) in ROIs (Fig. S3). In fact, levels of DCX were upregulated in response to bradykinin, and pre-treatment with U0126 prevented this bradykinin effect.

Under the same conditions, we also evaluated the expression of the TIS21 mRNA in hippocampus of P1–P2 mice by real-time PCR, given that this gene is specifically expressed in stem cells that, at their next division, will generate postmitotic neurons, but not in proliferating stem cells (Iacopetti et al., 1999). Expression of the TIS21 mRNA in the stem cells occurs specifically during G₁-phase of neuron-generating divisions (Calegari and Huttner, 2003; Iacopetti et al., 1999). We observed that TIS21 mRNA was markedly increased in bradykinin-treated mice (P=0.0225; Fig. 7F; 216% of control signal). Importantly, the responses of HOE-140 and U0126 plus bradykinin-treated mice were significantly lower than those of the vehicle-treated mice (P=0.0001 and P=0.0006, respectively; ∼30% of control signal). In other words, bradykinin-induced TIS21 expression required the ERK signaling pathway. Taken together, these data demonstrate that bradykinin induces neuron-generating divisions in vivo and suggest that ERK is involved in effector mechanisms of neurogenesis triggered by bradykinin.

Here, we studied the underlying mechanisms of bradykinin-induced neurogenesis. Among the main findings, we highlight the influence of bradykinin on division of undifferentiated NPCs. The change of type of cell division mediated by this kinin is dependent on ERK activity, as is shown by elongation of G₁-phase of NPCs cycle in vitro and TIS21 expression in vivo, characteristics of neuron-generating division (Calegari and Huttner, 2003; Iacopetti et al., 1999; Neganova et al., 2009). Bradykinin evokes sustained ERK activation with its concomitant translocation to the nucleus in NPCs and PC12 cells. This type of dynamics of ERK activation is well known for resulting in neurogenesis of PC12 cells (Marshall, 1995; Sasagawa et al., 2005; Traverse et al., 1992; Vaudry et al., 2002). The results presented in this study, including the modulation of ERK kinetics, followed by the redirection of phenotype induced by bradykinin, suggest that the sustained ERK activation and translocation to the nucleus is the mechanism of bradykinin action in fostering neurogenesis. Moreover, ERK inhibition blocked the bradykinin effect in downregulating expression of Hes1, a repressor of NPC neuronal differentiation (Imayoshi et al., 2008; Kageyama et al., 2008; Nakamura et al., 2000), and blocked the effect of bradykinin in upregulating expression of Ngn2, a neurogenic transcription factor. Additionally, p38 MAPK is involved in the action of bradykinin in promoting migration during differentiation (a schematic representation for the proposed mechanism of bradykinin action is demonstrated in Fig. S4).

It is known that bradykinin stimulates proliferation in quiescent cells, but suppresses it in cells under intense mitogenic stimulus (Dixon et al., 2002; Duchene et al., 2002). Therefore, our findings showing that bradykinin suppresses proliferation stimulated by growth factors (FGF2 plus EGF) are in agreement with previous studies. At the same time, it is known that growth factors maintain multipotent NPCs by stimulating proliferative divisions, but these cells remain subject to the influence of microenvironment, especially during the G₁ phase (Lukaszewicz et al., 2002; Reynolds and Weiss, 1996). We observed that bradykinin leads to an elongation of G₁ phase and cell cycle. Thus, given that the prolongation of G₁ phase results in commitment to neuronal differentiation by substituting proliferative divisions for neuron-generating divisions (Calegari and Huttner, 2003; Lako et al., 2009; Neganova et al., 2009), bradykinin could be fostering neurogenesis by increasing G₁-phase duration and thus removing NPCs from the proliferative division induced by growth factors. We also observed that bradykinin leads to neuron-generating division in vivo because it increases expression of the TIS21 mRNA and Ki67 staining of cells in hippocampal dentate gyrus. Thus, the apparent paradoxical results of global proliferation in the in vitro and in vivo model of bradykinin treatment can be explained because in the neurosphere model, cells are proliferating whereas in the hippocampal dentate gyrus they are in a quiescent state.

Many studies support the hypothesis that prolonged activation kinetics, together with nuclear translocation of ERK, provide specificity for this signal and constitute the dynamics of its activation, which is necessary and sufficient to result in the neurogenic response in PC12 cells and, possibly, in P19 cells (Bouschet et al., 2003; Ebisuya et al., 2005; Reffas and Schlegel, 2000; Santos et al., 2007; Sasagawa et al., 2005; Traverse et al., 1992). The present study is the first to investigate the kinetics and sublocalization of ERK after stimulation with bradykinin in undifferentiated neural cells (PC12 cells and NPCs), verifying a mechanism of sustained activation and nuclear translocation similar to that induced by neurogenic factors, such as NGF, PACAP and retinoic acid (Marshall, 1995; Reffas and Schlegel, 2000; Vaudry et al., 2002). Additionally, we investigated the involvement of ERK in the differentiation of NPCs and PC12 cells. ERK inhibition blocked the bradykinin effect in downregulating expression of transcriptional repressor Hes1 and upregulating expression of neurogenic transcription factor Ngn2. Moreover, we observed that both the direct inhibition of the MEK–ERK pathway with U0126 and PD98059 as well as the inhibition of one of its upstream pathways (Gö6973; cPKC inhibitor), prevented neurogenesis triggered by bradykinin. The inhibition of cPKC suggests the importance of the specific dynamics of ERK activation induced by bradykinin for neurogenic processes in PC12 cells. This experimental approach disregards a possible influence of cPKC in differentiation that is independent of the ERK pathway; however, it was sufficient to propose the role of sustained ERK activation by NGF in fostering differentiation (Santos et al., 2007). Considering the effects of bradykinin on neurogenesis in NPCs and PC12 cells, the dynamics of ERK activation is suggested to be implicated in the mechanisms of bradykinin-induced neurogenesis.

NPCs were isolated by dissection of embryo telencephalons (day 13.5) of wild-type and B2BkR^−/− C57BL/6 mice (provided by University of São Paulo and Federal University of São Paulo). Mice were killed with carbon dioxide. All animal experiments were carried out according to protocols of the ethics committee (authorization number: 10/2013). Embryo telencephalons were isolated and dissociated mechanically and by enzymatic action of trypsin. 2×10⁵ cells/ml were seeded in 2% (v/v) B-27 (Life Technologies, Carlsbad, CA), Dulbecco's modified Eagle's medium (DMEM) with Ham's F-12, 20 ng/ml EGF, 20 ng/ml FGF2 (Sigma-Aldrich, St Louis, MO) and antibiotics. Neurospheres were seeded on poly-L-lysine- and laminin-precoated plastic flasks and cultured in the presence of B-27 for differentiation (Reynolds and Weiss, 1996; Trujillo et al., 2012).

PC12 rat pheochromocytoma cells ATCC No. CRL-1721 (Tischler and Greene, 1975) were maintained in DMEM supplemented with 10% (v/v) FBS, 5% horse serum (Life Technologies), glutamine and antibiotics. For starvation conditions, cells were kept overnight in DMEM supplemented with 1% horse serum. For induction of differentiation, PC12 cells were then plated on adherent poly-L-lysine- and laminin-pre-coated cell culture grade dishes for 2 days.

Cells were treated with U0126 (10 μM; Cell Signaling Technology, Danvers, MA), PD98059 (20 μM; Sigma-Aldrich, St Louis, MO), Ly294002 (20 μM; Tocris Bioscience, Bristol, UK), SB203580 (10 μM; Cell Signaling Technology), PMA (1 μM; Tocris Bioscience), U73122 (10 μM; Sigma-Aldrich), Gö6973 (10 μM; Tocris Bioscience), YM254890 (10 μM; Astellas Pharma Ins., Miyukigaoka, Tsukuba, Japan), cholera toxin (0.5 μM; Sigma-Aldrich), pertussis toxin (200 ng/ml; Sigma-Aldrich), bradykinin (1 μM; Tocris Bioscience) and HOE-140 (1 μM; Tocris Bioscience). It is known that B2BkR sequestration is rapid (half-life of 5 min) (Lamb et al., 2001; Munoz et al., 1993), but in view of possible reappearance of accessible receptors, we added bradykinin into culture medium every day during the migration and differentiation processes.

NPCs were untreated or pretreated and lysed directly in culture wells. Cyclic AMP (cAMP) production was measured by the Amersham cAMP Biotrak Enzymeimmunoassay (EIA) System (GE Healthcare, Waukesha, WI) according to the manufacturer's procedures.

For measurements of changes in intracellular free Ca²⁺ concentration ([Ca²⁺]_i), NPCs were loaded with Fluo3-AM (5 µM) for 30 min at 37°C in DMEM-F12 plus 0.5% Me₂SO and 0.06% nonionic surfactant pluronic acid F-127 (Sigma-Aldrich) (Martins et al., 2005). After 30 min of incubation with Fluo-3AM, the cells were exposed to extracellular buffer and [Ca²⁺]_i imaging was immediately performed with an inverted research microscope ECLIPSE-TiS (Nikon, Melville, NY) equipped with a 14-bit high-resolution CCD camera CoolSNAP HQ2 (Photometrics, Tucson, AZ). Fluorescence emission was recorded at 515–530 nm with image acquisition rates of one frame per second; results were further analyzed with the NIS-Element AR software (Nikon). Forty cells were analyzed for each data point and [Ca²⁺]_i levels were estimated by mean variation between Fluo-3AM fluorescence intensities obtained upon stimulation with bradykinin (F) and in the resting state (F₀) divided by F₀ (ΔF/F₀).

Prior to making an intracranial injection of the drugs, newborn pups (P0–P3) were anesthetized by hypothermia, and mounted in stereotactic apparatus. 1 μl of the solution was injected using a 5 μl Hamilton syringe (32 gauge needle) in two injection sites, 1 mm from the midline between the Bregma and Lambda and 1–2 mm deep into the anterior lateral ventricles. The pups were then placed on a heating pad for 3–5 min and returned to their mother for further recovery. Mice (P7–P8) were deeply anesthetized with ketamine and xylazine and their brains were carefully dissected. After 48 h of incubation in 4% paraformaldehyde (PFA), brains were cryoprotected in 30% sucrose, and sectioned at a 40-µm thickness using a sledge microtome. Free-floating sections of the entire hippocampus were washed and incubated for 2 h in a blocking solution (3% FBS, 0.1% Triton X-100 in PBS). Next, sections were incubated overnight at 4°C with rabbit anti-DCX (1:500; cat. no ab18723, Abcam, Cambridge, MA), rabbit anti-Ki-67 (1:1000; cat. no ab15580, Abcam), chicken anti-MAP2 (1:2500; cat. no ab5392, Sigma-Aldrich, St-Louis, MO) and mouse anti-NeuN (1:1000; cat. no MAB377, Chemicon-Millipore, Billerica, MA) primary antibodies. After 24 h, sections were washed and incubated for 2 h at room temperature with secondary Alexa-Fluor-conjugated (1:1000; Life Technologies) antibodies. Sections were washed, incubated with DAPI and mounted with DPX. Images were obtained with a fluorescence microscope (Axiovert 200, Zeiss), and analyzed with the Zen 2 software (Oberkochen, Germany) or StrataQuest software (TissueGnostics GmbH, Vienna, Austria). Briefly, the StrataQuest software is part of the TissueFAXS™ Cytometry platform – a means of image cytometry of tissue sections, tissue- and cell-cultures with a workflow and data display similar to flow cytometry. The software provides cellular data in dot-plots to show multiple measurement parameters of single cells, multicellular structures and/or morphological features. Cut-off values based on negative controls can be set to determine positive or negative cell populations as observer-independent measurements. Gates are also available to identify or work with cellular sub-populations.

Differentiated neurospheres were fixed in 4% PFA for 20 min and then incubated for 20 min in PBS plus 3% FBS and 0.1% Triton X-100. After that, cells were incubated with rabbit anti-GFAP (1:500; cat. no. Z0334, DAKO Systems), rabbit anti-p-ERK (1:200; cat. no. 4370S, Cell Signaling Technology), mouse anti-β3-tubulin (1:500; cat. no. T8660, Sigma-Aldrich), rabbit anti-S100β (1:200; cat. no. G4546, Sigma-Aldrich), rabbit anti-MAP2 (1:500; cat. no. 4542S, Cell Signaling Technology) or mouse anti-Nestin (1:500; cat. no. MAB353, Millipore, Billerica, MA) antibodies for 2 h. Sequentially, samples were washed twice and incubated at room temperature with Alexa-Fluor-488-conjugated antibodies (1:1000; Life Techologies) for 1 h. After that, cells were incubated with DAPI solution (Sigma-Aldrich, 0.3 μg/ml) and mounted with DPX. Images were obtained with a fluorescence microscope (Axiovert 200, Zeiss) (Trujillo et al., 2012).

Experiments for quantification of neural marker proteins were performed according to Trujillo et al. (2012). Neurospheres were dissociated from the flasks using trypsin and FBS. Single cells were washed with PBS, fixed for 20 min in ice-cold 4% PFA, washed with PBS 3% FBS, and incubated with same antibodies described above in PBS plus 0.1% Triton X-100 and 3% FBS for 30 min and, then, analyzed in a flow cytometer.

Flow cytometry experiments for measurement of phosphorylated proteins were performed as previously described (Chow et al., 2001; Krutzik and Nolan, 2003; Santos et al., 2007). Briefly, 6-h-starved NPCs were stimulated and fixed for 10 min by adding 16% PFA directly into the culture medium to obtain a final concentration of 4% of PFA. Cells were then permeabilized by adding 90% ice-cold ethanol and incubated at 4°C for a least 10 min. Primary staining was performed with monoclonal antibodies against the phosphorylated proteins, rabbit anti-p-ERK (1:200; cat. no. 4370S, Cell Signaling Technology), rabbit anti-p-Akt (1:100; cat. no. 2965S, Cell Signaling Technology), rabbit anti-p-STAT3 (1:100; cat. no. 9145S, Cell Signaling Technology), mouse anti-p-JNK (1:200; cat. no. 9255S, Cell Signaling Technology) and anti-p-p38 (cat. no. 4511S 1:800; Cell Signaling Technology) followed by addition of secondary Alexa-Fluor-488-conjugated antibodies (1:1000; Life Technologies). The measurements were performed on an Attune Flow Cytometer (Life Technologies) and analyzed with Flowjo V10 software (Flowjo, Ashland, OR).

Cell proliferation and cell cycle were measured following incubation in culture for 30 min or 2 h with 0.2 mM BrdU (Sigma-Aldrich) (Cappella et al., 2008; Oliveira et al., 2013). Cells were fixed with ice-cold ethanol for at least 4 h, incubated in 1.5 M HCl for 30 min and 0.1 M borate buffer (pH 8.0) for 5 min. Cells were then incubated with antibodies rat anti-BrdU (1:200; cat. no. OBT0030, Accurate Chemical) or rabbit anti-Ki67 (1:500; cat. no. AB9260, Chemicon-Millipore) antibodies. Some samples (depending on the analysis) were incubated with propidium iodide (50 μg/ml; Life Techologies) and RNAse A (50 μg/ml; Life Techologies) for 20 min. Flow cytometry events were gated by FL2A versus FL2H for doublet exclusion.

To better characterize the influence of bradykinin on proliferation, the duration of cell cycle and its phases were examined by two methods with flow cytometry experiments. With the first analysis (A), according to Nowakowski et al. (1989) and Lukaszewicz et al. (2002), it is possible to calculate the duration of the cell cycle, S-phase and the combined duration of G₁, G₂ and M phases (G₁+G₂/M). These parameters were determined by computing the labeling index values (BrdU treatment for 30 min and 2 h) within the population of cycling cells (Ki67⁺; proliferation fraction). In the second analysis (B), adapted from immunohistochemistry experiments of Storey et al. (1989) and Wilcock et al. (2007), simultaneous equations were used to calculate the duration of the cell cycle and S-phase. That is, (1) length of S phase (denoted Ts)+[30 min (of treatment)−6 min]=percentage of S-phase cells labeled in pulse×total time of cell cycle (denoted Tc) and (2) Ts+(120 min−6 min)=percentage of S-phase cells labeled in pulse×Tc. Furthermore, the length of the G₂/M phase=percentage of cells in G₂/M phase (revealed by propidium iodide staining)×Tc. Finally, length of G₁ phase=Tc–Ts–time in G2 and M (denoted TG₂/M) or length of G₁ phase=time in G₁, G₂ and M phases (denoted TG₁+G₂/M)–TG₂/M. A minimum of 60,000 cells were analyzed per sample.

Separation of nuclear and cytosolic fractions was performed with the NE-PER^® Nuclear and Cytoplasmic Extraction Reagents kit (Thermo Scientific, Waltham, MA). Protein in sample buffer was separated by SDS-PAGE on a 10% polyacrylamide gel. Proteins were then transferred to a nitrocellulose membrane (Thermo Scientific) in a wet system for 1 h at constant amperage of 400 mA. The membranes were incubated with primary antibodies, such as rabbit anti-p-ERK1/2 (1:1000, cat. no. 9101, Cell Signaling Technology), mouse anti-ERK1/2 (1:1000, cat. no. SC-514302, Santa Cruz Biotechnology, Dallas, TX), goat anti-Histone H1 (1:1000, sc 34464, Santa Cruz Biotechnology) and mouse anti-β-actin (1:1000, cat. no. A5441, Sigma-Aldrich) plus 1% BSA in Tris-buffered saline with 0.03% Tween-20 (TBS-T) overnight at 4°C. Membranes were probed with the respective secondary antibodies for 1 h under agitation at room temperature. Membranes were scanned with a laser scanner for biomolecular imaging applications (Typhoon™, GE Healthcare). Detected bands were submitted to densitometric analysis with the ImageJ software (National Institutes of Health; this software can be obtained from http://imagej.net/Welcome).

Total RNA was obtained using the Trizol reagent (Life Technologies). After DNAse I treatment, cDNAs were produced by SuperScript II Reverse Transcriptase (Life Technologies). Specific mRNAs were analyzed by SYBR Green real-time PCR on the ABI StepOnePlus Instrument (Life Technologies). Primer sequences were: Hes1, forward primer 5′-TGCCTTTCTCATCCCCAACG-3′ and reverse primer 5′-AGGTGACACTGCGTTAGGAC-3′; GAPDH, forward primer 5′-TGCACCACCAACTGCTTAG-3′ and reverse primer 5′-GGATGCAGGGATGATGTTC-3′; TIS21, forward primer 5′-ATGAGCCACGGGAAGAGAAC-3′ and reverse primer 5′-GCCCTACTGAAAACCTTGAGTC-3′; Mash1, forward primer 5′-GCCTCCAATTGAAGCAACGTC-3′ and reverse primer 5′-AGAAGCAAAGACCGTGGGAG-3′; Ngn2, forward primer 5′-TCGGCTTTAACTGGAGTGCC-3′ and reverse primer 5′-GTGTGTTGTCGTTCTCGTGC-3′. Thermal cycling conditions consisted of 50°C for 2 min, 95°C for 10 min, and 50 cycles of 95°C for 15 s and 60°C for 1 min. The 2^−ΔΔCt parameter was used for quantification and GAPDH gene expression was used as internal standard (Lameu et al., 2012; Livak and Schmittgen, 2001).

The results are expressed as mean±s.e.m. from three or more independent experiments. Statistical comparisons between different treatments, with normal distribution, were performed by either a Student's t-test and/or two-way analysis of variance (ANOVA) with a Bonferroni post hoc test by using GraphPad Prism 5.1 software (La Jolla, CA). The criteria for statistical significance were set at P<0.05. For flow cytometry, a minimum of 30,000 cells was analyzed per sample. A minimum of 300 and up to 800 cells per sample was counted for quantification of immunolabeled cells or cells captured by differential interference contrast (DIC) microscopy.

We thank Dr Rupert Ecker (TissueGnostics GMBH, Vienna, Austria) for automatic analysis of hippocampal slices for Ki67 expression.