Differential Regulation of the B Cell Receptor-mediated Signaling by the E3 Ubiquitin Ligase Cbl*

The E3 ubiquitin ligase Cbl has been implicated in intracellular signaling pathways induced by the engagement of the B cell antigen receptor (BCR) as a negative regulator. Here we showed that Cbl deficiency results in a reduction of B cell proliferation. Cbl–/– B cells show impaired tyrosine phosphorylation, reduced Erk activation, and attenuated calcium mobilization in response to BCR engagement. The phosphorylation of Syk and Btk is also down-modulated. Interestingly, Cbl–/– B cells display enhanced BCR-induced phosphorylation of CD19 and its association with phosphatidylinositol 3-kinase. Importantly, Lyn kinase activity is up-regulated in Cbl–/– B cells, which correlates inversely with the Cblmediated ubiquitination of Lyn. Because Lyn has both negative and positive roles in B cells, our results suggested that Cbl differentially modulates the BCR-mediated signaling pathways through targeting Lyn ubiquitination, which affects B cell development and activation.

The Cbl family proteins are composed of an N-terminal tyrosine kinase binding domain, followed by a RING finger domain, and C-terminal multiple proline-rich regions (7). Cbl and Cbl-b function as adaptor proteins by interacting with proteintyrosine kinases and other crucial signaling molecules and function as E3 ubiquitin (Ub) ligases, in which the RING finger recruits Ub-bound E2 enabling the transfer of Ub to the substrate, resulting in its ubiquitination and subsequent downmodulation (7). Earlier studies identified Cbl as one of the rapidly tyrosine phosphorylated substrates upon the BCR engagement (5). Overexpression of Cbl leads to an inhibition of Syk kinase activity (8) but has no effect on the Lyn function (9). A later study showed that Cbl acts as an E3 Ub ligase to promote Syk ubiquitination (10). Experiments with Cbl-deficient chicken DT40 cells demonstrated that Cbl inhibits the BCR-mediated PLC-␥2 activation by regulating PLC-␥2 recruitment to B cell linker protein (BLNK) and Vav (11). However, a physiological function of Cbl in primary murine B cells remains unclear.
To understand the role of Cbl in B cells, we examined B cells from Cbl-deficient mice and investigated the mechanism by which Cbl regulates BCR signaling. It was found that a loss of Cbl resulted in reduced B cell proliferation and diminished phosphorylation of BCR proximal signaling molecules. However, the tyrosine phosphorylation of CD19 and CD19-mediated downstream signaling were increased. We provided evidence that Cbl differentially regulates the BCR signaling via targeting the ubiquitination of Lyn.

MATERIALS AND METHODS
Mice-Cbl Ϫ/Ϫ mice (12) were provided by M. Naramura and H. Gu. Cbl Ϫ/Ϫ mice on a mixed 129 and C57BL/6 background were further back-crossed with C57BL/6 mice for six more generations to obtain Cbl ϩ/ϩ , Cbl ϩ/Ϫ , and Cbl Ϫ/Ϫ B6 mice. The genotyping of the mutant mice was done by PCR using tail DNA. Mice between 6 and 8 weeks after birth were mostly used for the experiments. All the animal experiments were performed according to the institutional guidelines.
B Cell Proliferation-B cells from splenocytes were negatively selected with a B cells isolation kit following the manufacture's instructions (Miltenyl Biotec, Auburn, CA). Purified B cells (5 ϫ 10 5 cells/ml) were stimulated with increasing amount of LPS or anti-IgM F(abЈ) 2 (Jackson Immunoresearch Laboratories, West Grove, PA) or in the presence of recombinant mouse IL-4 (100 units/ml; Calbiochem, San Deigo, CA) either alone or with anti-CD40 monoclonal antibody (BD Biosciences). Proliferation was measured by pulsing the cells with 1 Ci of [ 3 H]thymidine for the final 16 h of the 2-day culture and then by counting the radioactivity uptake.
Calcium Influx Measurement-Splenic B cells were resuspended at 1 ϫ 10 7 /ml in Hanks' buffer (Invitrogen) containing 1% (w/v) bovine serum album and loaded with 1 M Indo-1-AM (Sigma-Aldrich) at 37°C for 45 min. The cells were washed twice at room temperature and resuspended at 2 ϫ 10 6 /ml. Continuous monitoring of the fluorescence ratio (525/405 nm) was performed using BD-LSR (BD Biosciences). Base-line fluorescence ratios were collected for 1 min before anti-IgM F(abЈ) 2 was added. The fluorescence ratios were collected at real time for 5 min following addition of antibodies.
In Vitro Kinase Assay-Lysates were prepared from splenic B cells in radioimmune precipitation assay buffer. The anti-Lyn precipitates were washed four times with the radioimmune precipitation assay buffer followed by washing two times with 20 mM HEPES, pH 8.0, containing 150 mM NaCl, 10 mM magnesium acetate, and 20 mM MnCl 2 (kinase assay buffer). The Lyn kinase activity was measured by the in vitro kinase assay system according to the manufacture's instructions (Chemicon, Temecula, CA). Briefly, the immunoprecipitates were incubated in kinase assay buffer including biotin-conjugated Src kinasespecific peptide substrates for 30 min at 30°C and then subjected to 96-well enzyme-linked immunosorbent assay-based colorimetric assay. All in vitro kinase activity experiments were repeated at least three times, and representative results are shown.
Plasmids and Cell Transfection-Plasmids containing wild-type Cbl or Cbl RING finger Cys to Ala mutant (Cbl-CA), Myc-Ub have been described previously (13). The pME plasmids containing Lyn or Lyn kinase-dead mutant (Lyn KR) cDNA were kindly provided by Dr. T. Kawakami and the cDNAs were subcloned into pCMV-FLAG mammalian expression vector (Sigma). Human embryonic kidney 293T cells were cultured in Dulbecco's modified Eagle's medium (Irvine Scientific, Santa Ana, CA) containing 10% fetal calf serum, 100 units/ml of penicillin, and 100 units/ml of streptomycin. Cells were transfected with appropriate plasmids (usually 5-10 g in total) using FuGENE 6 transfection reagent (Roche Applied Science). After 48 h, cells were collected and resuspended in 0.5 ml of Dulbecco's modified Eagle's medium. Cells were either untreated or treated with pervanadate for 30 min at 37°C. Cells were then pelleted and resuspended in 1% Nonidet P-40 lysis buffer. For the detection of ubiquitinated proteins, 0.1% SDS and 5 mM N-ethylmaleimide (Sigma-Aldrich) were added to the lysis buffer to disrupt nonspecific protein interactions (14).
Pulse-chase Experiment-Purified B cells (1 ϫ 10 7 ) from Cbl ϩ/ϩ or Cbl Ϫ/Ϫ mice were cultured in methionine-free Dulbecco's modified Eagle's medium containing 5% dialyzed fetal calf serum for 1 h at 37°C. The cells were stimulated with 10 g/ml anti-IgM F(abЈ) 2 and then labeled for 1 h with 100 Ci/ml Trans[ 35 S] (ICN Biomedicals, Costa Mesa, CA). Cells were then washed three times with cold phosphatebuffered saline and cultured for different time intervals in complete Dulbecco's modified Eagle's medium containing 10% fetal calf serum. At each time point, the cells were collected and lysed by Nonidet P-40 lysis buffer. The cell lysates were precleared with protein G-Sepharose for 30 min and then immunoprecipitated with anti-Lyn. The immune complexes were resolved by 12% SDS-PAGE and were subjected to autoradiography. The radiolabeled protein bands were quantified by using NIH Image 1.61 software.

RESULTS
Reduced Proliferation of Cbl Ϫ/Ϫ B Cells-To examine the effects of the Cbl deficiency on the biological responses of B cells, we assessed the proliferation of purified splenic B cells following anti-IgM, anti-CD40, anti-CD40 plus IL-4, or LPS stimulation. As shown in Fig. 1A, the proliferative response of Cbl Ϫ/Ϫ B cells to anti-IgM stimulation was severely impaired. However, no defect in proliferation was observed after stimulation with LPS in Cbl Ϫ/Ϫ B cells compared with wide-type B cells (Fig. 1B). The B cell proliferation was reduced after stimulation with anti-CD40 or anti-CD40 plus IL-4 ( Fig. 1, C and D). The data suggested that the proliferative responses induced by the engagement of the BCR, CD40, and IL-4, but not LPS, are affected by Cbl deficiency.
Impaired BCR Signaling in Cbl Ϫ/Ϫ B Cells-Although several biochemical studies have previously provided evidence about Cbl in BCR signaling (5), it is still unclear whether Cbl regulates BCR signaling in murine primary B cells in a similar manner as in cultured cell lines. To study Cbl-mediated downstream signaling events following the BCR engagement, we assessed the BCR-induced tyrosine phosphorylation patterns of cellular proteins in purified wild-type and Cbl Ϫ/Ϫ splenic B cells. In the absence of Cbl, the overall tyrosine phosphorylation in the whole cell lysates was reduced as compared with wild-type B cells during a period of 60-min stimulation ( Fig.  2A). We then examined the activation of downstream mitogenactivated protein kinases, Erk and JNK in those cells by using phospho-specific antibodies. The Erk phosphorylation in Cbl Ϫ/Ϫ B cells was obviously reduced after a 5-min stimulation and remained lower than in wild-type B cells even after a 60-min stimulation (Fig. 2B). However, the amount of phosphorylated JNK in Cbl Ϫ/Ϫ B cells was comparable with wild-type B cells under the same stimulation conditions (Fig. 2B). We next examined the BCR-triggered Ca 2ϩ influx and found that Cbl Ϫ/Ϫ B cells showed slightly reduced Ca 2ϩ mobilization in comparison with wild-type B cells (Fig. 2C).
Reduced Phosphorylation of Proximal Signaling Molecules in Cbl Ϫ/Ϫ B Cells-To further understand the molecular mechanisms that account for the BCR signaling defects in Cbl Ϫ/Ϫ B cells, we compared the BCR-mediated tyrosine-phosphorylation status of Syk, Btk, and PLC-␥2, the critical proximal signaling molecules for the initial of BCR signaling. It was observed that the phosphorylation of Syk and Btk was decreased in Cbl Ϫ/Ϫ B cells (Fig. 3, A and B). It has been shown previously that a Cbl deficiency in the DT40 cell line can affect the association of PLC-␥2 with BLNK and also PLC-␥2 phosphorylation (11). Surprisingly, there seemed to be no significant difference in the tyrosine phosphorylation of PLC-␥2 and the association of PLC-␥2 with BLNK between wild-type and Cbl Ϫ/Ϫ B cells (Fig. 3C). Thus our data from Cbl Ϫ/Ϫ mice suggested that Cbl may not directly play a negative role in the BCR-mediated phosphorylation of Syk or PLC-␥2 as previously proposed (8,11).
Up-regulation of CD19 Signaling Pathway-We next examined the potential contribution of Cbl in BCR-induced CD19 signaling pathway. Surprisingly, we found that CD19 phosphorylation was enhanced in Cbl Ϫ/Ϫ B cells upon BCR engagement (Fig. 4A). Because it has been reported that engagement of CD19 leads to the activation of Akt (15), we next examined the Akt phosphorylation by blotting the cell lysates with phosphospecific anti-Akt antibody. It was found that Akt phosphorylation was augmented in Cbl Ϫ/Ϫ B cells compared with wild-type B cells upon BCR stimulation (Fig. 4A). The increased CD19 phosphorylation was not because of the change of cell surface expression of CD19, which was equivalent between wild-type and Cbl Ϫ/Ϫ B cells (Fig. 4B). BCR stimulation did not further alter the expression levels of CD19 in those cells (data not shown). We also examined other CD19 downstream effector molecules such as the Bcl family proteins (16). We detected increased levels of Bcl-xL and, to a lesser extent, Bcl-2 in Cbl Ϫ/Ϫ B cells after anti-IgM stimulation for 6 -48 h (Fig. 4C).
It was shown previously that engagement of BCR, but not CD19 itself, induces an association between CD19 and the p85 subunit of PI3-kinase (17). The increased Akt phosphorylation in Cbl Ϫ/Ϫ B cells, a downstream target of PI3-kinase, could result from an altered association between CD19 and the p85. To test this hypothesis, we immunoprecipitated CD19 and ex-amined the associated p85 in anti-IgM stimulated wild-type and Cbl Ϫ/Ϫ B cells. Consistent with a previous observation (17), anti-IgM stimulation caused a rapid association between CD19 and p85 and gradual dissociation 5 min after stimulation (Fig.  4D). The association of CD19 with p85 was markedly increased in Cbl Ϫ/Ϫ B cells under the same stimulation conditions. Because Lyn also forms a complex with CD19 (18), we blotted the same membrane with anti-Lyn antibody and found that more Lyn protein was coimmunoprecipitated by anti-CD19 antibody in Cbl Ϫ/Ϫ B cells (Fig. 4D). Thus, in contrast to the positive role of Cbl in BCR-induced phosphorylation of Syk or Btk as described earlier, the results described here suggest that Cbl is a negative regulator of BCR-mediated CD19 signaling pathways.
Cbl Regulates Lyn Activation through Ubiquitination-Genetic and biochemical studies have shown that the tyrosine kinase Lyn plays both a negative and a positive role in BCR signaling (19). The reduced proximal BCR signaling, the upregulation of CD19 phosphorylation, and its association with Lyn in Cbl Ϫ/Ϫ B cells led us to speculate that Cbl may negatively regulate Lyn, which differentially affects the downstream signaling. To test such model, we examined the in vitro kinase activity of immunoprecipitated Lyn from wild-type and Cbl Ϫ/Ϫ B cells. Although the kinase activity was quite similar between wild-type and Cbl Ϫ/Ϫ B cells 2 min after BCR stimulation, higher kinase activity was observed after a 5-min stimulation (Fig. 5A). Cbl Ϫ/Ϫ B cells still showed higher kinase activity after a 15-min stimulation, whereas at the same time point the kinase activity in wild-type B cells decreased dramatically. Thus, the loss of Cbl caused a slower but stronger Lyn activation upon BCR engagement.
We next investigated whether Lyn is a target for Cbl E3 ligase activity and performed an in vivo ubiquitination assay in transiently transfected human embryonic kidney 293 cells by overexpressing wild-type Lyn, or kinase-dead Lyn, together with Myc-tagged Ub and Cbl. The overexpression of Cbl promoted Ub conjugation to wild-type Lyn (Fig. 5B). However, the kinase-dead Lyn KR mutant was not ubiquitinated by Cbl. Stimulation of the transfected cells with pervanadate, a phosphatase inhibitor, induced a stronger ubiquitination of wildtype Lyn. In addition, the ubiquitination of Lyn by Cbl requires an intact RING finger, because a mutation at a critical cysteine residue in the RING finger (Cbl CA) disrupted the Ub conjugation to Lyn (Fig. 5C). Interestingly, we observed that the immunoprecipitated Lyn from samples transfected with the Cbl CA mutant showed higher levels of tyrosine phosphorylation than from cells transfected with wild-type Cbl (Fig. 5C, bottom panel), suggesting that Lyn ubiquitination is inversely related to its phosphorylation. We further examined Cbl-promoted Lyn ubiquitination in a chicken DT40 B cell line and found that only wild-type Lyn, but not the Lyn KR mutant, was ubiquitinated by the expression of Cbl (Fig. 5D).
To further confirm the physiological relevance of Cbl-mediated Lyn ubiquitination, we examined the protein levels of Lyn in wild-type and Cbl Ϫ/Ϫ B cells following anti-IgM stimulation for different time periods. It was found that the protein amounts of Lyn in wild-type B cells decreased rapidly after BCR stimulation, becoming obvious 10 min after stimulation (Fig. 5E). However, the rate of reduction in Cbl Ϫ/Ϫ B cells was attenuated, and the Lyn protein was still detectable 3 h after BCR stimulation. We further performed a pulse-chase experiment to examine the protein stability of Lyn in wild-type and Cbl Ϫ/Ϫ B cells. Lyn in Cbl Ϫ/Ϫ B cells was more stable than in wild-type B cells (Fig. 5F). Thus, Lyn is a physiological target for Cbl-induced ubiquitination and subsequent degradation.
Altered B Cell Development in Cbl Ϫ/Ϫ Mice-To understand a functional implication of Cbl in the differential regulation of BCR signaling, we investigated the B cell development in wildtype and Cbl Ϫ/Ϫ mice, the B cell subpopulations in the spleen and peritoneal cavity were analyzed in wild-type and Cbl Ϫ/Ϫ littermates by flow cytometry. The analysis of splenocytes showed a normal ratio of B220 ϩ B cells versus CD3⑀ ϩ T cells in Cbl Ϫ/Ϫ mice, and the total cell numbers were comparable between wild-type and Cbl Ϫ/Ϫ mice (data not shown). As shown in Fig. 6A, double staining with anti-IgM and IgD of splenocytes showed an apparent increase in the population of IgM hi IgD low immature B cells, but a substantial decrease in the mature IgM low IgD hi B cells. We further examined the cell surface expression of CD21 and CD23 and found a marked reduction of CD21 int CD23 hi (mostly follicular B cells) in Cbl Ϫ/Ϫ mice (Fig.  6B). But the CD21 hi CD23 lo (enriched in the marginal zone) and CD21 low CD23 low (mostly newly formed B cells) populations were increased in Cbl Ϫ/Ϫ mice compared with wild-type mice.
We next examined the population of B1 B cells in the peritoneal cavities. B1 B cells share many phenotypic and functional properties with marginal zone B cells and are characterized by the cell surface expression of CD5 and IgM (20). We found that the population of IgM ϩ CD5 ϩ B1 B cells was markedly increased in Cbl Ϫ/Ϫ mice (Fig. 6C). These data collectively FIG. 3. Attenuated phosphorylation of Syk and Btk in Cbl ؊/؊ B cells. A, detection of Syk phosphorylation in wild-type and Cbl Ϫ/Ϫ splenic B cells stimulated with anti-IgM F(abЈ) 2 (10 g/ml) for the indicated time intervals. Total cell lysates (2 ϫ 10 7 cells/sample) were prepared and immunoprecipitated (IP) by anti-Syk antibody. The phosphorylation was determined by immunoblotting with anti-phosphotyrosine (pTyr). B, tyrosine phosporylation of Btk was carried out as in A. C, BCR-induced activation of PLC-␥2 and association of PLC-␥2 with BLNK. Total cell lysates were prepared as A and immunoprecipitated with anti-PLC-␥2 antibody. The phosphorylation and association were examined by immunoblotting with anti-phosphotyrosine or anti-BLNK antibody. The relative intensity for the phosphorylated proteins is labeled below each panel. The highest intensity in each row was given an arbitrary unit of 1. A-C are representatives of three repeated experiments.
suggest that a Cbl deficiency blocks the development of follicular B cells but promotes the development of the marginal zone and B1 B cells. DISCUSSION Although the Cbl Ϫ/Ϫ mice have been available for awhile, a vigorous investigation on Cbl in B cell development and activation has been lacking. Based on previous studies, it could have been predicated that a Cbl deficiency in B cells results in an up-regulation of BCR signaling. However, our data suggested that Cbl plays differential roles in the signal transduction pathways initiated by the engagement of the BCR. The reduced proximal BCR signaling and increased CD19 signaling in Cbl Ϫ/Ϫ B cells prompted us to speculate that Cbl regulates a further upstream molecule or molecules that act as switches between CD19 and BCR signaling pathways. The present study leads us to conclude that Lyn is such a candidate as the physiological target for Cbl in B cells, which is supported by the increased Lyn kinase activity in Cbl Ϫ/Ϫ B cells, Cbl-mediated ubiquitination of Lyn in transfected cells, and augmented Lyn stability upon BCR stimulation. An early study showed that Lyn associates with Cbl and induces its phosphorylation in B cells (9). Another recent study showed that Lyn is ubiquitinated by Cbl and/or Cbl-b in mast cells upon stimulation of the high affinity IgE receptor (21). It was previously reported that Lyn Ϫ/Ϫ B cells are hyperproliferative and show enhanced Erk activation (22,23). In the present study, we observed that Cbl deficiency results in increased Lyn kinase activation and reduced Erk activation, which is the opposite of Lyn Ϫ/Ϫ B cells. Thus, our data showing reduced proliferation in Cbl Ϫ/Ϫ B cells are consistent with the hypothesis that Cbl regulates B cell proliferation via modulating Lyn activity.
The finding that Lyn is a target of Cbl in B cells is further supported by another interesting observation of the increased CD19 phosphorylation in Cbl Ϫ/Ϫ B cells upon BCR stimulation.
It was previously shown that Lyn plays a positive role in phosphorylating CD19 tyrosine residues and the phosphorylated CD19 in turn positively regulates Lyn activation (18). To be consistent with this observation, we indeed observed an increased complex formation between CD19 and Lyn in anti-IgM-stimulated B cells from Cbl Ϫ/Ϫ mice. However, opposite observations have also been reported that Lyn and CD19 independently transduce downstream signaling (24). It is possible that in addition to Lyn, other Src kinases may compensate for the function of Lyn for the CD19 phosphorylation. The observation of increased CD19 signaling in Cbl Ϫ/Ϫ B cells is further supported by the augmented association of CD19 with p85 of PI3-kinase, which then induces the Akt phosphorylation and probably the expression of Bcl proteins.
Although previous biochemical studies have clearly established a negative role of Cbl in regulating Syk either through inhibiting its kinase activity (8) or inducing Syk ubiquitination (10), it is surprising to find that Syk activation is not upregulated, but rather down-modulated in Cbl Ϫ/Ϫ B cells. The results shown here may imply that Syk is not a primary target for Cbl under physiological conditions. The reduced Syk activation in Cbl Ϫ/Ϫ B cells correlated with the attenuated B cell proliferation and decreased Erk phosphorylation. A possible explanation for the reduced Syk activation is that Cbl functions as an adaptor to facilitate the phosphorylation of Syk by Src family kinases, which becomes defective in the absence of Cbl. This explanation is consistent with previous biochemical studies that Syk interacts with Cbl via the N-terminal tyrosine kinase binding domain, whereas Src family kinases associate with Cbl via its C-terminal proline-rich sequences (7). Similarly, we could not detect a change of tyrosine phosphorylation of PLC-␥2 or its association with BLNK, as observed in DT40 cells (11). The discrepancy could be the difference of B cell origin. To support this notion, it was observed that Cbl-b Ϫ/Ϫ murine B cells display enhanced BCR signaling through the ubiquitination of Syk (6), whereas a decreased PLC-␥2 activation was observed in Cbl-b Ϫ/Ϫ DT40 B cells (25). In any case, the present genetic study using Cbl Ϫ/Ϫ primary B cells allows us to revisit the issue of Cbl in the regulation of BCR signaling.
In this study, we also showed that Cbl Ϫ/Ϫ mice display an which includes the strength of BCR signal, chemokine-mediated cell migration, or Notch-regulated differentiation (26,27). Our results partially supported the idea that BCR signaling strength is involved in follicular B cell development, because we detected decreased total tyrosine phosphorylation and Erk activation in Cbl Ϫ/Ϫ splenic B cells in response to BCR engagement. Intriguingly, we also observed an increase in marginal zone and B1 B cells in Cbl Ϫ/Ϫ mice, which does not fit with the signal strength model. Interestingly, both CD19 and PI3-kinase signaling are important in marginal zone and B1 B cell development (28 -30). To be consistent with these observations, it was found that in Cbl Ϫ/Ϫ splenic B cells, CD19 phosphorylation and the downstream PI3-kinase activation were up-regulated. Thus, we propose that Cbl plays distinct roles in regulating BCR signaling pathways, which differentially affects B cell development.
In summary, we presented evidence that Cbl plays relatively complicated roles in the signaling transduction emanated from the BCR engagement under physiological conditions. The present study focused on the analysis of splenic B cell development and activation and revealed unexpected findings different from previous biochemical studies. It is obvious that Cbl is ubiquitously expressed in other cells or tissues such as T cells or dendritic cells, which might also influence the B cell phenotype in Cbl Ϫ/Ϫ mice, although our adoptive transfer study showed that Cbl has an intrinsic role in B cell development (data not shown). Further dissection of BCR signaling in Cbl Ϫ/Ϫ B cells using biochemical and genetic approaches is needed to fully understand the diverse roles of Cbl in B cell signaling and B cell-mediated immune responses.
FIG. 6. Altered B cell differentiation in Cbl ؊/؊ mice. A, splenocytes from 6 -8week-old mice were analyzed for the cell surface expression of B220, IgM, and IgD by fluorescence-activated cell sorter after staining. The percentages of each lymphocyte population of B220 ϩ -gated cells are shown. WT, wild-type. B, splenocytes from wild-type and Cbl Ϫ/Ϫ were stained by antibodies for B220, CD21, and CD23 to determine the cell surface expression of CD21 and CD23 of B220 ϩ -gated cells. C, cells from the peritoneal cavity were examined to determine the expression of IgM and CD5 of B220 ϩ cells. A-C are representatives of three repeated experiments.