| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* GI Cell Biology, Children’s Hospital, and the Harvard Digestive Diseases Center, Boston, Massachusetts, USA;
Department of Pediatrics, Harvard Medical School, Boston, Massachusetts, USA; and
Teine Keijinkai Hospital, Sapporo, Japan
1 Correspondence: Children’s Hospital, 300 Longwood Ave., Enders 720, Boston, MA 02115, USA. E-mail: david.saslowsky{at}childrens.harvard.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: sphingomyelinase • sphingomyelin • potassium channel
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3
4
5
6
7)
. Dysregulation of SMase isoforms has been implicated in inflammatory bowel disease (8)
, Niemann-Pick type A and B syndrome (9)
, pulmonary edema (10)
, Alzheimer’s (11)
, and atherosclerosis (12)
. Membrane SM may also affect infection by HIV and prion protein, and the cellular dynamics of endogenous glycolipids (13
14
15
16)
.
Our interest in SMases arose from the possibility that these enzymes might affect the structure and function of SM- and cholesterol-rich plasma membrane (PM) microdomains, termed lipid rafts. These microdomains harbor the glycolipid receptors for the AB5-subunit Shiga and cholera toxins and are thought to affect bacterial pathogenesis by mediating toxin uptake into host epithelial cells. In testing this idea, we discovered that the bacterial SMase from Staphylococcus aureus (β-hemolysin) acts on the apical (lumenal) brush border membrane of T84 intestinal cells to cause resistance to microbial invasion (unpublished results) and cholera toxin (16)
. We also found that SMase applied to the basolateral (serosal) membrane of the same cells acted independently to repress the cAMP-induced Cl– secretory response that typifies this cell type. How the suppression of Cl– secretion might occur remains unknown.
Cl– secretion is the primary ion transport event that drives salt and water secretion across the intestine and other mucosal surfaces and requires the coordinated activity of membrane transporters and channels differentially distributed in the apical and basolateral PM of epithelial cells (17)
. Cl–, Na+, and K+ ions enter the basolateral PM of intestinal epithelia together by passing through the basolateral Na+, K+, 2Cl– cotransporter (NKCC) following the inwardly directed Na+ electrochemical gradient established by the basolateral Na+/K+ ATPase. Cl– then exits the apical PM by passing through the cAMP-activated cystic fibrosis transmembrane conductance regulator (CFTR) or the Ca2+-activated chloride channel (CaCC). Export of K+ across the basolateral membrane maintains the electrochemical potential required to sustain apical Cl– secretion. In the intestine, this is likely accomplished by the KCNQ1/KCNE3 and KCNN4 K+ channels, which are gated open by intracellular cAMP and Ca2+, respectively (18
, 19)
. The T84 intestinal cell line used in this study models all aspects of this transport physiology (17)
.
The conversion of SM to ceramide by extracellular SMases can alter cell function in two general ways. First, the reaction can directly affect the structure of the membrane so as to induce ceramide-rich membrane microdomain formation and disrupt the association between SM and cholesterol (20)
. This has already been shown to affect the distribution and function of some membrane proteins, including ion channels and other transporters (21
22
23
24)
. In another way, the reaction can initiate signaling cascades mediated by ceramide itself or its metabolites such as sphingosine-1-phosphate (S1P) (25)
. It is also possible that signal transduction might occur via diacylglycerol either by direct hydrolysis of phosphatidyl choline, a lower efficiency, off-target effect of the enzyme, or as a consequence of the regeneration of SM via the action of SM-synthase (25
, 26)
.
In the present work, we report on studies that define a novel ceramide-induced signal transduction pathway mediated by JNK for regulation of salt and water secretion at mucosal surfaces. Because intestinal epithelial cells do not express sphingomyelinase on their basolateral membranes (16
, 27)
, the pathway is predicted to depend on enzymatic crosstalk with adjacent cells in the subepithelial space.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
S. aureus supernatants
S. aureus harboring a viral inactivating mutation in the SMase gene (DU5719; ref. 28
) and the wild-type isotype control (NCTC832S-4) were kind gifts from T. Foster (Trinity College, Dublin, Ireland). Briefly, bacterial cultures grown in TSB medium rotating overnight at 37°C were pelleted at 12,000 rpm for 10 min. Conditioned culture supernatants or sterile TSB were diluted into HBSS buffer as indicated prior to basolateral application to T84 monolayers.
Cell culture
Polarized monolayers of T84 cells were cultured on 0.33 or 5 cm2 polyester Transwell® inserts (Corning, Acton, MA, USA) as described previously (29)
. Polarized T84 monolayers were used for these experiments because alterations in cAMP-dependent ion transport across the PM of these cells can be measured in real time using standard methods of electrophysiology, resulting in an experimental system with a high degree of sensitivity and temporal resolution. In addition, T84 monolayers possess high transepithelial resistance, resulting in an inability of either charged or protein-sized molecules to passively diffuse to the contralateral pole.
Electrophysiology
Short-circuit current (Isc) and resistance measurements in electrophysiological studies on polarized T84 monolayers (0.33 cm2 inserts) were performed as described previously (30)
. Measurements of monotypic channel functions (apical chloride and basolateral potassium channels) in permeabilized T84 monolayers were as described previously (31)
. All measurements (n=2 monolayers/experiment unless otherwise stated) in the illustrated time-courses are representative of at least 3 independent experiments.
Phospho-protein isolation and immunoblot analysis
Phosphorylated proteins were isolated by immunoprecipitation (IP) from crude extracts prepared from T84 monolayers grown on 5 cm2 inserts as described previously (32)
, except a mixture of phosphatase inhibitors (Cocktail Set II, CalBioChem) was included in the lysis buffer and specific anti-phospho antibodies were used for IP. The p38 activity assay was performed per the manufacturer’s protocol (Cell Signaling Technology), except that each T84 5 cm2 insert was lysed in 450 µl lysis buffer containing 1x protease complete (Roche Applied Sciences, Indianapolis, IN, USA).
SDS-PAGE and standard immunoblotting methodologies were as described previously (29)
using the manufacturer’s suggested antibody concentrations. Where indicated, nitrocellulose membranes were stripped of antibodies for 1 h in stripping buffer (62.5 mM Tris, pH 6.7; 100 mM β-mercaptoethanol; 2% SDS) and reprobed with β-actin mAb (Sigma) (1:4000). For total JNK detection (
Fig. 5D
), equal volumes of pre-IP supernatant from each sample were separated by SDS-PAGE and immunoblotted using a rabbit anti-JNK antibody. The results of all immunoblots were reproduced in 3 independent experiments.
|
|
|
|
|
Conditioned medium (CM) from human fibroblasts
CM was obtained from commercially available normal and Niemann-Pick disease type B human fibroblasts (reference GM00498 and GM11097, respectively; Coriell Cell Repositories, Camden, NJ, USA) as described previously (27)
, with the following modifications: 100 mM ZnCl2 was added to sterile and CM prior to 10-fold concentration using a Centriprep 30 concentrator (Amicon, Beverly, MA, USA). Concentrated samples were buffer-exchanged 3 times with HBSS buffer, pH 6.5; diluted 1:50 in HBSS, pH 6.5; and applied basolaterally to T84 monolayers (sterile HBSS buffer, pH 6.5, was in the apical chamber for these electrophysiological experiments).
Statistical analysis
Standard error (SE) was used as a measure of variance in each electrophysiological experiment. All bar graphs are summaries of at least 3 independent time-course experiments (only maximal Isc for each condition is shown) with replicates (n) indicated under each bar. Significance was determined using Student’s t test.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). SMase had no effect on resting transepithelial potential (Fig. 1A, B, D, F
) or on transepithelial resistance (TER; data not shown). Inhibition of Isc was also observed when cells were exposed to a highly enriched preparation of S. aureus SMase (aka β-hemolysin, Supplemental Fig. S1A) or SMase isolated from B. cereus, which closely typifies the mammalian neutral SMase (data not shown) (33)
. To confirm that SMase was responsible for the secretory defect, an S. aureus mutant (DU5719) harboring an inactivating viral insertion in the SMase gene was utilized (28)
. No effect on forskolin-induced Isc was observed in monolayers incubated with conditioned growth medium from the mutant (solid circles; Fig. 1B
), whereas CM from the isogenic wild-type S. aureus attenuated forskolin-induced Isc to a similar degree as the purified enzyme (solid triangles; Fig. 1B
). Similar results were obtained when the microbes were applied directly to the cell monolayers (not shown).
The inhibition of forskolin-induced Isc by SMase was dose dependent, with 
0.1 U/ml SMase causing maximal repression (Fig. 1C
), and this dose was used for all other experiments. The addition of SMase to apical membranes had no effect at any concentration tested (Fig. 1C
). The decrease in forskolin-induced Isc in SMase-treated monolayers was completely reversed when the SMase was boiled at 100°C for 1 h prior to incubation with cells (Supplemental Fig. S1B), suggesting that the folded protein, presumably harboring an enzymatic activity, is required. SMase had only marginal effects on Isc induced by the Ca2+-dependent agonists carbachol (Fig. 1D
) or thapsigargin (not shown), indicating specificity for inhibition of the cAMP-dependent pathway of Cl– secretion. The result also shows that the effects of SMase on T84 cells cannot be explained by inhibition of NKCC or the Na+/K+ ATPase that are required for a coordinated Cl– secretory response to both cAMP- and Ca2+-dependent agonists. Secretagogue-induced Isc in SMase-treated monolayers was attenuated equally when the cAMP agonists forskolin, vasointestinal peptide (VIP), or membrane-permeant 8-bromo-cAMP were used (Fig. 1E
). Thus, the SMase-induced effects on Cl– secretion must be downstream of adenylate cyclase. SMase rapidly (3 min) suppressed forskolin-induced Isc in monolayers treated with forskolin either before (Fig. 1F
) or after (Supplemental Fig. S1B) SMase treatment. Accordingly, all subsequent SMase preincubations were for 5 min unless otherwise indicated (including Fig. 1B
). Attenuation of cAMP-dependent Cl– secretion was still observed when SMase was washed out for 30 min prior to forskolin addition (Supplemental Fig. S2).
SMase causes inhibition of a basolateral cAMP-gated K+ conductance
To explain how SMase acts to inhibit cAMP-induced Isc, we first studied K+ conductance across the basolateral membrane. This was done by selectively permeabilizing the apical membranes of T84 monolayers to monovalent ions by application of amphotericin B and using buffers containing K+ as the primary charge-carrying ion (31)
. Under these conditions, the transepithelial Isc recorded represents K+ transport across the basolateral membrane (termed Isc-blK) (31)
. Monolayers pretreated with SMase displayed 3-fold lower forskolin-induced Isc-blK as compared to controls (Fig. 2A
, compare solid circles with open squares; data from 5–7 independent experiments are summarized in B). The cAMP-induced K+ conductances were sensitive to BaCl2, consistent with cAMP-gated K+ channels as the source of these currents. To test whether the Ca2+-gated K+ conductance was affected, we added the Ca2+-dependent agonist carbachol after Ba2+ treatment. Carbachol rapidly induced a strong Isc-blK above the newly established baselines that were equal in magnitude for monolayers treated or not treated with SMase (Fig. 2A, B
). This result shows specificity for SMase action on the cAMP-gated K+ current, as predicted by our studies on intact cells (Fig. 1A, D
).
Isc-blK-voltage relationships in control monolayers measured at steady state before and during forskolin treatments showed linear forskolin-activated outward currents (apical to basolateral) (Fig. 2C
, top panel). Reversal potentials were shifted after forskolin treatment to more negative values, consistent with a strong activation of basolateral cAMP-gated K+ conductances. In SMase-treated monolayers, however, the magnitude of Isc-blK induced by forskolin at all transepithelial voltages tested were attenuated, and the reversal potential was shifted to a less negative value, consistent with inhibition of basolateral cAMP-gated K+ conductance, as indicated by our studies described above (Fig. 2C
, bottom panel).
We next tested whether basolateral SMase treatment might affect apical membrane Cl– channel function, as it appears to do in Calu3 airway epithelial cells (6)
. To do this, we permeabilized the basolateral membrane of T84 monolayers in the presence of a symmetrical 142 mM choline chloride solution, where Cl– was the major charge-carrying ion. Under these conditions, the transepithelial Isc recorded represents Cl– transport across the apical membrane (termed Isc-apCl). No significant difference of forskolin-induced Isc-apCl was found between SMase treated and untreated monolayers (Fig. 2D
). Thus, basolateral SMase treatment has no detectable effect on apical Cl– channels in polarized T84 cells.
Altogether, these data indicate that in intact intestinal cells, the attenuation of the chloride secretory response induced by SMase can be explained by down-regulation of a cAMP-gated basolateral K+ conductance.
The production of ceramide explains the repression of Isc by SMase
We first considered the possibility that SMase may act on Isc by activating protein kinase C (PKC) (34
35
36)
, which can inhibit the cAMP-induced Cl– secretory response (unpublished results and ref. 37
). This could occur either by the promiscuous hydrolysis of phosphatidyl choline and subsequent production of diacylglycerol, or through downstream products of ceramide metabolism (25
, 26)
. To test this, we used the PKC inhibitor calphostin C. If PKC were involved in the SMase response, calphostin C should rescue the forskolin-induced Isc in SMase-treated monolayers. Figure 3A
shows that calphostin C by itself had no effect on forskolin-induced Isc (compare open circles to solid squares), and inclusion of calphostin C in the SMase treatment resulted in the same degree of forskolin-induced Isc repression as SMase by itself (compare solid and open triangles). Thus, PKC activation cannot be responsible for the SMase-induced attenuation of cAMP-dependent Cl– secretion.
The major action of SMase is to cleave SM to ceramide and phosphocholine. To discern which is the causative factor for the repression of cAMP-dependent Cl– secretion, we utilized SMase D from Corynebacterium pseudotuberculosis. Unlike SMase, SMase D hydrolyzes SM to ceramide-1-phosphate and choline (38)
. Ceramide-1-phosphate is more hydrophilic than ceramide and more closely resembles a phospholipid in structure. Monolayers pretreated with SMase (S. aureus) showed a strongly inhibited forskolin-induced Isc as before (Fig. 3B
, compare solid triangles with open squares). In contrast, SMase D pretreatment enhanced forskolin-induced Isc (150% of control; Fig. 3B
, compare solid circles with open squares). Thus, neither the depletion of SM from the basolateral membrane, nor the generation of ceramide-1-phosphate or choline, can explain the action of SMase on the Cl– secretory response.
We next tested whether phosphocholine is the active molecule by treating T84 monolayers basolaterally with increasing doses of phosphocholine. In contrast to the inhibition of Isc induced by SMase, the addition of phosphocholine alone caused a strong dose-dependent enhancement of forskolin-induced Isc (Fig. 3C
). Thus, the release of ceramide into the membrane explains the repressive action of SMase on cAMP-dependent Isc. To confirm this result, we used the retinoic acid analog 4-HPR. 4-HPR increases cellular ceramide levels by activating serine palmitoyltransferase and ceramide synthase, two enzymes involved in ceramide metabolism (39
, 40)
. Treatment of T84 monolayers with 4-HPR had no effect on resting Isc or TER, but the compound dose-dependently inhibited forskolin-induced Isc similar to that seen after SMase treatment (Fig. 3D
). This result and its behavior in experiments described below are fully consistent with the identification of ceramide as the active molecule mediating the SMase-induced response.
If ceramide is the active molecule, it is possible that S1P, a soluble second messenger that can be formed from ceramide catabolism (41)
, might act downstream to initiate the repression of forskolin-induced Isc by ligating an S1P receptor. To test this, S1P was applied basolaterally to both buffer and SMase-treated monolayers prior to forskolin addition. The forskolin response of S1P-treated monolayers was similar to buffer-treated controls, and the repression of forskolin-induced Isc observed in SMase-treated samples was unaffected by the inclusion of S1P (Fig. 3E
). Thus, S1P-mediated signaling events initiated in the exoleaflet of the basolateral PM cannot be responsible for the attenuation of forskolin-induced Isc observed in SMase treated cells. Altogether, the results from this series of experiments show that SMase represses cAMP-induced Isc by liberating ceramide in the basolateral PM.
The signal transduction pathway: studies on Erk1/2 and p38
The rapidity with which SMase acts to suppress cAMP-dependent Cl– secretion suggested to us that the effect might be mediated by signal transduction events initiated by ceramide. Ceramide can activate several signal transduction pathways, including those mediated by Erk1/2, p38, and JNK (42
43
44)
. We first tested Erk1/2. If ERK is involved, the Erk1/2 inhibitor PD98059 should rescue the SMase-induced repression of the forskolin secretory response. Figure 4A
shows that monolayers pretreated with PD98059 and then incubated with SMase (open diamonds) displayed the same forskolin response as those incubated with SMase alone (solid triangles, no rescue). PD98059 was active under these conditions because it strongly inhibited Erk1/2 phosphorylation induced by osmotic stress (Fig. 4B
, sorbitol treatment, compare lanes 3 and 4). Likewise, we tested the p38-mediated signal transduction pathway using the p38 inhibitor, SB203580. Monolayers pretreated with SB203580 and then incubated with SMase (open diamonds) displayed the same forskolin response as those incubated with SMase alone (Fig. 4C
, solid triangles, no rescue). SB203580 was active under these conditions because it inhibited the in vitro phosphorylation of ATF2 (downstream target of p38) by phospho-p38 immunoprecipitated from T84 cell extracts (Fig. 4D
; compare lanes 3 and 4, bottom two panels). Thus, neither Erk1/2 nor p38 mediate the ceramide-induced repression of basolateral K+ transport (measured as Isc).
JNK mediates the SMase-induced inhibition of basolateral K+ conductance
To test whether JNK action is responsible for the SMase effect on basolateral K+ conductance, we used two different JNK inhibitors. Here we find that the small compound JNK inhibitor SP600125 (45)
rescued the SMase-induced repression of Isc (Fig. 5A
, compare solid diamonds and triangles). SP600125 treatment had no effect on forskolin-induced Isc in the absence of SMase (Fig. 5A
, compare open circles and solid squares), indicating that the compound cannot be acting by off-target inhibition of the cAMP-gated K+ conductance as reported for Kv channels in small cell lung cancer and Jurkat cells (46)
. The ability of SP600125 to rescue the SMase-induced repression of the Cl– secretory response was dose dependent (Fig. 5B
), with 30 µM displaying near maximal effects, consistent with the potency of SP600125 in other systems (45)
. To show that SP600125 acted on JNK as expected, we measured SP600125 effects on the phosphorylation of c-Jun, a downstream target of JNK. Considerably less phospho-c-Jun was detected in T84 monolayers treated with SP600125 as compared to controls (Fig. 5C
; compare lanes 2 and 1, respectively). Thus, SP600125 effectively inhibits JNK activation in the T84 cell model.
To confirm this result, we incubated T84 monolayers with an inactivating JNK binding domain (JBD) peptide fused to the HIV-TAT48–57 peptide so as to allow cell entry. The HIV-TAT48–57 peptide alone was used as a negative control. While neither of these peptides directly affected cAMP-dependent secretion in buffer-treated T84 monolayers (Fig. 5D
, bars 1–3), only the JBD peptide caused rescue of the forskolin-induced Isc in SMase-treated monolayers (Fig. 5D
, compare bars 4 and 6, and 4 and 5). These data show that JNK inhibition by two independent methods rescues the SMase-induced inhibition of cAMP-dependent Isc.
To show directly that SMase activates JNK, we treated T84 monolayers with SMase or buffer alone and analyzed cell extracts for phosphorylated (activated) JNK (Fig. 5E
). Monolayers treated with SMase showed a marked increase in phosphorylated JNK (p-JNK, lane 2) as compared to untreated monolayers (lane 1). We also observed that forskolin alone induced a modest increase in p-JNK (Supplemental Fig. S3A) and, accordingly, a combined treatment of both SMase and forskolin resulted in an even greater increase in p-JNK than SMase alone (Fig. 5E
, lane 3). Maximal activation of JNK was demonstrated by treating T84 cells with sorbitol (Fig. 5E
, lane 4). Total JNK and β-actin protein levels were similar for all samples (bottom two panels of Fig. 5E
and Supplemental Fig. S3A). Since a small amount of JNK phosphorylation was observed after forskolin treatment (Supplemental Fig. S3A), it is predicted from the above studies that, in addition to stimulation of cAMP-sensitive channels, forskolin should also exert an opposing effect on basolateral K+ conductance via JNK activation. We tested this hypothesis by preincubating monolayers with SP600125 or buffer before forskolin addition. We found that both the rate and maximal Isc achieved were greater in the SP600125-treated monolayers (Supplemental Fig. S3B, C), further corroborating the connection between JNK activation and the attenuation of a basolateral K+ conductance.
Finally, we reasoned that if JNK mediates signal transduction downstream of ceramide, SP600125 should also rescue the effects of the retinoic analog 4-HPR on forskolin-induced Isc. This is what we find (Fig. 5F
; compare open triangles and circles). The results of multiple independent experiments are summarized in Fig. 5G
. Thus, ceramide accumulation in the basolateral PM of T84 cells induces JNK activation, and this mediates the inhibition of a basolateral cAMP-gated K+ conductance.
Signal transduction in trans at mucosal surfaces
Intestinal cells lack any evidence for endogenous SMase or its enzymatic activity on their basolateral membranes (16)
. Also, if SMase is to act at mucosal surfaces in vivo the enzyme must act in trans via another cell type. To show that such signal transduction can occur in trans, we used normal human fibroblasts, which secrete a processed form of acid-SMase (s-SMase) (12
, 16
, 27)
, and human fibroblasts isolated from a patient with Niemann-Pick disease type B, which are defective in the gene encoding acid-SMase. Incubation of T84 cells for 5 min with CM from normal fibroblasts caused the expected inhibition of forskolin-induced Isc, consistent with the action of SMase described above (Fig. 6A, B
). In contrast, no inhibition of Isc was detected in T84 cell monolayers incubated with CM from Niemann-Pick type B disease fibroblasts, or medium alone. Thus, SMase released from nonepithelial cell types that typify the lamina propria of the intestinal mucosa can act in trans to regulate epithelial salt and water secretion.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 47
, 48)
. Because ceramide is produced by an SMase that must be donated by another cell type (see below), the results suggest a way that cells resident in the lamina propria, invading microbes, or migrating inflammatory cells might act beneath the epithelial barrier so as to affect intestinal salt and water homeostasis. This pathway has implications for infectious and inflammatory diseases of the intestine.
Ceramide-induced activation of JNK
We identify ceramide as the active molecule that initiates suppression of cAMP-dependent Cl– secretion. This finding is supported by our studies using the atypical isoform SMase D, which demonstrate that the hydrolysis of basolateral SM to C1P has an opposite (enhanced) effect on cAMP-dependent secretion to that of SMase. This result is consistent with similar studies in oocyte membranes (49)
. The other metabolite of the SMase reaction, phosphorylcholine, also enhances Isc when applied alone. Thus, ceramide appears to be the decisive factor. Because the phosphorylcholine also liberated from the SMase reaction has opposing effects on cAMP-dependant Isc, our results may underestimate the actual inhibitory effects of ceramide.
We find that ceramide activates JNK in T84 cells, consistent with the activation of stress-activated protein kinases by ceramide in other experimental systems (42
43
44
, 50)
. Ceramide could act on JNK either by ligating a signaling receptor or indirectly by self-assembling into ceramide-rich membrane microdomains that displace or otherwise influence the activity of membrane-associated signal transduction proteins (20)
. So far, no conclusive evidence indicates that ceramide binds a high-affinity receptor that initiates JNK signaling (25
, 51)
, but these two paradigms seem equally possible, and both may operate in parallel. We also observed that stimulation of cAMP production by forskolin results in modest JNK activation in the T84 model. This finding is consistent with studies that found the guanine nucleotide exchange factor Epac to mediate cAMP-dependent activation of JNK in HEK-293T cells (52)
. However, the mechanism of JNK activation by cAMP in polarized T84 cells remains unknown. This finding and that the JNK inhibitor SP600125 augmented the kinetics and maximal Isc of the forskolin response are fully consistent with our conclusion that JNK activation mediates the down-regulation of cAMP-dependent secretion in intestinal epithelia.
Ceramide and K+ channels
Consistent with our results, ceramide has been shown to activate signal transduction cascades that inhibit K+ channels in other cell types. This action includes inhibition of an outwardly directed K+ current in rat pinealocytes via PKC activation (1)
, Kv1.3 in Jurkat T cells via tyrosine phosphorylation (3)
, and the HERG channel in HEK 293 cells by inducing ubiquitin-mediated degradation (4)
. The multiplicity of pathways affected by ceramide in different systems might be explained by differences in structure of the ceramide signaling molecule, the site of ceramide production, or the site of ceramide action (20)
. The other possibility that ceramide can directly affect K+-channel function by forming ceramide-rich microdomains in the PM (22
23
24)
do not fit our results as well because of the strong evidence for JNK as an upstream mediator.
We also observed a minor inhibition of Ca2+-dependent Isc in intact T84 monolayers that were treated basolaterally with SMase. It is unlikely that this condition represents repression of Ca2+-dependent K+ currents because no differences were observed in permeabilized monolayers that were clamped to K+. The small effect of ceramide on Ca2+-dependent Isc is reproducible and might be caused by inhibition of Ca2+-sensitive Cl– channels, but this possibility remains untested.
How might JNK activation affect K+-channel function? Although numerous nuclear and non-nuclear substrates for phosphorylation by JNK have been described (53)
, we do not yet understand how JNK functions to down-regulate basolateral K+ currents in T84 cells. Perhaps JNK may act through one (or more) of the described regulatory mechanisms that influence KCNQ1 function. These include PKA phosphorylation (54)
, modulation of PIP(2)-dependent activation of KCNQ channels (55)
, Rab5-mediated endocytosis of KCNQ1 (56)
, Nedd4/Nedd4-like-dependent ubiquitylation (57)
, or though interactions with regulatory β-subunits (presumably KCNE3).
Strict dependence on epithelial polarity
The action of ceramide on cAMP-dependent Isc in T84 cells depends strictly on its formation in the basolateral membrane. Conversion of SM to ceramide in the apical membrane has no effect on cAMP-dependent Isc (ref. 16
and the present study). Thus, ceramide produced at opposite surfaces of the mucosal barrier has different consequences for epithelial cell function. It is possible that the structure of the ceramide produced in the basolateral PM of intestinal cells is different from that produced apically, and the apical ceramides cannot initiate signaling in the same way. Another possible explanation is that the machinery required for ceramide to activate JNK (as yet unidentified) is sequestered near or within the basolateral membrane and is, therefore, not influenced by ceramide generated at the apical surface.
We recently found that the alkaline isoform of SMase (alk-SMase) expressed by T84 cells is an apically localized ectoenzyme that can use endogenous apical SM as substrate [(16)
and see (58)
]. Neither alk- nor acid-SMase proteins nor any SMase activity is detected at the basolateral cell surface. Such polarity of cell surface expression for SMase in the intestine in vivo has been confirmed by immunostaining in mice (16)
and humans (59)
. As such, we speculate that polarized intestinal epithelial cells actively maintain a basolateral PM that is both free of endogenous SMase activity and sensitive to the action of exogenous SMases donated by other cell types located in, or migrating into the lamina propria (the subepithelial space). That such crosstalk can occur between human intestinal epithelia and another cell type resident to the lamina propria is demonstrated by our use of normal and Niemann-Pick type B human fibroblasts possessing or impaired in acid-SMase activity, respectively.
SMases in the human intestine
How might SMases gain access to the serosal PM of intestinal enterocytes so as to affect the secretory response in vivo? One possibility is that intestinal injury or loss of barrier function could allow microbe-derived SMases access to these surfaces. This would not necessarily require the microbe to penetrate into the mucosa, as bacterial SMases are secreted enzymes (60)
, although such transepithelial migration of microbes is known to occur in the intestine, even in the absence of injury. Our studies with S. aureus and mutant strains lacking SMase show that such crosstalk between microbes and host can occur in vitro. S. aureus is known to colonize the human intestine (61
, 62)
and thus may typify a general principle for the effects of other SMase-producing microbes that inhabit the human intestine. Other intestinal microbes that express SMase include Helicobacter pylori, Clostridium perfringens (63)
, B. cereus (64)
, Listeria species (65
, 66)
, and Saccharomyces cerevisiae (67)
.
Another possibility is that immune or nonimmune cells endogenous to the lamina propria, or migrating through the lamina propria during states of inflammation or immune surveillance, might present SMases to serosal membranes of intestinal epithelial cells so as to modulate intestinal salt and water secretion as reported here. This could happen in a regulated way, mediated by diffusion of secretory SMases or by localized and direct contact with cells expressing SMase as an ectoenzyme on the cell surface (12
, 68)
. Lymphocytes have been shown to rapidly translocate SMase to the cell surface from intracellular locations in response to ligation of CD95 receptor by FasL (69)
, and recruitment of such activated lymphocytes to the intestine in states of inflammation is predicted by our results to affect intestinal function.
Overall, our results define a new mechanism for the regulation of Cl– secretion that is predicted to depend on enzymatic crosstalk with adjacent cells in the subepithelial space. This finding suggests a novel way that commensal and invading microbes, or migrating inflammatory cells endogenous to the host, might act beneath the epithelial barrier so as to affect mucosal biology.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 1, 2008. Accepted for publication August 28, 2008.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Isc (mean±SE, normalized for baseline) induced by forskolin or carbachol in buffer (–) or SMase treated (+) monolayers treated as in A. Baseline currents prior to agonist additions were subtracted from the maximal signal for each sample in each independent experiment. C) Steady-state current-voltage (Isc-blK/V) relation of K+ transport in apically permeabilized T84 monolayers pretreated with buffer (top panel) or SMase (bottom panel). Currents were recorded at indicated voltage-clamped potentials. Ordinate indicates Isc-blK measured for the same monolayer after addition of the indicated reagents. D) T84 monolayers were treated as in A, except HBSS was exchanged for high-chloride buffer (see ref. 31
