Paneth cell cryptdins act in vitro as apical paracrine regulators of the innate inflammatory response.

Intestinal-specific antimicrobial alpha-defensins, termed cryptdins, are secreted into the intestinal lumen by mouse Paneth cells in response to microbial pathogens. Cryptdins kill microbes by forming pores in their limiting membranes. The cryptdin isoforms 2 and 3 also can form anion-conductive pores in eukaryotic cell membranes, thus affecting cell physiology. Here, we find that when applied to apical membranes of the human intestinal cell line T84, cryptdin 3 (Cr3) induces secretion of the proinflammatory cytokine interleukin 8 (IL-8) in a dose-dependent manner. The induction of IL-8 secretion is specific to the cryptdins that form channels in mammalian cell membranes because cryptdin 4, which does not form pores in T84 cells, does not induce IL-8 secretion. Cr3 induces inflammatory cytokine secretion by activating NF-kappaB and p38 mitogen-activated protein kinase in a Ca2+-dependent manner, but influx by extra-cellular Ca2+ is not involved. Unlike other known inflammatory agonists, signal transduction by Cr3 occurs slowly, suggesting a novel mechanism of action. These results show that selective cryptdins may amplify their roles in innate immunity by acting as novel paracrine agonists to coordinate an inflammatory response with the antimicrobial secretions of Paneth cells.

Paneth cells, located at the base of small intestinal crypts, participate in innate immunity against invading pathogens by secretion of lysozyme and the microbicidal ␣-defensins termed cryptdins (1). At least 20 cryptdin isoforms have been described in the mouse (cryptdins [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] and two, HD-5 and HD-6, in humans (2)(3)(4)(5). Cryptdins act by forming anion-conductive channels in microbial cell membranes that depolarize and kill the microbe (6 -9). We also find that certain cryptdins, cryptdins 2 and 3, can form apical anion-conductive channels in eukaryotic cell membranes. The formation of such channels in the apical membrane of crypt epithelial cells causes a salt and water secretory response that flushes the intestinal crypt of noxious agents (10). Crypt epithelial cells may further contribute to innate intestinal host defense by orchestrating the recruitment of immune cells via the secretion of chemokines into the lamina propria, such as the neutrophil chemokine interleukin 8 (IL-8). 1 IL-8 secretion initiates the first step in neutrophil recruitment into the mucosa and ultimately in the formation of the crypt abscess, which represents the hallmark of acute and chronic intestinal inflammation. Although Paneth cells respond to microbial pathogens by discharging their granule contents, it is not known whether the Paneth cell or the secreted cryptdins function as intrinsic components of the intestinal inflammatory response.
Pore formation in eukaryotic cells by a variety of agents causes many biological effects including cytokine release (11,12). The complement membrane attack complex (C5-9), for example, forms pores on endothelial cells that induce IL-8 secretion via NF-B activation (13). The pore-forming bacterial toxins such as staphylococcal ␣-toxin, streptolysin O, and Escherichia coli ␣-hemolysin activate NF-B and cause IL-8 secretion from mammalian cells when applied at non-cytotoxic doses (12, 14 -16). In most cell types, E. coli ␣-hemolysin and staphylococcal ␣-toxin act by conducting extracellular Ca 2ϩ influx into the cell (11,14). In some cell types, however, staphylococcal ␣-toxin forms a pore that conducts only monovalent ions, but it still can induce release of the cytokine IL-1␤ or the induction of apoptosis (11,17).
Here, we report that the Paneth cell ␣-defensin cryptdin 3 (Cr3) induces IL-8 secretion from intestinal T84 cells via the Ca 2ϩ -dependent activation of the p38 mitogen-activated protein kinase (MAPK) and NF-B signaling cascades. The activity is specific to the pore-forming activity of Cr3 in mammalian cell membranes because cryptdin 4 (Cr4) that does not form ion-conducting pores in T84 cells fails to induce an inflammatory response. Unlike most other pore-forming agonists, however, Cr3 does not act by inducing an influx of extracellular Ca 2ϩ . Signal transduction occurs through activation of NF-B and MAPK, but the time course of activation and transcription is very slow. These results suggest that certain cryptdins may act in vivo via a novel mechanism of paracrine signal transduction to amplify their role in innate immunity by coordinating the antimicrobial activities of Paneth cell secretions with the induction of an intestinal inflammatory response.

EXPERIMENTAL PROCEDURES
Cell and Bacterial Culture-T84 cells, an intestinal cell line originally obtained from metastatic human colorectal carcinoma, were used to model intestinal epithelia. Confluent monolayers of T84 cells were grown on collagen-coated inserts (Costar Corning, Cambridge, MA) as described previously (18). Wild-type Salmonella typhimurium strain x3306 was maintained and prepared for use via non-agitated microaerophilic conditions as described previously (19).
IL-8 Secretion-Agonist-induced IL-8 secretion from T84 cells was measured by enzyme-linked immunosorbent assay as described previously (22,23). Briefly, T84 cells on 0.33-cm 2 inserts were washed and incubated in 50 l of apical and 300 l of basolateral HBSS (unless otherwise indicated) for at least 10 min. Appropriate agonists were added for 5 h and then basolateral solutions were collected for IL-8 measurement.
Drug Treatments-To chelate intracellular Ca 2ϩ , T84 cells were treated with 30 M BAPTA-AM (Molecular Probes Inc., Eugene, OR) for 45 min before agonist addition in HBSS with reduced Ca 2ϩ as described previously (24). To chelate extracellular Ca 2ϩ , T84 cells were treated with 10 mM EGTA (Sigma) 1 h prior to agonist addition. Recombinant TNF-␣ (R&D Systems Inc., Minneapolis, MN), carbachol, and ionomycin (Sigma) were added as described in the figure legends.
Western Blot Analysis-After experimental treatment, T84 cells (0.33 cm 2 ) were washed in ice-cold HBSS. Then cells were harvested, and whole cell extracts were prepared by removing the cell-coated filters and placing them directly into SDS loading buffer for rapid lysis. Lysates were analyzed by SDS-PAGE and immunoblot using antiphospho-IB-␣ (New England Biolabs Inc., Beverly, MA), anti-IB-␣ (Santa Cruz Biotechnology Inc., Santa Cruz, CA), anti-phospho-p38 (Cell Signaling Technology Inc., Beverly, MA), and anti-p38 antibodies (Cell Signaling Technology Inc.). Because phospho-IB-␣ is rapidly degraded in vivo, we pretreated monolayers analyzed for phospho-IB-␣ with the cell-permeant proteasome inhibitor MG-262 (Calbiochem-Novabiochem Corp.).
p65 Immunostaining-Following experimental treatment, T84 cells were fixed and immunostained for the p65 subunit of NF-B as described previously (25).
Real-time Quantitative Reverse Transcription-PCR-After experimental treatment, T84 cells (0.33 cm 2 ) were washed in ice-cold HBSS and RNA was extracted using TRIzol reagent (Invitrogen). Total RNA was reverse transcribed from random hexamer primers using Multiscribe reverse transcriptase (Applied Biosystems, Foster City, CA). Real-time quantitative PCR analysis (SYBR Green real-time PCR assay, Applied Biosystems) was performed on the reverse transcription cDNA products using primers for IL-8 (Primer Express software, Applied Biosystems) and 18 S ribosomal RNA (Taqman ribosomal RNA control reagents kit, Applied Biosystems) as described previously (26). IL-8 expression level was normalized to the 18 S rRNA level of the same sample. Fold difference was the ratio of the normalized value of each sample to that of untreated control cells. (IL-8 primer sequences are: forward, 5Ј-aaaccaccggaaggaaccat; reverse, 5Ј-gccacgttggaagtcatgt.) RayBio TM Inflammation Antibody Array Analysis-After 5 h agonist treatment, the basolateral-conditioned media from stimulated T84 cells (0.33 cm 2 , 4 inserts per condition) was isolated and assessed for cytokine secretion. Antibody array membranes were incubated in the conditioned media, and cytokines were detected per the manufacturer's guidelines (inflammation antibody array III, Ray Biotech, Inc., Norcross, GA).
Statistical Methods-Data were analyzed for significance by Student's t test.

Cr3
Induces IL-8 Secretion from T84 Cells-To test whether certain cryptdins may participate differentially in an in vitro intestinal inflammatory response, we utilized the human in- testinal cell line T84 and purified refolded synthetic or recombinant mouse Cr3 and Cr4. The inflammatory response was assessed by immunoassay for IL-8 secretion. When applied to apical membranes of T84 cells, Cr3 induced IL-8 secretion in a dose-dependent manner (Fig. 1a, first through third columns). The IL-8 secretory response induced by Cr3 is comparable to that induced by the proinflammatory apical agonist S. typhimurium (Fig. 1a, compare third and fourth columns). Because pore-forming toxins and complement also induce cytokine secretion from target cells, we tested whether signal transduction by Cr3 depended on the ability of Cr3 to form anion-conductive channels in eukaryotic cell membranes. Consistent with this idea, Cr4, which cannot form pores in T84 cells (10), had no effect on IL-8 secretion (Fig. 1b, second column). In contrast, both Cr3 and the proinflammatory basolateral agonist TNF-␣ induced robust IL-8 secretory responses of comparable magnitudes (Fig. 1b, third and fourth columns). These results show that Cr3 can act apically to induce IL-8 cytokine secretion from T84 intestinal cells and suggest that the mechanism of action may require channel formation in the apical membrane.
Cr3 Induces IL-8 Secretion via Activation of the Transcription Factor NF-B and MAPK-To examine the mechanism of signal transduction by Cr3, we tested for the induction of phospho-IB-␣ by immunoblot (Fig. 2a). Phospho-IB-␣ was first detected after 2 h of incubation with Cr3 (Fig. 2a, lane 3) and increased with time ( compare lanes 1-4). In contrast, the inflammatory agonist TNF-␣ applied to basolateral membranes of T84 cells induced phosphorylation of IB-␣ within 5 min (Fig. 2a, lane 5). This study shows that Cr3 acts via the apical membrane of intestinal cells to induce phosphorylation of IB-␣, but the time course of activation was slow compared with the classic inflammatory cytokine TNF-␣.
Next we tested whether Cr3 induced the translocation of NF-B from the cytosol to the nucleus as would be predicted for an inflammatory agonist acting via the phosphorylation of IB-␣. Translocation of NF-B from the cytosol to the nucleus in T84 cells was assessed by immunofluorescence microscopy for the p65 subunit of NF-B (Fig. 2, b-g). Unstimulated cells had cytoplasmic staining but no nuclear staining for the p65 subunit (Fig. 2b). In cells treated basolaterally with TNF-␣, p65 was readily visualized inside the nucleus 1 h after application of the agonist (Fig. 2c). In cells treated apically with Cr3, the p65 subunit also was visualized inside the nucleus, but translocation to the nucleus was first visualized only 3-4 h after addition of Cr3 (Fig. 2, d-g). These data are consistent with the slower time course of IB-␣ phosphorylation induced by Cr3 and confirm that Cr3 acts as an inflammatory agonist via the transcription factor NF-B.
To determine whether Cr3 may also act via activation of p38 MAPK, we tested for agonist-induced phosphorylation of p38 MAPK by immunoblot (Fig. 3). Cr3 causes phosphorylation of p38 that was first observed 30 min after application to T84 cells and peaked at 2 h (Fig. 3, lanes 1-6). In contrast, Cr4, which does not form pores in T84 cells, had no effect (lanes 13-18). Like Cr3, TNF-␣ also induced an increase in phospho-p38, but the phosphorylation occurred more rapidly with peak levels observed at 30 min (lanes 7-12).
To confirm the slower time course of signal transduction and also the transcriptional regulation of IL-8 secretion by Cr3, we assessed agonist-induced IL-8 mRNA levels by quantitative Cr3 exhibited no synergy with carbachol (mean Ϯ S.D., representative of three independent experiments performed in duplicate). b, IL-8 secretion induced in T84 cells stimulated with 100 g/ml apical Cr3, 10 ng/ml basolateral TNF-␣, or Cr3 with TNF-␣. Cr3 and TNF-␣ acted synergistically to induce more IL-8 secretion than either agonist alone (mean Ϯ S.D., representative of three independent experiments performed in duplicate). c, IL-8 secretion induced in T84 cells stimulated with 10 ng/ml basolateral TNF-␣, 100 M basolateral carbachol, or 100 g/ml apical Cr3 in the presence (gray bars) or absence (white bars) of the intracellular Ca 2ϩ chelator BAPTA-AM (30 M) (mean Ϯ S.E., representative of three independent experiments performed in triplicate; *, p Ͻ 0.05).

FIG. 3. Cr3 induces activation of p38 MAPK.
Western blot analysis of phospho-p38 levels in T84 cells stimulated with Cr3 (100 g/ml), TNF-␣ (10 ng/ml), or Cr4 (100 g/ml) at the indicated time points. Equal loading was confirmed by assessment of total p38 on each Western blot. real-time PCR (Fig. 4). T84 cells stimulated with TNF-␣ expressed increasing IL-8 mRNA levels over time with strong expression at 2 h and peak expression 3 h after the application of the agonist (Fig. 4, dark bars). IL-8 mRNA levels also increased over time after stimulation by Cr3 but with almost no expression until 3 h and peak expression at 4 h (Fig. 4, light bars), consistent with the slower time course of Cr3-induced activation of NF-B and p38 MAPK. These results show that Cr3, like TNF-␣, induces IL-8 secretion in intestinal T84 cells via signaling cascades that involve both NF-B and p38 MAPK but with slower kinetics.
Proinflammatory Signal Transduction by the Pore-forming Cr3 Does Not Require Influx of Extracellular Ca 2ϩ -Because other pore-forming proteins may induce proinflammatory signaling cascades by permeabilizing the plasma membrane to extracellular Ca 2ϩ , we tested whether Cr3 acts on intestinal cells by this mechanism. First we asked whether Cr3 acts by increasing intracellular Ca 2ϩ . Cr3 was applied to T84 intestinal cells alone or together with the muscarinic agonist carbachol, which induced IL-8 secretion by elevating [Ca 2ϩ ] i , or with TNF-␣, which acted independently of [Ca 2ϩ ] i (Fig. 5, a and b). Here, we found that Cr3 acts synergistically with TNF-␣ to induce a proinflammatory response (Fig. 5b, compare the fourth column with the second and third). In contrast, Cr3 did not act synergistically with the Ca 2ϩ -dependent agonist carbachol (Fig. 5a, compare the fourth column with the second and third). These results suggest that the signaling cascade induced by Cr3 overlaps with that induced by carbachol and may indeed involve the induction of intracellular Ca 2ϩ transients.
To test this idea, we used the membrane-permeant Ca 2ϩ chelator BAPTA-AM. BAPTA-AM had no effect on IL-8 secretion from T84 cells induced by the Ca 2ϩ -independent agonist TNF-␣ (Fig. 5c, compare the third and fourth columns) but inhibited the IL-8 secretory response induced by the Ca 2ϩ -dependent agonist carbachol (Fig. 5c, compare the fifth and sixth  columns). BAPTA-AM also inhibited IL-8 secretion induced by Cr3 (Fig. 5c, the seventh and eighth columns). These results suggest that Cr3 causes an increase in intracellular Ca 2ϩ in intestinal cells that activates the proinflammatory signaling cascade involving NF-B and p38 MAPK.
To test whether Cr3 acts by permeabilizing the apical plasma membrane to Ca 2ϩ , we examined the IL-8 secretory response to inflammatory agonists in the absence of extracellular Ca 2ϩ . Ca 2ϩ was removed from the apical buffer by using the non-membrane-permeant Ca 2ϩ chelator EGTA. As predicted, apical EGTA strongly inhibited IL-8 secretion induced by application of the Ca 2ϩ ionophore ionomycin to apical membranes of T84 cells (0.7 versus 0.1 ng/ml) (Fig. 6a). Unexpectedly, however, EGTA also caused a 50% reduction in the IL-8 secretory response to the basolateral application of the Ca 2ϩindependent agonist TNF-␣. The reason for this effect on TNF-␣ signal transduction was not explained. Nonetheless, apical EGTA had no detectable effect on the IL-8 secretory response induced by apical Cr3. Thus, Cr3 cannot act by permeabilizing the apical membrane to Ca 2ϩ and inducing extracellular Ca 2ϩ influx.
To confirm these results, we measured IL-8 secretion induced by Cr3 in T84 cells bathed apically and basolaterally in nominally free Ca 2ϩ buffer (Fig. 6b). Consistent with our results using EGTA, IL-8 secretion induced by the Ca 2ϩ ionophore ionomycin (0.8 versus 0.2 ng/ml) was strongly inhibited by the removal of extracellular Ca 2ϩ . The removal of extracellular Ca 2ϩ , however, had no effect on IL-8 secretion induced by the Ca 2ϩ -independent agonist TNF-␣ (1.9 versus 1.6 ng/ml) or by Cr3 (1.1 versus 0.8 ng/ml) (Fig. 6b). Thus, signal transduc-tion by Cr3 does not require influx of extracellular Ca 2ϩ into the cell.
Cr3 Induces Secretion of Multiple Proinflammatory Cytokines from T84 Cells-To determine whether the Paneth cell cryptdins act in a general pathway to initiate an inflammatory response, using a commercially available immunoblot array to 40 known proinflammatory proteins, we tested for the production and release of other cytokines. Apically applied Cr3 induced the basolateral secretion of 6 proinflammatory cytokines: macrophage inhibitory protein (MIP)-1␣ (Fig. 7,

DISCUSSION
The results of this study show that Paneth cells may play a critical role in coordinating the innate immune response to invading pathogens in the intestine. Somehow, Paneth cells respond to the presence of microbial pathogens by discharging their granule contents (1). The exocytosis of Paneth cell granules delivers high concentrations of antimicrobial ␣-defensins, termed cryptdins, into the intestinal crypt lumen. At these concentrations, the cryptdins disrupt microbial cell membranes, thus exerting their anti-microbial activities (1). Cryptdins 2 and 3 also can form pores in eukaryotic cell membranes in vitro (10). Such activity may be cytolytic in vivo. In our first studies on these peptides, however, we discovered that the cryptdins have their own cytoprotective mechanism that could prevent such potential damage to the intestine precisely be- cause they can insert anion-conducting pores into the apical membrane of intestinal epithelial cells in vitro and presumably in epithelial cells lining the intestinal crypt in vivo (10). Such pore formation that causes a Cl Ϫ secretory response would flush the crypt lumen and reduce the concentration of cryptdins below cytotoxic levels. We now find that this same activity allows the cryptdins to act as novel apical paracrine factors that induce an inflammatory response.
The proinflammatory activity of the cryptdins depends on their pore-forming activities. The cryptdins that cannot form pores in eukaryotic cell membranes do not induce an inflammatory response. Here, it is interesting to point out that Cr4, which cannot form pores in T84 cells and does not induce an inflammatory response, is nonetheless the most bactericidal of the cryptdin peptides and is highly membrane disruptive in bacteria (20). Although the structural differences between Cr3 and Cr4 are known, still we do not understand why these potent pore-forming peptides in bacteria display such strict specificity in their action on eukaryotic cell membranes.
We also find that signal transduction by Cr3 occurs through Ca 2ϩ -dependent activation of NF-B and p38 MAPK, but formation of the cryptdin pore does not induce the influx of extracellular Ca 2ϩ , which is not required for the inflammatory response. These results are consistent with our previous studies on the ion-conducting properties of the Cr3 channel in intestinal T84 cells and by patch clamp technique in 293 HEK cells (10,27).
There are other pore-forming proteins such as complement and certain bacterial toxins that induce IL-8 secretion from mammalian cells. Most of these act by permeabilizing the target cell membrane to Ca 2ϩ (11,14), but the cryptdins do not act in this way. Presumably, the cryptdins induce an increase in Ca 2ϩ by release of intracellular stores. We also find that the cryptdins induce a proinflammatory response very slowly even when compared with other Ca 2ϩ -dependent agonists. S. typhimurium, for example, induces a Ca 2ϩ -dependent inflammatory response from T84 cells by ligand binding to TLR5 on the basolateral membrane. Like Cr3, Salmonella also acts by activation of NF-B and p38 MAPK, but signal transduction by Salmonella is faster, causing phosphorylation of IB-␣ within 30 min and phosphorylation of p38 within 1 h (24,28). Salmonella, also unlike Cr3, causes an influx of extracellular Ca 2ϩ as well as the release of Ca 2ϩ from intracellular stores (24). Thus, the mechanism of signal transduction by Cr3, although similar to other proinflammatory agonists in activating NF-B and p38 MAPK, is in other ways unique.
In certain cell types, some bacterial toxins, when applied at low doses, form pores in target cell membranes that do not conduct Ca 2ϩ and yet, like Cr3, induce an inflammatory response. The induction of an inflammatory response correlates with efflux of K ϩ through the toxin pore and presumably on depolarization of the target cell membrane (11,17). Perhaps Cr3 acts by a similar mechanism but via the induction of an anion rather than cation conductance.
Finally, the ability of Cr3 to act as a paracrine regulator of the intestinal inflammatory response raises the possibility that dysregulation of Paneth cell function may contribute to the development of disease. Although normal Paneth cell function may act in host defense, the secreted Paneth cell products are intrinsically cytotoxic and, as we show in this study, potentially proinflammatory. Thus, the regulation of Paneth cell secretions may run a fine line between health and disease. In patients with inflammatory bowel disease, for example, high colonic levels of the proinflammatory cytokines IL-8, IL-6, IL-1␤, and TNF-␣ (29) induced increased neutrophil activity that correlates with disease severity (30). It also is possible that the developmentally normal deficiency of Paneth cells and their secretory products in the neonatal period of the human predisposes some premature infants to necrotizing enterocolitis, an inflammatory disease of the intestine related to deficiency in innate immunity (31)(32)(33). Here again, the complex interplay among cell types of the intestinal mucosa in the inflammatory response may critically involve the antimicrobial and proinflammatory activities of the pore-forming Paneth cell ␣-defensins.