Proteolytic Processing of the Epithelial Adherens Junction Molecule E-Cadherin by Neutrophil Elastase Generates Short Peptides With Novel Wound-Healing Bioactivity.

We report the novel observation that the inflammatory protease, neutrophil elastase (NE), present in high abundance in inflamed tissue in inflammatory bowel disease (IBD) patients, is capable of cleaving the cellular adherens junction protein, E-cadherin. Proteolysis of E-cadherin by NE generates a variety of short peptides, several of which were observed in patient tissue samples, showing biological activity to promote wound closure in an in vitro model system. This effect is independent of proliferation either in a wounded monolayer or under subconfluent conditions, suggesting a primarily migratory activity upon colonic epithelial monolayers. We report the novel observation that inflammatory proteases post-translationally modify cellular junction proteins to create signaling peptides that contribute to the wound healing response and identifies a new mechanism of mucosal healing to be examined further in the context of chronic inflammatory diseases.

IBDs, including ulcerative colitis and Crohn’s disease, comprise a spectrum of chronic inflammatory gastrointestinal diseases of complex etiology. Although there is no one defined cause or trigger for IBD, the unifying feature across the spectrum of IBD is the concept of chronic relapsing and remitting inflammatory disease, primarily in the colon (although in Crohn’s disease inflammation may occur anywhere along the gastrointestinal tract). Mucosal healing now is considered the current gold standard in assessing IBD therapeutic remission, however, our understanding of how and why many patients fail to achieve healing remains poorly elucidated. Upon an inflammatory stimulus, the intestinal epithelial monolayer is compromised by bacterial insult at the luminal surface as collateral damage from degranulation and an oxidative burst from lamina propria granulocytes, primarily neutrophils, and from cytokines released from leukocytes. Neutrophils are the first immune cells recruited to areas of inflammation in IBD and sustained high infiltration of activated neutrophils in inflamed tissue is a hallmark of disease. However, neutrophils now also are considered to be important players in the resolution phase of the inflammatory response.5, 6 Neutrophils can interact directly with epithelial cells by transmigrating through epithelia and interacting with apical intercellular adhesion molecule 1 to enhance wound healing through activation of the Akt and β-catenin pathways. Damage to the intestinal epithelium causes a shift from a tight barrier to a migratory/repair phenotype, a process that involves the proteolytic cleavage of junctional proteins such as E-cadherin. Proteases can cleave epithelial junctional proteins to generate peptides that have biological activity that can affect the intestinal mucosa, and recently NE was shown to cleave E-cadherin in bronchial epithelial cells, although the potential effects of E-cadherin degradation peptides was not assessed in that study. Of particular interest is the recent observation that NE can be internalized by cells and is thus capable of processing both intracellular and extracellular substrates.10, 11 In this context, we examined whether the inflammatory protease NE could proteolytically cleave the adherens junction protein E-cadherin and, specifically, whether the peptides resulting from this cleavage event could affect epithelial wound healing.

We first used a cell-free system to digest human recombinant E-cadherin with purified human NE and used liquid chromatography tandem mass spectrometry (LC-MS/MS) to identify the resulting peptides and cleavage events that occurred. We found that NE was capable of cleaving E-cadherin efficiently, with 48 peptides identified with high confidence using mass spectrometry. Of these 48 peptides, we focused on 24 of these based on their frequency of occurrence, P value scoring, and accessibility of peptide cleavage site to proteases (Supplementary Table 1 and Supplementary Figure 1A and B). To further focus on the most physiologically relevant peptides, we sought to confirm the presence of these peptides in patient tissues. By using a modified extraction protocol, we obtained protein/peptide fractions from banked formalin-fixed, paraffin-embedded IBD and control samples, and identified peptides originating from E-cadherin using mass spectrometry enriched in IBD samples. Six of these peptides showed substantial overlap with 6 peptides identified from our cell-free digest, and we chose these 6 peptides for biological activity screening (Table 1). First, we tested our peptides for effects on wound healing capacity using a scratch assay, under both 10% serum and serum-free conditions, over 48 hours at concentrations of 1, 10, and 100 μg/mL using a high-throughput protocol we designed for the Incucyte live cell imaging system (EssenBiosciences Inc, Ann Arbor, MI) (see the Supplementary Methods section for detailed methodologies). Three peptides, designated E-cadherin peptide (EP)-15, EP-17, and EP-22, at concentrations of 100, 1, and 10 μg/mL, respectively, showed increased wound closure compared with untreated and vehicle controls, in both 10% serum (Figure 1) and serum-free (Supplementary Figure 1C) conditions. The peptides appeared to have a synergistic effect with 10% serum.

Supplementary Table 1

Sequences and Characteristics of Peptide Fragments of Human E-Cadherin Generated by NE In Vitro

Peptide name	Peptide sequence	Position	Domain
EcadP-1	FDYEGSGSEAA	833-843	α β γ catenin binding domain
EcadP-2	TVTDQNDNKPEFT	251-263	Between cadherin domains 1 and 2 (calcium binding)
EcadP-3	QAADLQGEGLSTTAT	346-360	Cadherin domain 2
EcadP-4	SSNGNAVEDPMEI	236-248	Cadherin domain 1
EcadP-5	NNDGILKTA	431-439	Cadherin domain 3
EcadP-6	QYNDPTQESI	641-650	Cadherin domain 5
EcadP-7	SLTTSTATV	465-473	Cadherin domain 3
EcadP-8	SEDFGVGQEI	496-505	Cadherin domain 4
EcadP-9	TNPVNNDGILKT	427-438	Cadherin domain 3
EcadP-10	ATDADDDVNTYNAAI	286-300	Cadherin domain 2 (calcium binding)
EcadP-11	TTNPVNNDGILKTA	426-439	Cadherin domain 3
EcadP-12	LNDDGGQFVV	417-425	Cadherin domain 3
EcadP-13	TYKGQVPENEANV	378-391	Cadherin domain 3 (calcium binding)
EcadP-14	RNDVAPTLM	784-792	Cytoplasmic
EcadP-15	KAADTDPTAPPYD	816-828	α β γ catenin binding domain
EcadP-16	SGIQAELL	133-140	Propeptide
EcadP-17	LPPEDDTRDNV	741-752	Cytoplasmic
EcadP-18	TGQGADTPPVGV	193-204	Cadherin domain 1
EcadP-19	LLILILLLLL	721-729	Cytoplasmic
EcadP-20	ALIIATDNGSPV	563-574	Cadherin domain 4
EcadP-21	EAGLQIPA	702-709	Ectodomain/transmembrane junction
EcadP-22	NRNTGVISVV	320-329	Cadherin domain 2
EcadP-23	TVTDTNDNPPIFNPT	364-377	Between cadherin domains 2 and 3 (calcium binding)
EcadP-24	NNDGILKT	431-438	Cadherin domain 3

NOTE. Samples were sequenced by mass spectroscopy to show 24 distinct peptides. Sequences are shown, as well as the regions of the full-length E-cadherin that they represent. Ecad, E-cadherin.

Supplementary Figure 1

( (B) Sequence logo cleavage sites of E-cadherin by NE as generated by WebLogo Software. The height of the amino acid 1-letter code illustrates the relative observed frequency. The dotted red line indicates the NE cleavage site. The local cleavage site residue pattern is [PVLTA]-[VITAD]↓[STA]-X-D-D. (C) E-cadherin peptides (designated EP-15, 17, and 22) significantly and synergistically enhanced healing in scratch-wounded Caco-2 monolayers over 48 hours in the absence of serum. *P < .05, **P < .01, ***P < .001 compared with vehicle and untreated controls (using a 2-way analysis of variance with the Bonferroni post-test). (D) Low-dose NE enhances wound healing in Caco-2 cells. Purified human NE showed a concentration- and activity-dependent wound-healing effect in scratch-wounded Caco-2 monolayers over 48 hours in the presence of serum (left panel), but not in the absence of serum (right panel). (E) The proteolytic activity of NE was confirmed to be reduced significantly after heat denaturing. *P < .05, compared with vehicle and untreated controls (using a 1-way analysis of variance with the (D) Dunnett post-test or (E) unpaired t tests with Welch correction). (C–E) N = 3–10 independent experiments, each with 3 or more technical replicates. AFU, arbitrary fluorescence unit; SF, serum free.

Table 1

Alignment of E-Cadherin Peptides Generated by Neutrophil Elastase In Vitro With Peptide Fragments Isolated From IBD Patient Tissue

Peptide	Position	Overlap
TAYFSLDTR	66-74	None
VTEPLDR	216-222	None
NTGVISVVTTGLDR	332-335	EP-22	KNMFTINRNTGVISVVTTGLDRESFPTYTL
GQVPENEANVVITTLK	382-397	EP-13, partially EP-23	IFNPTTYKGQVPENEANVVITTLKVTDADAPNTVTDTNDNPPIFNPTTYKGQVPENEANVVITTLKVTDADAPN
DTANWLEINPDTGA	528-541	None
ISTRAELDR	542-550	None
TIFFCER	559-565	None
MALEVGDYK	656-664	None
EPLLPPEDDTR	739-749	EP-17	LRRRAVVKEPLLPPEDDTRDNVYYYDE
GLDARPEVTR	775-784	Partially EP-14	FDLSQLHRGLDARPEVTRNDVAPTLM
NDVAPTLMSVPR	785-796	EP-14	DARPEVTRNDVAPTLMSVPRYLPRPANP
PANPDEIGNFIDENLK	801-816	Partially EP-15	SVPRYLPRPANPDEIGNFIDENLKAADTDPTAPPYD

NOTE. The left column indicates the peptides identified in patient tissue, the second column shows its position in the full E-cadherin protein, and the right columns indicate overlap with in vitro neutrophil elastase digestion of E-cadherin. Highlighted text indicates peptides identified in patient samples, and underlined text indicates peptides before trypsin treatment. Boxed text indicates E-cadherin peptides generated by NE cleavage.

Figure 1

E-cadherin peptides enhance wound healing in Caco-2 cells. E-cadherin peptides (designated EP-15, EP-17, and EP-22) significantly and synergistically enhanced healing in scratch-wounded Caco-2 monolayers over 48 hours in the presence of 10% serum. *P < .05, **P < .01, and ***P < .001 compared with vehicle and untreated controls (using a 2-way analysis of variance with the Bonferroni post-test). N = 5 independent experiments, each with 3 or more technical replicates. Tx, treatment.

The pro-healing effect was replicated by NE in a concentration-dependent manner, with 1 ng/mL having a small but significant effect in the presence of serum (Supplementary Figure 1D). The effect of NE was dependent on its catalytic activity because the effect on wound healing was blocked with heat denaturation, which was confirmed to significantly reduce NE activity (Supplementary Figure 1E). To determine whether this effect was due to an increase in the proliferation of cells in response to the presence of E-cadherin peptides, we used the 5-Ethynyl-2′-deoxyuridine (EdU) system to identify actively dividing cells and found that there was no significant difference in proliferation under serum-free or serum conditions. Concurrently, we used the terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling enzyme to identify cell death to determine any cytotoxicity of our peptides at their effective doses and found no cytotoxicity of these peptides on Caco-2 monolayers (Supplementary Figure 2). To examine whether these peptides may have a mitogenic effect under nonwounding conditions, peptides also were tested for biological activity under subconfluent conditions. Caco-2 cells were transfected with a green fluorescent protein construct, seeded at medium density and exposed to the 6 peptides at the concentrations described earlier. Cell number and morphology were tracked over 48 hours. No significant changes in cell number or cell spreading were seen in response to any of the 6 peptides in either serum-free or 10% serum conditions (data not shown), suggesting that these peptides do not have mitogenic properties, and that the biological activity of the E-cadherin peptides is primarily to increase the migratory capacity under wound-healing conditions. To assess whether E-cadherin peptides could enter Caco-2 cells to potentially evoke intracellular signaling pathways, 7-amino-4-methylcoumarin (AMC) fluorescently tagged versions of EP-15 and EP-17 were synthesized and shown to transmigrate across the plasma membrane and into the cytosol (Supplementary Figure 3).

Supplementary Figure 2

E-cadherin peptides do not significantly affect Caco-2 proliferation or apoptosis. Caco-2 cells were grown on plastic and scratch-wounded and stained with EdU as a marker of cell proliferation (green), terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) stain for apoptotic cells (red), and 4′,6-diamidino-2-phenylindole to show nuclei (blue) under (A and B) serum-free and (C and D) 10% serum conditions. Representative images (A and C) of wounded monolayers and summary data (B and D) for EdU staining and TUNEL staining at 48 hours. E-cadherin peptides had no effect on either parameter. Similar results were seen at 30-minute and 24-hour treatment times (data not shown). *P < .05, ***P < .001 (using a 1-way analysis of variance with the Dunn post-test). N = 3 independent experiments, each with 3 or more technical replicates.

Supplementary Figure 3

E-cadherin peptides are capable of crossing the lipid bilayer to enter the cytosol of Caco-2 cells. (A) EP-15 and (B) EP-17 entered the cytosol of scratch-wounded Caco-2 monolayers 5 days after confluency. *P < .01 compared with vehicle and untreated controls (using 1-way analysis of variance with the Dunnett post-test). N = 4 independent experiments, each with 3 or more technical replicates. AFU, arbitrary fluorescence unit.

In this research letter, we show the ability of an inflammatory protease, neutrophil elastase, to process the adherens junction protein E-cadherin to generate short peptide fragments with effects on epithelial function, and a novel role for low levels of NE being pro-resolution. These peptide fragments are present in IBD patient tissues and appear to enhance the wound-healing response of intestinal epithelial cell monolayers independently of cellular proliferation. Our study raises important questions about the cellular mechanism whereby NE-derived peptide fragments of E-cadherin stimulate an epithelial wound-healing response. Is the site of action of E-cadherin peptides on epithelial cells extracellular or intracellular, and what is the mechanism of transport across the cell membrane? What intracellular pathway(s) are altered by these peptides to modify epithelial cell behavior? Are other bioactive peptides proteolytically produced by NE or other inflammatory proteases as a resolution response, and what are the cellular targets of these peptides? Thus, our work provides the impetus for further research that will determine the signaling mechanisms underlying this phenomenon, identify potential new peptide biomarkers of inflammatory diseases, and develop new therapeutic targets. Overall, our discovery adds a new layer of complexity to our understanding of the signaling mechanisms underlying mucosal repair after inflammatory insult and suggests a new potential arm of repair signaling that may be dysregulated during IBD and other chronic inflammatory diseases.