CXCR4 and CXCR7 Mediate TFF3-Induced Cell Migration Independently From the ERK1/2 Signaling Pathway

Abstract

Purpose: Trefoil factor family (TFF) peptides, and in particular TFF3, are characteristic secretory products of mucous epithelia that promote antiapoptosis, epithelial migration, restitution, and wound healing. For a long time, a receptor for TFF3 had not yet been identified. However, the chemokine receptor CXCR4 has been described as a low affinity receptor for TFF2. Additionally, CXCR7, which is able to heterodimerize with CXCR4, has also been discussed as a potential TFF2 receptor. Since there are distinct structural similarities between the three known TFF peptides, this study evaluated whether CXCR4 and CXCR7 may also act as putative TFF3 receptors.

Methods: We evaluated the expression of both CXCR4 and CXCR7 in samples of human ocular surface tissues and cell lines, using RT-PCR, immunohistochemistry, and Western blot analysis. Furthermore, we studied possible binding interactions between TFF3 and the receptor proteins in an x-ray structure-based modeling system. Functional studies of TFF3–CXCR4/CXCR7 interaction were accomplished by cell culture–based migration assays, flow cytometry, and evaluation of activation of the mitogen-activated protein (MAP) kinase signaling cascade.

Results: We detected both receptors at mRNA and protein level in all analyzed ocular surface tissues, and in lesser amount in ocular surface cell lines. X-ray structure-based modeling revealed CXCR4 and CXCR7 dimers as possible binding partners to TFF3. Cell culture–based assays revealed enhanced cell migration under TFF3 stimulation in a conjunctival epithelial cell line, which was completely suppressed by blocking CXCR4 and/or CXCR7. Flow cytometry showed increased proliferation rates after TFF3 treatment, while blocking both receptors had no effect on this increase. Trefoil factor family 3 also activated the MAP kinase signaling cascade independently from receptor activity.

Conclusions: Dimers CXCR4 and CXCR7 are involved in TFF3-dependent activation of cell migration, but not cell proliferation. The ERK1/2 pathway is activated in the process, but not influenced by CXCR4 or CXCR7. These results implicate a dependence of TFF3 activity as to cell migration on the chemokine receptors CXCR4 and CXCR7 at the ocular surface.

Trefoil factor family peptide 3 (TFF3, intestinal trefoil factor) is one of the three known trefoil factor family peptides first detected in the mammalian intestine. At mucosal surfaces, TFF peptides have been shown to carry out different functions (for review, see Ref. 1) like proliferation and migration enhancement, antiapoptosis, and wound healing. However, catabolic functions such as activation of matrix metalloproteinases and proapoptotic effects have also been described outside of mucosae.² The three known TFF peptides (TFF1, -2, and -3) are very consistent in structure, all of them contain a particular trefoil domain consisting of 40 to 41 amino acids with three preserved cysteine residues.³ 

Trefoil factor family peptides are associated with mucus. Human TFF3 has been found in nearly all mucosae and several glands (e.g., the gastrointestinal, respiratory, and urogenital tracts; the mammary and salivary glands), but also in nonepithelial structures such as articular cartilage.^2,4,5 The natural concentration of TFF3 in human body fluids varies between 12 nmol/L (∼94 ng/mL, gastric juice) and 870 nmol/L (∼6.8 μg/mL cervical mucus).⁶ 

With regard to the eye, which our group uses as a model system, TFF3 has been described in goblet cells of the conjunctiva, the tear film, and the epithelium of the nasolacrimal ducts.^7,8 It is not detectable in healthy cornea, but is induced under pathologic conditions like keratoconus and herpetic keratitis,⁹ as well as after experimental epithelial damage in the remaining epithelial cells at the defect rim.¹⁰ In human epithelial cell lines, TFF3 is additionally induced under exposure to UV-B light, after stimulation with proinflammatory cytokines (tumor necrosis factor [TNF]-β, interleukin [IL]-1β, interferon γ) and under hyperosmolar conditions.¹¹ 

The functions of TFF3 in the body are various. Its dimeric more than its monomeric isoform is thought to interact directly with mucins, increasing viscosity and elasticity of mucin-containing fluids like the tear film.^8,12 It has been hypothesized that this may be their leading mechanism of action. The motogenic effects of TFF3 play an important role during epithelial restitution and ensure rapid sealing of the epithelial layer after injuries.¹⁰ 

For a long time receptors for TFFs have been unknown. In 2009 the chemokine receptor type 4 (CXCR4, also stromal cell–derived factor 1 [SDF-1] receptor) was described as a low affinity receptor for TFF2.¹³ Chemokine receptor type 4 is an almost ubiquitously present membrane receptor that, in combination with its main substrate SDF-1 (CXCL12), is involved in a variety of chemotactic processes, thus regulating precisely the fate of migrating cells.¹⁴ Its role in stem cell trafficking, mobilization, and targeted migration to their reservoirs during development and adulthood is strongly illustrated by its various effects on organogenesis, regeneration, damage repair, tumorigenesis and metastasis.¹⁵ In the eye, presence of CXCR4 has previously been reported in limbal and corneal epithelial and stromal cells, where it contributes to the attraction of stromal niche cells to epithelial stem cells to prevent their differentiation.¹⁶ It is also present in giant papillae in atopic keratoconjunctivitis.¹⁷ 

Chemokine receptor type 4 consists of seven transmembrane helices and signals mainly through activation of G proteins. While SDF-1 was long considered the only CXCR4 ligand, there has been recent evidence that it has in fact multiple substrates (e.g., the macrophage migration inhibitory factor [MIF]).¹⁸ Beta defensin 3 is also a competitor with SDF-1 for CXCR4, and acts as a natural antagonist at the receptor.¹⁹ 

Chemokine receptor type 7 (CXCR7) is of the same subfamily and also contains seven transmembrane helices. It is present in T lymphocytes and regulates SDF-1–mediated chemotaxis.²⁰ High levels of CXCR7 were found in the heart, brain, spleen, kidney, lung, bladder, skeletal muscles, Langerhans islets, cartilage, synovia, testes, ovary, and placenta.^21–23 Just like CXCR4, it is overexpressed in a variety of tumor cells and facilitates tumorigenesis in multiple different tissues.²⁴ 

Chemokine receptor type 7 also has been shown to be expressed by corneal and limbal epithelial and stromal cells.¹⁶ It has been described to bind SDF-1 and ITAC (CXCL11), the latter with a 10-fold lower affinity than SDF-1. In contrast to CXCR4, CXCR7 has been shown to be unable to recruit G-proteins and mobilize calcium influx after ligand binding.²⁴ For a considerable amount of time, its only function had been assumed to be modulating the CXCR4 effects, probably by scavenging or sequestering SDF-1, thus creating SDF-1 gradients leading to altered CXCR4 signaling.²⁵ This view had to be revised, as its activation by SDF-1 alone was described as inducing intracellular signaling events, with CXCR7 acting as a “biased receptor,” (e.g., through β-arrestins), thus activating MAP kinases.^26,27 Receptors CXCR4 and CXCR7 may dimerize, homologously as well as heterologously, creating a variety of binding possibilities.^27,28 Chemokine receptor type 7 has been suggested as a possible binding partner for TFF2 as well²⁹ and may therefore also qualify as a putative TFF3 receptor. 

In addition to the knowledge that CXCR4 acts as a low-affinity receptor for TFF2, there is evidence that it also engages with the sister peptide TFF3. Xue et al.³⁰ demonstrated the inhibitory effect of CXCR4 blockade via the competitive antagonist AMD3100 on gastric wound healing induced by TFF3 in TFF2 null mice. 

The present study uses ocular surface tissues as a model system in which the presence and effects of TFF3 have been extensively studied, while until now the expression of CXCR4 and CXCR7 has only been investigated marginally. Here, we investigate whether there is a dependence of TFF3 activity on the chemokine receptors CXCR4 and CXCR7. 

Materials and Methods

Modeling of TFF3 and CXCR4/CXCR7 Interaction

The x-ray structure of the homodimer CXCR4 (pdb-entry: 3ODU)³¹ crystallized as lysozyme chimera was used to study interactions with TFF3. Since the lysozyme part is located in the intracellular domain, it was assumed that it does not influence the docking behavior of TFF3, which is expected to be located on the extracellular side. The x-ray structure of TFF3 has also been determined previously (pdb-entry: 1PE3).³² Hydrogen atoms were added to both proteins by applying the 3D-protonate option of a molecular operating environment (MOE, 2012.10; Chemical Computing Group, Inc., Montreal, QC, Canada). Since there is no x-ray structure for CXCR7, a homology model for the dimeric structure was built using the x-ray structure of CXCR4 (3ODU) as a template. This structure was modeled by using MOE with the new force field (AMBER12 EHT, new amber version implemented in MOE; Chemical Computing Group, Inc.). The quality of the model was checked with PROCHECK³³ and fulfilled all stereochemical criteria for a reasonable model. Subsequently, rigid body protein–protein docking studies were performed with PLANTS³⁴ by choosing an atom in the center of the extracellularly located loops as origin, with a radius of 40Å, to define the putative binding site of TFF3. Fifty different poses were generated for each docking run. Finally, the best poses were energy minimized with the force field (Chemical Computing Group, Inc.) to allow induced fit and final formation of hydrogen bonds and salt bridges which could not be reached perfectly during the rigid body docking. 

Ocular Sample Preparation

Tissues of the ocular surface and the lacrimal apparatus as well as lung tissue for positive controls were obtained from cadavers donated to the Department of Anatomy and Cell Biology, Martin Luther University Halle-Wittenberg, Germany. The donors (age 67–82 years) were free of recent trauma, eye and nasal infections, and diseases involving or affecting lacrimal function. All tissues were dissected from the cadavers within a time frame of 4 to 24 hours post mortem. After dissection, the tissues were prepared for paraffin embedding (right eye) by 4% paraformaldehyde fixation or were used for molecular biological investigation (left eye) and were immediately frozen at −80°C. 

Cell Lines

Simian vacuolating virus 40–transformed human corneal epithelial (HCE) cells (provided by Kaoru Araki-Sasaki, Tane Memorial Eye Hospital, Osaka, Japan),³⁵ as well as a spontaneously immortalized epithelial cell line from normal human conjunctiva (IOBA-NHC, here referred to as HCjE cells; provided by Yolanda Diebold, University Institute of Applied Ophthalmobiology, University of Valladolid, Valladolid, Spain)³⁶ were cultured as monolayers and used for stimulation experiments. We cultivated HCE cells in Dulbecco's modified Eagle's medium (DMEM)/HAM'S F12 1:1 medium, supplemented with 10% fetal calf serum (FCS; growing medium). The growing medium for HCjE cells also contained 1 μg/mL bovine insulin and 5 μg/mL hydrocortisone. The human breast cancer cell line MCF7 that served as a positive control was obtained and cultivated as described.³⁷ 

RNA Extraction, cDNA Synthesis, and Polymerase Chain Reaction

For reverse transcriptase polymerase chain reaction (RT-PCR), the following samples of ocular surface tissue and lacrimal apparatus were investigated: cornea, conjunctiva, and lacrimal gland (n = 3 for each tissue). Tissue samples were first crushed in an agate mortar under liquid nitrogen and subsequently homogenized (Polytron, Norcross, GA, USA). Total RNA was extracted following the manufacturer's protocol (RNeasy Mini Kit; Qiagen, Hilden, Germany). Reverse transcription of the RNA samples to first-strand cDNA was performed according to the manufacturer's protocol (RevertAid H Minus M-MuL V Reverse Transcriptase Kit; Fermentas, St. Leon-Rot, Germany). For each reaction, 2 μg total RNA and 10 pmol Oligo (dT) primer (Fermentas) were used. Each PCR reaction was prepared with 2 μL cDNA, 9.8 μL H₂O, 2 μL 50 mM MgCl₂, 2 μL 10 mM dNTPs, 2 μL 10x PCR buffer, 0.2 μL (5 U) Taq-Polymerase (Invitrogen, Darmstadt, Germany) and 2 μL 10 pmol primer mix. The following primers were used: 

1.  
   forward 5′-GGTGGTCTATGTTGGCGTCT-3′ and reverse 5′-TGGAGTGTGACAGCTTGGAG-3′ (product: 227 base pairs [bp]); and
2.  
   forward 5′-TCGTCTGCATCCTGGTGTGG-3′ and reverse 5′-CTGTGCTTCTCCTGGTCACTGG-3′ (product: 259 bp).

For PCR control samples, 2 μL DNase-free water was used instead of cDNA. Lung tissue served as a positive control. The polymerase chain reaction included an initial cycle at 95°C for 5 minutes followed by 35 cycles of 95°C for 15 seconds, 64°C for 30 seconds, 72°C for 25 seconds, and a final elongation at 72°C for 5 minutes, followed by a temperature hold at 4°C. We loaded 10 μL PCR product on a 2% agarose gel containing ethidium bromide, and the amplified products were visualized under UV light after electrophoresis. Base pair values were compared with GenBank data. We confirmed PCR products by DNA sequencing using a sequencing kit (BigDye Terminator Cycle Sequencing Kit; Applied Biosystems, Inc., Foster City, CA, USA). 

Western Blot Analysis

For Western blot analysis, tissue samples of cornea, conjunctiva and lacrimal gland (n = 4) were crushed in an agate mortar under liquid nitrogen, then homogenized in 300 μL 1% Triton buffer (Triton X-100; Carl Roth, Karlsruhe, Germany) with a protease and phosphatase inhibitor cocktail (Fermentas). The samples were centrifuged at 17.99g for 30 minutes, and the supernatant was stored at −80°C. The total protein concentration was measured based on the Bradford dye-binding procedure (Bio-Rad, Hercules, CA, USA). Total protein (40 μg) was analyzed by Western blot according to the following protocol: Samples were diluted in reducing Laemmli buffer, denatured for 5 minutes at 95°C, separated by SDS-PAGE on 10% acrylamide separating gels, and transferred to Hybond-ECL nitrocellulose membranes (Amersham Life Technologies, Arlington Heights, IL, USA). Protein blots were then incubated with anti-CXCR4 or anti-CXCR7 antibodies (CXCR4: rabbit polyclonal to CXCR4, ab2074; Abcam, Inc., Cambridge, UK; CXCR7: rabbit polyclonal to GPCR RDC1, ab72100, Abcam, Inc.) diluted 1:500 in PBS containing 5% milk overnight at 4°C. After a 2-hour incubation period at room temperature with the peroxidase-conjugated goat anti-rabbit IgG secondary antibody (1:10,000; DAKO, Glostrup, Denmark), positive binding was detected using the ECL chemiluminescence substrate (ECL Western blotting detection kit, Amersham RPN 2106; Amersham Life Technologies). 

Immunohistochemistry

Immunohistochemistry on paraffin-embedded sections (7 μm) of human lacrimal gland, cornea, conjunctiva, and lung tissue (positive control; n = 5) was performed with a monoclonal mouse antibody to human CXCR4 (1:10 in TBS, MAB173; R&D Systems, Inc., Minneapolis, MN, USA) and a monoclonal mouse antibody to human CXCR7 (1:20 in TBS, MAB4227, R&D Systems, Inc.). These were applied with a standard peroxidase-labeled streptavidin-biotin technique, using conventional methods with trypsinization. Nuclei were counterstained with hematoxylin and slides finally mounted in aqueous medium (Aquatex; Roche, Mannheim, Germany). Two negative control sections, one incubated with the secondary antibody only, the other with the primary antibody only, were used for each tissue. Furthermore, control sections were incubated with nonimmune IgG to determine possible nonspecific binding of mouse or rabbit IgG. All slides were examined and photomicrographs were taken (Keyence Biozero BZ-8100; Keyence Corporation, Osaka, Japan). 

Quantitative RT-PCR of Cell Lines

We isolated RNA using TRIzol reagent (Life Technologies, Karlsruhe, Germany), DNA was digested by DNase, and cDNA was synthesized. For quantitative RT-PCR TaqMan master mix and primer probes were used (Applied Biosystems, Inc.): hGAPDH (Hs99999905_m1), hCXCR4 (Hs00607978_s1) and hCXCR7 (Hs00664172_s1). Analyses (n = 3) were performed with an ABI 7500 Fast detection system. Values of ΔC_T = C_TGene of interest − C_TGAPDH (glyceraldehyde-3-phosphate dehydrogenase, housekeeping gene). A value of ΔC_T equal to 3.33 corresponds to one magnitude lower gene expression compared with GAPDH. 

Immunocytochemistry of Cell Lines

Cells were seeded on poly-D-lysine coated cover slips, grown for 24 hours and differentiated for 24 hours (n = 2). Medium was changed to DMEM + 1% bovine serum albumin (BSA), and cells kept for 1 hour at 37°C, then gradually cooled down to 4°C. Antibodies against CXCR4 (rabbit anti-CXCR4, 1:150; Imgenex, San Diego, CA, USA) and CXCR7 (mouse anti-CXCR7, 1.100; R&D Systems, Wiesbaden, Germany) were applied in DMEM + 4% horse serum at 4°C for 1 hour. Cells were carefully washed with ice-cold PBS + 1% BSA and incubated with the secondary antibodies (donkey anti-mouse AlexaFluor 488 and donkey anti-rabbit AlexaFluor 555, 1:800; Life Technologies) for 1 hour at 4°C. Cells were washed again and fixed with Zamboni fixative (freshly prepared, 4% paraformaldehyde, 17.5% picrinic acid in phosphate buffer) at room temperature for 30 minutes, washed, and nuclei were counterstained with DAPI (Sigma-Aldrich, Munich, Germany). Cover slips were washed, desalted in distilled water, and embedded with immumount (ThermoShandon, Frankfurt, Germany). 

Migration Assay (Scratch Assay)

We cultivated HCjE cells in DMEM/HAM'S F12 1:1 medium, supplemented with 10% FCS, 1 μg/mL bovine insulin, and 5 μg/mL hydrocortisone. After reaching confluence, the medium was switched to equally supplemented, but FCS-free medium (starvation medium) for 3 hours. Where applicable, the receptor antagonists AMD3100 (CXCR4, 100 ng/mL; Sigma-Aldrich Corp., St. Louis, MO, USA) or CCX733 (CXCR7, 10 nM; provided by ChemoCentryx, Inc., Mountain View, CA, USA) were added to the media for the last 30 minutes of the starvation time. Using pipette tips, the cell layer was scratched several times, creating “wounds” of similar width. The cells were washed twice with PBS to remove debris, and starvation medium (with or without antagonists) was applied. Images of wounded areas were taken (Keyence Biozero BZ-8100) and areas were marked for later observation. Cells were subsequently stimulated with recombinant human (rh) TFF3 (10, 30, 100, 300, 1000 μg/mL). Controls contained no rhTFF3, but adequate volumes of rhTFF3/AMD3100/CCX773 solvents (PBS or DMSO). Bovine serum albumin (300 μg/mL) served as a protein control. In order to ensure only migration, but not proliferation, mitomycin C was added to the cells at a concentration of 10 μg/mL. The previously imaged areas were again photographed after 24 hours of stimulation. The wounded area was determined at 0 hours as well as 24 hours using graphics editing software (Adobe Photoshop; Adobe Systems, Mountain View, CA, USA). Stimulated samples were compared with control values. Statistical significance was determined by 1-way ANOVA and Tukey's multiple comparisons test using commercial scientific statistics software (InStat; GraphPad Software, Inc., San Diego, CA, USA); n = 6 for each condition. 

Proliferation Assay (Flow Cytometry)

We cultivated and treated HCjE cells as mentioned above (migration assay). After the respective treatment, cells were detached using trizol, and stained with propidium iodide (PI) using PI/RNase staining buffer (BD Biosciences, San Jose, CA, USA) following the manufacturer's protocol. Resulting PI staining was subsequently read using a flow cytometer (BD FACSCalibur; BD Biosciences); n = 4 for each condition. 

Western Blot on ERK1/2 Phosphorylation

We differentiated HCjE cells (n = 3) for 24 hours, washed serum-free (2 × 1 hour with DMEM + 1% BSA), and preincubated for 1 hour with inhibitors: 1 μM AMD3100 or 0.1 μM CCX733 or 0.1% DMSO (solvent control). Cells were stimulated with 10 μg/mL rhTFF3 (as this concentration proved most efficient in the migration assay) for 15 minutes while inhibitor concentrations were maintained. Samples were lysed and subjected to SDS-PAGE with subsequent immunoblotting with antiphosphorylated ERK1/2 (pERK, Tyr202/Tyr204, 1:500; Cell Signaling, Danvers, MA, USA) and horseradish peroxidase (HRP)–labeled secondary antibody (anti-rabbit, 1:30,000; Santa Cruz Biotechnologies, Dallas, TX, USA). Signals were detected with Amersham ECL Advance and Amersham Hyperfilm ECL (GE Healthcare Life Sciences, Uppsala, Sweden). To ensure equal loading, membranes were stripped (Reblot Plus Strong Stripping solution; Millipore, Darmstadt, Germany) and reprobed with anti-ERK2 (1:500; Santa Cruz Biotechnologies) and HRP-labeled secondary antibody (1:30,000 anti-mouse; Santa Cruz Biotechnologies). 

Results

Dimers of CXCR4 or CXCR7 Show High Affinity for TFF3 in X-Ray Structure-Based Modeling

To define possible interactions between the chemokine receptors and TFF3 we used published x-ray structure data and performed 3D modeling structural analysis. The most probable protein–protein docking arrangement between TFF3 and CXCR4 is displayed in Figure 1A. This complex is strongly stabilized by the formation of several strong salt bridges between amino acid side chains of TFF3 and CXCR4 (R34:D193 chain B, D1:K271 chain A, E30:K271 chain B), hydrogen bonds (N33:E277 chain B, Y23:E268 chain B), and some hydrophobic interactions (e.g., F59:F189 chain B). These interactions of TFF3 mainly occur with the extracellular loops of CXCR4 but also partly in the pore formed by the trans-membrane helices. The docking behavior of TFF3 with the model of CXCR7 (Fig. 1B) is also characterized by the formation of several strong salt bridges and hydrogen bonds (R41:E202A, S40:E202B, N33:N191B, F99C terminus, K40B and N191B.E56: R288B. R18:D39B) and occurs also with the extracellular loops at the surface. These software modeling results indicate that CXCR4 and CXCR7 may have the possibility to interact with the TFF3 molecule. 

Figure 1

 

(A) Docking arrangement of TFF3 (above, space-fill representation) to the x-ray structure of CXCR4 (below, secondary structure visualization of the seven trans-membrane helices in red). (B) Docking arrangement of TFF3 (above, space-fill representation) to the model of CXCR7 (below, secondary structure visualization).

CXCR4 and CXCR7 Are Present at the Ocular Surface

The presence of mRNA for chemokine receptors CXCR4 and CXCR7 was demonstrated by RT-PCR in human lacrimal gland, cornea, and conjunctiva (Fig. 2A, n = 3 for each tissue). Additionally, Western blot analysis revealed protein biosynthesis in the named tissues (n = 4). Immunohistochemistry (n = 5 for each tissue) showed the following distribution of both proteins: in the cornea, the reactivity of the CXCR4 antibody is visible as membrane-bound as well as partly cytoplasmic throughout the epithelium, while the CXCR7 antibody shows primarily a weak cytoplasmic binding. The conjunctiva shows a stronger apical localization of both receptors, again membrane-bound as well as intracellular. For the lacrimal gland we detected a weak cytoplasmic signal of both CXCR4 and CXCR7 inside the acini as well as strong reactivity with myoepithelial cells, while the signal of ductal cells had a more apical localization (Fig. 2B). 

Figure 2

 

(A) Detection of CXCR4 and CXCR7 in tissues of the ocular surface. Analysis of RT-PCR (n = 3) shows receptor transcription in all tested tissues. Protein synthesis is shown by Western blot (n = 4); one lacrimal gland, two cornea, three conjunctiva. (B) Immunohistochemistry of CXCR4 and CXCR7 in human lacrimal gland, cornea, and conjunctiva (n = 5). Lung tissue served as a positive control. The CXCR4 antibody shows reactivity throughout the corneal epithelium—both membrane-bound (arrowheads) and cytoplasmic (arrows)—while binding in conjunctival tissue occurs predominantly in the apical cell layer. The lacrimal gland synthesizes CXCR4 weakly, mostly in acinar and myoepithelial cells (asterisks). The CXCR7 receptor is produced by corneal cells as well, showing mostly intracellular localization. In the conjunctiva, the receptor is mainly expressed intracellularly by apical cell layers. In the lacrimal gland it is produced by myoepithelial cells, ductal cells with an apical localization and a cytoplasmic localization inside the acini. Ac, acinus; Alv, alveolus; B, Bowman's layer; D, duct; Ep, epithelium; FC, fat cell; St, stroma.

CXCR4 and CXCR7 Are Not Expressed in the HCE Cell Line, but in the HCjE Cell Line

Human immortalized corneal epithelial cells did not express CXCR4 and CXCR7, as evaluated by quantitative RT-PCR and immunocytochemistry (data not shown). Human immortalized conjunctival epithelial cells expressed both receptors. However, the expression level of CXCR4 as well as CXCR7 in HCjE cell line was not pronounced compared with the expression in MCF7 cell line (a human breast cancer cell line), which was used as control cell line. In HCjE, quantitative RT-PCR yielded signals at the detection limit (Fig. 3A, n = 3) and immunocytochemistry revealed only few positive signals at the surface of cultivated cells (Fig. 3B, n = 2). The control cell line MCF7 showed strong signals for CXCR4 and CXCR7 in both analyses. 

Figure 3

 

(A) Quantitative RT-PCR of CXCR4 and CXCR7 in human conjunctival epithelial cell line (HCjE) and MCF7 cell line (positive control; n = 3). Expression is detectable in both cell lines; in HCjE, however, in a lower relative concentration at the detection limit. (B) Only few receptor antibody signals are detectable by immunocytochemistry in a cultured HCjE cell compared to a cultured MCF7 cell that was used as positive control (n = 2).

rhTFF3 Promotes Cell Migration and Proliferation of HCjE Cells in a Concentration-Dependent Manner

For the following experiments, only the receptor expressing cell line HCjE was further investigated. The cell culture–based migration assays revealed increased migration of HCjE cells in the presence of rhTFF3 (Figs. 4A, 4B; n = 6). The effect seen with lower concentrations, however, decreased with increasing concentrations of rhTFF3. When adding the specific chemokine receptor antagonists, the observed rhTFF3 effect was completely neutralized (Fig. 4C; n = 6). Significant differences were seen between rhTFF3 stimulation and the addition of AMD3100 and CCX733 in every rhTFF3 concentration tested. There are no measurable differences in migration rates between the two antagonists: The remaining wound areas in both cases resemble those of controls. Simultaneous blockade of both receptors has been tested for a rhTFF3 concentration of 300 μg/mL and yielded a migration impairing effect comparable to blocking either one receptor (data not shown). Only migratory effects, but not proliferation, seem to be mediated by CXCR4 and/or CXCR7. 

Figure 4

 

(A) Scratch assay on HCjE cells (representative pictures, n = 6). Confluent cell layers were scratched with a pipette tip, wounded area was photographed and measured (0 hours). After 24 hours of incubation, remaining wound area was again photographed and measured (24 hours). Values of cells incubated with different concentrations of rhTFF3, with or without addition of AMD3100 or CCX733, were compared with control values. (B) Restored wound area after scratch and 24-hour incubation with rhTFF3, compared with control values. The wound healing rates are significantly higher under stimulation with 10, 30, 100, and 300 μg/mL and 1 mg/mL rhTFF3 compared with no TFF3 control as well as BSA protein control. Lower concentrations seem to have better effects on cell migration, while effect regresses with increasing concentrations. *P < 0.05; **P < 0.01; ***P < 0.001 compared with control; °P < 0.05; °°P < 0.01; °°°P < 0.001 compared with BSA protein control. (C) Restored wound area after scratch and 24-hour incubation with rTFF3 and the CXCR4 antagonist AMD3100 or the CXCR7 antagonist CCX733, compared with control values. Enhancement of migration, seen with rTFF3, is significantly inhibited by addition of either antagonist. No significant differences in migration rates between the two antagonists were seen. *P < 0.05; **P < 0.01; ***P < 0.001.

Cell proliferation assays testing for cycling cells using PI staining and subsequent flow cytometry showed increased proliferation rates (higher number of cells undergoing S- and G2/M-phase) in HCjE cells after rhTFF3 stimulation (Fig. 5A, n = 4). The addition of the specific receptor antagonists AMD3100 (for CXCR4) and/or CCX733 (for CXCR7) did not significantly affect the increase in cycling cells, neither alone nor added in combination (Fig. 5B). 

Figure 5

 

Flow cytometry on PI stained HCjE cells (n = 4). (A) Shows cell cycle distribution in G0/G1-, S- and G2/M-phase. Representative graphs are shown for control cells as well as after stimulation with rTFF3 or rTFF3 + AMD3100. FL2-A: emitted fluorescent light of the DNA dye (FL2) measured as pulse-area (FL2-A). (B) Number of cells in S- or G2/M-phase compared with control after the addition of rhTFF3 and/or the cytokine receptor antagonists. Addition of rhTFF3 alone leads to a significant increase of cells undergoing S- and G2/M-phase. Addition of AMD3100 and/or CCX733 does not significantly influence this activation of the cell cycle. **P < 0.01; °°°P < 0.001 compared with respective control.

rhTFF3 Stimulation Results in Activation of the MAP Kinase Signaling Cascade Independently From CXCR4 or CXCR7

To evaluate whether rhTFF3 had any effect on the MAP kinase pathway, we investigated the phosphorylation status of ERK1/2 proteins by Western blot analysis in HCjE cells (n = 3). This revealed an upregulation of pERK after stimulation of HCjE cells with rhTFF3 (Fig. 6). This increase in pERK by rhTFF3 was also seen in the presence of the specific CXCR4 and CXCR7 antagonists AMD3100 and CCX733. There was also an increase in signal intensity after AMD3100 treatment alone. 

Figure 6

 

Western blot of ERK1/2 phosphorylation (upper panel) and respective densitometry (lower panel; n = 3). Stimulation of HCjE cells with rhTFF3 results in significantly higher levels of pERK, an effect that is not abrogated by addition of the specific receptor antagonists AMD3100 or CCX733. *P < 0.05 compared with control. DMSO, solvent control; NS, not significant.

Discussion

Chemokine receptors CXCR4 and CXCR7 are present in a variety of tissues throughout the human body. At the ocular surface, their presence had only been shown in corneal and limbal epithelial and corneal stromal cells¹⁶ and, under pathologic conditions, during atopic keratoconjunctivitis.¹⁷ In the present study, we demonstrated the expression and synthesis of CXCR4 and CXCR7 at the healthy ocular surface. Under physiological conditions, both receptors were detected in human cornea, conjunctiva and lacrimal gland, and in lesser amounts in the conjunctival epithelial cell line used. The presence of TFF3 in these tissues has already been shown.^7,8 Their similar expression patterns in these tissues allow for initial speculation about putative interactions. 

Chemokine receptor type 4 is present on most hematopoietic cell types, where it plays an important role in recruitment of immune cells during inflammation. It is known that CXCR4, but not SDF-1, is upregulated by proinflammatory cytokines, like TNF-α or IL-1β.³⁸ A comparable situation was found for gastric ulcer healing, where SDF-1 levels decrease in the process, while CXCR4 levels increase.³⁹ At the ocular surface, TFF3 is induced during corneal infection⁹ and after corneal damage.¹⁰ Corneal epithelial cells upregulate TFF3 when stimulated with TNF-α and IL-1β.⁴⁰ The fact that CXCR4, but not its main ligand SDF-1, is upregulated during inflammation leads to the hypothesis that another ligand may be activating the receptor under these conditions. The described upregulation of TFF3 leaves room for further theorizing about a possible CXCR4-TFF3 interplay. 

As stated earlier, CXCR4 and CXCR7 have been discussed as low affinity receptors for TFF2.^13,29 The well-studied structural similarities between the different TFF peptides allow for speculation about a possible interaction with TFF3 as well. As for CXCR4, this has been shown recently by Xue et al.³⁰: In wild-type mice, AMD3100 significantly slowed restitution of gastric mucosa after experimental injury. In TFF2 knockout animals, topic addition of rat TFF3 peptide partly compensated for the loss of TFF2, an effect that was also strongly inhibited by adding AMD3100. This suggests that both trefoils (i.e., TFF3 as well), engage the CXCR4 receptor during wound healing.³⁰ 

To elucidate these theoretic interactions, we used an x-ray based computer modeling system to determine possible binding conformations of both receptors with the TFF3 peptide. The modeling identified the CXCR4 homodimer and the CXCR7 homodimer as probable binding partners for TFF3 in each case, showing that the main TFF3-receptor interaction was taking place between the substrate TFF3 and basically only one of the involved monomers (referred to as chain B in each receptor, see above). Therefore, an interaction of TFF3 with the heterodimer also seems possible, as TFF3 may interact independently with each monomer involved. 

The most often described effect of TFF3 is enhancement of wound healing. This complex process can be understood as a synergy of, among others, cell proliferation and migration in order to cover and repair an injured tissue. Here, we investigated both subprocesses independently to evaluate a putative chemokine receptor function. In general, it is to be noted that the expression levels of CXCR4 and CXCR7 in the used HCjE cells were low, so that interpretation of the in vitro results with this cell line needs to be done with caution. However, according to the immunohistochemistry results obtained in conjunctiva, the in vivo receptor density is markedly higher than in the HCjE cell line. Therefore, the obtained reactions to stimulation with rhTFF3 may prove even stronger in an experimental system that mimics the in vivo situation more precisely. Nevertheless, our results are of great interest as they indicate an interplay of TFF3 with CXCR4 and CXCR7 that can be more pronounced in other tissues and cell lines which need to be determined in the future. 

In our model system, we found opposite results: investigating proliferation rates by analyzing cell cycle parameters with flow cytometry, we found elevated levels of cycling cells after rhTFF3 stimulation with or without addition of AMD3100 or CCX733, suggesting independence from the chemokine receptors. On the contrary, our cell culture–based migration assay yielded most different results: stimulation of wounded ocular surface cells with rhTFF3 resulted in accelerated cell migration. In accordance with the findings of Xue et al.,³⁰ the blockage of CXCR4 completely abrogated the rhTFF3 effect when compared with control values. In addition, we show here that antagonizing CXCR7 with CCX733 had a comparable effect: Wound healing rates that had increased under rhTFF3 stimulation were significantly impaired when CXCR7 was blocked. The migration assay revealed that the effect of rhTFF3 was fully diminished not only by blocking both receptors at once, but by blocking either one of the receptors, despite the fact that the other receptor in each case was still fully functional. 

There have been previous investigations in which a blockade of either receptor, CXCR4 or CXCR7, resulted in impaired effects of a substrate, or even in complete abrogation of the effect, respectively. Yan et al.⁴¹ showed that blocking CXCR7 with CCX733 impaired the SDF-1 induced adhesion of endothelial progenitor cells to active human umbilical vein endothelial cells (HUVECs). Antagonist AMD3100 completely abolished the adhesion, while the inhibitory effect of blocking both receptors resembled that of AMD3100 alone.⁴¹ Moreover, Watanabe et al.⁴² demonstrated results even more congruent with those gathered in the current study: SDF-1 was confirmed to induce tube-forming activity in HUVECs, which was completely suppressed by blocking either CXCR4 or CXCR7, and to a comparable extent. Blocking both receptors resulted in similarly low levels of tube formation compared to blocking only one receptor alone.⁴² In the latter study, the authors suggest that signaling from both receptors was necessary for the SDF-1 effect to take place. 

Another possible explanation involves the previously studied ability of CXCR4 and CXCR7 to form functional heterodimers. Using FRET analysis, Sierro et al.⁴³ demonstrated the presence of preformed CXCR4-CXCR7 dimers on the cell membrane of human embryonic kidney cells in the absence of a ligand, with additional evidence of further intracellular heterodimer pools. Levoye et al.²⁸ confirmed the constitutive formation of heterodimers, and indicated the same efficiency for homo- and heterodimerization, suggesting that the levels of CXCR4 and CXCR7 biosynthesis determine the incidences of homo- and heterodimers. Heterodimerization on the cell surface and combined internalization of both receptors has also been shown by fluorescence light and electron microscopy on MCF-7 breast cancer cells.³⁷ Besides detecting CXCR4-CXCR7 heterodimers using immunoprecipitation, Décaillot et al.²⁷ described the ability of CXCR7 to alter CXCR4/SDF-1–mediated inhibition of adenylyl cyclase and subsequent cAMP production, and instead switch to constitutively recruiting β-arrestin complexes. In fact, β-arrestin recruiting was notably stronger in CXCR4 and CXCR7 coexpressing cells than in cells expressing CXCR7 alone. So a strong β-arrestin response is dependent on the presence, function, and heterodimerization of both CXCR receptors. 

In summary, we found that cell proliferation of rhTFF3 stimulated HCjE cells occurred independently from the functionality of CXCR4 and CXCR7, while the observed migratory effects were dependent on both chemokine receptors, and possibly on their homo- or heterodimerization. 

A closer look into the signaling chains of CXCR4, CXCR7 and TFF3 further elucidates the possibility of interaction. The receptor CXCR4 signals mainly through G_i-protein mediated inhibition of adenylyl cyclase and IP3 dependent mobilization of intracellular calcium, but to a lesser extent also through ERK1/2, PI3-K, JAK/STAT and NFκ-B.^44,45 In contrast, the preferential mediators of CXCR7 signaling are β-arrestin–dependent cell signaling pathways like ERK1/2, MAPK, Akt, p38, and PKCα/β.^26,46 As stated before, the CXCR4-CXCR7 heterodimer also preferentially signals β-arrestin dependently, including ERK1/2, p38 MAPK, JAK2/STAT3 and JNK.^27,47 Looking at TFF3, its intracellular signaling pathways have been proven to function through ERK1/2, p38 MAPK, JAK2/STAT3, JNK and PI3-K.^46,48,49 Obviously, the CXCR4/CXCR7 heterodimer and TFF3 share common pathways, at least suggesting possible interaction and joint activation. 

Therefore, we further investigated one of TFF3's best studied signaling pathways for CXCR4/CXCR7 dependence: the MAP kinase pathway, in particular the phosphorylation and activation of ERK1 and 2. This pathway has been described as mediating cellular migration: For instance, the ERK pathway inhibitors PD98059 and U0126 inhibit migration of various cell types in response to growth factors such as VEGF, fibroblast growth factor, EGF, and insulin.⁵⁰ Moreover, it has been extensively studied in the context of cell proliferation. In fact, Orime et al.⁵¹ have previously described the TFF2/CXCR4 axis to induce cell proliferation through ERK1/2 phosphorylation. In our study we found a strong ERK phosphorylation signal after rhTFF3 treatment, which was not significantly altered by adding the specific CXC receptor antagonists. So while ERK1/2 signaling seems to be involved in the TFF3 signaling pathway, its activation is not CXCR4/CXCR7 mediated. This suggests the presence of another target for TFF3 binding. Also, though the MAP kinase pathway has been described to facilitate cell migration, it does not seem to be the mediating pathway in this given model system, as our cell migration assay showed CXCR4/CXCR7 dependence. 

In conclusion, our study revealed the dependence of TFF3 action on more than one receptor, depending on the respective process investigated. This highlights the probability of a complex substrate-receptor interplay with multiple cellular signaling targets involved, with CXCR4 and CXCR7 just being two of them. In the given model system, they promote cell migration in a MAP kinase–independent manner. Further experiments are needed to determine alternative receptors that activate the MAP kinase pathway, as well as the subsequent signaling chain after CXCR4/CXCR7 activation by TFF3. 

Acknowledgments

The authors thank Susann Möschter, Stephanie Beilecke, Martin Schicht, and Katja Dieckow, Tamara Grummet, and Andrea Sehr for their helpful assistance and advice. 

Supported by the German Research Foundation (DFG, Program Grant PA738/9-2 and PA738/11-1, and ME758/10-1, RM, and SE1995/1-1, SS); a special fund for scientific work at Friedrich Alexander University of Erlangen-Nürnberg (FP); an MD fellowship of the Boehringer Ingelheim Fonds, Foundation for Basic Research in Medicine, Heidesheim, Germany (JD); and Sicca Forschungsförderung of the Association of German Ophthalmologists (JD). The present work was performed in fulfillment of the requirements for obtaining the degree “Dr med.” 

Disclosure: J. Dieckow, None; W. Brandt, None; K. Hattermann, None; S. Schob, None; U. Schulze, None; R. Mentlein, None; P. Ackermann, None; S. Sel, None; F.P. Paulsen, None