Chemical Synthesis of TFF3 Reveals Novel Mechanistic Insights and a Gut-Stable Metabolite

Abstract

TFF3 regulates essential gastro- and neuroprotective functions, but its molecular mode of action remains poorly understood. Synthetic intractability and lack of reliable bioassays and validated receptors are bottlenecks for mechanistic and structure-activity relationship studies. Here, we report the chemical synthesis of TFF3 and its homodimer via native chemical ligation followed by oxidative folding. Correct folding was confirmed by NMR and circular dichroism, and TFF3 and its homodimer were not cytotoxic or hemolytic. TFF3, its homodimer, and the trefoil domain (TFF3_10-50) were susceptible to gastrointestinal degradation, revealing a gut-stable metabolite (TFF3_7-54; t_1/2 > 24 h) that retained its trefoil structure and antiapoptotic bioactivity. We tried to validate the putative TFF3 receptors CXCR4 and LINGO2, but neither TFF3 nor its homodimer displayed any activity up to 10 μM. The discovery of a gut-stable bioactive metabolite and reliable synthetic accessibility to TFF3 and its analogues are cornerstones for future molecular probe development and structure-activity relationship studies.

Figure 1

Synthesis of TFF3 and its homodimer. (A) TFF3 sequence and disulfide bond connectivity. Sequence highlighted in green and blue represents the N- and C-terminal fragments used for native chemical ligation, respectively, with Cys⁵⁷ (red) used for dimerization. The trefoil domain is underlined. (B) Synthetic strategy used to produce TFF3 and its homodimer. Highlighted in gray is full-length TFF3. (C) Ligation reaction at time 0 h (left) and 24 h (right). (D) Analytical RP-HPLC chromatogram and MS of folded TFF3 monomer (left) and homodimer (right). TFF3 monomer displays a two-peak RP-HPLC profile due to its conformational complexity (see also Figure S3). TFF3 PDB: 1E9T; TFF3 homodimer PDB: 1PE3.

Figure 2

Comparison of secondary Hα chemical shifts of (A) TFF3(C⁵⁷Acm) and (B) TFF3 homodimer produced by chemical synthesis with recombinant homologues and reported values from the Biological Magnetic Resonance Data Bank (BMRB: 5771). Secondary Hα chemical shifts were determined by subtracting the shifts observed in random coil peptides from the shifts determined from the two-dimensional (2D) NMR analysis. An α-helical region is highlighted in blue in the sequence and NMR structure. CD spectra of synthetic (C) TFF3(C⁵⁷Acm) and (D) TFF3 homodimer. (E) Co-elution of synthetic and recombinant TFF3 homodimer (1:2 ratio) on a C₃-RP-HPLC (1% gradient).

Figure 3

Biological characterization of TFF3(C⁵⁷Acm) and TFF3 homodimer. (A) Cell death was induced with etoposide (10 μM), and cells were treated with 10 μM of TFF3. Etoposide and TFF3 were added at the same time. Caspase-3/7 activity was measured after 6 h. The results are expressed (mean ± standard error of the mean (SEM)) of n ≥ 3 independent experiments as the number of green fluorescent caspase-3/7 active objects generated by caspase-3/7 reagent added in media. Z-DEVD-FMK (50 μM; caspase-3 inhibitor) was used as the positive control. E: etoposide, E + C3I: etoposide + caspase-3 inhibitor. One-way analysis of variance (ANOVA) followed by Dunnett correction was performed to assess differences between treated cells and etoposide only. *p < 0.05, **p < 0.01. (B) TFF3 effect on the viability of HEK-293 cells after 20 h. Tamoxifen was used as a positive control for cell growth inhibition. Data are representative of two independent experiments shown as mean ± SEM. (C) Hemolytic activity of TFF3 after 1 h on erythrocytes. Melittin was used as a positive hemolytic control. Data are representative of two independent experiments shown as mean ± SEM. (D) Effect on CXCR4. Agonistic or antagonistic effects were assessed by the IP1 accumulation assay obtained through homogeneous time-resolved fluorescence (HTRF). CXCL12 was added for antagonistic studies after a 5 min preincubation of TFF3(C⁵⁷Acm) or homodimer. For agonistic evaluation, all compounds were used individually for stimulation in the given concentration range. Neither TFF3(C⁵⁷Acm) nor TFF3 homodimer were able to activate or inhibit the signal transduction of CXCR4 at concentrations up to 10 μM, indicating that these peptides are neither agonists nor antagonists at this receptor. The results are expressed as mean ± SEM of n ≥ 2 independent experiments. (E) Effect on LINGO2. HEK-293 coexpressing LINGO2-YFP and LINGO2-Rluc were stimulated with increasing amounts of TFF3(C⁵⁷Acm) and TFF3 homodimer (1 nM–10 μM), but no significant increase in the bioluminescence resonance energy transfer (BRET) signal was observed. No ligand able to alter LINGO2 dimerization is known; therefore, no positive control could be used. Results are expressed as mean ± SEM for n = 4. One-way ANOVA followed by Dunnett correction was performed to assess differences between treated and nontreated cells.

Figure 4

Identification and characterization of a stable and bioactive TFF3 metabolite (TFF3_7–54). (A) The observed cleavage sites are highlighted in the three-dimensional structure and linear sequence of TFF3. The trefoil domain is underlined and displayed in blue in the sequence and NMR structure. (B) A table listing the observed fragments including their masses. The shortest stable metabolite (TFF_7-54) is underlined. (C) Intestinal stability of the trefoil domain TFF3_10–50 and the gut-stable metabolite TFF3_7–54. (D) Comparison of the secondary Hα chemical shifts of TFF3 (BMRB: 5771) and the trefoil domain TFF3_10–50, and TFF3 (BMRB: 5771) and the gut-stable metabolite TFF3_7–54. Secondary Hα chemical shifts were determined by subtracting the shifts observed in random coil peptides from the shifts determined from the 2D NMR analysis. (E) Antiapoptotic effects of TFF3_7–54 on SH-SY5Y cells. E: etoposide. Etoposide and TFF3_7–54 were added at the same time. Caspase-3/7 activity was measured after 6 h. Results are expressed (mean ± SEM) of n ≥ 3 independent experiments as the number of green fluorescent caspase-3/7 active objects generated by caspase-3/7 reagent added in media. One-way ANOVA followed by Dunnett correction was performed to assess differences between treated cells and etoposide only. *p < 0.05, **p < 0.01.