Selective targeting of collagen IV in the cancer cell microenvironment reduces tumor burden

Abstract

Goodpasture antigen-binding protein (GPBP) is an exportable1 Ser/Thr kinase that induces collagen IV expansion and has been associated with chemoresistance following epithelial-to-mesenchymal transition (EMT). Here we demonstrate that cancer EMT phenotypes secrete GPBP (mesenchymal GPBP) which displays a predominant multimeric oligomerization and directs the formation of previously unrecognized mesh collagen IV networks (mesenchymal collagen IV). Yeast two-hybrid (YTH) system was used to identify a 260SHCIE264 motif critical for multimeric GPBP assembly which then facilitated design of a series of potential peptidomimetics. The compound 3-[4''-methoxy-3,2'-dimethyl-(1,1';4',1'')terphenyl-2''-yl]propionic acid, or T12, specifically targets mesenchymal GPBP and disturbs its multimerization without affecting kinase catalytic site. Importantly, T12 reduces growth and metastases of tumors populated by EMT phenotypes. Moreover, low-dose doxorubicin sensitizes epithelial cancer precursor cells to T12, thereby further reducing tumor load. Given that T12 targets the pathogenic mesenchymal GPBP, it does not bind significantly to normal tissues and therapeutic dosing was not associated with toxicity. T12 is a first-in-class drug candidate to treat cancer by selectively targeting the collagen IV of the tumor cell microenvironment.

Keywords: EMT; GPBP; collagen IV; drug-resistant cancer; tumor microenvironment.

Conflict of interest statement

CONFLICTS OF INTEREST J.S, F.R and F.R-R are co-founders and stockholders of Fibrostatin, S.L. that is expanding GPBP inhibitors including T12 and hN26 into regulatory development. J.S acts as Chief Scientific Officer of Fibrostatin, S.L. The other authors declare that they have no competing interest other than being Fibrostatin, S.L. employees when indicated.

Figures

Figure 1. GPBP directs the formation of two independent mesh collagen IV networks in chemoresistant EMT cancer cell phenotypes

(A) Left, the threshold-cycle (CT) for the indicated mRNA and cell lines was determined by RT-qPCR. Denoted in red characters are predominantly expressed mRNA (*murine breast cancer cell lines). Right, WB analysis of the indicated cell cultures and polypeptides. (B) Left, represented with bars are the relative quantity (RQ) expressed as mean ± standard deviation (SD) of the indicated mRNA in c4T1 cultures (circulating) in comparison (versus, vs) with 4T1 cultures (control). Middle, WB analysis of the indicated 4T1 cell extracts to assess GPBP (e11-2), GPBP and GPBP-2/CERT (N27) and GAPDH (glyceraldehyde 3-phosphate dehydrogenase) expression. Right, RQ of the indicated mRNA and 4T1 cultures is represented as at left. Statistics: Student’s t-test, n = 2. (C) The indicated 4T1 spheroids and polypeptides were analyzed by IF-CM. In this and following figures denoted in arbitrary units (AU) are fluorescence intensities (FI) expressed as mean ± SD of five representative fields (n = 5). Statistics: Student’s t-test. (D) RQ of the indicated mRNA is represented as in B comparing A549-derived clone C79 (progenitor) or G7A (mesenchymal) expressing GPBP-EYFP with an A549 clone expressing EYFP (EYFP). Shown is a representative analysis of two independent experiments. Statistics: Student’s t-test, n = 2. In all the figures: n, sample size per group; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; DAPI (4′,6-diamidine-2′-phenylindole dihydrochloride) stained cell nucleus; WCIF ImageJ was used for quantification of WB; unless otherwise indicated scale bars were 50 μm.

Figure 2. GPBP and mesh collagen IV networks maintain chemoresistant EMT cancer cell phenotypes in A549 spheroids

(A) Left, A549 2D cultures form tumors in nude mice when inoculated with Matrigel® or spheroids when seeded in non-adherent plates (3D culture conditions). Spheroids exhibit epithelial characteristics or, upon EMT-induction with TαTβ, they display an EMT phenotype. Right, spheroids observed by transmitted light microcopy (top, bar = 200 μm) or IF-CM with antibodies against the indicated polypeptides. (B) Graph, plotted is RQ of indicated mRNA over time in stimulated vs unstimulated A549 spheroids. Arrows denote stimuli pulses. Image, WB analysis of the indicated spheroids (24 h) and polypeptides. Statistics: one-way ANOVA and Dunnett’s (1WA-D) comparison respect t = 0 in each series. Results are expressed as mean ± SD of three assays. (C) A549 cultures were transfected with the indicated small interfering (si)RNA (siCONT = control siRNA) seeded in 3D conditions and following TαTβ stimulation, lactate dehydrogenase (LDH) measured over time in culture media (mean ± SD). A siRNA efficacy >75% was determined by RT-qPCR at the end of the studies. Statistics: 2WA-Sidak’s (2WA-S), n = 2. (D) Left, dose-response curve (mean ± SD) (graph) and IC50 (table) of doxorubicin, and GAPDH-normalized GPBP expression (WB and histogram) were determined for the indicated cultures. Right, IF-CM analysis of indicated spheroids and polypeptides. Statistics: 2WA-S, n = 4. Shown are representative analysis of a total of three (A, D) and two (C) assays.

Figure 3. GPBP and mesh collagen IV networks associate with EMT phenotypes in human NSCLC

(A) HE analysis of cell types in small A549 tumors. Left to right: poorly differentiated, epithelial, signet ring-like and fusiform cells. (B) Frozen (IF-CM) and paraffin-embedded (α-SMA and Alcian Blue) sections of small A549 tumors were analyzed. Upper, IF-CM detection of the indicated polypeptides in stromal and nodular tumor regions. Lower, from left to right the indicated polypeptides were detected by IF-CM (bar = 10 μm), horseradish peroxidase immunohistochemical method (brown) (bar = 100 μm) in tumor peri-nodular stroma, IF-CM (bar = 10 μm), and Alcian blue staining (bar = 20 μm) of the mucus (blue) in tumor nodular regions. (C) Paraffin-embedded sections of a NSCLC patient tumor at diagnosis prior to initiation of chemotherapy (Initial tumor) and at relapsing after surgery and chemotherapy (Relapsing tumor) were analyzed by IF-CM to visualize the indicated proteins. Indicated are FI measured, expressed and compared as in Figure 1. Statistics: Student’s t-test. Similar conclusions were obtained when comparing untreated vs after treatment chemoresistant NSCLC specimens.

Figure 4. The druggable GPBP isoform is a multimer stabilized by self-interacting 260SHCIE264 motif

(A) The indicated recombinant GPBP pools were analyzed by fast protein liquid chromatography (FPLC)-SEC. Calibrators, MW (kDa) and elution volumes (mL) were: thyroglobulin (669, 6.86), aldolase (158, 12.35) and serum human albumin (66, 13.89). In all chromatograms, the positions of the main GPBP oligomers are denoted. (B) Media (2.5 L) from the indicated cultures were affinity-purified using single-chain variable fragment (scFv)N26-column. Bound material was eluted in native conditions using N26-competing synthetic peptide (PS132) and analyzed as in A. Represented in AU are the FI of individual fractions in indirect ELISA analysis using anti-GPBP monoclonal antibodies (mAb). Shown is a representative assay of three. Image, mAb N27-developed WB of similarly purified (25 mL) GPBP eluted under denaturing conditions. Numbers and bars here and in following WB denote kDa and position of MW markers, respectively. (C) Structural features such as PH (empty box) and Start (scratched box) domains are indicated and represent GPBP and deletion mutants thereof. Indicated are spanning residues (numbers) and YTH interaction using colony-lift filter assays (+ or −). The core interactive motif and its position are denoted. (D) The interaction of the indicated pairs of fused polypeptides (BD, binding domain; AD, activation domain of GAL4 transcription factor) was assayed as in C (GIP130, GPBP-interacting protein 130-kDa, residues 86–350). Blue color (+). (E) The indicated FLAG-tagged polypeptides were expressed, purified and analyzed by high-performance liquid chromatography (HPLC)-SEC. (A, E) Chromatograms are representative of at least three independent experiments.

Figure 5. T12, a peptidomimetic of the GPBP region comprising 260SHCIE264 motif disturbs multimerization

(A) Chemical structure of T12 and dose-response effects on GPBP autophosphorylation and A549 doxorubicin IC50. Phosphorylation mixtures were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), blotted and exposed (32P) prior to detect GPBP with N27 (WB). Specific activities (histogram) were calculated using WCIF ImageJ and represented as percentage (mean ± standard error of the mean, SEM) of control mixture (−) that was set at 100%. Statistics: 1WA and Fisher’s (1WA-F), n = 2. Graph and Table show the efficacy and IC50 of doxorubicin on A549 cultures at the indicated T12 concentrations. Efficacy values are mean ± SD. The 2WA and Tukey’s (2WA-T) and IC50 are indicated, n = 4. Representative of five assays. (B) Recombinant (multimer or trimer, n = 3) or extracellular native (A427, n = 2) GPBP were analyzed as in A at 50 µM T12. Statistics: Student’s t-test. (C) Similar amounts (125 μg) of recombinant extracellular GPBP and bovine serum albumin (BSA) were exposed to T12 (10 mM) during 15 min at room temperature and further FPLC-SEC analyzed essentially as in previous Figure. Represented is one out of three independent analysis performed from two different GPBP preparations.

Figure 6. T12 selectively targets mesenchymal GPBP produced by cancer EMT phenotypes

(A) A549 cultures (−) were stimulated (TαTβ) for 24 h and analyzed by IF-CM to visualize in situ T12 binding (bioT12) and the indicated polypeptides (bar = 20 μm). Indicated are FI measured, expressed and compared as in Figure 1. Statistics: Student’s t-test. (B) Frozen sections of a small A549 tumor were analyzed by IF-CM to visualize in situ T12 binding (bioT12) and the indicated polypeptides at tumor areas populated by the indicated cells (bar = 20 μm). (C) Frozen sections of patient adenocarcinoma were IF-CM analyzed to visualize in situ T12 binding (bioT12) and the indicated polypeptides in a representative tumor region. In the bottom row, the squared area in the top row is displayed at higher magnification (bottom row bar = 10 μm).

Figure 7. T12 impairs mesh collagen IV formation and reduces tumor burden in mouse models

(A) TαTβ-stimulated A549 and (B) A7C11 spheroids were treated (48 h) and analyzed by IF-CM to visualize the indicated polypeptides (images) and cell death measured as in Figure 2C and represented with bars (histograms). Statistics: Student’s t-test, n = 3 (A) n = 2 (B). Indicated are FI measured, expressed and compared as in Figure 1 Statistics: Student’s t-test. (C) The extracellular matrix of A7C11 spheroids subjected to the indicated treatments (48 h) was collagenase digested and analyzed by WB using mAb 202. (D) The RQ of the indicated mRNA in treated (10 µM, 36 h) A7C11 spheroids (T12) vs untreated (UT) is represented as in previous figures. Statistics: Student’s t-test, n = 2. (E) A549 mouse-model bearing the indicated xenografts was untreated (Control) or treated (T12) and the indicated variable determined over time and plotted (mean ± 95% confidence interval, CI). Tumor volume at treatment initiation was set to 100%. Statistics: 2WA-S, n = 15. (F) Plotted are indicated variables (mean± 95% CI) in 4T1 mouse-model untreated (Control) or treated (T12) over time (left) or at day-25 (right). Statistics: Left, 2WA-S, n = 10. Representative of three assays. Right, Student’s t-test, n = 30.

Figure 8. T12 impedes tumor epithelial-mesenchymal transdifferentiation

(A) Bars represent RQ (mean ± SD) of the indicated mRNA in small sensitive (S, n = 14) vs large refractory (R, n = 8) A549 tumors and in T12-treated (n = 6) vs untreated (control, n = 6) S-A549 tumors. Statistics: Student´s t-test. (B–D) Paraffin-embedded sections from untreated (Control, n = 6) or treated (T12, n = 6) S-A549 tumors were stained with the indicated reagents or subjected to IF-CM or standard immunohistochemical procedures to visualize the indicated polypeptides. Immunohistochemical specimens were counterstained with hematoxylin. Shown are selected representative images. Bar = 100 μm in B and C (Alcian Blue); Bar = 200 μm in C (Masson, IF-CM) and D.

Figure 9. Low-dose doxorubicin sensitizes epithelial cancer progenitor cells to T12

(A) WB analysis with N27 of the indicated immunoprecipitates (IP). A549 cells were treated as indicated (42 h) and normalized caspase activity plotted (mean ± SD). Statistics: 2WA-S, n = 3. Representative of three and two assays, respectively. (B) Left, A549 cultures were treated as indicated (40 h) and analyzed by flow cytometry following doxorubicin fluorescence in SP and non-SP. Ordinate: Event numbers normalized to mode. Abscise: Phycoerythrin channel autofluorescence intensity (PE-A). Shown are mean FI. Statistics: Student’s t-test, n > 1,000. Representative of four assays. Right, relative SP content (mean ± SD) in treated respect untreated (Control, SP = 1) cultures. Statistics: 1WA-T, n = 4. (C) A549 cultures were doxorubicin-treated (0 h) and further cultured in doxorubicin-free medium with the indicated inhibitors (6 h) and analyzed as in B. Doxorubicin content (mean ± SD) was determined subtracting cell autofluorescence. Statistics: 1WA-T, n = 3. (D) A549 cultures were treated (40 h) with the indicated reagents and WB analyzed. Plotted is GAPDH-normalized GPBP expression (mean ± SD). Statistics: 1WA-D, n = 2. (E) Nude mice bearing large A549 tumors were treated and the indicated variable (mean ± 95% CI) plotted over time. Initial tumor size: 100%. Statistics: 2WA-S, n = 14. Groups in this and Figure 7E (Large) were from the same assay. Images were selected from the indicated analysis and tumors. (F) Representative images of tumors in E analyzed by IF-CM. Indicated are FI measured and expressed as in Figure 1. Statistics: 1WA-D comparison respect to Control. (G) FLAG-GPBP autophosphorylation mixtures were analyzed and represented as in Figure 5A. Statistics: 1WA-F, n = 2. (H) 4T1 mouse-model was treated from day-9 (arrow) and the indicated variable (mean ± 95% CI) plotted over time (left) or at day-25 (right). Statistics: Left, 2WA-S, n = 10. Right, 1WA-T, n = 10. (I) Kaplan–Meier survival curves of 4T1 mouse-model T12-treated after surgical removal of primary tumor at day-16. Statistics: Log-rank (Mantel-Cox) test, P = 0.15. (J) Pictures of mouse-1 at the indicated days post-inoculation. Unless otherwise indicated, culture treatments were: doxorubicin, 1 µM; T12, 50 µM; and 172i, 25 µM.

Figure 10. Extracellular GPBP instigates pathogenesis

(A) Represented are the indicated variables in 4T1 mouse-model untreated (Control) or treated (N26) over time (left) or at day-25 (right). Statistics: Left, 2WA-S, n = 10. Representative of three assays. Right, Student’s t-test, n = 30. (B) Tumor growth over time in A549 mouse-model untreated (Control) or treated (hN26). Initial tumor volume (200–300 mm3) was set as 100%. Statistics: 2WA-S, n = 14. (A, B) Values are mean ± 95% CI. (C) cGPBP levels (mean ± SD) were monitored over time in 4T1 mouse-model using the preclinical EMTEST. Macroscopic metastases appear on day-17 (arrow). Statistics: 1WA-D. Two mice were sacrificed per day. Representative of three assays. (D) Time course flowchart of cGPBP levels and relevant clinical events for Patient 1. cGPBP measurements were done on fresh serum samples using preclinical EMTEST and samples stored at –80° C. MRI, magnetic resonance imaging; CT, computerized tomography. (E) Percentage of mice with primary tumor and percentage of mice with primary tumor also developing lung metastases at day-28 in LLC mouse-model. Statistics: Fisher’s exact test. Tumor: n = 17; Metastases: WT n = 14, GPBP−/−n = 4.

Figure 11. Epithelial cells clear tumor mesenchymal cGPBP levels and generate a previously unrecognized form of GPBP

(A) HEK 293 cells were transfected with pcDNA3 (–) or the indicated derived constructs coding for FLAG-GPBP or FLAG-GPBP-Ala132. After 36 h, equivalent amounts of the indicated materials were analyzed by WB with anti-GPBP mAb 14 [3]. (B) FLAG-GPBP-Ala132 transfected HEK 293 cells were analyzed by IF-CM using specific antibodies for the indicated polypeptides. FLAG-GPBP-Ala132 was visualized with anti-FLAG. MPR, mannose-6-phosphate receptor (bar = 10 μm). (C) Frozen lung sections from 4T1 mouse-model sacrificed at the indicated days (n = 2) were analyzed by IF-CM to visualize in situ T12 binding (bioT12) and the indicated polypeptides. FI was measured and expressed as in Figure 1. Statistics: 1WA-D comparison respect to day 1. (D) The phosphorylation mixtures containing the indicated FLAG-polypeptides and reagents were analyzed and represented as in previous figures. WB was developed with α-FLAG antibody. (E) A549 cultures were supplemented with BM40-FLAG-GPBP for 6 h in the presence (+) or absence (–) of dorsomorphin. BM40-FLAG-GPBP purified from cellular extracts was phosphorylated with (+) or without (–) T12 and similarly analyzed and represented. Phosphorylation mixtures of BM40-FLAG-GPBP (Input) and anti-FLAG immunoprecipitated material from mock A549 cultures (Control) are also displayed. WB, was developed with N27 antibody. Where indicated in D and E, dorsomorphin and T12 were used at 40 and 50 µM, respectively. Statistics: Student´s t-test, n = 4 (D) n = 2 (E). (F) cGPBP from NSCLC patient’s plasma (0.5 mL) was immunopurified with scFvN26, released using appropriated blocking peptide and subjected to in vitro phosphorylation assay. Resulting phosphorylation mixtures were analyzed and represented as in Figure 5B. Statistics: Student’s t-test, n = 3. Similar results were obtained when pull-down assays were done using scFvN27 and the corresponding blocking peptide, suggesting that cGPBP multimeric structures contained a minority of N27-reactive material that allowed immunoprecipitation and limited detection by EMTEST (<1 ng/mL).