The human tumor cell lines A549 (lung adenocarcinoma), C33A (cervical carcinoma), DLD-1 (colorectal adenocarcinoma), MDA-MB-231 (breast adenocarcinoma) and PC-3 (prostate carcinoma) were purchased from American Type Culture Collection. These cell lines were selected because the respective tumors are known to be affected by the CXCL12 pathway (3). The cell line authenticity was verified by the supplier using short tandem repeat profiling. Cell lines were regularly tested for mycoplasma contamination using MycoSpy^® Master Mix (Biontex Laboratories GmbH). Cells were plated on 12-well culture plates (TPP Techno Plastic Products AG), and grown in either DMEM (4.5 g/l glucose) for C33A, MDA-MB-231 and PC-3 or RPMI-1640 for A549 and DLD-1, supplemented with 0.05% gentamycin and 10% fetal bovine serum (FBS) (all from Gibco; Thermo Fisher Scientific, Inc.) at 37°C in a water-saturated atmosphere of 95% air and 5% CO₂. For experiments, cell cultures at 60% confluence were starved for 24 h in their corresponding serum-free culture medium at 37°C, supplemented with or without pertussis toxin (PTX; 100 ng/ml pre-dissolved; cat. no P2980; Sigma-Aldrich; Merck KGaA), followed by addition of the following compounds for 1 h: CCX771 (CXCR7 antagonist; 100 nM dissolved in DMSO; Amgen, Inc.), AMD3100 (CXCR4 antagonist; 10 µM dissolved in double-distilled water; Sigma-Aldrich; Merck KGaA), Tyrphostin AG1478 (EGFR inhibitor; 2 µM dissolved in DMSO; cat. no. T4182; Sigma-Aldrich; Merck KGaA), Src inhibitor-1 (Src-I1; Src kinase inhibitor; 10 µM dissolved in DMSO; Sigma-Aldrich; Merck KGaA). For control purposes, untreated cells were supplemented with appropriate concentrations of DMSO.

For western blot analysis, cultured A549, C33A, DLD-1, MDA-MB-231 and PC3 cells were harvested and immediately placed in liquid nitrogen. Total protein was isolated in RIPA protein extraction buffer (cat. no. 89901; Thermo Fisher Scientific, Inc.) supplemented with protease and phosphatase inhibitors (cat. no. 78442; Thermo Fisher Scientific, Inc.) to prevent protein dephosphorylation. Protein content was determined using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Proteins (10–20 µg/lane) were separated using 10% SDS-PAGE and transferred to nitrocellulose membranes by electroblotting. After blocking non-specific binding sites with Pierce Fast Blocking Buffer (cat. no. 37575; Thermo Fisher Scientific, Inc.) for 60 min at room temperature, the membranes were incubated overnight at 4°C with one of the following primary antibodies: Rabbit anti-EGFR (1:1,000; cat. no. 4267; Cell Signaling Technology Europe, B.V.), rabbit anti-β-arrestin2 (1:1,000; cat. no. 3857; Cell Signaling Technology Europe, B.V.) and mouse monoclonal anti-β-arrestin1 (1:1,000; cat. no. MA1-183; Thermo Fisher Scientific, Inc.). Subsequently, the membranes were incubated at room temperature for 1 h with HRP-conjugated anti-mouse (1:10,000; cat. no. PI-2000-1; Vector Laboratories, Inc.) or anti-rabbit (1:10,000; cat. no. 711-035-152; Jackson ImmunoResearch Laboratories Europe, Ltd.) secondary antibodies. Antibody-labeling was visualized using Pierce™ ECL Western Blotting Substrate (Thermo Fisher Scientific, Inc.). To control for protein loading, membranes were re-probed with rabbit anti-β-actin (1:5,000; cat. no. 4970; Cell Signaling Technology Europe, B.V.) or mouse anti-GAPDH (1:5,000; cat. no. 10R-G109A; Fitzgerald Industries International) antibodies. Chemiluminescence was captured on a Biostep Celvin S imager (Biostep; Bionis) and immunoreactive protein bands were quantified using the Snap and Go 1.8.3 software (Biostep; Bionis).

Chemotactic responses of A549, C33A, DLD-1, MDA-MB-231 and PC3 cells to CXCL12 were determined in a modified 12-well or 48-well Boyden chamber (Neuro Probe Inc.), in which the upper and lower wells were separated by polyornithine-coated Nucleopore^® PVP-free polycarbonate filter (Whatman plc; Cytiva; 8-µm pore size). For seeding, pretreated cells were counted using an improved Neubauer chamber and ~10,000 cells were placed in serum-free DMEM or RPMI medium into the upper well of the Boyden chamber. A total of 150 µl of the respective serum-free medium supplemented with CXCL12 (100 ng/ml; PeproTech, Inc.) were added to the lower chamber. The chamber was incubated at 37°C in a water-saturated atmosphere of 95% air and 5% CO₂ for 4 h. After incubation, non-migrated cells attached to the upper part of the membrane were wiped off, and migrated cells attached to the lower part of the membrane were fixed with ice-cold methanol (100%) at room temperature for 5 min, stained with DAPI (1:5,000; AAT Bioquest, Inc.; 5 min, room temperature), and counted on an Olympus BX40 microscope using the Olympus cellSens Dimension software (Olympus Corporation) at a final magnification of 50×. The number of cells migrating in the absence of CXCL12 was set to 1 and the migration index was calculated as the ratio of cells migrating in the presence and absence of CXCL12.

Predesigned human β-arrestin1 small interfering (si)RNA (cat. no. 4392420; Assay ID, s1624) and human β-arrestin2 (cat. no. 4392420; Assay ID, s1625), as well as control siRNA (cat. no. 4390844) were purchased from Thermo Fisher Scientific, Inc. The sequences of the siRNAs are presented in Table I. Transfection of A549, C33A, DLD-1, MDA-MB-231 and PC3 cells with siRNA (75 pmol/well) was performed in 6-well plates (TPP Techno Plastic Products AG) at 37°C in the presence of serum-free DMEM or RPMI-medium using the siPORT Amine Transfection Agent (cat. no. AM4503; Thermo Fisher Scientific, Inc.) for 16 h. Transfected cells were further maintained with serum-free DMEM or RPMI-medium and subjected to experiments after 24 h. The success of RNA interference was validated by western blotting and reverse transcription-quantitative (RT-q)PCR. Cells treated with PTX (100 ng/ml, 24 h) were used as a positive control for inhibited cell migration.

Briefly, total RNA was extracted from A549, C33A, DLD-1, MDA-MB-231 and PC3 cells using InviTrap Spin Universal RNA Mini Kit (Invitek Diagnostics), followed by reverse transcription of 1 µg RNA using Protoscript First Strand cDNA Synthesis Kit (New England BioLabs, Inc.) as per the manufacturer's instructions. For quantification of gene expression, qPCR analysis was performed with PowerUp SYBR Green Master Mix (cat. no. A25742; Thermo Fisher Scientific, Inc.) on a CFX96 thermal cycler system (Bio-Rad Laboratories, Inc.). PCR conditions consisted of an initial denaturation step at 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec, 59.5°C for 30 sec and 72°C for 30 sec. Gene expression was calculated using the 2^−ΔΔCq method (32) and normalized to β-actin. Primer sequences are listed in Table II.

Data are presented as the mean ± SD Experiments were replicated at least three times. When results showed large variations, a total of up to 15 replicates were used. GraphPad Prism 9 (Dotmatics) was used for statistical analysis. One-way ANOVA followed by Tukey's post hoc test was employed for multiple comparisons. P<0.05 was considered to indicate a statistically significant difference.

Consistent with our previous study (6), CXCL12 at 100 ng/ml induced chemotaxis of all tumor cell lines as evidenced by a 2- to 3-fold increase in their migration index (Fig. 1). In addition, the CXCR4 antagonist AMD3100 and the CXCR7 antagonist CCX771 attenuated the CXCL12-induced migration of C33A, MDA-MB-231 and PC-3 cells (Fig. 1), suggesting that the chemotactic responses of these cells to CXCL12 are mediated by both CXCR4 and CXCR7. Furthermore, the CXCL12-induced migration of DLD-1 cells was sensitive to CCX771, but not to AMD3100, confirming the previously suggested primary role of CXCR7 in the CXCL12-dependent migration of DLD-1 cells (6). In further accordance with our previous study, CXCL12-induced migration of A549 cells was sensitive to CCX771. However, in contrast to the previous observation showing a tendency of AMD3100 to attenuate CXCL12-induced migration of A549 cells (6), a statistically significant decrease in the numbers of migrating A549 cells treated with AMD3100 and CXCL12 compared with CXCL12 alone was observed in the present study (Fig. 1). This discrepancy likely reflects the use of a 12-well Boyden chamber for chemotactic analysis in the present experiments compared with a 48-well Boyden chamber plated with 5,000 cells per well in our previous study (6). Comparison of the chemotactic responses of A549 cells to CXCL12 in the presence and absence of receptor antagonists under the different experimental set-ups indicated more pronounced inhibitory responses in the 12-well Boyden chamber compared with the 48-well Boyden chamber assay, the reason of which is currently unknown (Fig. S1). The mechanism responsible for this discrepancy is currently unknown, but may well reflect differences in cell adhesion in the different sized chambers. These data imply that contrary to the conclusion of our previous study, CXCL12-dependent chemotaxis of A549 cells is mediated by both CXCR7 and CXCR4 and not solely CXCR7.

Tumor cells have been reported to express CXCL12 (5). The present RT-qPCR analysis revealed low levels of expression of CXCL12 mRNA in C33A cells which became detectable only after 30 PCR cycles. By contrast, in A549, DLD-1, MDA-MB-231 and PC3 cells, CXCL12 mRNA was virtually undetectable after 30 PCR cycles (data not shown). These findings suggest that CXCL12 in respective tumors is predominantly derived from non-tumor cells, such as CAFs or endothelial cells (5). These findings further argue against the possibility that the current analyses are biased by endogenous CXCL12.

It is has been demonstrated that CXCR4 and CXCR7 initiate cell signaling through either G proteins (preferentially Gα protein) or β-arrestin1/β-arrestin2, depending on the cell type and context (7–10,12). To assess the role of Gα proteins in CXCL12-dependent tumor cell migration, the migratory responses of tumor cells previously treated with PTX (100 ng/ml) for 24 h were analyzed. PTX partially or completely prevented the CXCL12-induced chemotaxis of A549, C33A, DLD-1, MDA-MB-231 and PC-3 cells (Figs. 1 and S2). To determine whether arrestins are additionally involved in CXCL12-induced tumor cell migration, cells were transfected with siRNA recognizing β-arrestin1 mRNA, β-arrestin2 mRNA or both. Western blot and RT-qPCR analyses confirmed a substantial 60–98% decrease in both β-arrestin1 and β-arrestin2 protein as well as mRNA levels on day 2 post transfection (Figs. S3 and S4). The depletion of β-arrestins had no apparent effect on the chemotactic responses of tumor cells to CXCL12, which were similar to those observed with wild-type cells (Figs. 2 and S5). Furthermore, in A549, C33A, MDA-MB-231 and PC-3 cells neither PTX treatment nor β-arrestin depletion affected the sensitivity of CXCL12-induced chemotaxis to AMD3100 and/or CCX771 (Figs. 1 and 2). Similarly, the sensitivity of CXCL12-induced chemotaxis to these inhibitors was unaffected after transfection of cells with control siRNA (Figs. S6 and S7). Collectively, these findings suggested that inhibition of CXCR4 or CXCR7 is not associated with a switch from the inactive to the active state of the receptor downstream CXCL12 signaling. However, it was also observed that transfection of DLD-1 cells with control siRNA or β-arrestin siRNA rendered CXCL12-dependent chemotaxis partially sensitive to AMD3100, the reason of which is unclear (Figs. 2 and S6). Collectively, these findings indicated that CXCL12-induced tumor cell migration mediated by CXCR4 and/or CXCR7 is typically dependent on Gα_i/o signaling.

Among others, G proteins activate cell signaling through Src-dependent transactivation of EGFR family members, including EGFR/ErbB1 (18). Western blotting demonstrated that the tumor cell lines used in the present study expressed EGFR at varying levels, with the lowest levels observed in C33A cells (Fig. S8). To assess the putative role of EGFR transactivation in the chemotactic responses of A549, C33A, DLD-1, MDA-MB-231 and PC-3 cells to CXCL12, chemotaxis was re-analyzed after pretreatment of the cells with the EGFR inhibitor AG1478 (2 µM) for 1 h. AG1478 completely abolished the CXCL12-dependent chemotactic responses of the different cell lines (Figs. 3 and S9). These inhibitory effects were not affected by the additional treatment of the cells with PTX (Fig. 3). Transactivation of EGFR typically occurs via Src according to previous reports (18–23). The chemotactic responses of the different cell lines to CXCL12 were also prevented after pretreatment with the Src kinase inhibitor Src-I1 (10 µM) for 1 h (Figs. 4 and S10). Again, these inhibitory effects were not further modulated by PTX treatment (Fig. 4). To evaluate the putative effects of the different inhibitors on EGFR expression, tumor cells were treated with either PTX (100 ng/ml) for 24 h or AG1478 (2 µM) and Src-I1 (10 µM) for 1 h, and EGFR expression levels were then analyzed by western blotting. None of the treatments resulted in obvious changes in EGFR protein, implying that the observed inhibitory effects are not due to the loss or reduction of EGFR expression (Fig. S8).

Taken together, these findings identified Src-dependent transactivation of EGFR as a common mechanism underlying CXCL12-induced migration of tumor cells from a number of organs. In addition, the present results showed that both CXCR4 and CXCR7 signal through EGFR transactivation in tumor cells.