TIMP1 promotes cell proliferation and invasion capability of right-sided colon cancers via the FAK/Akt signaling pathway

2.1 Patients and tissue samples

We enrolled 15 patients with seven right-sided and eight left-sided CRCs and obtained 14 paired biopsy specimens from each cancer and surrounding normal epithelia at least 5 cm away from the lesion to establish an organoid (PDO). Of these, we used only eight cancer PDOs (four right-sided and four left-sided CRCs), which carried WT KRAS, WT BRAF, mutated TP53, and microsatellite stable (MSS) and the corresponding normal epithelial PDOs to perform in vitro experiments under genetically similar conditions.

2.2 Crypt isolation and organoid culture

Crypt isolation and organoid culture were performed as we previously described.¹⁹ In brief, epithelial fragments were adequately washed with cold PBS and crypts were isolated by incubation in 0.5 mM EDTA with shaking at 4°C. Isolated crypts were incubated in digestion buffer consisting of DMEM with 2.5% FBS, collagenase type IX, and dispase type II, and epithelial cells were mechanically dissociated by pipetting. Isolated cells were embedded in Matrigel (Corning) and seeded in 48-well plates, followed by polymerization at 37°C. Subsequently, embedded crypts were cultured with the culture medium as described previously.¹⁸

2.3 Cell viability assay

Organoid cell viability was determined by Cell Titer Glo assay (Promega), as previously described.²⁰

2.4 RNA extraction and quantitative RT-PCR

RNA extraction and quantitative RT-PCR were performed as previously described.²¹ The simultaneous assay of mRNA gene expression of organoids was performed in custom-designed TaqMan Array Plates (Applied Biosystems), consisting of four endogenous controls (18 s, GAPDH, HPRT, and GUSB) and 184 target genes associated with the prognosis of CRC, as previously reported.^22-25 The TaqMan primer/probes used are listed in Table S2.

The detailed methods for the invasion assay, inhibitor experiments, lentiviral transduction, western blotting, The Cancer Genome Atlas data analysis and statistical analysis are described in Appendix S1.

3.1 Higher cell proliferation activity in right-sided colorectal cancer compared with left-sided colorectal cancer

To compare the malignant potential of cancer cells between right-sided and left-sided CRCs, we first established a total of eight paired cancer and normal epithelial organoids from four right-sided and four left-sided CRCs. The patient characteristics are presented in Table S1. We cultured those PDOs for 1 – 7 days and directly compared cell proliferation between them by Cell Titer Glo assay. Cell viability in right-sided cancer PDOs was significantly higher than in left-sided cancer PDOs at day 7 (8.98 ± 0.95 vs 5.18 ± 1.08; p = 0.0019; Figure 1A). Moreover, the viability of right-sided and left-sided cancer PDOs was significantly higher than in the respective corresponding normal PDOs (8.98 ± 0.95 vs 1.61 ± 0.46 and 5.18 ± 1.08 vs 1.87 ± 0.43; p < 0.0001 and p = 0.0013, respectively; Figure 1A). No significant difference was observed between right-sided and left-sided normal PDOs.

To confirm the difference in cell proliferation activity in PDOs, we measured the diameter of organoid clusters in six randomly selected areas in right-sided and left-sided cancer PDOs. Representative images are shown in Figure 1B. The diameters of most clusters of right-sided cancer PDO-8 increased rapidly from day 1 to day 7, whereas the diameters of clusters of left-sided cancer PDO-12 increased slowly over time. Quantitative analysis showed significantly higher diameters in right-sided cancer PDO-8 than in left-sided cancer PDO-12 at day 3 (p = 0.0009), day 5 (p = 0.0344), and day 7 (p = 0.0403; Figure 1C).

3.2 Higher cell invasion capability in right-sided colorectal cancer than in left-sided colorectal cancer

To further investigate the malignant potential of right- and left-sided CRCs, we performed an invasion assay using four right-sided and four left-sided paired PDOs to compare their invasion capability. Representative images of paired PDOs are shown in Figure 1D. Right-sided cancer PDO-8 exhibited many more invading cells than corresponding normal PDO-8. However, left-sided cancer PDO-15 exhibited slightly more invading cells than corresponding normal PDO-15. Similar images were obtained in other right-sided and left-sided PDOs (Figure S1). The quantitative analysis of invading cells revealed significantly more invading cells in right-sided CRC PDOs than in left-sided CRC PDOs (324.00 ± 25.34 vs 82.25 ± 49.49; p = 0.0003; Figure 1E). The number of invading cells in right-sided and left-sided cancer PDOs was significantly greater than in respective corresponding normal PDOs (p = 0.0003 and p = 0.0323, respectively; Figure 1E). Interestingly, the number of invading cells in right-sided normal PDOs was higher than in left-sided normal PDOs (p = 0.0333; Figure 1E).

3.3 Predominant activation of tissue inhibitor matrix metalloproteinase 1-related signaling pathway in right-sided colorectal cancer

To investigate the mechanism of higher cell proliferation activity and higher cell invasion capability of right-sided CRCs, we performed gene expression assays of 184 genes in 30 signal transduction pathways associated with cell proliferation and invasion capability using three paired cancer and normal PDOs. The relatively higher mRNA expression in right-sided versus left-sided CRC PDOs was ranked, and the top five genes identified by this approach were S100P, DHH, SDC2, TIMP1, and MSN (Figure 2A). We further verified the mRNA levels of these five genes using all four paired PDOs from both sides. The relative mRNA level of TIMP1 in all four right-sided cancer PDOs was higher than that of all four left-sided cancer PDOs (1.00 ± 0.23 vs 0.09 ± 0.06; p = 0.0002; Figure 2B). The mRNA levels of MSN and DHH in right-sided cancer PDOs tended to be higher compared with left-sided PDOs, although the difference was not statistically significant (p = 0.0638 and p = 0.1262; respectively; Figure 2B). Regarding S100P and SDC2, the difference in the relative mRNA levels between the four right-sided and four left-sided PDOs did not reach statistical significance.

We then examined TIMP1 protein expression by western blotting in cancer and normal PDOs on both sides. The TIMP1 protein signals in all four right-sided cancer PDOs were stronger than those in the corresponding normal PDOs. In contrast, the TIMP1 protein signals in all four left-sided PDOs were much lower than in the corresponding normal PDOs (Figure 2C). Quantitative analysis of TIMP1 protein signals, normalized according to the β-actin signal and corresponding normal PDO TIMP1 signal, revealed that TIMP1 protein expression in right-sided CRCs was significantly higher than in left-sided CRCs (1.00 ± 0.18 and 0.05 ± 0.02; p < 0.0001; Figure 2D). These results suggest that the TIMP1-related signal transduction pathway is predominantly activated in right-sided CRC.

3.4 Tissue inhibitor matrix metalloproteinase 1 knockdown inhibits cell proliferation activity and invasion capability of right-sided colorectal cancer

To investigate the role of TIMP1 in right-sided CRCs, we knocked down the TIMP1 gene using two different lentiviral shRNAs (TRCN299345 and TRCN303682; hereafter defined as sh#1 and sh#2, respectively) in two right-sided and two left-sided cancer PDOs that were randomly selected. TIMP1 mRNA levels in right-sided PDO-4 transfected with TIMP1 shRNAs (PDO-4sh#1 and sh#2) were inhibited by 94.5% and 96.9%, respectively, compared with PDO-4 transfected with scramble shRNA (PDO-4NC) (Figure 3A). TIMP1 protein expression in PDO-4sh#1 and sh#2 was also markedly inhibited compared with PDO-4NC (Figure 3B). Similarly, TIMP1 mRNA levels in left-sided cancer PDO-12 transfected with shRNAs (PDO-12sh#1 and sh#2) were inhibited by 92.5% and 96.0%, respectively (Figure 3A), and no TIMP1 protein was observed in those PDOs (Figure 3B). When cell proliferation activity was examined from day 1 to day 7 by Cell Titer Glo assay, the viability of right-sided PDO-4sh#1 and sh#2 cells was significantly lower than that of PDO-4NC cells (p = 0.0001 and p < 0.0001, respectively; Figure 3C). However, the viability of left-sided PDO-12sh#1 and sh#2 cells was similar to that of PDO-12NC cells with no statistically significant difference (Figure 3C).

We next investigated the invasion capability of right-sided and left-sided PDO cells with TIMP1 knockdown. Representative images of invasion assays with PDO-4sh#1 and sh#2 as well as PDO-4NC cells are shown in Figure 3D. Invading cells in PDO-4sh#1 and sh#2 were noticeably fewer compared with PDO-4NC. Quantitative analysis revealed a significantly lower number of invading cells in PDO-4sh#1 and sh#2 than in PDO-4NC (p < 0.0001 for each; Figure 3E). Representative images from the invasion assay of left-sided PDO-12sh#1/#2 and PDO-12NC cells are shown in Figure 3D. There were fewer invading cells in PDO-12sh#1 and sh#2 than in PDO-12NC. Quantitative analysis revealed a significantly lower number of invading cells in PDO-12sh#1 and sh#2 than in PDO-12NC (p < 0.0001 for each; Figure 3E).

We also confirmed similar results for cell proliferation activity and invasion capability using right-sided cancer PDO-8 and left-sided cancer PDO-18 with TIMP1 knockdown (Figure S2). Thus, our results clearly indicate that TIMP1 plays an important role in cell proliferation activity and invasion capability, predominantly in right-sided CRCs.

3.5 Tissue inhibitor matrix metalloproteinase 1 activates FAK/Akt signal transduction pathway in right-sided colorectal cancer

To investigate TIMP1-related signal transduction pathways involved in cell proliferation activity and invasion capability, we examined the protein expression of key molecules in the FAK/Akt and MAPK pathways downstream of TIMP1 using four right-sided and left-sided TIMP1 knockdown PDOs. The expression of phosphorylated FAK (pFAK) protein in right-sided PDO-4sh#1 and sh#2 was manifestly lower compared with PDO-4NC (Figure 4A). Similar results were obtained for right-sided PDO-8sh#1/sh#2 compared with PDO-8NC. However, pFAK protein levels in the two left-sided knockdown PDOs were similar to those in each of the PDO negative controls. Similarly, downstream pAkt and pBad protein levels were decreased in right-sided TIMP1 knockdown PODs, whereas the levels were almost unchanged in left-sided TIMP1 knockdown PODs (Figure 4A). Quantitative analysis of band signals revealed that pFAK, pAkt, and pBad protein levels were significantly decreased by TIMP1 knockdown in right-sided PDOs, whereas those in the left-sided PDOs were unchanged by TIMP1 knockdown (Figure 4B), suggesting predominant activation of the FAK/Akt/Bad signaling pathway in right-sided but not left-sided PDO. Interestingly, higher expression of pAkt was observed in three of four right-sided cancer PDOs than in any of the four left-sided cancer PDOs (Figure S3). Quantitative analysis revealed that pAkt levels in right-sided cancer PDOs were significantly higher than those in left-sided cancer PDOs (p = 0.0210, Figure S3).

We then investigated the inhibitory effects of selective FAK and Akt inhibitors on cell proliferation and invasion capability in right-sided PDOs as well as left-sided PDOs. When the four right-sided PDOs (PDO-4/8/9/24) were treated with the selective Akt inhibitor MK-2206 for 7 days according to the methods described in Appendix S1, the viability of PDOs was significantly reduced compared with respective control PDOs, which were treated with vehicle alone; the reduction was 76.3% in PDO-4, 87.0% in PDO-8, 80.0% in PDO-9, and 79.5% in PDO-24, with an overall average (±SD) of 80.7 ± 4.5%, while the average reduction in the four left-sided PDOs was 53.8% ± 13.5%, which was significantly lower than that in right-sided PDOs (p = 0.0092) (Figure S4). In the invasion assay, the number of invading cells in right-sided PDOs treated with MK-2206 was significantly reduced by 92.1% ± 4.2%, while the reduction for four left-sided PDOs was 70.7% ± 11.7%, which was significantly lower than that in right-sided PDOs (p = 0.0137) (Figure S4). These data indicate that a selective Akt inhibitor had a greater effect on the cell proliferation and invasion capability of right-sided CRC PDOs than those of left-sided CRC PDOs. Similar results were obtained in experiments using the selective FAK inhibitor GSK2256098 (Figure S5).

In addition, we examined pERK protein expression, a downstream molecule of the MAPK signaling pathway, which is the other TIMP1-related pathway, in both-sided TIMP1 knockdown PDOs. However, the pERK levels in right-sided PDO-4 and PDO-8 were unchanged by TIMP1 knockdown. In contrast, pERK levels in left-sided PDO-12 and PDO-18 were increased by TIMP1 knockdown (Figure 4C). Quantitative analysis revealed that the pERK protein level in right-sided PDO-4 and PDO-8 were unchanged by TIMP1 knockdown, but the pERK level in left-sided PDO-8 and PDO-18 were conversely increased by TIMP1 knockdown (Figure 4D).

These results suggest that TIMP1 plays a crucial role in cell proliferation and invasion capability in right-sided CRC via FAK/Akt/Bad signaling.

3.6 Tissue inhibitor matrix metalloproteinase 1 expression and survival in high- and low-tissue inhibitor matrix metalloproteinase 1 colorectal cancers

To evaluate TIMP1 expression in a large number of right- and left-sided CRCs, we analyzed the public database of TCGA-COAD and TCGA-READ, which includes mRNA gene profiling data and tumor location information from a cohort of 587 patients. The median TIMP1 expression level in 251 right-sided CRCs was significantly higher than in 366 left-sided CRCs (9.2364 vs 8.9833, p = 0.0046; Figure 5A).

To further evaluate the association between TIMP1 expression and CRC prognosis, we compared OS between the high- and low-TIMP1 groups by Kaplan–Meier analysis. In the analysis of all patients (n = 587), OS in the high-TIMP1 group was significantly shorter than in the low-TIMP1 group (19.27 vs 21.42 months; HR = 1.482; 95% Cl, 1.017–2.158; p = 0.0398; Figure 5B). Similarly, in the analysis of right-sided CRCs (n = 251), OS in the high-TIMP1 group was significantly shorter than in the low-TIMP1 group (18.30 vs 23.52 months; HR = 1.727, 95% Cl, 1.022–2.916; p = 0.0445; Figure 5C). However, in the analysis of left-sided CRCs (n = 336), no significant difference in OS was observed between the two groups (24.27 vs 21.23 months; HR = 1.205, 95% Cl, 0.7010–2.072; p = 0.4953; Figure 5D). These results clearly indicate that TIMP1 plays an important role in determining the poor prognosis of right-sided CRC.

3.7 Activation of epidermal growth factor receptor signaling is predominant in left-sided colorectal cancer

It has been reported that left-sided CRC with RAS WT is more sensitive to anti-EGFR agents,²⁶ and that expression of key molecules in the EGFR signaling pathway is higher in left-sided CRC tissues.^{27, 28} Therefore, we investigated the expression of the EGF receptor family (EGFR, ERBB2, and ERBB3) and key ligands in the EGFR signaling pathway using PDOs. EGFR mRNA levels in left-sided PDOs were significantly higher than in right-sided PDOs (1.01 ± 0.27 vs 1.94 ± 0.44; p = 0.0361; Figure 6A). Similarly, the levels of ERBB2 and ERBB3 mRNA in left-sided PDOs tended to be higher than those in right-sided PDOs, although the difference did not reach statistical significance. In the analysis of ligands, the amphiregulin (AREG) mRNA level in left-sided PDOs was significantly higher than in right-sided PDOs (0.60 ± 0.35 vs 1.49 ± 0.12; p = 0.0139; Figure 6B). However, the TGFα (TGFA) level was significantly higher in right-sided PDOs compared with left-sided PDOs (0.94 ± 0.15 vs 0.18 ± 0.04; p = 0.0009). The levels of mRNA for the other ligands, including epiregulin (EREG), neuregulin 2 (NRG2), NRG4, and EGF, did not show any significant difference between the two sides.

To verify the clinical benefit of anti-EGFR agents in left-sided CRCs, we examined the sensitivity of PDOs to cetuximab by Cell Titer Glo assay. The IC₅₀ values of cetuximab against left-sided PDOs were much lower than any of the right-sided PDOs (Figure 6C). The mean IC₅₀ value in left-sided PDOs was significantly lower than in right-sided PDOs (54.45 ± 21.18 vs 205.57 ± 31.92 μg/ml; p = 0.0027; Figure 6D), indicating significantly higher sensitivity of left-sided PDOs compared with right-sided PDOs.

We then analyzed total EGFR and phosphorylated EGFR (pEGFR) protein levels before and after treatment with cetuximab by western blotting. Before treatment, pEGFR levels in all three left-sided PDOs were obviously higher than those in any of the three right-sided PDOs, although total EGFR protein levels on both sides were similar to each other. However, after treatment with cetuximab for 72 h, only faint expression was observed in right- and left-sided PDOs (Figure 6E). Quantitative analysis revealed that the pEGFR level in left-sided PDOs was significantly higher than in right-sided PDOs before treatment, but cetuximab treatment markedly inhibited pEGFR levels to similarly very faint levels (Figure 6F). That is, the reduction of pEGFR levels following cetuximab treatment in left-sided PDOs (90.3% ± 1.1%) was significantly higher than that in right-sided PDOs (86.3% ± 1.2%; p = 0.0136). Because pEGFR levels are reportedly correlated with cell proliferation activity and/or malignant potential in CRCs,^{29, 30} these results suggest that the greater reduction in left-sided PDOs causes stronger tumor cell inhibition (i.e. higher susceptibility) to anti-EGFR agents. Thus, the EGFR signaling pathway is more predominantly activated in left-sided cancer PDOs.

To investigate the correlation between activation of the EGFR signaling pathway and the status of TIMP1, we then examined protein levels of pEGFR and total EGFR in TIMP1 knockdown PDOs and PDO NCs by western blotting (Figure S6). The pEGFR levels in left-sided PDO NCs (PDO-12/18 NC) were obviously higher than in right-sided PDO NCs (PDO-4/8 NC), similar to the findings for left-sided and right-sided PDOs in Figure 6E. However, the pEGFR levels in TIMP1 knockdown PDOs were similar to those of the respective PDO NCs for both left-sided PDOs (PDO-12sh#1/sh#2, PDO-18sh#1/sh#2) and right-sided PDOs (PDO-4sh#1/sh#2, PDO-8sh#1/sh#2). Although cetuximab treatment inhibited pEGFR levels to a greater degree in left-sided PDOs than in right-sided PDOs, the inhibitory effects of cetuximab on pEGFR levels in TIMP1 knockdown PDOs were similar to those of the respective PDO NCs in both left- and right-sided PDOs. These results suggest no significant direct correlation between the status of TIMP1 and activation of the EGFR signaling pathway or susceptibility to anti-EGFR agents.