PEPDG278D inhibits CRC cells resistant to EGFR MABs
We compared the response of four human CRC cell lines to cetuximab, panitumumab, and PEPDG278D, including HCT116, HT29, SW48, and SW620 cells. Both EGFR and HER2 were expressed in HCT116, HT29, and SW48 cells, but their expression levels varied greatly among the cell lines, whereas neither EGFR nor HER2 could be detected in SW620 cells (Fig. 1a). HCT116 cells carry activating mutations of KRAS (G13D) and PIK3CA (H1047R), and HT29 cells carry activating mutations of BRAF (V600E) and PIK3CA (P449T) (Suppl. Fig. 1). Mutated KRAS, BRAF and PIK3CA are widely believed to drive resistance of CRC cells to EGFR MABs by rendering the cells independent of EGFR.
Both cetuximab and panitumumab inhibited the growth of SW48 cells in a time- and concentration-dependent manner, but neither agent is active in HCT116 cells and HT29 cells (Fig. 1b and Suppl. Fig. 2a). Cetuximab was also evaluated in SW620 cells and was inactive (Fig. 1b), which is expected, since these cells do not express EGFR. PEPDG278D strongly inhibited the growth of SW48, HCT116, and HT29 cells in a time- and concentration-dependent manner but was ineffective in SW620 cells (Fig. 1b). We previously showed that PEPDG278D specifically targets EGFR and HER2 [19, 20, 22]. Although both PEPDG278D and the EGFR MBAs inhibit SW48 cells, the former is more potent than the latter. We next analyzed EGFR, several other RTKs, including HER2, HER3, MET, IGF1R, and several key downstream signaling proteins, including AKT, ERK, and SRC. EGFR and HER2 form heterodimeric signaling units with various RTKs to diversify their oncogenic signaling [28, 29]. The growth-inhibitory activities of the EGFR MABs in SW48 cells were accompanied by decrease in expression and phosphorylation of EGFR as well as decrease in phosphorylation but not expression of MET and ERK (Fig. 1c and Suppl. Fig. 2b). p-AKT and p-SRC were undetectable in SW48 cells. Decreased phosphorylation of MET and ERK induced by the EGFR MABs in SW48 cells apparently resulted from EGFR inhibition, as neither cetuximab nor panitumumab had any effect on MET and ERK in HCT116 cells and HT29 cells in which EGFR was not inhibited as well as in SW620 cells which do not express EGFR. Cetuximab was also evaluated against HER3 and IGF1R in SW48 cells and showed no effect on the RTKs. Cetuximab showed no effect on any of the signaling proteins in HCT116 and HT29 cells (Fig. 1c). PEPDG278D obliterated both expression and tyrosine phosphorylation of EGFR and HER2 in SW48, HCT116 and HT29 cells (Fig. 1c). PEPDG278D is more effective than the EGFR MABs in obliterating EGFR in SW48 cells, even though cells were treated by PEPDG278D at 25 nM but by the MABs at 275–277 nM. In HCT116, HT29 and SW48 cells, PEPDG278D had no effect on the expression of other signaling proteins but markedly decreased their phosphorylation (Fig. 1c). PEPDG278D inhibition of phosphorylation of HER3, IGF1R and MET but not their expression is consistent with its disruption of EGFR association with HER3, MET or IGF1R (Suppl. Fig. 3). We previously showed that PEPDG278D also disrupts the association of HER2 with each of these RTKs in HER2-positive breast cancer cells [21]. In SW620 cells that lack EGFR and HER2, PEPDG278D had no effect on the phosphorylation or expression of HER3, MET, IGF1R, SRC, AKT, and ERK (Fig. 1c). These results show that by depleting EGFR and HER2 in CRC cells, PEPDG278D not only directly suppresses both RTKs but also indirectly suppresses other RTKs by disrupting their association with EGFR or HER2, thereby causing extensive inhibition of oncogenic signaling. Indeed, siRNA knockdown of EGFR or HER2 also significantly inhibited the growth of HCT116 cells and HT29 cells (Suppl. Fig. 4a-b). Our results indicate that the inability of EGFR MABs to downregulate EGFR in HCT116 cells and HT29 cells is primarily responsible for their failure to inhibit the growth of these cells, rather than compensatory signaling driven by mutated KRAS, BRAF and PIK3CA. Our results also show that PEPDG278D is active in CRC cells overexpressing different levels of EGFR and HER2.
PEPDG278D abolishes RAS-ERK and PI3K-AKT signaling despite activating mutations in KRAS, BRAF, and PIK3CA
Although HCT116 cells carry activating mutations of KRAS and PIK3CA, and HT29 cells carry activating mutations of BRAF and PIK3CA, and both cell lines are resistant to cetuximab and panitumumab, as described above, the inhibitory activities of PEPDG278D in HCT116 cells and HT29 cells were similar to that in SW48 cells whose KRAS, BRAF and PIK3CA are not mutated. ERK and AKT are downstream of KRAS and PIK3CA, respectively. PEPDG278D caused marked loss of phosphorylation of ERK and AKT in both HCT116 and HT29 cells (Fig. 1c). MEK is upstream of ERK, and PEPDG278D also markedly decreased MEK phosphorylation in both HCT116 and HT29 cells (Fig. 2a). The loss of phosphorylation of MEK, ERK and AKT apparently resulted from PEPDG278D targeting EGFR and HER2, as PEPDG278D was inactive in SW620 cells and cetuximab was active only in SW48 cells (Figs. 1c and 2a).
It was previously shown that oncogenic RAS mutants regulate basal effector pathway signaling, while WT RAS in the same cells mediates signaling downstream of activated RTKs [30]. It was also shown that the gain of function of PIK3CA mutants is enabled by activated RAS or PI3K/p85-mediated binding to activated RTKs [31, 32]. Notably, only one allele of each of the KRAS and PIK3CA genes in HCT116 cells is mutated, and only one allele of each of the BRAF and PIK3CA genes in HT29 cells is mutated (Suppl. Fig. 1). Moreover, HRAS and NRAS are also expressed in both HCT116 and HT29 cells as well as in SW48 and SW620 cells (Fig. 2a). These results provide an explanation for why PEPDG278D is able to strongly inactivate MEK, ERK and AKT in HCT116 cells and HT29 cells. We also found that PEPDG278D strongly downregulates the expression of both HRAS and NRAS in HCT116, HT29 and SW48 cells but not in SW620 cells, whereas cetuximab had no effect on the expression of HRAS and NRAS in any of the cell lines (Fig. 2a). Since cetuximab only targets EGFR, the above results suggested that downregulation of HRAS and NRAS by PEPDG278D might result from HER2 depletion. Indeed, siRNA silence of HER2 but not EGFR resulted in loss of HRAS and NRAS (Suppl. Fig. 4a). However, neither PEPDG278D nor cetuximab regulated the expression of WT or mutated KRAS (Fig. 2a).
Consistent with the changes in the signaling molecules described above, total RAS and PI3K activities were strongly inhibited by PEPDG278D in HCT116 and HT29 cells as well as in SW48 cells but not in SW620 cells (Fig. 2b-c). There was little difference among HCT116, HT29 and SW48 cells with regard to percentage of inhibition of RAS and PI3K by PEPDG278D, as RAS activity was inhibited by 73% in HCT116 cells and 75% in both HT29 and SW48 cells, and PI3K activity was inhibited by 74% in HCT116 cells, 71% in HT29 cells, and 76% in SW48 cells. Notably, basal RAS activity is much higher in HCT116 and SW620 cells than in HT29 and SW48 cells, consistent with HCT116 cells carrying KRASG12D (Suppl. Fig. 1) and SW620 cells carrying KRASG12V [27], and basal PI3K activity is much higher in HCT116 and HT29 cells than in SW48 and SW620, consistent with HCT116 cells carrying PIK3CAH1047R and HT29 cells carrying PIK3CAP449T (Suppl. Fig. 1). Also, the remaining RAS and PI3K activities after PEPDG278D treatment were still higher in HCT116 and HT29 cells than in SW48 cells. Collectively, our results show that PEPDG278D strongly inhibits the RAS-MEK-ERK and PI3K-AKT signaling pathways by depleting EGFR and HER2, even if CRC cells harbor activating mutations of KRAS, BRAF and/or PIK3CA. Our results also indicate that PEPDG278D accomplishes this feat by abolishing both canonical function (tyrosine kinase) and non-canonical function (scaffolding – heterodimerization with other RTKs) of EGFR and HER2 as well as abolishing HER2 regulation of HRAS and NRAS.
PEPDG278D also targets EGFR mutants that occur in CRC patients
While EGFR is not mutated in SW48, HCT116 and HT29 cells [27], several acquired mutations in the extracellular domain of EGFR have been reported in CRC patients following cetuximab treatment, including R451C, K467T, and S492R, each of which prevents cetuximab binding and confers resistance to cetuximab [33, 34]. EGFRR451C and EGFRK467T also bind poorly to panitumumab [33]. However, these mutations locate far from the site (amino acids #166–310) to which PEPDG278D binds [20]. Because SW620 cells do not express EGFR, we transfected each EGFR mutant as well as WT EGFR into these cells and then treated the cells with solvent or PEPDG278D (25 nM for 48 h). Each EGFR mutant was strongly downregulated by PEPDG278D, showing loss of both expression and phosphorylation, and the extent of downregulation of each mutant by PEPDG278D is very similar to that of WT EGFR (Fig. 2d). Thus, mutations in EGFR which occur in CRC patients do not interfere with PEPDG278D targeting of EGFR.
EGFR ligands slow PEPDG278D induction of EGFR internalization and lysosomal degradation
PEPDG278D causes depletion of both EGFR and HER2, but HER2 depletion was much faster than that of EGFR in cells cultured in medium with 10% serum. In HCT116, HT29, and SW48 cells, HER2 level decreased markedly after 3 h of PEPDG278D treatment, whereas EGFR level showed no decrease even after 6 h of PEPDG278D treatment, although it showed profound decrease at 24 h (Fig. 3a). However, if the cells were cultured in serum-free medium, both EGFR and HER2 showed marked decrease after 3 h treatment with PEPDG278D (Fig. 3a). We previously showed that epidermal growth factor (EGF), a high affinity EGFR ligand, competes with PEPD for binding to EGFR [26]. Notably, EGFR ligands bind to subdomains 1 and 3 in EGFR extracellular domain [35, 36], whereas PEPDG278D binds to subdomain 2 in EGFR [20]. Adding HB-EGF, another high affinity EGFR ligand, to culture medium without serum mimicked the effect of serum on EGFR depletion induced by PEPDG278D, while HB-EGF itself did not modulate the expression of EGFR or HER2 (Fig. 3a). No HER2 ligand, other than PEPD or PEPDG278D, is known. These results suggest that EGFR ligands from serum interfere with PEPDG278D targeting of EGFR. We also found that cetuximab attenuates the growth-inhibitory activity of PEPDG278D in both HCT116 cells and HT29 cells (Suppl. Fig. 5a), which likely resulted from cetuximab competing with PEPDG278D for EGFR binding. Notably, cetuximab binds to extracellular subdomain 3 of EGFR [37].
We previously showed that PEPDG278D induces HER2 internalization and lysosomal degradation [21]. Here, we show that PEPDG278D also induces EGFR internalization and lysosomal degradation. We focused on SW48 cells, taking advantage of their high EGFR level. SW48 cells were cultured in serum-free medium. PEPDG278D binding to EGFR and subsequent EGFR trafficking were analyzed by immunofluorescence staining and confocal microscopy. PEPDG278D bound abundantly to cell membrane and colocalized with EGFR after 15 min of treatment, but at 6 h, neither PEPDG278D nor EGFR remained on cell membrane, with residual amount of PEPDG278D but no EGFR detected intracellularly (Fig. 3b). Next, cells were treated with PEPDG278D and/or chloroquine, the latter of which is a lysosome inhibitor. In the absence of chloroquine, PEPDG278D induced EGFR internalization, and the internalized EGFR colocalized with LAMP1, a lysosome marker, but at 6 h of treatment, almost no EGFR could be detected (Fig. 3c). However, chloroquine blocked EGFR degradation induced by PEPDG278D (Fig. 3c). Collectively, our results show that PEPDG278D induces EGFR internalization and degradation in the lysosome but EGFR ligands slow this process by interfering with PEPDG278D binding to EGFR.
PEPDG278D fails to inhibit tumors that overexpress a high-affinity EGFR ligand
We next compared the antitumor activities of PEPDG278D and cetuximab in vivo. PEPDG278D is degraded in vivo by coagulation proteases, but EP, a clinically used anticoagulant, inhibits PEPDG278D degradation [38]. EP itself has no antitumor activity but combining EP with PEPDG278D allows therapeutically relevant plasma concentrations of PEPDG278D to be achieved for inhibition of tumors overexpressing EGFR and/or HER2 [20, 22]. We inoculated human CRC cells to immunocompromised mice subcutaneously, and the tumor-bearing mice were randomized for treatment with EP, EP plus PEPDG278D, or cetuximab. Based on previous studies, EP was administered to the mice at 0.5 mg/kg daily by intraperitoneal injection (ip); PEPDG278D was administered at 4 mg/kg ip three times weekly; and cetuximab was administered at 15 mg/kg ip twice weekly. Both PEPDG278D and cetuximab strongly inhibited the growth of SW48 tumors and at the end of treatment inhibiting tumor growth by 91.5 and 85.8%, reactively, but neither agent inhibited the growth of HCT116 tumors and HT29 tumors (Fig. 4a). Escalating PEPDG278D to 8 mg/kg did not inhibit tumor growth either (Suppl. Fig. 5b). Tumors were collected 24 h after the final dose of each agent and select signaling proteins were analyzed. In SW48 tumors, PEPDG278D decreased the expression and phosphorylation of both EGFR and HER2, while cetuximab only decreased the expression and phosphorylation of EGFR, and both agents also decreased ERK phosphorylation (Fig. 4b). Similar results were shown in cultured SW48 cells as described before. However, neither PEPDG278D nor cetuximab had any effect on EGFR, HER2 and ERK in HCT116 and HT29 tumors (Fig. 4b). Yet, in cultured HCT116 and HT29 cells as described before, while cetuximab was inactive, PEPDG278D strongly reduced the expression and phosphorylation of both EGFR and HER2 and decreased ERK phosphorylation.
Lack of inhibitory activity of PEPDG278D in HCT116 tumors and HT29 tumors was not due to lack of PEPDG278D delivery, as plasma concentrations of PEPDG278D were high and similar in mice bearing SW48 tumors and HCT116 tumors (Fig. 4c). We analyzed all seven known EGFR ligands (soluble form) in the tumor tissues but detected only AREG and HB-EGF. AREG is a low affinity EGFR ligand, and its affinity for EGFR is approximately 50 fold lower than that of HB-EGF [39]. AREG level was high in all three types of tumors (Fig. 4d). HB-EGF level was 13.6–14.6 fold higher in HT29 and HCT116 tumors than in SW48 tumors (Fig. 4e). This suggests that excessive tumor-generated HB-EGF might prevent PEPDG278D from binding to EGFR, and also suggests that EGFR signaling remains important to the tumors carrying activating mutations of KRAS, BRAF and/or PIK3CA. PEPDG278D also failed to downregulate HER2 in tumors expressing high level of HB-EGF. HB-EGF does not bind to HER2, and we showed previously that PEPDG278D disrupts the HER2-EGFR heterodimer even when EGF is bound to EGFR [22]. However, it is possible that without inhibiting EGFR, the impact of PEPDG278D on HER2 may be negated by rapid tumor growth.
Aderbasib restores the antitumor activity of PEPDG278D in tumors overexpressing HB-EGF
Aderbasib inhibits the shedding of all EGFR ligands by inhibiting ADAM10 and ADAM17 as mentioned before. We evaluated the antitumor activity of aderbasib as a single agent or in combination with EP plus PEPDG278D in HCT116 and HT29 tumors. Tumor-bearing mice were treated with EP, aderbasib, or the combination of aderbasib with EP and PEPDG278D. As in previous experiments, EP was administered at 0.5 mg/kg ip daily, and PEPDG278D was administered at 4 mg/kg ip three times weekly. Aderbasib was administered at 60 mg/kg by gavage daily. In HCT116 and HT29 tumors, aderbasib alone was ineffective, but combining aderbasib with EP plus PEPDG278D inhibited tumor growth by 63.3 and 54.4% respectively at the end of treatment (Fig. 5a). The combination treatment was less effective when aderbasib was reduced to 30 mg/kg and became ineffective when it was reduced to 15 mg/kg (Suppl. Fig. 6a-c). Mice treated with aderbasib alone or the combination regimen did not show signs of toxicity.
Aderbasib caused marked decrease in soluble HB-EGF level in the tumor tissues (Fig. 5b). In both tumor models, neither EP nor aderbasib had any effect on the expression or phosphorylation of the proteins analyzed, but combining aderbasib with EP and PEPDG278D caused profound loss of both expression and phosphorylation of EGFR and HER2, loss of phosphorylation but not expression of all other RTKs and downstream signaling proteins analyzed, including HER3, IGF1R, MET, SRC, AKT, ERK, and MEK, loss of expression of NRAS and HRAS, and activation of caspase 3 (Fig. 5c). These results show that by blocking shedding of HB-EGF from tumor cells, aderbasib enables PEPDG278D to engage its targets and to exert its antitumor activity. Although not measured, aderbasib probably also blocked the shedding of AREG from the tumor cells.
Notably, in tumors treated by the triple combination (EP, aderbasib and PEPDG278D), low levels of p-AKT, p-ERK and p-MEK remained despite profound loss of both expression and phosphorylation of EGFR and HER2 induced by PEPDG278D, suggesting that mutated PIK3CA, KRAS and BRAF may sustain a low level of signaling despite depletion of EGFR and HER2. Likewise, in cultured HCT116 and HT29 cells, despite profound loss of both EGFR and HER2 and marked inhibition of both RAS and PI3K activities by PEPDG278D, residual RAS and PI3K activities remain as mentioned before.
Adding 5-FU to the PEPDG278D-based combination treatment enhances therapeutic outcome
Because 5-FU, an antimetabolite, is commonly used in CRC treatment, we asked whether combining 5-FU with the PEPDG278D-based combination regime described above enhances treatment outcome. Thus, tumor-bearing mice were treated with EP, or the combination of EP, 5-FU, aderbasib and PEPDG278D. As in other experiments, EP was administered at 0.5 mg/kg ip daily, PEPDG278D was administered at 4 mg/kg ip three times weekly, and aderbasib was administered by gavage at 60 mg/kg daily. 5-FU was administered to mice at 35 mg/kg ip once every 3–4 days, which was not toxic in a dose-finding experiment. We first evaluated the combination regimen in mice bearing subcutaneous HCT116 and HT29 tumors. The combination treatment was highly effective against both types of tumors, inhibiting tumor growth by 72.4% (HCT116 tumors) and 69.4% (HT29 tumors) at the end of treatment (Fig. 6a-b), which is more efficacious than the combination minus 5-FU as described before. 5-FU as a single agent or 5-FU in combination with EP and PEPDG278D without aderbasib was ineffective (Suppl. Fig. 7). The combination regimen was also evaluated in a CRC PDX (PDX14650) which harbors KRASG12D (homozygous) and generates high level of HB-EGF (Suppl. Fig. 8). The combination regime markedly decreased tumor soluble HB-EGF level (Suppl. Fig. 8) and inhibited tumor growth by 83.2% at the end of treatment (Fig. 6c). Adverse effects were not detected in the mice in any of the tumor models. Tumors in the different models were collected 24 or 48 h after final treatment. In all the tumor models, the combination treatment caused profound loss in both expression and phosphorylation of EGFR and HER2, loss of phosphorylation but not expression of all other RTKs and downstream signaling proteins measured, including HER3, IGF1R, MET, SRC, AKT, MEK, ERK, loss of expression of HRAS and NRAS, and activation of caspase 3 (Fig. 6d). These results are similar to that obtained from tumors treated by aderbasib plus EP and PEPDG278D and from cultured cells treated by PEPDG278D as a single agent, as described before. Thus, 5-FU enhances tumor inhibition when combined with the other agents but does not interfere with the inhibition of oncogenic signaling by PEPDG278D and does not interfere with aderbasib inhibition of EGFR ligand shedding by tumor cells. 5-FU is known to exert its antitumor activity by causing misincorporation of fluoronucleotides into RNA and DNA and inhibiting the nucleotide synthetic enzyme thymidylate synthase.
We also evaluated the combination regimen in mice bearing orthotopic HCT116 tumors. HCT116 cells stably expressing firefly luciferase were inoculated to the cecum of mice. Tumor growth was monitored by bioluminescence imaging (Suppl. Fig. 9a). Mice were randomized to EP or the combination regimen. Treatments with EP, aderbasib, PEPDG278D and 5-FU were the same as described before and were started on 22, 26, 27 and 28 days after cell inoculation, respectively. Tumor burden was not significantly different between the control and combination treatment group at the beginning of treatment (day 25), but tumor burden became significantly and consistently lower in the combination treatment group (Fig. 7a). The mean tumor bioluminescence intensity in the combination treatment group was consistently nearly 2 orders of magnitude lower than that in the control. The experiment was terminated on day 57, when several mice in the control group became moribund. Necropsy showed primary tumors in the cecum, local metastasis (peritoneal tumors), and liver metastasis. However, macroscopic tumors were found only in mice showing bioluminescence signals of > 3.2 × 108 photons per second. Six of the 11 mice in the control (54.6%) and 2 of the 12 mice in the combination treatment (16.7%) showed cecum and/or peritoneal tumors (Fig. 7b). Average tumor weight (cecum and peritoneal tumors) in the combination treatment was only 8.8% of that in the control (Fig. 7c). Four of the 11 mice (36.4%) in the control showed liver metastasis, but only 1 of the 12 mice (8.3%) in the combination treatment showed liver metastasis (Fig. 7d). Representative primary and metastatic tumors are shown in Suppl. Fig. 9b. All tumors were verified by histological analysis, and representative images are shown in Suppl. Fig. 9c. Adverse effects of the treatments were not detected. As in other experimental models described before, the combination treatment caused profound loss in both expression and phosphorylation of EGFR and HER2, loss of phosphorylation but not expression of HER3, IGF1R, MET, SRC, AKT, MEK and ERK, loss of expression of HRAS and NRAS, and activation of caspase 3 (Fig. 7e).