Research ArticleCancer

Blocking EGFR palmitoylation suppresses PI3K signaling and mutant KRAS lung tumorigenesis

See allHide authors and affiliations

Science Signaling  03 Mar 2020:
Vol. 13, Issue 621, eaax2364
DOI: 10.1126/scisignal.aax2364

An EGFR pathway switch

Epidermal growth factor receptor (EGFR) signaling through generally two pathways, RAS-MAPK and PI3K-AKT, stimulates cell proliferation and survival; as such, EGFR is an attractive therapeutic target to inhibit tumor growth. However, EGFR inhibitors are rarely effective in tumors with KRAS mutations, which are common and result in MAPK pathway stimulation independently of EGFR activity. One exception is when the intracellular tail region of EGFR is palmitoylated. Kharbanda et al. found that this may be because the palmitoylated EGFR interacts preferentially with a PI3K subunit rather than a MAPK adaptor protein. Blocking palmitoylation reduced PI3K signaling activity, sensitizing cells to PI3K inhibitors. These findings may help clinicians determine which KRAS-mutant patients could benefit from EGFR inhibitors and, with future palmitoyltransferase inhibitors, better restrict these two pathways that commonly and reciprocally drive drug resistance.

Abstract

Non–small cell lung cancer (NSCLC) is often characterized by mutually exclusive mutations in the epidermal growth factor receptor (EGFR) or the guanosine triphosphatase KRAS. We hypothesized that blocking EGFR palmitoylation, previously shown to inhibit EGFR activity, might alter downstream signaling in the KRAS-mutant setting. Here, we found that blocking EGFR palmitoylation, by either knocking down the palmitoyltransferase DHHC20 or expressing a palmitoylation-resistant EGFR mutant, reduced activation of the kinase PI3K, the abundance of the transcription factor MYC, and the proliferation of cells in culture, as well as reduced tumor growth in a mouse model of KRAS-mutant lung adenocarcinoma. Knocking down DHHC20 reduced the growth of existing tumors derived from human KRAS-mutant lung cancer cells and increased the sensitivity of these cells to a PI3K inhibitor. Palmitoylated EGFR interacted with the PI3K regulatory subunit PIK3R1 (p85) and increased the recruitment of the PI3K heterodimer to the plasma membrane. Alternatively, blocking palmitoylation increased the association of EGFR with the MAPK adaptor Grb2 and decreased that with p85. This binary switching between MAPK and PI3K signaling, modulated by EGFR palmitoylation, was only observed in the presence of oncogenic KRAS. These findings suggest a mechanism whereby oncogenic KRAS saturates signaling through unpalmitoylated EGFR, reducing formation of the PI3K signaling complex. Future development of DHHC20 inhibitors to reduce EGFR-PI3K signaling could be beneficial to patients with KRAS-mutant tumors.

INTRODUCTION

Non–small cell lung cancers (NSCLCs) account for 15% of all cancer-related deaths in the United States and are characterized by mutually exclusive, activating mutations in epidermal growth factor receptor (EGFR; 26%) or the guanosine triphosphatase KRAS (37%) (13). EGFR is one of four members of the ErbB family and is known to facilitate tumorigenesis and cancer progression. EGFR is structurally composed of an extracellular ligand binding domain, a transmembrane region, a tyrosine kinase domain, and an unstructured C-terminal tail that harbors receptor autophosphorylation sites (4). Ligand binding induces activation of the tyrosine kinase domain, leading to autophosphorylation of tyrosine residues in the C-terminal domain. The phosphorylated tyrosine residues serve as docking sites for adaptor proteins that link the receptor to the downstream signaling pathways RAS/mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K)/AKT, which promote cell growth and survival. Activating mutations in EGFR increase both MAPK and PI3K signaling and promote oncogenesis. Mutations in KRAS, predominantly an amino acid substitution at codon 12 or 13, lead to up-regulation of MAPK signaling. However, in the mutant KRAS setting, tumor growth requires increased PI3K signaling through a mechanism dependent on a KRAS-PI3K interaction mediated by the RAS-binding domain of PIK3CA (5). Therefore, essential mechanisms are in place to maintain levels of PI3K signaling during tumorigenesis in the mutant KRAS background.

Our laboratory previously reported that EGFR is palmitoylated on the C-terminal tail by the palmitoyltransferase DHHC20 (6). Reduction of DHHC20 increases MAPK signaling by a mechanism that is independent of EGFR kinase activity, suggesting that palmitoylation modulates assembly of the MAPK signaling complex on the C-terminal tail. Furthermore, a palmitoylation-defective EGFR point mutant (EGFRC1025A) activates downstream MAPK signaling with increased Grb2 receptor association, confirming that the mechanism is through the palmitoylated cysteine residues. Although MAPK signaling was increased in cells with reduced DHHC20, we observed a cell proliferation defect in KRAS-mutant cells caused by reduced EGFR palmitoylation (7). Here, we investigated the mechanism by which unpalmitoylated EGFR hinders KRAS-mutant growth.

RESULTS

DHHC20 inhibition reduces tumor burden in KRAS-mutant mice

We previously demonstrated that inhibition of the palmitoyltransferase DHHC20, or mutation of the palmitoylation site Cys1025 on EGFR, induces EGFR activation and increases downstream signaling to MAPK in KRAS-mutant (KRASG12V) SW1573 lung cancer cells (6). We therefore asked whether DHHC20 loss affects KRAS-mediated lung de novo tumorigenesis in vivo. To study KRAS-mutant tumor initiation, we used the genetically engineered KrasLSL-G12D/+;p53flox/flox;Rosa26LSL-YFP (KPY) autochthonous mouse model of lung cancer. In this model, tumors are initiated by endotracheal delivery of viral particles that transduce lung epithelial cells to express Cre recombinase to activate KrasG12D and delete p53 expression. To additionally ablate DHHC20 expression in vivo, we transduced KPY mice with LentiCRISPRv2Cre, a construct that expresses Cre recombinase, Cas9, and a single-guide RNA (sgRNA) targeting DHHC20 (sgDHHC20#1/#2) or an inert sgRNA targeting β-galactosidase (sgInert) (Fig. 1, A and B) (8). The expression of Cre recombinase in the KPY cells will remove the stop cassette flanked by two LoxP sites, turning on the expression of KRASG12D and YFP (yellow fluorescent protein). Cre recombinase will also remove the p53 gene flanked by two LoxP sites. As a result, the cells in the mouse lung infected by the LentiCRISPRv2Cre construct will have the following genotype: KRASG12D;p53 null;YFP positive. Twelve weeks after tumor initiation, animals transduced with LentiCRISPRv2Cre targeting DHHC20 harbored a ~10-fold lower tumor burden compared to sgInert (Fig. 1, C and D). For subsequent analyses, sgDHHC20#1 was solely evaluated, as the resulting knockdown was more efficient than that of sgDHHC20#2. Furthermore, at 24 weeks after transduction with sgDHHC20#1, tumor burden still had not increased further than tumor burden measured at 12 weeks (fig. S1A). To rule out possible differences in infection efficiency between sgInert and sgDHHC20 viruses, we quantified the number of regions in the lungs where YFP expression was activated, marking the sites of Cre-mediated recombination of the Rosa26LSL-YFP locus, and found similar numbers of infected areas in the sgInert and sgDHHC20 mice (Fig. 1, E and F). We observed epithelial cells specifically in the lung airways that were YFP positive and DHHC20 negative (Fig. 1G), indicating that nontumor cells may be viable in the absence of DHHC20. Consistent with the lack of tumor outgrowth, YFP-labeled sgDHHC20#1-infected cells persisted 12 weeks after transduction, but a lower percentage expressed the proliferation-associated antigen Ki67 than did tumors expressing the sgInert vector (Fig. 1, H and I). Inhibition of DHHC20 increases MAPK signaling in human KRAS-driven cancer cells (6). Consistently, we observed a visible increase in phosphorylated extracellular signal–regulated kinase (ERK) in focal areas of lung tissue from sgDHHC20#1 mice compared to those from sgInert mice (Fig. 1J).

Fig. 1 Inhibition of DHHC20 leads to decreased tumor burden in mutant KRAS-driven tumorigenesis.

(A) Diagram of the LentiCRISPRv2Cre vector. (B) Depiction of lentiviral lung tumor induction by intratracheal intubation. (C and D) Stitched images (4×) of hematoxylin and eosin (H&E)–stained lung tissue from sgInert-, sgDHHC20#1-, and sgDHHC20#2-infected lungs from KPY mice (C) and quantification of tumor burden (D) at 12 weeks after LentiCRISPRv2Cre infection. Data are means ± SEM of n = 6, 8, and 6 mice, respectively. ***P < 0.001 and **P < 0.01, Student’s t test. (E) Representative images of immunohistochemical (IHC) staining for YFP (left, at ×20 magnification; scale bar, 75 μm) in lungs from KPY mice described in (C) 12 weeks after LentiCRISPRv2Cre infection (arrows indicate infection sites). (F) Quantification of YFP-positive regions in 4× stitched immunohistochemical images represented in (E). Data are means ± SD of five mice per cohort. (G) Immunofluorescence staining for YFP (red) and DHHC20 (green) (magnification; scale bar, 10 μm). (H and I) Representative images (H) and quantified colocalization (I) of YFP (green) and Ki67 (red) in sgInert- and sgDHHC20#1-infected lung tissue sections. DAPI counterstain, blue. Scale bars, 25 μm. Data are means ± SEM of n = 40 fields per cohort and n = 4 mice from each cohort. ***P < 0.001, Student’s t test. (J) Representative images of immunohistochemical staining for phosphorylated ERK (pERK) in lungs from KPY mice described in (C) 12 weeks after LentiCRISPRv2Cre infection. Right images show ×20 magnification of region indicated by the red box in the corresponding left image. n = 40 fields. Scale bars, 250 μm (left) and 75 μm (right).

The palmitoylation-resistant EGFR point mutant reduces KRAS-driven tumor growth

As previously discussed, expression of the palmitoylation-deficient EGFR cysteine point mutation, EGFRC1025A, also induces EGFR and subsequent MAPK activation. We sought to determine whether specific loss of EGFR palmitoylation blocks mutant KRAS tumorigenesis in vivo, phenocopying the result seen with inhibition of DHHC20. To assess whether EGFRC1025A is sufficient to block mutant KRAS tumorigenesis in vivo, we transduced KPY mice with lentiviral vectors to express Cre recombinase and stably express either mCherry or palmitoylation-defective EGFRC1025A, or oncogenic EGFRL858R as a positive control (Fig. 2A), which has been previously shown to induce synthetic lethality in the presence of mutant KRAS in NSCLC cell lines (9, 10). EGFRC1025A- and EGFRL858R-expressing cohorts stained positively for EGFR compared to the minimal endogenous expression of EGFR in the control cohort (fig. S2, A and B). Expression of EGFRC1025A or EGFRL858R reduced the tumor burden by greater than 10-fold compared to control (0.9%, 3.7% versus 11.7%) (Fig. 2, B and C). Despite the lack of tumor formation in KPY mice with enforced EGFRC1025A or EGFRL858R, a similar frequency of YFP-positive transduction sites was evident in lungs (Fig. 2, D and E). These results demonstrate that it is the unpalmitoylated form of EGFR that is incompatible with oncogenic KrasG12D-driven tumor formation, similar to the synthetic lethality observed with expression of EGFRL858R in the mutant KRAS background (10).

Fig. 2 Specific loss of EGFR palmitoylation blocks KRAS-driven tumor growth.

(A) Diagram of pCREator lentiviral EGFR overexpression construct introduced directly to the lungs of KPY mice via intratracheal intubation. (B) Average tumor burden of mCherry (n = 9 mice), EGFRC1025A (n = 10 mice), and EGFRL858R (n = 9 mice) in KPY mice 12 weeks after infection with Lenti-pCREator. ****P < 0.0001 and ***P < 0.001, Student’s t test. (C) H&E staining at ×4 magnification of lungs from KPY mice expressing mCherry, EGFRC1025A, and EGFRL858R quantified in (B). (D and E) Representative images (D) and quantification (4× stitches) (E) of immunohistochemical staining detecting YFP in mCherry-, EGFRC1025A-, or EGFRL858R-expressing cells in lungs from KPY mice 12 weeks after Lenti-pCREator infection. Scale bars, 25 μm; ×10 magnification. Data are mean number of YFP-positive sites ± SEM of n = 5 mice per cohort.

DHHC20 inhibition reduces PI3K/AKT signaling and Myc expression

Although the increase in MAPK signaling upon DHHC20 inactivation in vivo is consistent with our previous in vitro results, the pro-proliferative function of MAPK signaling is in conflict with the observed inhibition of tumor growth. Similarly, when we examined cell proliferation in KRAS-mutant, EGFR-positive cancer cell lines [H23 (KRASG12V) and MDA-MB-231 (KRASG12D)], we found that cell proliferation decreased significantly in both cell lines when DHHC20 was silenced by short hairpin RNA (shRNA) (Fig. 3, A and B). We therefore examined PI3K-AKT, a parallel branch of the EGFR signaling pathway, and found a marked decrease in AKT phosphorylation at Thr308 (the primary PI3K-mediated activating phosphorylation site) when DHHC20 was inhibited by shRNA (Fig. 3C). The PI3K-AKT pathway inactivates glycogen synthase kinase 3β (GSK3β) by phosphorylating it on Ser9, preventing GSK3β-mediated phosphorylation of Myc and its subsequent proteasomal degradation (11, 12). Consistent with a reduction in the PI3K-AKT pathway, DHHC20 silencing in both MDA-MB-231 and H23 cells decreased GSK3β phosphorylation at Ser9 and severely reduced Myc expression (Fig. 3C). Treating MDA-MB-231 and H23 shDHHC20 cells with MG132, a proteasome inhibitor, partially restored Myc expression, indicating that inhibiting DHHC20 promotes, to an extent, Myc proteasomal degradation (Fig. 3D). Similarly, pharmacologic inhibition of GSK3β with CHIR-90021 also partially restored Myc protein levels (Fig. 3E), confirming that Myc degradation caused by loss of DHHC20 involves GSK3β activity. Treatment with CHIR-90021 decreased glycogen synthase phosphorylation in both control and shDHHC20-expressing MDA-MB-231 cells, confirming its ability to inhibit GSK3β (fig. S3A). Silencing DHHC20 did not decrease Myc mRNA levels (fig. S3), indicating that the decrease in Myc expression is not a result of decreased Myc transcription. Unexpectedly, although Myc protein levels were reduced, Myc mRNA levels were increased when DHHC20 was silenced (fig. S3B). Last, we examined whether restoring Myc expression could rescue the DHHC20-induced growth defect. Mutating the GSK3β phosphorylation site on Myc (MycT58A) inhibits Myc degradation (12). When MycT58A was expressed in H23 shDHHC20 cells, the growth rate was fully restored to that of the shCon cells (Fig. 3, F and G), indicating that the reduction in cell growth caused by DHHC20 inhibition is due to reduced Myc expression.

Fig. 3 Inhibition of DHHC20 reduces Myc expression in KRAS-mutant cells that is restored by inhibition of the proteasome or GSK3β.

(A and B) Immunoblots for DHHC20 (left) and cell proliferation in H23 (A) and MDA-MB-231 cells (B) (right) stably expressing control scrambled shRNA (shCon) or shRNA targeting DHHC20 (shD20). Data are means ± SEM of n = 3 experiments with three replicates each. ***P < 0.001, Student’s t test. (C) Immunoblots for pAKT, pGSKβ, and Myc in H23 and MDA-MB-231 shDHHC20 cells. Quantification of blot density for pAKT/AKT (*P < 0.05 and ***P < 0.001), pGSK3B/GSK3B (*P < 0.05), and Myc (*P < 0.05 and ***P < 0.001). Data are means ± SEM of n = 3 experiments. (D and E) Immunoblot for Myc in MDA-MB-231 and H23 shDHHC20 cells treated with 5 μM MG132 (*P < 0.05) (D) or 3 μM GSK inhibitor (CHIR-99021) (**P < 0.01) (E) for 6 hours. Quantification of blot density for Myc/actin. Data are means ± SEM of n = 3 experiments (P values from Student’s t test). (F) Representative immunoblots for Myc and DHHC20 in H23 shControl (shCon) and shDHHC20 cells (shD20) stably transduced with a MycT58A mutant. (G) Cell proliferation in H23 shCon and shD20 cells, and H23 shCon/MycT58A and H23 shD20/MycT58A cells, expressed as mean cell number ± SEM of n = 3 experiments, each with three replicates. ***P < 0.001, Student’s t test. ns, not significant.

Palmitoylation-resistant EGFR antagonizes oncogenic KRAS signaling and cell growth

The results thus far suggest that PI3K signal activation is inhibited in the presence of unpalmitoylated EGFR. To test this hypothesis, EGFR was partially silenced by shRNA in MDA-MB-231 and H23 cells additionally transfected with control (shCon) or DHHC20 shRNA (shDHHC20) (Fig. 4A). We found that when EGFR expression was reduced, the amount of phosphorylated AKT and Myc increased and cell growth was restored at the 72-hour time point in both MDA-MB-231– and H23 DHHC20–deficient cells (Fig. 4, A to C). Therefore, the loss of Myc and the growth defect in KRAS-mutant cells caused by the DHHC20 loss seem to be dependent on the presence of EGFR (Fig. 4, B and C).

Fig. 4 DHHC20 loss–induced reduction of Myc requires palmitoylated EGFR.

(A) Representative anti-EGFR, anti-pAKT, anti-AKT, anti-Myc, and anti-actin immunoblots of MDA-MB-231 and H23 shCon and shDHHC20 cells stably infected with lentiviruses expressing a control scrambled shRNA (shCon) or an shRNA targeting EGFR (shEGFR). Quantification of blot density for pAKT/AKT and Myc. Data are means ± SEM of n = 3 experiments. *P < 0.05. (B and C) Cell proliferation of MDA-MB-231 (B) and H23 (C) cells expressing shCon (black) and shDHHC20 (red) coexpressing shCon (left) or shEGFR (right). Data are mean cell number ± SD of three experiments. ***P < 0.001, Student’s t test. (D) Representative immunoblots for FLAG, EGFR, pAKT, AKT, pGSK3β, GSK3β, Myc, and actin in MDA-MB-231 cells expressing inducible wild-type (WT) or palmitoylation-defective mutant (C1025A; “C-A”) EGFR or an empty vector (EV) control and treated with doxycycline (DOX) (1 μg/ml). Quantification of blot density for pAKT/AKT, pGSK3B/GSK3B, and Myc. Data are means ± SEM of n = 3 experiments. *P < 0.05 and ***P < 0.001, Student’s t test.

We next asked whether the unpalmitoylated form of EGFR is sufficient to reduce Myc expression. We used a conditional system for expressing wild-type EGFR (EGFRWT) or EGFRC1025A, the palmitoylation-deficient mutant (7). Although treatment of the cells with doxycycline for 72 hours induced equal levels of EGFRWT and EGFRC1025A, Myc protein levels were reduced in the EGFRC1025A-expressing cells compared to those expressing EGFRWT (Fig. 4D). Similarly, EGFRC1025A expression decreased the levels of pAKT(Thr308) and pGSK3β(Ser9) compared to EGFRWT, phenocopying the results observed with DHHC20 knockdown (Fig. 4D).

Thus far, all the cell contexts examined were in an activated mutant KRAS background. We therefore asked whether the reduction in cell growth and Myc expression mediated by EGFRC1025A specifically requires mutant KRAS by measuring Myc levels after inducing EGFRWT or EGFRC1025A in NIH3T3 cells expressing KRASWT or activated KRASG12V. When EGFRC1025A was induced in the presence of KRASG12V, there was a decrease in pAKT(Thr308) and Myc that was not observed with KRASWT (Fig. 5A). Expression of EGFRC1025A in the presence of KRASG12V also significantly decreased cell growth within 72 hours (Fig. 5B and fig. S4A), suggesting that the mechanism for reducing Myc expression and cell proliferation might involve oncogenic KRAS.

Fig. 5 Mutant KRAS reduces PI3K-AKT pathway signaling, Myc expression, and cell proliferation in the absence of EGFR palmitoylation.

(A) Representative immunoblots for FLAG, pAKT, AKT, pGSK3β, GSK3β, Myc, RAS, and actin in NIH3T3 cells expressing wild-type KRAS (KRASWT) or mutant KRAS (KRASG12V) and stably infected with lentivirus to express doxycycline-inducible EGFRWT (WT), EGFRC1025A (C-A), or the EV. Quantification of blot density for pAKT/AKT, pGSK3B/GSK3B, and Myc. Data are means ± SEM of n = 3 experiments. *P < 0.05 and **P < 0.01, Student’s t test. (B) Growth curve of NIH3T3 cells coexpressing KRASWT (black squares) or KRASG12V (red triangles) with doxycycline-inducible EGFRWT (left) or EGFRC1025A (right). Cell number is expressed as the mean ± SEM of n = 3 experiments, each with three replicates. ***P < 0.001, Student’s t test. (C) Representative immunoblots for FLAG, pAKT, AKT, pGSK3β, GSK3β, Myc, and PI3K in NIH3T3 cells expressing wild-type PIK3CA (PIK3CAWT) or PIK3CA harboring the activating mutation E545K (PIK3CAE545K) and stably infected with lentivirus to express doxycycline-inducible EGFRWT, EGFRC1025A, or empty vector; n = 3. (D) Growth curve of NIH3T3 cells coexpressing either PIK3CAWT (black squares) or PIK3CAE545K (red triangles) with doxycycline-inducible EGFRWT (left) and EGFRC1025A (right). Cell number is expressed as the mean ± SEM of n = 3 experiments, each with three replicates. (E) Representative immunoblots for FLAG, pAKT, pGSK3β, Myc, and actin in SW1573 cells harboring both mutant KRASG12C and activated mutant PIK3CAK111E and stably infected with lentivirus to express doxycycline-inducible EGFRWT, EGFRC1025A, or empty vector; n = 3. (F) Growth curve of SW1573 cells expressing doxycycline-inducible EGFRWT (triangles), EGFRC1025A (diamonds), or empty vector (squares). Cell number is expressed as the mean ± SEM of n = 3 experiments, each with three replicates.

These results may indicate that expression of EGFRC1025A in a mutant KRAS background leads to a deficit of PI3K signaling because of lack of recruitment of PIK3R1 to the membrane where it is required to facilitate downstream signaling. The PI3K catalytic subunit p110α is also often mutated in lung cancer but, unlike KRAS mutations, does not appear to be mutually exclusive with EGFR mutations in NSCLC patients (13, 14). We therefore investigated whether expressing a constitutively active mutant of PIK3CA (p110) in EGFRC1025A-expressing cells can bypass the requirement of PIK3R1 at the membrane. Coexpression of EGFRWT or EGFRC1025A with the oncogenic mutant PIK3CAE545K in NIH3T3 cells (15) did not induce a detectable change in levels of pAKT(Thr308) and pGSK3β(Ser9) (Fig. 5C). In contrast to the effect seen in KRAS-mutant cells coexpressing EGFRC1025A, KRAS wild-type cells coexpressing the mutant PIK3CAE545K with EGFRC1025A had markedly increased Myc expression (Fig. 5C). However, there was no difference in cell proliferation over 72 hours between cells expressing EGFRC1025A and those expressing EGFRWT, suggesting that the Myc protein is not limiting for cell growth in the mutant PIK3CAE545K-expressing cells (Fig. 5D and fig. S4B). We next examined the lung adenocarcinoma cell line SW1573, which harbors both activating KRAS and PIK3CA mutations. We found that induction of EGFRC1025A expression in SW1573 was unable to reduce AKT or GSK3β phosphorylation (Fig. 5E) and increased Myc expression (Fig. 5E), but did not significantly affect cell proliferation over 72 hours (Fig. 5F), indicating that PIK3CA-activating mutations are sufficient to restore PI3K signaling and cell growth in mutant KRAS cells expressing unpalmitoylated EGFR.

Palmitoylated EGFR recruits PI3K signaling components to the plasma membrane

We reasoned that, in the presence of oncogenic KRAS, EGFR palmitoylation promotes PI3K complex formation at the membrane, biasing downstream signaling toward PI3K-AKT signaling. The PI3K heterodimer is made up of a regulatory subunit PIK3R1 (p85) and the catalytic subunit PIK3CA (p110α) (14). To detect recruitment of signaling components to the membrane, we isolated the membrane fraction of NIH3T3 cells expressing either EGFRWT or EGFRC1025A together with either KRASWT or KRASG12V. Immunoblotting of the membrane fractions revealed a decrease in PI3K regulatory subunit p85 in EGFRC1025A-expressing cells compared to EGFRWT-expressing cells in the presence of KRASG12V but not KrasWT (Fig. 6A and fig. S4C). We did not see a corresponding increase in p85 in the cytoplasm, which may be caused by higher turnover of p85 when not associated with the membrane. We thus investigated whether the preferential binding EGFRC1025A exhibited for Grb2 over p85 was also observed in the KRAS-mutant cell line and again found a consistent decrease in p85 and an increase in Grb2 association in the plasma membrane fractions from cells expressing EGFRC1025A compared to those expressing EGFRWT (Fig. 6, B and C). We then probed the membrane fraction for the presence of KRAS and found that EGFRC1025A increased the membrane association of KRASWT beyond that of EGFRWT, but we found equally high levels of KRASG12V at the membrane in all three conditions (empty vector, EGFRWT, and EGFRC1025A). This may indicate that, in contrast to constitutively guanosine triphosphate (GTP)–bound mutant KRAS, the cycling of wild-type KRAS between the GTP- and guanosine diphosphate (GDP)–bound forms causes its transient association with EGFRC1025A, allowing both Grb2 and PI3K to associate with EGFRC1025A, but this hypothesis requires further investigation.

Fig. 6 Expression of unpalmitoylated EGFR in the presence of mutant KRAS decreases PIK3R1 (p85) recruitment to the membrane.

(A) Representative immunoblots and analysis (means ± SEM, n = 3) of the abundance of PIK3R1 (p85) in the membrane fraction from NIH3T3 cells stably coexpressing wild-type KRAS (KRASWT) or mutant KRAS (KRASG12V) and doxycycline-inducible EGFRWT (WT), EGFRC1025A (C-A), or EV. Abundance was normalized to that of β-catenin (used as membrane enrichment marker); α-tubulin served as cytosol marker. **P < 0.01, Student’s t test. (B and C) Representative immunoblots (B) and analysis (C) of the endogenous abundance of PIK3R1 (p85), Grb2, and RAS in the membrane fraction from MDA-MB-231 cells expressing EGFRC1025A or EGFRWT. Abundance was normalized as described in (A). Data are means ± SEM of n = 3 experiments. *P < 0.05 and **P < 0.01, Student’s t test. (D) Representative blots of a coimmunoprecipitation (IP) study of PIK3R1 (p85) and EGFR or PI3K p110 in NIH3T3 cells stably coexpressing KRASG12V and doxycycline-inducible EGFRWT (WT), EGFRC1025A (C-A), or EV. (E) Lysates from NIH3T3 cells expressing V5-tagged PIK3R1 were incubated for 15 hours with increasing concentrations of biotinylated peptide encompassing the sequence around either palmitoylated or unpalmitoylated Cys1025 or a palmitoylated scrambled (Scr) control. Peptides were isolated on streptavidin beads, the washed beads were boiled, and proteins were separated by SDS–polyacrylamide gel electrophoresis and both pulldown (PD) and input immunoblotted for PIK3R1 (p85).

Our findings thus far show that palmitoylated EGFR promotes PI3K activation and Myc abundance in the presence of oncogenic KRAS, whereas unpalmitoylated EGFR dampens PI3K signaling because of a reduction of p85 at the membrane, but what remained to be determined is how palmitoylated EGFR recruits the PI3K complex and whether EGFR engages PI3K directly. To explore this, we immunoprecipitated endogenous PIK3CA (p110α) in NIH3T3 cells expressing EGFRWT or EGFRC1025A and probed for EGFR. We found that PIK3CA (p110α) coprecipitated with EGFRWT and the p85 subunit but not with EGFRC1025A, suggesting that EGFR palmitoylation (or this residue elsewise) may be required to promote formation of the PI3K complex in the vicinity of EGFR (Fig. 6D).

To measure interactions between PI3K and the palmitoylated region of EGFR, we synthesized biotinylated peptides encompassing Cys1025, in both palmitoylated and unpalmitoylated forms, and incubated them with lysates from PIK3R1-expressing cells. Although weakly detectable, PIK3R1 associated with the wild-type palmitoylated peptide but not apparently the unpalmitoylated peptide or a palmitoylated peptide with a scrambled sequence (Fig. 6E). These results suggest that sequence-specific palmitoylated motifs can be recognized as docking sites on proteins that may modulate the signal complex formation.

Loss of DHHC20 sensitizes KRAS-mutant cells to PI3K inhibitors

The results thus far indicate that blocking EGFR palmitoylation reduces PI3K signaling; consequently, KRAS-mutant cancer cells are unable to proliferate because of a diversion of EGFR signaling to MAPK and away from PI3K signaling that is needed to stabilize Myc levels. Therefore, we hypothesized that KRAS-mutant cancer cells, wherein DHHC20 is reduced by shRNA, will be sensitive to further reduction of the residual PI3K signaling by treating cells with an EGFR inhibitor or, more so, a PI3K inhibitor. Stable knockdown of DHHC20 in KRAS-mutant SW1573 cells increased sensitivity to the pan-PI3K inhibitor BKM120, also known as buparlisib (Fig. 7A). Furthermore, constitutive knockdown of DHHC20 in SW1573 and H23 cells reduced the IC50s (median inhibitory concentrations) of BKM120 by twofold, further indicating the increased sensitivity of the KRAS-mutant cell lines to inhibition of PI3K signaling (Fig. 7B and fig. S5A, right). Constitutive knockdown of DHHC20 in KRAS-mutant SW1573 and H23 cells also moderately increased sensitivity to inhibition of EGFR upstream of PI3K using the EGFR tyrosine kinase inhibitor gefitinib, as evidenced by the reduced IC50 concentration of gefitinib (Fig. 7, A and B, left, and fig. S5A, left). This result indicates a general sensitivity to inhibition of downstream EGFR signaling in a mutant KRAS setting in the absence of DHHC20. The effect is more pronounced when PI3K is inhibited directly. Relevant to a potential clinical application, simulating the acute nature of drug inhibition, knockdown of DHHC20 using doxycycline to induce expression of shRNA targeting DHHC20 in A549-GFP-Luciferase (A549-GL) and H23-GFP-Luciferase (H23-GL) cells increased sensitivity to BKM120 (Fig. 7C and fig. S5B, left) and reduced the IC50 concentrations of BKM120 by three- and fivefold, respectively (Fig. 7D, right, and fig. S5B, right). Similar to the result with constitutive knockdown of DHHC20, the acute knockdown of DHHC20 in A549-GL and H23-GL cells moderately increased sensitivity to EGFR inhibition by gefitinib (Fig. 7, C and D, left, and fig. S5B). These results corroborate the finding that KRAS-mutant cells become sensitive to the loss of PI3K signaling when EGFR is not palmitoylated. Although the knockdown/inhibitor combination was not tested here in vivo, this facet of the mechanism might one day be explored for translation to the clinic by developing a pharmacologic inhibitor of DHHC20 to use in combination with a PI3K inhibitor.

Fig. 7 Loss of DHHC20 sensitizes KRAS-mutant cells to an EGFR inhibitor and a PI3K inhibitor.

(A) Cytotoxicity in control (CN) or DHHC20-silenced (shD20) SW1573 cells treated with gefitinib (5 μM; gray) or BKM120 (500 nm; black). Data are means ± SEM of n = 3 experiments with triplicate wells for each condition. **P < 0.01, Student’s t test. (B) Dose-response curves and IC50 values (boxed, legend inset) calculated for cells described in (A) treated with the indicated range of doses for 72 hours. Data are from a representative of n = 3 experiments with triplicate wells for each data point. P < 0.001 for each IC50 pair by two-way ANOVA (R2 = 0.98). (C and D) A549-GFP-Luciferase (A549-GL) cells stably infected with lentivirus for doxycycline-inducible expression of shControl or shDHHC20. Treated as described in (A) and (B). (E and F) Bioluminescence (BLI)–based assessment (E) and representative images (F) of tumor volume of doxycycline-inducible shControl A549-GL xenografts (left, black; n = 4 mice, 8 tumors; P < 0.01 on day 8, Student’s t test) and doxycycline-inducible shDHHC20 A549-GL xenografts (right, red; n = 5 mice, 10 tumors) over 8 days normalized to day 0.

Inhibition of DHHC20 slows growth of established KRAS-mutant lung tumors

To determine the potency of targeting DHHC20 in existing tumors, we generated xenografts using the KRAS-mutant cell line A549 expressing GFP and luciferase and doxycycline-inducible shRNA targeting either a control scrambled sequence or DHHC20. Doxycycline treatment was initiated when tumors reached 100 mm3 in size and was administered every day for 10 days. Tumor growth was measured by luciferase imaging. Induction of shRNA targeting DHHC20 abrogated the growth of all xenograft tumors by day 2 of treatment (Fig. 7, E, right, and F), whereas induction of shRNA targeting control scrambled was unable to inhibit tumor growth (Fig. 7F) with the exception of one tumor that experienced slowed growth at day 8 (Fig. 7E, right, light blue) and one tumor that experienced moderate regression at day 8 (Fig. 7E, right, orange). These in vivo data are validated by the analysis of large lung adenocarcinoma patient datasets that reveal a strong correlation between low DHHC20 expression and improved probability of survival (fig. S6) (16). Although a longer study is needed to determine the durability of tumor suppression, these results suggest that developing specific, small-molecule inhibitors to the enzyme DHHC20 might be beneficial to treating KRAS-mutant lung tumors.

DISCUSSION

Together, our findings indicate that in a mutant KRAS background, the PI3K heterodimer associates with palmitoylated EGFR at the plasma membrane to activate PI3K-AKT signaling. As a result, there is an increase in Myc expression that, we assume, activates pro-proliferative transcription programs supporting cancer cell proliferation (Fig. 8A, left). Using a genetically engineered mouse model for KRAS-driven lung cancer, we found that in the absence of this mechanism, tumorigenesis is markedly impeded.

Fig. 8 Mechanistic model.

Proposed model wherein EGFR palmitoylation by DHHC20 (left) promotes PI3K-AKT signaling, leading to stable Myc production and cell proliferation. Loss of EGFR palmitoylation (right) promotes binding of Grb2-SOS, leading to hyperactivation of KRAS-MAPK signaling, but impedes PI3K/AKT signaling, thereby causing Myc depletion and reduced cell proliferation.

Our previous studies examining EGF stimulation of shDHHC20 cells showed that without EGF stimulation, basal ERK phosphorylation is increased, whereas the increase in AKT phosphorylation is dependent on ligand stimulation. EGF stimulation also increases EGFR palmitoylation at Cys1025, but phosphorylation at Tyr1068, the main Grb2 binding site that mediates MAPK signaling, is reduced considerably when Cys1025 is palmitoylated (6), which suggested that EGF stimulation–mediated EGFR palmitoylation at Cys1025 antagonizes MAPK signaling. Our follow-up study presented here examines the consequence of losing EGFR palmitoylation in either a wild-type or oncogenic KRAS setting. In a wild-type KRAS setting, we observed an increase in KRAS abundance at the membrane upon expression of EGFRC1025A (a palmitoylation-deficient mutant), but we propose that the rapid cycling of KRAS activity allows association of the PI3K complex at the membrane. However, in the KRAS-mutant setting, KRAS is locked in the active GTP-bound state; therefore, when EGFR is not palmitoylated, there is constitutive binding of Grb2 and hyperactivation of the downstream KRAS-MAPK pathway. Under these conditions, PI3K is unable to interact with EGFR to form a functional signaling complex, and the PI3K-AKT signaling cascade is impeded. As a result, GSK3β is active and promotes rapid degradation of Myc, leading to a loss of pro-proliferation signals and attenuation of cancer cell proliferation (Fig. 8A, right).

To undergo oncogenic transformation, cancer cells are dependent on signaling mechanisms that, in some cases, may be dispensable in normal cells (5). For example, mutations in the Ras-binding domain of PIK3CA have no discernible effects on mouse development or cell homeostasis, but this mutation reduces oncogenic KRAS-driven tumor formation and maintenance through a loss of PI3K signaling (5). This demonstrates that oncogenic KRAS requires interaction with PIK3CA for downstream signaling to MAPK and PI3K to initiate tumorigenesis. We found a similar requirement for EGFR palmitoylation to maintain PI3K signaling during KRAS-driven tumorigenesis, because expressing EGFRC1025A had a weaker effect on reducing Myc abundance in cells expressing wild-type KRAS compared to those expressing mutant Kras. In addition, our previous study showed that although knockdown of DHHC20 reduces cell proliferation of the breast cancer cell line MDA-MB-231, knockdown of DHHC20 has no effect on the growth of a transformed but nonmalignant breast epithelial cell line MCF10A (7).

The growth defect in the shDHHC20 cells was restored by inhibiting EGFR expression, indicating that unpalmitoylated EGFR is itself inhibitory in the presence of oncogenic KRAS. In general, the mechanism of PI3K activation by EGFR is unclear because EGFR lacks the canonical PI3K binding motif (pYXXM) present in other receptor tyrosine kinases, and it has been proposed that another adaptor (like Gab1) mediates PI3K signaling by EGFR (17). We propose an alternative mechanism whereby cells expressing EGFR may be dependent on DHHC20-mediated palmitoylation to sustain PI3K signaling in the presence of oncogenic KRAS.

Our findings show that reducing DHHC20 levels impairs tumor formation in a KRAS-mutant genetically engineered mouse model and arrests the growth of existing human, xenografted KRAS-mutant tumors in mice. One would predict that low DHHC20 levels in KRAS-mutant NSCLC tumors would predict improved prognosis of survival of patients. An analysis of large lung adenocarcinoma patient datasets revealed a strong correlation between low DHHC20 expression and improved probability of survival, further showing the potential clinical benefit of inhibiting DHHC20 in lung cancer patients (fig. S6) (16). The potential clinical impact of these findings is strengthened by the improved efficacy of the pan-PI3K inhibitor buparlisib that we observed upon knockdown of DHHC20. Buparlisib monotherapy has modest efficacy in the clinic so far; thereby, the focus of clinical trials rests in combination therapy. These findings reveal that the enzyme DHHC20 may be a susceptible drug target for use in combination with clinically available PI3K inhibitors to treat KRAS-driven adenocarcinoma.

MATERIALS AND METHODS

Cell culture

MDA-MB-231 and NIH3T3 cells [American Type Culture Collection (ATCC)] were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS). H23 and SW1573 cells (ATCC) were maintained in RPMI supplemented with 10% FBS. GSK3β inhibitor (CHIR-99021), gefitinib, and BKM120 were purchased at Selleck Chemicals. 2-Bromopalmitate (catalog no. 238422-10G) was purchased from Sigma-Aldrich. Cell lines were tested for mycoplasma contamination using MycoAlert Mycoplasma Detection Kit (Lonza). All cell lines were free of contaminants.

Silencing of human DHHC20

The oligonucleotides for shControl and shDHHC20 constructs were synthesized (Integrated DNA Technologies) and inserted into the pLKO.1 vector. shControl encodes the nontargeting sequence of SHC002 (Sigma-Aldrich); the shRNA target sequence of human DHHC20 is 50-GAGCTCTGCGTGTTTACTATT-30. MDA-MB-231 and H23 cells were transduced with lentivirus encoding shControl or shDHHC20 and selected by puromycin treatment (1 mg/ml) for several passages.

Plasmids and generation of stable cell lines

Human mutant MycT58A from pLV-tetO with a hemagglutinin (HA) tag (Addgene, plasmid no. 19763) (18) was gateway-cloned into the pLX304 backbone with a blasticidin resistance marker and V5 tag (Addgene). Lentivirus of pLX305-MycT58A was generated using human embryonic kidney (HEK) 293T cells with Gag, VSVG, and Rev plasmids using TransIT-LT1 (Mirus) according to the manufacturer’s instructions. H23 cells that were transduced with lentivirus encoding shControl or shDHHC20 and selected by puromycin treatment were subsequently infected with MycT58A-V5. H23 shControl or shDHHC20 and Lenti-pLX304-MycT58A were selected with puromycin (1 μg/ml) and blasticidin (10 μg/ml) together for several passages. To generate inducible cell lines, wild-type EGFR and EGFRC1025A complementary DNA (cDNA) were first subcloned into the inducible pTRIPZ backbone with a puromycin resistance marker and FLAG tag. Empty pTRIPZ, which expresses the rtTA3, puromycin resistance marker, and FLAG tag, was used as a negative control. Virus production was performed by transfecting HEK293T cells with the pTRIPZ constructs, psPAX2, and pMD2.G plasmids (Addgene) using TransIT-LT1 (Mirus) according to the manufacturer’s instructions. MDA-MB-231 and NIH3T3 were infected with pTRIPZ virus using polybrene and incubated for 24 hours. After infection, fresh medium was added on infected cells and incubated for an additional 48 hours before selection. Cells infected with the pTRIPZ constructs were selected with puromycin (1 μg/ml) for several passages. Expression of EGFR cDNA was induced with doxycycline (1 μg/ml) overnight before proceeding with experiments. Lentivirus of human KRAS4B(WT) or mutant KRAS4B(G12V) in pLenti-PGK-hygromycin resistance with a HA tag (Addgene plasmid no. 35633) (18) was generated using HEK293T cells with Gag, VSVG, and Rev plasmids using TransIT-LT1 (Mirus) according to the manufacturer’s instructions. NIH3T3 cells infected with pTRIPZ constructs and selected with puromycin were subsequently infected with KRAS4B(WT)-HA or KRAS4B(G12V)-HA. NIH3T3 pTRIPZ-plenti-KRAS4B (WT) or (G12V) cells were selected with puromycin (1 μg/ml) and hygromycin (500 μg/ml) together for several passages. Lentivirus of human mutant PI3KCA-E545K in pcw107-PGK-puromycin resistance (Addgene plasmid no. 64605) (19) was generated using HEK293T cells with Gag, VSVG, and Rev plasmids using TransIT-LT1 (Mirus) according to the manufacturer’s instructions. NIH3T3 cells infected with pTRIPZ constructs and selected with puromycin were subsequently infected with PI3KCA-E545K (19).

Immunoblot analysis

Cell lysates were prepared in 1% Triton X-100 buffer, including tris-HCl (pH 7.5) and sodium chloride (NaCl) solution. Lysates were analyzed by immunoblotting with the following antibodies: Antibody to DHHC20 (HPA014702) was purchased from Sigma-Aldrich. Antibodies to EGFR-XP (catalog no. 4267S), phosphorylated ERK (pERK, catalog no. 4370S), ERK (catalog no. 4695S), pS473-AKT (catalog no. 4060S), pT308-AKT (catalog no. 13038S), AKT (catalog no. 4691S), pS9-GSK3β (catalog no. 5558S), GSK3β (catalog no. 12456S), PI3K (p110α) (catalog no. 4255S), PIK3R1 (p85) (catalog no. 4292S), Ras (catalog no. 3339S), β-catenin (catalog no. 8480S), Myc (catalog no. 2272S), and β-actin (catalog no. 4970S) were obtained from Cell Signaling Technology (CST). Additional antibodies to Myc and α-tubulin were purchased from Santa Cruz Biotechnology. Primary antibodies were diluted 1:1000 in 5% bovine serum albumin dissolved in tris-buffered saline and 0.01% Tween (TBST) and incubated overnight at 4°C. Immune complexes were detected with horseradish peroxidase–conjugated secondary antibodies diluted in 5% nonfat milk dissolved in TBST and enhanced chemiluminescence (Thermo Scientific).

Membrane fractionation

Cell lysates were prepared in hypotonic lysis buffer including tris-HCl (pH 8), magnesium chloride (MgCl2), potassium chloride (KCl), and dithiothreitol (DTT). Lysates were disrupted using passage through 25-gauge needle. Lysates were subject to centrifugation at 800g for 10 min at 4°C to pellet nuclei. The resulting supernatant was then subject to centrifugation with a Beckman ultracentrifuge at 50,000 rpm for 1 hour at 4°C using a Beckman type 70.1 Ti rotor. The resulting supernatant was kept as the cytosolic fraction and pellet contained the membrane fraction. Membrane fraction was resuspended in 1% NP-40 lysis buffer including tris-HCl (pH 8), NaCl, SDS, and glycerol. Sample loading buffer was added to the membrane samples, and the samples were boiled for 8 min followed by Western analysis and immunoblotting.

Cell viability assays

Cells were treated with gefitinib (5 μM) or BKM120 (500 nM) for 72 hours, and viability was measured by trypan blue staining. Quantification was performed using a one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post-test.

Drug-treated cell proliferation assays

Cells were plated in a solid white 96-well plate at 5000 cells per well in 90 μl of 10% FBS RPMI medium. Serial dilutions of gefitinib and BKM120 were made in medium, and 10 μl of the diluted compounds was transferred to the cells. After 72 hours, cell viability was measured using CellTiter-Glo (Promega) according to the manufacturer’s instructions. Luminescent readout was normalized to dimethyl sulfoxide (DMSO)–treated control cells and empty wells. Data were analyzed by nonlinear regression curve fitting on Prism 8, and IC50 values were reported.

Myc quantitative polymerase chain reaction

Total RNA was isolated using the RNeasy Extraction Kit (Qiagen). To quantify Myc expression levels, equal amounts of cDNA were synthesized using the SuperScript III First-Strand Synthesis System (Invitrogen) and mixed with the Power SYBR Green PCR master mix (Applied Biosystems, Carlsbad, CA) and 5 pmol of both forward and reverse primers. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was amplified as an internal control. The sequences of the human primers used for quantitative polymerase chain reaction (qPCR), listed from 5′ to 3′, were as follows: MYC, CCTACCCTCTCAACGACAGC (forward) and CTCTGACCTTTTGCCAGGAG (reverse); ACTB (β-actin), AATCTGGCACCACACCTTCTAC (forward) and ATAGCACAGCCTGGATAGCAAC (reverse).

Peptide streptavidin pull-down

Cell lysates were made as described above. Lysates were incubated with various concentrations of biotinylated scrambled-palmitoylated, unpalmitoylated C1025 containing EGFR C-terminal tail, and palmitoylated C1025 containing EGFR C-terminal tail peptides dissolved in DMSO overnight at 4°C. Palmitoylated peptides were synthesized with diaminopimelic acid (DAP) substituted for cysteine, which was conjugated to palmitate by an ether linkage, making the modification resistant to thioesterase cleavage (Biomatik, USA). The streptavidin agarose beads (Thermo Scientific) were washed three times with aforementioned lysis buffer. Lysates with peptides were incubated with 20 μl of the prewashed streptavidin agarose beads for 2 hours at 4°C with rotation. The beads-lysate-peptide mix was spun down at 6000g for 1 min. The beads were washed three times with cold lysis buffer. Loading sample buffer containing β-mercaptoethanol was added to beads, and the beads were then boiled for 10 min at 100°C. The boiled sample was centrifuged at 16,000g for 1 min, and the supernatant was collected for Western blotting.

Vector design and production

LentiCRISPRv2Cre is described by Walter et al. (8) (Addgene plasmid no. 82415; http://n2t.net/addgene:82415; RRID:Addgene_82415). DHHC20 sgRNAs were designed to target exons in the first one-third of the gene using the CRISPR Design Tool (http://crispr.mit.edu/about) to minimize off-target effects in the mouse genome. DHHC20 sgRNAs were cloned into LentiCRISPRv2Cre vector by Golden Gate assembly using Bsm BI (New England BioLabs, R0580S). The sgRNAs used for targeting Cas9 are as follows: DHHC20 1, 5′-CACCGAGTACGTGGAACTTTGCGCTGTTT-3′; DHHC20 2, 5′-CACCGGCGCTGCTGCCAACGCGTGGGTTT-3′. The sensor assay reporter was generated by synthesizing sgRNA targets in series and cloning them upstream of mCherry in the pCHK-mCherry vector using Gibson assembly as discussed by Walter et al. (8). pCREatorBsmBI was constructed by synthesis of a gene block encoding a Kpn I cloning site, the eukaryotic elongation factor short promoter followed by a 2xBsmBI golden gate cloning site, P2a peptide sequence, and CreNLS. The fragment was obtained from Genescript and cloned into pUC57mini vector. The Kpn I–Cla I fragment of the gene block was subsequently subcloned into the vector backbone portion of a Kpn I–Cla I digested pLentiCRISPRv2Cre to create pCREatorBsmBI. mCherry and EGFR mutants were PCR-amplified with primers containing Bsm BI tails and appropriate restriction sequences for Golden Gate cloning such that the 5′ (left) overhang is 5′-CACC-3′ and the 3′ (right) overhang is 5′-ATCC-3′ after Bsm BI digest.

Lentivirus production

HEK293FT cells were transfected with LentiCRISPRv2Cre or pCREator EGFR WT, C1025A, L858R, and Δ8.2, and VSV-G plasmids in a 4:3:1 ratio using polyethylenimine. Twenty-four hours after transfection, the media were replaced with fresh DMEM supplemented with 25 mM Hepes (Gibco, 15630-080) and 3 mM caffeine (Sigma-Aldrich, C0750). Lentivirus-containing supernatant was collected from the cells at 48 and 72 hours following transfection, filtered through 0.45-μm filters (Thermo Scientific, 723-2545), and centrifuged at 107,000g. The viral pellet was soaked in 100 μl of phosphate-buffered saline (PBS) for 16 hours at 4°C, triturated, vortexed for 15 min at 4°C, and finally centrifuged at 16,000g for 30 s to remove insoluble debris. Lentivirus was then aliquoted and frozen at −80°C for later use. Lentivirus was titered on Green-Go cells, an NIH3T3 derivative harboring an integrated Cre-dependent GFP reporter. These cells were validated for reporter activity by flow cytometry during viral titering. A total of 2 × 105 cells were plated in six-well plates, and 24 hours later, lentivirus was added at 10, 1, and 0.1 μl per well. Cells were analyzed by flow cytometry for GFP expression 48 hours after infection, and viral titer was calculated accordingly.

Animal work

KPY mice were maintained on a mixed C567B6/129Sv4 background and were treated as previously described (20). Mice were given lentivirus at 6 × 104 pfu per mouse by intratracheal intubation at 6 to 10 weeks of age as described previously (21). Mouse lungs were harvested at 12 weeks.

Histologic analysis

Tumor number was counted on H&E-stained slides. Tumor area was quantified using ImageJ software. Tumor burden percentage was calculated as tumor area over total lung area multiplied by 100.

Immunohistochemistry and immunofluorescence

Mouse tissues were fixed in 10% neutral buffered formalin (Fisher Scientific, 23-245-685) for 16 hours and dehydrated in a series of ethanol washes up to 100%. Samples were paraffin-embedded and sectioned at 4-μm thickness. For immunohistochemistry, slides were deparaffinized in xylene and rehydrated with a series of ethanol washes. Antigen retrieval was performed using citrate buffer (Electron Microscopy Sciences, 62706-10), and slides were stained using antibodies against GFP that cross-react with YFP (1:200; Abcam), pERK (1:500; CST), and EGFR-XP (1:200; CST). Primary antibody was incubated on slides for 16 hours at 4°C, and biotinylated secondary antibody (Vector Laboratories) was incubated at room temperature for 1 hour. ABC reagent and ImmPACT DAB were prepared as directed (Vector Laboratories, PK-4001 and SK-4105). Slides were analyzed on a Leica DMI6000B inverted microscope. For immunofluorescence, slides were again deparaffinized in xylene and rehydrated with a series of ethanol washes. Antigen retrieval was performed using citrate buffer (Electron Microscopy Sciences, 62706-10), and slides were stained using antibodies against GFP (1:200; Abcam), Ki67 (1:200; Abcam), and EGFR-XP (1:200; CST) and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) using Fluoro-Gel II with DAPI (EMS, catalog no. 17985-50). Detection was by Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary antibodies (Molecular Probes). Slides were analyzed on a Leica DMI6000 B inverted microscope.

NSCLC xenografts

SCID (severe combined immunodeficient) beige mice (CB17.Cg-PrkdcscidLystbg-J/Crl), aged 4 to 6 weeks, purchased from Charles River, were injected subcutaneously with 5 × 106 cells in 1:1 solution with Matrigel in both flanks. Mice with established A549-GFP-Luciferase tumors (100 mm3) were treated intraperitoneally each day with doxycycline (20 μg in 0.5-cm3 water) to induce expression of shRNA targeting either a control scramble sequence or DHHC20. Tumors were measured on days −2, 0 (day of doxycycline addition), 5, and 8 using in vivo bioluminescent imaging. To image, mice were anesthetized and intraperitoneally injected with d-luciferin (GoldBio, LUCNA-1G) in PBS at 150 mg/kg. Luminescent signals were acquired 15 min after injection with the IVIS Spectrum (Caliper Life Sciences). Analysis was performed using Living Image 4.5 (PerkinElmer).

Statistical analysis

Statistical analyses were performed using Prism software version 7.0 (GraphPad). Experiments are reported as means ± SEM as noted in the legends. Data were analyzed using a two-tailed Student’s t test for comparison between two datasets. Multiple comparisons were analyzed by two-way ANOVA, followed by Tukey’s multiple-comparison correction. A P value of less than 0.05 was considered statistically significant. Every immunoblot experiment was performed three times. Drug response curves and cell viability experiments were performed three times with three replicates each for each dose.

Study approval

All experiments involving live animals were performed in compliance with the guidelines set forth in the Public Health Service Policy on the Humane Care and Use of Laboratory Animals. Mice were housed in a pathogen-free facility at the American Association for Laboratory Animal Science–accredited Animal Facility at the University of Pennsylvania Perelman School of Medicine. All studies were performed under protocols approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania Perelman School of Medicine (#804774).

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/13/621/eaax2364/DC1

Fig. S1. Tumor burden remains low with extended loss of DHHC20.

Fig. S2. pCREater (EGFR)-Cre expresses EGFR mutants equivalently in KPY mice.

Fig. S3. Myc mRNA levels increase upon inhibition of DHHC20.

Fig. S4. Presence of mutant KRAS is required to reduce cell growth from loss of EGFR palmitoylation.

Fig. S5. Inhibition of DHHC20 in H23 cells induces sensitivity to PI3K inhibitor.

Fig. S6. Low levels of DHHC20 correlate with better survival probability in patients.

REFERENCES AND NOTES

Acknowledgments: We would like to thank D. Brady (University of Pennsylvania) for useful discussion and advice regarding the experiments and manuscript. We would also like to thank K. Wood (Duke University) for the PIK3CA (E545K) lentiviral plasmid. Funding: This work was supported by NIH grant R01CA181633 (to E.S.W.) and by ACS grant RSG-15-027-01 (to E.S.W.). This project is funded, in part, under a grant with the Pennsylvania Department of Health. The Department specifically disclaims responsibility for any analyses, interpretations, or conclusions. Author contributions: E.S.W. and A.K. conceived and coordinated the study; designed, performed, and analyzed the experiments; and wrote the paper. A.K., D.M.W., A.A.G., and N.S. performed experiments. D.M.F. designed and developed lentiviral constructs and provided mice and animal facilities. All authors reviewed the results and approved the final version of the manuscript. Competing interests: The authors declare that they have no conflicts of interest. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article