Increased EGFRvIII Epitope Accessibility after Tyrosine Kinase Inhibitor Treatment of Glioblastoma Cells Creates More Opportunities for Immunotherapy

Abstract

The number of glioblastoma (GB) cases is increasing every year, and the currently available therapies remain ineffective. A prospective antigen for GB therapy is EGFRvIII, an EGFR deletion mutant containing a unique epitope that is recognized by the L8A4 antibody used in CAR-T (chimeric antigen receptor T cell) therapy. In this study, we observed that the concomitant use of L8A4 with particular tyrosine kinase inhibitors (TKIs) does not impede the interaction between L8A4 and EGFRvIII; moreover, in this case, the stabilization of formed dimers results in increased epitope display. Unlike in wild-type EGFR, a free cysteine at position 16 (C16) is exposed in the extracellular structure of EGFRvIII monomers, leading to covalent dimer formation in the region of L8A4-EGFRvIII mutual interaction. Following in silico analysis of cysteines possibly involved in covalent homodimerization, we prepared constructs containing cysteine-serine substitutions of EGFRvIII in adjacent regions. We found that the extracellular part of EGFRvIII possesses plasticity in the formation of disulfide bridges within EGFRvIII monomers and dimers due to the engagement of cysteines other than C16. Our results suggest that the EGFRvIII-specific L8A4 antibody recognizes both EGFRvIII monomers and covalent dimers, regardless of the cysteine bridging structure. To summarize, immunotherapy based on the L8A4 antibody, including CAR-T combined with TKIs, can potentially increase the chances of success in anti-GB therapy.

Keywords: CAR-T; EGFRvIII; dimerization; glioblastoma; immunotherapy.

Figure 1

Characterization of research model according to different EGFRvIII expression profiles. (A) The cell lines DK-MG^parental, DK-MG^low, MDA-MB-468, and primary GB1 and GB2 culture were selected for testing L8A4 specificity in recognition and binding to the EGFRvIII epitope. For this purpose, cells were immunocytochemically stained with mouse L8A4 antibody (red) (A). There was no positive signal in the case of MDA-MB-468 cells, in which there is no EGFRvIII expression. About 5–10% and 50% of DK-MG^low and DK-MG^parental cell populations, respectively, interact with the L8A4 antibody, which corresponds with their profile of EGFRvIII expression. Specific binding of the antibody was also observed in primary GB cultures, with a previously confirmed expression of EGFRvIII. Taken together, these results demonstrate that the L8A4 antibody specifically recognizes EGFRvIII. Observations were performed under 40× magnification, except for GB2-20x, for the following exposition times: DAPI, 10 ms; L8A4, 600 ms. (B,C) Confirmation of EGFR (wt and vIII) expression levels in extralow, low, parental, and extrahigh DK-MG sublines performed at both mRNA (B) and protein (C) levels. Results for real-time PCR of (D) GB1 and GB2 and (E) MDA-MB-468 compared with AD293^vIII. (F,G) The influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, and 10 µM lapatinib) on total EGFR amount was investigated by immunocytochemical staining of DK-MG^extrahigh cell line. In both analyzed cases, erlotinib and afatinib increased, whereas lapatinib decreased the accumulation of EGFR protein. The fluorescence density ratio was measured in DK-MG^extrahigh using ImageJ software (E). (H,I) The influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, and 10 µM lapatinib) on total EGFR amount was also investigated by immunocytochemical staining of AD293^vIII cell line. Signals for the A10 (total EGFR) and the L8A4 (EGFRvIII-specific) antibodies were also compared. ICC on AD293^vIII shows that erlotinib and afatinib increased whereas lapatinib decreased the accumulation of EGFRvIII protein (H). The fluorescence density ratio was measured using ImageJ software (I). Statistical significance was calculated using a two-tailed paired t-test from three independent experiments, ns > 0.05, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001. Observations were performed under 40× magnification for the following exposition times: DAPI, 10 ms; L8A4, 600 ms; and A10, 300 ms.

Figure 2

Characteristics of cell lines with different EGFR expression profiles. (A,B) An influence of particular TKIs—10 µM erlotinib and 0.5 µM afatinib as well as EGF 20 ng/mL on total EGFR amount was investigated by immunocytochemical staining on MDA-MB-468 cell line. In this case erlotinib and afatinib no significant effect on the EGFRwt accumulation was observed, whereas EGF increased it (A). The fluorescence density ratio was measured using ImageJ software (B). (C,D) The influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, and 10 µM lapatinib) on total EGFR amount was also investigated by immunocytochemical staining of DK-MG^parental cell line (C). Similarly to DK-MG^extrahigh, the EGFR signal in DK-MG^parental was increased by erlotinib and afatinib, whereas lapatinib decreased the accumulation of EGFR protein. The fluorescence density ratio was measured using ImageJ software (D). (E,F) The influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, as well as EGF 20 ng/mL) on total EGFR amount was also investigated by immunocytochemical staining of the H1975 cell line. ICC on H1975 shows that erlotinib and afatinib increased, whereas lapatinib decreased the accumulation of EGFR protein (G). The fluorescence density ratio was measured using ImageJ software (H). (G,H) Influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, as well as EGF 20 ng/mL) on total EGFR amount was also investigated by immunocytochemical staining of the AD293^wt cell line. ICC on AD293wt shows that erlotinib and afatinib increased whereas lapatinib decreased the accumulation of EGFRwt protein (G). The fluorescence density ratio was measured using ImageJ software (H). Statistical significance was calculated using a two-tailed paired t-test from three independent experiments: ns > 0.05, * p ≤ 0.05 and ** p ≤ 0.01. Observations were performed under 40× magnification for the following exposition times: DAPI, 10 ms; L8A4, 600 ms; A10, 300 ms; and sc03, 300 ms.

Figure 3

Mechanism of EGFR dimerization. (A) Semi-native Western blot demonstrating that EGFRvIII dimerization occurs via a homodimerization process that is not dependent on interaction between EGFRvIII and EGFRwt monomers. The addition of 10 µM erlotinib for 1 h further increased the dimerization of EGFRvIII. (B,C) Semi-native Western blots were performed using tEGFR antibody with prior treatment of SOV and TKIs: erlotinib (10 µM), afatinib (0.5 µM), gefitinib (10 µM), and lapatinib (10 µM) (B). Afatinib and gefitinib demonstrate similar activity to erlotinib, whereas lapatinib significantly decreases the stability of EGFRvIII homodimers. Using ImageJ, bands were quantified via densitometry and normalized to actin (C). (D,E) Semi-native Western blot with the parallel use of A10 (D) and L8A4 (E) antibodies in lines with different EGFRvIII expression profiles (endogenous DK-MG^extrahigh, exogenous DK-MG^vIII-exovIII, and AD293^vIII) or with no expression of EGFRvIII-AD293 parental was performed to verify the potential differences in dimerization and antibody binding. The results indicate the same level of dimerization independent of the expression profile. Differences in EGFRvIII expression also do not affect the binding of antibodies. The addition of 10 µM erlotinib increased dimer formation. (F) Substitution of cysteine with serine in position 16 of EGFRvIII protein expressed in AD293 cell line significantly decreased the number of formed dimers and indicated that this cysteine is involved in disulfide bridge formation. In both cases, EGFRvIII and C16S, treatment of cells with TKIs such as 10 µM erlotinib or 0.5 µM afatinib for 1 h enhanced and stabilized dimerization. However, the addition of 1 µM SOV for 1 h alone or in combination with the abovementioned TKIs had no effect on this phenomenon, indicating that the dimerization process is not dependent on the phosphorylation status of EGFRvIII. (G,H) Statistical analysis of monomer (G) and dimer (H) formation in the case of AD293^vIII and AD293^c16s cells lines incubated by 1 h with 10 µM erlotinib, 0.5 µM afatinib, and 1 µM SOV (alone or in combination with the abovementioned TKIs) demonstrates that C16S substitutions impede dimer formation and favor the formation of monomers. (I) Schematic representation of factors that favor the formation of monomers (wild-type EGFR, C16S substitution, and lapatinib treatment) or dimers (EGFRvIII, erlotinib, and afatinib). Densitometric analysis (C,H,I) was obtained from three independent experiments, calculated by ImageJ, and data were analyzed using a two-tailed paired t-test, * p ≤ 0.05.

Figure 4

In silico analysis of dimer formation and the discovery of other cysteines were involved in this process in in vitro model development. (A) Monomer and dimer schematics. (B) All predicted disulfide bonds at the time of cysteine–serine substitution. (C–E) Predicted structures of interacting monomers. (F) According to the in silico models, the different cysteine–serine mutants, as well as EGFRvIII constructs, were prepared by site-directed mutagenesis of EGFRvIII cDNA and subsequent transformation into the AD293 cell line. Confirmation of expression levels of EGFRvIII and specific EGFRvIII cysteine–serine (C16S, C20S, C35S, C38S, and C42S) in established AD293 cell lines at the mRNA level.

Figure 5

Validation of in silico models performed by in vitro testing of the different EGFRvIII cysteine–serine mutants. (A) Semi-native Western blot analysis of dimer formation ability in the case of AD293 cell line expressing EGFRvIII and EGFRvIII cysteine–serine mutants, C16S, C20S, C35S, C38S, and C42S in the presence of erlotinib, SOV, and a combination of both. (B,C) The influence of particular TKIs (10 µM erlotinib, 0.5 µM afatinib, and 10 µM lapatinib) on total EGFR amount was also investigated by immunocytochemical staining of AD293^C16S (B) and AD293^C35S (C) cell lines. In both cases, erlotinib and afatinib increased, whereas lapatinib decreased the accumulation of EGFRvIII protein. (D,E) Semi-native Western blot on AD293 parental, vIII, C16S, C20S, C35S, C38S and C42S cell lines was performed to elucidate whether different disulfide bridge arrangements may impede L8A4 antibody recognition of the EGFRvIII epitope (D). In addition to L8A4, the A10 antibody was tested (E). AD293^vIII cells were additionally treated with 10 µM erlotinib to verify the analysis. The results demonstrate that variables in interactions between EGFRvIII monomers do not impact on L8A4 mode of action. Moreover, there were no statistically significant differences in dimer creation when binding with either A10 or L8A4 antibodies (F). WB from (D,E) was calculated by ImageJ from three independent experiments, and data were analyzed using a two-tailed paired t-test, ns p > 0.05. (G) Schematic of introduced mutations. MUT: any of the following substitutions of cysteine by serine in positions 16 (C16S), 20 (C20S), 35 (C35S), 38 (C38S), or 42 (C42S). TM—transmembrane; TK—tyrosine kinase; RD—regulatory/phosphorylation domain.