Steric disruption of EGFR oligomerization overcomes therapy resistance in non-small cell lung cancer.

INTRODUCTION
Lung cancer ranks as the foremost cause of cancer-related deaths in the United States, with non–small cell lung cancer (NSCLC) constituting 85% of cases (1). Within NSCLC, prevalent oncogenic driver mutations involve epidermal growth factor receptor (EGFR), which plays a pivotal role in fundamental cellular processes (1–4). Upon ligand binding, EGFR becomes activated and undergoes subsequent phosphorylation, precisely regulating cell proliferation, differentiation, survival, and migration in normal cellular contexts (5). However, mutations in EGFR confer sustained activation, amplification, or overexpression, resulting in uncontrolled cellular proliferation, a hallmark of cancer (6). Activating mutations within EGFR exons 18 to 21, governing its catalytic domain and phosphorylation regulation, occur in 10 to 15% of individuals of European descent and up to 50% of Asian NSCLC patients (2–4). Over the past 18 years, extensive efforts have focused on designing EGFR mutant–selective tyrosine kinase inhibitors (TKIs), which competitively bind to the adenosine triphosphate (ATP)–binding site of EGFR with higher affinity than ATP, thereby blocking EGFR activity (7). Although TKIs are initially effective in NSCLC, resistance due to mutations in the drug-binding pocket of EGFR arises after about 1 year of treatment (8–10). This has motivated the continued development of additional drugs to overcome acquired TKI resistance, including allosteric inhibitors that target non-ATP sites on EGFR (11–14).
The effect of TKI-resistant mutations on EGFR oligomerization and the possible role of oligomerization in the mechanism of TKI resistance are unclear. EGFR oligomerization, a critical step in its autophosphorylation and the activation of downstream effectors, profoundly affects EGFR inhibitor efficacy as well (11, 14). EGFR dimerization can obstruct target site binding of both ATP-competitive TKIs and allosteric EGFR inhibitors, diminishing their therapeutic effectiveness (11, 14). Remarkably, cotreatment with cetuximab, which disrupts EGFR dimerization, restores TKI efficacy for some resistance mutants (11). Despite the importance of EGFR oligomerization, its analysis in TKI evaluation remains scarce due to the methodological complexity involved (15–17). Therefore, elucidating the oligomeric states of resistant EGFR mutants and their interactions with TKIs is imperative. This underscores the urgent need for advanced screening techniques that accurately reflect EGFR oligomerization states, enhancing drug efficacy evaluations.
Here, we developed a fluorescent protein–based single-probe system, which enables simultaneous measurement of EGFR oligomerization and intracellular functional signaling in real time. Oligomer levels and activated signaling were quantitatively monitored on oncogenic EGFR mutants, including those conferring resistance to existing TKIs for NSCLC, and with diverse TKIs and allosteric inhibitors. Analysis data consistently indicated a direct correlation between the oligomeric level and EGFR signaling activity as well as subsequent cell proliferation. EGFR mutants conferring TKI resistance formed substantially higher levels of oligomers than did wild-type (WT) EGFR in the absence and presence of EGF stimulation. Notably, EGFR inhibition by TKIs on its active site also suppressed receptor oligomerization. Additionally, we artificially induce or disrupt EGFR oligomerization to investigate the effects of receptor oligomers on functional signaling, regardless of natural ligand binding, mutations, and active-site inhibitions. Physical EGFR clustering induced receptor signaling in a clustering degree–dependent manner. Steric disruption of oligomerization, by fusing a bulky protein, completely abrogated downstream signaling in all tested EGFR mutants including the most drug-resistant triple mutants. Protease-assisted removal of the bulky protein reinstated the hyperactive oncogenic phenotype of these mutants in real time, presenting disrupted EGFR-mutant oligomerization as a previously unexplored avenue for drug design strategies. Moreover, an externally administered bulky EGFR-binding protein sterically disrupted oligomerization and suppressed downstream signaling even in cetuximab-resistant mutants. These results highlight the therapeutic potential of extracellular steric hindrance in NSCLC and support revisiting EGFR monomerization, a previously underexplored mechanism, as a strategy to overcome TKI resistance.

RESULTS

Real-time analysis of EGFR-mutant oligomerization unveils enhanced signaling in NSCLC
Activating EGFR mutations in NSCLC drive tumor growth by hyperactivating the Ras-Raf-MEK-ERK signaling pathway, which promotes cell proliferation and survival. These genetic alterations primarily occur within the tyrosine kinase domain, specifically in exons 19 and 21 of EGFR (18). Notably, mutations like L858R and exon 19 deletions (Ex19Del) were initially identified in NSCLC patients, with subsequent mutations such as T790M and C797S emerging in patients treated with ATP-competitive TKIs, leading to therapeutic resistance against these inhibitors (4, 11, 18).
Given the constitutive activity of oncogenic EGFR mutants, we aimed to compare their oligomerization levels and signaling activity with WT EGFR in living cells, both at baseline and after EGF stimulation. To achieve this, we designed all-in-one probes that allow for simultaneous visualization of EGFR oligomerization and its downstream signaling within a single construct (Fig. 1A). A green fluorescent protein (GFP)–based homomolecular fluorescence complementation probe (HomoFC) was fused to the C terminus of EGFR to visualize oligomerization. HomoFC generates GFP signals only when fused receptors form oligomers (19). For quantitative analysis, the HomoFC-to-mCherry (HomoFC/mCherry) ratio was used to normalize expression levels. Additionally, ERK-kinase translocation reporter (KTR) (20) was introduced to monitor the real-time effect of EGFR-mutant oligomerization on downstream signaling by tracking ERK activity, using cyan fluorescent protein–labeled ERK-KTR (ERK-KTR-mTurq2) as a kinase activity reporter. Upon phosphorylation by active ERK, ERK-KTR translocates from the nucleus to the cytosol, while decreased ERK activity results in nuclear localization of ERK-KTR. We measured the cytosol-to-nucleus (C/N) ERK-KTR ratio using mTurq2 fluorescence intensity.
WT EGFR–expressing human embryonic kidney (HEK) 293A cells exhibited low oligomerization signals (~0.08 HomoFC/mCherry ratio) at baseline, with most ERK-KTR-mTurq2 localized in the nucleus (C/N ratio of 0.31) (Fig. 1B). Upon EGF stimulation, however, EGFR oligomerization was induced, with the HomoFC green signal gradually increasing to 0.95 over 60 min. Additionally, as ERK activity increased due to EGF-induced Ras-Raf-MEK-ERK signaling, ERK-KTR translocated from the nucleus to the cytosol, resulting in a C/N ratio of 2.1 within 30 min. This multiplexing system, using HomoFC and ERK-KTR, provided real-time insights into EGFR oligomerization and downstream signaling (Fig. 1C, top and middle panels). While accumulating HomoFC signals steadily increased, kinase signaling rapidly peaked at 15 to 20 min, as previously reported (21). EGFR-expressing HEK293A cells responded to EGF stimulation with increased proliferation over the subsequent 48 hours, suggesting that they are suitable for modeling the relationship between EGFR activity and proliferative outcomes (Fig. 1C, bottom panel). Thus, the combination of HomoFC and ERK-KTR assays enables live-cell assessment of how well EGFR oligomerization corresponds to downstream kinase signaling that drives cell proliferation.
Next, using the same dual-indicator system, we investigated the oligomerization levels of oncogenic EGFR mutants identified in NSCLC (Fig. 1D). L858R or Ex19Del causes basal EGFR autophosphorylation and drives cell proliferation. Ex19Del and the TKI-resistant double-mutant L858R/T790M were previously reported to promote proliferation even when dimerization was disrupted by the additional mutation of L704N or I941R (22). Using HomoFC, we found that L858R and Ex19Del exhibited higher basal oligomer levels (0.31 to 0.36 HomoFC/mCherry ratios) compared to WT EGFR (0.08 HomoFC/mCherry ratio) (Fig. 1, E and F, and fig. S1A), consistent with previous studies (23). Double mutants (L858R/T790M) and triple mutants (L858R/T790M/C797S, Ex19Del/T790M/C797S) maintained high basal oligomerization (~0.38 to 0.46 HomoFC/mCherry ratio). In the absence of L858R or Ex19Del, the TKI-resistant mutations T790M or T790M/C797S, whose effects alone on dimerization have not previously been investigated, also displayed higher basal oligomer levels than did WT EGFR (fig. S1A). These results indicate that both EGFR-activating and TKI-resistant mutations strongly promote EGFR oligomerization, and suggest that analyzing oligomerization could help understand other NSCLC-related and TKI-resistant EGFR mutants.
To verify that the observed differences in baseline EGFR oligomerization were not due to variation in expression levels, we first confirmed that HomoFC fluorescence (GFP) scaled linearly with EGFR expression (mCherry) (fig. S1D). HomoFC signals were then normalized as HomoFC/mCherry ratios to control for expression effects (fig. S1E). EGFR mutants consistently exhibited higher baseline HomoFC/mCherry ratios than did WT EGFR across the entire expression range, and all quantitative analyses were performed on cells with mCherry intensities of 10,000 to 30,000 arbitrary units (a.u.) to avoid low-signal noise and GFP saturation.
Oncogenic EGFR mutants also produced up to 5.5-fold higher basal ERK activity compared to WT EGFR, as assessed by the C/N ERK-KTR ratio (Fig. 1F, gray bars in the middle panel). The L858R/T790M mutant, which exhibited the highest baseline oligomerization, showed the highest C/N ERK-KTR ratio (2.72), again highlighting a strong correlation between EGFR-mutant oligomerization and Ras/Raf/MEK/ERK signaling. Other EGFR-mutant combinations observed in NSCLC patients also exhibited a strong correlation between oligomerization levels and ERK activity. Consistent with oligomerization patterns, EGFR mutants with only T790M or C797S mutations—absent L858R or Ex19Del—demonstrated elevated basal ERK activity (fig. S1B). HomoFC reports oligomerization through temporal signal integration because split-GFP complementation is irreversible, reflecting the cumulative signal of receptor interactions over time rather than capturing real-time binding events. This property enables highly sensitive detection of weak or transient interactions, particularly under baseline conditions. The cumulative HomoFC readout is effectively complemented by the real-time dynamics of the ERK-KTR reporter, together providing a comprehensive view of EGFR activation states. Consistent with this dual-readout approach, quantitative analysis revealed a strong correlation between EGFR oligomerization and basal ERK activity (Pearson r = 0.95, P < 0.0001), underscoring the physiological relevance of receptor clustering in NSCLC-associated EGFR mutants (fig. S1C).
We then analyzed changes in oligomerization levels of each EGFR mutant 30 min after EGF stimulation (Fig. 1F, green bars in top panel, and fig. S2). All HomoFC-fused EGFRs showed increased oligomer signals within 30 min, with EGFR mutants displaying 1.37- to 1.64-fold higher oligomerization levels compared to WT EGFR. EGFR mutants also showed a 1.48- to 1.83-fold higher C/N ERK-KTR ratio (Fig. 1F, cyan bars in middle panel) and a 2.53-fold increase in cell proliferation compared to WT EGFR (Fig. 1F, bottom panel). It is noteworthy that signaling activities enhanced by EGFR mutations were further increased by EGF treatment, resulting in significantly accelerated cell proliferation. It is also interesting that the maximum oligomerization levels for EGFR mutants were higher than those for WT EGFR at comparable receptor densities. This could be due to the ability of mutant EGFRs to evade regulatory mechanisms, such as internalization and degradation by proteins like Cbl (24), allowing them to remain on the cell membrane for a longer period, which may lead to increased oligomerization levels. In summary, investigating EGFR oligomerization offers valuable insights into predicting downstream signaling in NSCLC-related EGFR mutants.

Mutant EGFR oligomerization predicts drug resistance
These findings raised the possibility of whether EGFR-mutant oligomerization in either basal or EGF-stimulated conditions might predict the efficacy of TKIs against oncogenic EGFR mutants. To test this possibility, we administered first-generation TKIs (gefitinib and erlotinib) (25, 26), a second-generation TKI (afatinib) (8, 27, 28), and a third-generation TKI (osimertinib) (29, 30) to four types of EGFR mutants and compared oligomerization, ERK activity, and cell proliferation after EGF stimulation (Fig. 2, A and B). First-generation TKIs gefitinib and erlotinib effectively suppressed oligomerization of the single-mutant EGFR L858R, maintaining a low HomoFC/mCherry ratio of approximately 0.3 even with EGF stimulation (Fig. 2C). In contrast, in the absence of TKIs, EGFR L858R exhibited increased oligomerization, reaching a HomoFC/mCherry of 1.20. This suggests that oligomer levels can serve as indicators of TKI efficacy on EGFR mutants. Notably, suppression of oligomerization by gefitinib or erlotinib was again closely associated with a limited increase in ERK activity, as indicated by a low C/N ERK-KTR ratio (~1.5) (Fig. 2D). Gefitinib or erlotinib also suppressed cell proliferation driven by EGFR L858R (Fig. 2E).
To further investigate the relationship between TKI efficacy and oligomerization suppression, we next treated double-mutated EGFR mutants with TKIs. The T790M mutation confers resistance to gefitinib and erlotinib due to enhanced ATP-binding affinity with EGFR (11, 31). Gefitinib or erlotinib was ineffective in inhibiting EGF-induced oligomerization in double-mutant EGFR L858R/T790M, which exhibited a high HomoFC/mCherry ratio (~1.5). In contrast, osimertinib or afatinib, each designed to overcome resistance to first-generation TKIs, successfully reduced EGF-induced oligomerization of EGFR L858R/T790M, resulting in a HomoFC/mCherry ratio of 0.2 to 0.5, repressed ERK activity, and reduced cell proliferation (Fig. 2, D and E). While afatinib more potently reduced cell proliferation in L858R/T790M mutants than osimertinib (Fig. 2E, third graph), its suppression of EGFR oligomerization and ERK signaling was significantly weaker (Fig. 2, C and D). This disparity suggests that afatinib’s antiproliferative effect may arise in part from nonspecific cytotoxicity, likely due to its broader ErbB-family inhibition and covalent binding to Cys797. These findings are consistent with previous reports describing afatinib’s preclinical potency against T790M-mutant EGFR and its limited clinical efficacy due to dose-limiting toxicity (32–34). Both osimertinib and afatinib failed to inhibit EGF-induced oligomerization (HomoFC/mCherry ratio 1.0 to 1.3) in triple-mutant EGFR L858R/T790M/C797S or EGFR Ex19Del/T790M/C797S, consistent with the C797S mutation arising in resistant tumors after osimertinib treatment (35). High ERK activity with a C/N ERK-KTR ratio of 3.2 to 3.6, similar to non–drug-treated cells, were observed, indicating resistance to all tested ATP-competitive TKIs. Cell proliferation assays confirmed significant proliferation under EGF conditions, irrespective of ATP-competitive TKI treatment in triple-mutant EGFR cells (Fig. 2E). These results indicate that none of the conventional competitive TKIs tested inhibited the activity of C797S-containing EGFR mutants, consistent with existing literature (36, 37), and this correlated with high EGFR oligomerization. A comprehensive assessment of oligomer levels provides additional insight into the resistance of EGFR mutants to specific TKIs. This underscores the potential of using oligomerization propensity information in early-stage, high-throughput TKI screening to evaluate the efficacy of TKI candidates on each EGFR mutant. To evaluate the suitability of our assay system for TKI screening, we adapted it from a cell imaging–based setup to a more high-throughput microplate reading system, also enabling collective signal acquisition from a large number of cells. Results in a 96-well microplate reader were consistent with fluorescent measurements, demonstrating that oligomerization can be robustly assessed in wells of only ~10,000 cells (fig. S3).

Allosteric inhibitors restrain oligomerization of EGFR mutants including C797S
To address drug resistance in ATP-competitive TKIs against triple EGFR mutants (L858R/T790M/C797S) and (Ex19 deletion/T790M/C797S), researchers have developed allosteric inhibitors such as EAI045 (11), JBJ-04-125-02 (12), and JBJ-09-063 (14). These inhibitors bind to sites distinct from the ATP-binding domain, thus avoiding the mutations that typically confer resistance to competitive TKIs. This makes them a promising therapeutic strategy for targeting kinases with resistant driver mutations, including the C797S variant in NSCLC patients. However, the impact of allosteric inhibitors on EGFR oligomerization remains underexplored, as most efforts have concentrated on overcoming resistance to existing TKIs.
We examined three allosteric EGFR inhibitors—EAI045, JBJ-04-125-02, and JBJ-09-063—on cells expressing highly TKI-resistant triple EGFR mutants (Fig. 3A). Thirty minutes after EGF stimulation, EAI045-treated cells showed modestly reduced oligomerization with a HomoFC/mCherry ratio of 0.9 to 1.1 compared to osimertinib-treated cells (1.0 to 1.3 HomoFC/mCherry ratio) and nontreated cells (1.2 to 1.3 HomoFC/mCherry ratio) (Fig. 3B). EAI045 is known to require coadministration of another drug to be effective against EGFR mutants (11, 38). Consistent with the oligomerization results, EAI045 treatment did not significantly affect ERK activation, with a C/N ERK-KTR ratio of 2.6 to 3.0, although it was still more effective than osimertinib (3.4 to 3.6 C/N ERK-KTR ratio) (Fig. 3C). In contrast, JBJ-04-125-02 and JBJ-09-063, more recently developed allosteric TKIs, significantly inhibited oligomerization after EGF stimulation, with HomoFC/mCherry ratios of 0.3 to 0.6. JBJ-09-063 demonstrated the highest efficacy in disrupting EGF-induced oligomerization with a HomoFC/mCherry ratio of 0.3 to 0.4, effectively reducing ERK phosphorylation to a C/N ERK-KTR ratio of 0.6 to 0.8 and cell proliferation to approximately 46% (Fig. 3, C and D).
Previous studies revealed that combining allosteric inhibitors with osimertinib enhanced their efficacy (11, 14, 39). Cotreatment with EAI045 and osimertinib significantly inhibited EGF-induced oligomerization, with HomoFC/mCherry ratios dropping from 0.87 and 1.06 to 0.45 and 0.64, respectively (Fig. 3B). In contrast, oligomerization signals were less affected by osimertinib cotreatment with JBJ-04-125-02 and JBJ-09-063, likely because EGF-induced oligomerization was already sufficiently suppressed by these drugs. However, JBJ-09-063 and osimertinib resulted in the lowest ERK activation (~0.3 C/N ERK-KTR ratio) and effectively reduced EGF-induced cell proliferation by approximately 34% in L858R/T790M/C797S EGFR mutants and 25% in Ex19Del/T790M/C797S mutants (Fig. 3D). These findings suggest that allosteric drugs could provide synergistic treatment approaches that act through both oligomerization-dependent and oligomerization-independent mechanisms. Consistent results were also observed in a well-plate format, supporting the potential for high-throughput screening (fig. S4).

Disruption of EGFR oligomerization with a bulky attachment overcomes therapeutic resistance
We next examined the effects of artificially induced EGFR oligomerization on downstream signaling in the absence of EGF, exploring the potential to control EGFR oligomerization without disrupting its inherent signaling dynamics. Using the high-affinity biotin-avidin interaction [dissociation constant (Kd) ≈ 10 fM] (40), we generated physically tethered EGFR oligomers by three types of multimeric avidin with varying valencies to control the degree of oligomerization: rhizavidin (dimeric) (41), streptavidin (tetrameric) (40), and enhanced monoavidin (eMA)–fused ferritin (24-meric) (42) (fig. S5A). Treatment of biotinylated AP-mCh-EGFR-HomoFC with multimeric avidin induced oligomerization, increasing the HomoFC/mCherry ratio by up to 5-fold compared to controls without avidin treatment (fig. S5B) with increased proliferation by up to 3.2-fold, as avidin valency increase (fig. S5C). EGFR oligomerization by multimeric avidins was only moderately reduced when cetuximab or gefitinib was pretreated (fig. S5, D to F), suggesting that this approach enables activity-independent oligomerization control. Cetuximab and even the ATP-competitive TKI gefitinib suppressed avidin-induced EGFR oligomerization to 67% of the avidin-only condition (fig. S5, D to F), highlighting that these clinically used drugs can modulate receptor oligomerization even under enforced, ligand-independent conditions. Conversely, enforced EGFR oligomerization—despite the presence of these inhibitors—was sufficient to promote downstream signaling and cell proliferation. These findings underscore the critical role of the extracellular domain in mediating EGFR oligomerization and suggest that therapeutic manipulation of this region could overcome limitations of current inhibitors. This rationale led us to explore whether direct disruption of EGFR oligomerization through extracellular steric hindrance could suppress aberrant signaling in drug-resistant mutants.
On the basis of these findings, we examined whether artificially inhibiting EGFR-mutant oligomerization could deactivate downstream signaling or enhance the TKI’s efficacy against drug-resistant EGFR mutants. To sterically hinder EGFR mutants’ oligomerization at the outer membrane, we introduced a bulky protein, BICD2N (1 to 594 amino acids), which is the N-terminal portion of human bicaudal D2, a cytoplasmic α-helical coiled-coil dynein adapter protein (43–45), at the extracellular N terminus of EGFR (Fig. 4A). Previously, light-induced protein oligomerization was effectively inhibited by BICD2N fusion (43). While BICD2N’s primary contribution may stem from its extended size, its rigid and elongated α-helical coiled-coil architecture likely provides a spatially defined steric barrier that blocks receptor-receptor interactions at the membrane interface. BICD2N-fused WT EGFR–expressing cells exhibit only 15% of the basal oligomeric level compared to BICD2N-free EGFR. Moreover, BICD2N-fused EGFR mutants demonstrate significantly reduced oligomeric levels ranging from 5.2% to 10.7% of those observed in their nonfused counterparts (Fig. 4B and fig. S6A). Even highly active triple EGFR (Ex19Del/T790M/C797S) mutants exhibited an 18.8-fold reduced oligomer signal by BICD2N fusion. Moreover, this BICD2N-fused triple mutants showed a 3.3-fold lower oligomer signal than WT EGFR at the basal level.
While all constructs included N-terminal fused mCherry for expression normalization, we sought to rigorously confirm that the observed suppression was not an artifact of the fluorescent protein fusion. To this end, we generated BICD2N-EGFR-HomoFC constructs lacking mCherry, which exhibited similarly low oligomerization and ERK activity (fig. S7, A and B) compared to mCherry-fused BICD2N–EGFR–HomoFC constructs (Fig. 4, B and C), indicating that the suppressive effect of BICD2N fusion is independent of mCherry. Moreover, mCherry-free EGFR-HomoFC constructs (without BICD2N) displayed comparable cellular localization pattern, oligomerization, and ERK activity (fig. S7, A and B) to their mCherry-tagged counterparts in (Fig. 1F), confirming that N-terminal fluorescent protein fusion does not interfere with EGFR function. Together, these results validate that EGFR oligomerization can be potently disrupted by bulky protein–mediated steric hindrance.
Despite EGF stimulation on BICD2N-fused triple EGFR (Ex19Del/T790M/C797S) mutants, oligomerization did not significantly increase, with a HomoFC/mCherry ratio of 0.12 (Fig. 4C). Furthermore, consistent with the low HomoFC/mCherry ratio, ERK-KTR primarily localized to the nucleus with a C/N ERK-KTR ratio of ~0.64 even after EGF stimulation, indicating that disrupting oligomerization inactivated EGF-induced EGFR-Raf-Ras-MEK-ERK downstream signaling. Time-lapse images showed that EGF stimulation activated ERK in both WT EGFR and EGFR mutant within approximately 20 min, while BICD2N-fused EGFR mutants did not exhibit a noticeable increase with the ~0.7 C/N ERK-KTR ratio (Fig. 4D). Several allosteric and ATP-competitive inhibitors were tested on this BICD2N-fused EGFR triple mutant (fig. S6B). Many drugs slightly reduced EGFR oligomerization and cell proliferation. Notably, osimertinib, which had previously shown no effect on this EGFR triple mutant (Fig. 2, C and D), meaningfully reduced the signals (fig. S6C). Further investigation is needed to extensively explore whether disrupting oligomerization could enhance the efficacy of existing drugs and to elucidate the underlying mechanisms driving this effect. Nonetheless, these findings suggest that targeting receptor oligomerization at the outer membrane could be a previously unexplored strategy for developing potent NSCLC therapies to overcome drug resistance.

Removal of a bulky attachment restores activity of oncogenic EGFR mutants
To further investigate the role of bulky proteins in disrupting EGFR oligomerization, we introduced a tobacco etch virus (TEV) cleavage site (ENLYFQ/G, TEVcs) between mCherry-labeled BICD2N and EGFR-HomoFC. This TEV protease–cleavable construct (mCh-BICD2N-TEVcs-EGFR-HomoFC) would enable the controlled removal of BICD2N by TEV protease in live cells (Fig. 5A). Upon TEV treatment, a gradual increase in oligomerization signals was observed in cells expressing mCh–BICD2N–TEVcs–EGFR (L858R) mutants–HomoFC, beginning around 40 min (Fig. 5B). This suggests that the removal of steric hindrance by TEV protease allowed EGFR mutants to restore baseline oligomerization activity. As oligomerization increased, ERK-KTR rapidly translocated from the nucleus to the cytosol, with the C/N ERK-KTR ratio rising from 0.4 to approximately 1.3 over 120 min, indicating gradual ERK activation. No significant changes in oligomerization or ERK activation were observed in cells when lacking the TEV site between BICD2N and EGFR (fig. S8). These data strongly suggest that the observed inhibition of EGFR oligomerization and signaling was due to steric hindrance by the fused BICD2N, rather than irreversible mislocalization or misfolding of the BICD2N-fused EGFR in cells.
When oligomerization of the triple EGFR mutant (L858R/T790M/C797S) was restored by TEV-mediated BICD2N removal, both HomoFC/mCherry and C/N ERK-KTR ratios increased more rapidly and strongly than in the single EGFR L858R mutant (Fig. 5C), indicating the high basal activity of this TKI-resistant triple mutant. These patterns of oligomerization and signaling increases were similar to those observed in WT EGFR following EGF stimulation (Fig. 1C).
To further verify that the rescued EGFR mutants retained their ability to respond to upstream stimuli, we stimulated TEV protease–treated cells with EGF at 120 min post-treatment (fig. S9, A and B). EGF stimulation robustly increased both EGFR oligomerization and ERK activity in WT and L858R constructs, reaching levels comparable to those observed in mCherry–EGFR (WT or L858R)–HomoFC constructs in Fig. 1F. These findings confirm that EGFR mutants not only regain baseline activity after BICD2N removal but also retain intact responsiveness to ligand stimulation, ruling out irreversible disruption of receptor function. In addition, to confirm that the observed suppression was not attributable to fluorescent protein fusion or fusion orientation, we repeated the TEV cleavage experiment using mCherry-free BICD2N–TEVcs–EGFR (L858R) constructs, which exhibited similar restoration of oligomerization and ERK activity 120 min after TEV protease treatment (fig. S10).
The TEV treatment assay was also applied in a 96-well plate format, where microplate analysis confirmed restored oligomerization of BICD2N-fused EGFR mutants upon TEV treatment, with no effects on EGFR-mutant constructs without BICD2N or TEV site (Fig. 5, D and E). Additionally, the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] cell proliferation assay showed that BICD2N-fused EGFR mutants had reduced proliferation, which was restored following BICD2N removal by TEV protease. Notably, the oligomerization and cell proliferation levels of BICD2N-fused EGFR mutants recovered to the same levels as BICD2N-free EGFR mutants, highlighting the strong reversibility of bulky protein–based steric disruption of EGFR oligomerization.

External treatment of a bulky fused DARPin improves inhibition of mutant EGFR
To further assess the therapeutic potential of steric disruption, we evaluated an extracellularly administered designed ankyrin repeat protein (DARPin) targeting EGFR. We used E01, an orthosteric DARPin that binds to the EGF-binding site to inhibit cell proliferation, and E69, an allosteric DARPin with limited efficacy (46). To explore whether steric hindrance could enhance inhibition, we fused these DARPins with the bulky BICD2N protein, generating YFP–DARPin (E01)–BICD2N and YFP–DARPin (E69)–BICD2N constructs. The effects of these constructs on EGFR-mutant oligomerization and ERK activation were compared to those of YFP-BICD2N and YFP-DARPin alone (Fig. 6, A and B).
Following a 4-hour post-transfection period, cells were treated with YFP-DARPin-BICD2N, YFP-BICD2N, or YFP-DARPin and allowed to express EGFR mutants for 24 hours. YFP-BICD2N alone, in the absence of DARPin, had no effect on EGFR oligomerization, ERK activation, or cell proliferation in any tested mutants, including L858R/T790M, L858R/T790M/C797S, and Ex19Del/T790M/C797S, with levels comparable to untreated controls (Fig. 6, C and D). Externally delivered DARPin (E01) alone moderately inhibited the oligomerization of L858R/T790M and L858R/T790M/C797S mutants (HomoFC/mCherry ratio 0.21 to 0.28) and ERK activation (C/N ratio 0.44 to 1.22) but lost its inhibitory effect on Ex19Del/T790M/C797S mutants (HomoFC/mCherry ratio: 0.35, C/N ratio: 1.77). In contrast, YFP–DARPin (E01)–BICD2N strongly suppressed EGFR oligomerization (HomoFC/mCherry ratio 0.11 to 0.12) and ERK activation (C/N ratio 0.27 to 0.38), even in highly resistant triple mutants harboring C797S, surpassing the inhibitory effect of YFP–DARPin (E01) alone. We also treated YFP–DARPin (E01)–BICD2N to cells expressing mCherry-free EGFR (Ex19Del/T790M/C797S)–HomoFC. Consistent with results from mCherry-fused constructs (Fig. 6, B to D), strong suppression of both oligomerization and ERK activity was observed (fig. S11), further supporting that the inhibition is independent of the presence of N-terminal mCherry fusion. Notably, only YFP–DARPin (E01)–BICD2N effectively suppressed oligomerization and ERK activation in Ex19Del/T790M/C797S mutants, resulting in a 28% reduction in cell proliferation (Fig. 6E). These findings suggest that steric hindrance directly disrupts EGFR oligomerization, impairing downstream signaling and inhibiting cancer cell proliferation.
The allosteric binder YFP–DARPin (E69) alone showed minimal inhibition of oligomerization in the L858R/T790M, L858R/T790M/C797S, or Ex19Del/T790M/C797S mutants, with HomoFC/mCherry ratios of 0.35 to 0.37 (Fig. 6, C and D). These also showed minimal inhibition of downstream activity, with C/N ERK-KTR ratios of 1.75 to 1.87. In contrast, YFP–DARPin (E69)–BICD2N exhibited enhanced inhibition compared to YFP–DARPin (E69) alone, particularly against the L858R/T790M mutant, reducing oligomerization by 50% (HomoFC/mCherry ratio 0.18). Furthermore, the C/N ERK-KTR ratio decreased from 1.87 with E69 alone to 0.32 with YFP–DARPin (E69)–BICD2N. Nevertheless, its efficacy remained lower than that of the E01 constructs (Fig. 6, C to E). This disparity likely arises from E69’s allosteric binding mechanism, which partially modulates receptor activation, in contrast to the full inhibition observed with E01 binding at the EGF-competitive site. Our findings demonstrate that steric hindrance enhances the inhibition of EGFR mutants by both orthosteric and allosteric DARPins, including the highly drug-resistant C797S variant.
The AlphaFold3-predicted structure of the YFP–DARPin (E01)–BICD2N fusion protein reveals that BICD2N forms an elongated coiled-coil structure (fig. S12). This spatial arrangement maximizes steric bulk when DARPin (E01) binds to the EGFR extracellular domain, likely interfering with lateral EGFR-EGFR interactions at the plasma membrane. Functionally inert in the extracellular context due to the absence of its known partner dynein (47), BICD2N acts as a purely steric blocker. Additionally, consistent with the proposed steric inhibition mechanism, oligomerization measurements in a 96-well format using a plate reader showed inhibition patterns consistent with confocal imaging, suggesting that this approach could be adapted for high-throughput screening of EGFR-targeting therapeutics (fig. S13). Furthermore, this study provides the first confirmation that DARPin (E01) effectively inhibits triple EGFR mutants when fused with a bulky domain, highlighting steric disruption as a promising therapeutic strategy.

DISCUSSION
Here, we investigated the increased propensity for oligomerization observed in diverse EGFR mutants associated with NSCLC, establishing it as a prevalent phenomenon. Oncogenic EGFR mutants, particularly L858R, exhibit a fourfold increase in oligomerization compared to WT EGFR, with further enhancement upon EGF stimulation. Double mutants (L858R/T790M) and triple mutants (L858R/T790M/C797S), which are linked to acquired resistance to ATP-competitive TKIs, showed up to a sixfold increase in oligomerization at baseline compared to WT EGFR. These elevated oligomerization levels were associated with increased ERK activity and enhanced cellular proliferation. Previous structural and biochemical studies have demonstrated that the cancer-associated mutations L858R and Ex19Del promote EGFR activation by stabilizing the active conformation of the receiver kinase within the canonical asymmetric dimer (48, 49). This mechanism likely underlies the enhanced EGF-independent oligomerization observed in the nine mutants tested, which contain either one of these mutations. Our study also highlights that suppression of EGFR-mutant oligomerization could be a consistent indicator of drug efficacy.
While some oncogenic EGFR mutants are known to exhibit elevated basal dimerization (23), the link we observed between TKI resistance and oligomerization was unexpected. This finding is surprising, as one might assume that resistance mutations primarily evolve to hinder drug binding at the active site, rather than to affect oligomerization. Additionally, it is intriguing that all active-site inhibitors also suppressed EGFR oligomerization. Although it is well established that EGFR oligomerization is required for activation, the extent to which kinase inhibition suppresses oligomerization was not previously appreciated. Similarly, artificial clustering of EGFR was shown to activate both signaling and oligomerization, while physically preventing oligomerization abolished downstream signaling. A previous report that EGFR mutated at the asymmetric dimerization interface could still promote cell proliferation (22) has been interpreted to imply their ability to signal in the monomeric state (11). However, the original observations actually showed these mutations to reduce proliferation, and it has not been ruled out that these mutations may have effects on EGFR signaling other than affecting dimerization. These findings underscore the need for further extensive studies to fully elucidate the intricate relationship between EGFR oligomerization and its activity. One limitation of HomoFC is that it cannot distinguish between dimer and higher-order oligomers. EGFR is known to assemble into higher-order complexes mediated by extracellular domain IV (5, 50), and it would be interesting to determine if any mutations shift the equilibrium between oligomeric species. This question would be better addressed by single-molecule imaging or fluorescence cross-correlation spectroscopy (FCCS) using direct fluorescent labels. Nevertheless, our findings suggest that analyzing EGFR-mutant oligomerization could be a valuable approach for predicting drug efficacy and the emergence of drug resistance.
TKIs targeting oncogenic EGFR mutants are often initially effective in NSCLC, but the development of drug resistance remains a substantial clinical challenge. Our study demonstrates that analysis of EGFR-mutant oligomerization can serve as a robust predictor of drug resistance and efficacy in the early stages of drug development. The high-throughput, functional assay we developed provides a distinct advantage by directly assessing receptor oligomerization, thereby reducing the risk of false positives often encountered in downstream signaling assays, such as ERK-KTR or cell proliferation measurements. Furthermore, we show that TEV protease–mediated removal of the bulky protein nearly fully restored the oncogenic properties of EGFR mutants. This TEV-mediated release of steric inhibitors offers a powerful tool for selectively reactivating hyperactive EGFR mutants in live cells, facilitating the evaluation of related drug candidates. Additionally, the system is adaptable for the design of screening strategies targeting oligomer-specific allosteric drugs. For example, it can be used to assess drug efficacy in artificially induced EGFR oligomerization, allowing for more focused screening of oligomer-specific compounds. Moreover, it can be integrated with computational methods for structure-based drug design, particularly targeting EGFR dimers and higher-order oligomers. This makes the platform well-suited for the development of more effective EGFR inhibitors for NSCLC and other EGFR-driven cancers.
By introducing a bulky protein (BICD2N) to the extracellular domain of EGFR, we effectively inhibited oligomerization and subsequent signaling, even in highly drug-resistant triple mutants. Previous structural studies show that EGF binding to extracellular domains I and III positions domain II to mediate receptor homodimerization, which in turn drives asymmetric dimer formation between the intracellular kinase domains (51). However, the extent to which the extracellular region contributes to dimerization in oncogenic EGFR carrying cytosolic mutations remains unclear. We found that fusing the bulky protein BICD2N to domain I suppressed mutant receptor clustering and downstream ERK activation in both the presence and absence of EGF, suggesting that extracellular homodimerization may also contribute to mutant receptor activation. Building on these findings, we assessed whether BICD2N could enhance the efficacy of existing therapies or restore the effectiveness of previously ineffective treatments. To this end, we externally administered BICD2N-fused DARPins, demonstrating their ability to overcome EGFR-mutant resistance. These DARPin fusions achieved robust suppression of downstream signaling and proliferation in EGFR-mutant models, including the Ex19Del/T790M/C797S triple mutant, when administered as single agents. We note that cetuximab was previously observed not to inhibit growth of cells expressing an EGFR mutant that we found to be oligomeric (Ex19Del/T790M), so it is possible that stronger disruption of dimerization by a sterically extended and directionally targeted extracellular binder is beneficial for suppressing EGFR-driven cancer growth.
Together, our findings support the canonical model in which oncogenic EGFR mutations stabilize the active kinase conformation and promote asymmetric dimerization, but further suggest that extracellular domain interactions may also play a critical role in oncogenic EGFR activation. Along with the discovery that disruption of receptor tyrosine kinase (RTK) trafficking can also effectively inhibit NSCLC cells with drug-resistant EGFR (52, 53), these results suggest that the development of therapeutics with mechanisms of action other than inhibition of kinase catalytic activity may be useful for drug-resistant EGFR mutants. The HomoFC platform offers the ability to find extracellular determinants of oncogenic activation via structure-function analysis or testing of extracellular domain binders, allowing refinement of current models to account for the full-length receptor’s structural complexity and its therapeutic implications.
In summary, our findings suggest that precisely targeted disruption of receptor oligomerization represents a mechanistically distinct and potentially more efficacious approach for overcoming EGFR-driven resistance mechanisms that remain refractory to conventional antibody- or TKI-based therapies. Future research can address whether this strategy may be generalizable to other receptor-driven pathologies, such as those involving human epidermal growth factor receptor 2 (HER2) or vascular endothelial growth factor receptor (VEGFR), whose oligomerization is also essential for receptor activation.

MATERIALS AND METHODS

Plasmid preparation
The EGFR genes were synthesized by Bioneer (South Korea). A −15 charged variant of the superfolder GFP-based HomoFC (19) construct was used as a fluorescent reporter. The immunoglobulin κ (Igκ)–chain leader sequence (ETDTLLLWVLLLWVPGSTGD) was amplified by polymerase chain reaction (PCR) from the pDisplay vector (Invitrogen). NSCLC-related EGFR mutants were generated via site-directed mutagenesis using the NEBuilder HiFi DNA Assembly kit (New England Biolabs). The ERK-KTR construct was amplified from pENTR-ERKKTR Clover, a gift from M. Covert (Addgene plasmid #59138), and BICD2N was cloned from GFP-BICDN-CIBN, a gift from B. Cui (Addgene plasmid #102254). The E01 DARPin-fused constructs were cloned from pAWP78-PVCpnf_pvc13-E01DARPin, a gift from F. Zhang (Addgene plasmid #198287), and the E69 DARPin (54) and monomeric hyperfolder YFP (mhYFP) (55) were synthesized by IDT. All constructs used for mammalian cell–based studies were cloned into the pcDNA3.1 (+) vector (Invitrogen), while DARPin-fused constructs for bacterial expression were cloned into the pET21a vector (Invitrogen). The protein sequence is available in the Supplementary Materials.

Cell culture
HEK293A cells (Invitrogen, R70507), used within 25 passages, were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin-streptomycin (Gibco, USA) at 37°C in a humidified incubator with 5% CO2.

EGFR oligomerization and ERK activity analysis
HEK293A cells (undetectable endogenous EGFR, <25 passages) were seeded into ibidi 8-well chamber slides (80826) at 2 × 104 cells per well for confocal imaging or 96-well plates at 1 × 104 cells per well for plate-reader analysis and incubated for 18 hours. Transient transfection was performed with Lipofectamine 3000 (Thermo Fisher Scientific), and cells were analyzed 24 hours post-transfection. For EGF-stimulated conditions, cells were treated with EGF (10 ng/ml) (Abcam, ab9697), and oligomerization dynamics were monitored for 0 to 120 min. Live-cell imaging was conducted in a 37°C, 5% CO2 incubation chamber using an LSM 980 confocal microscope (Zeiss) equipped with a 63× oil immersion objective. Imaging used identical laser/detector settings: mTurq2 (445 nm, 0.6 mW; 450 to 485 nm), GFP (488 nm, 0.2 mW; 495 to 566 nm), and mCherry (561 nm, 0.03 mW; 565 to 708 nm).
HomoFC-to-mCherry (HomoFC/mCherry) oligomerization signals and ERK-KTR responses were quantified in real time by confocal imaging, and oligomerization measurements were independently validated using a Varioskan LUX Multimode Microplate Reader (Thermo Fisher Scientific). Fluorescence images were analyzed using Zen 2.5 (Carl Zeiss) and ImageJ [National Institutes of Health (NIH)]. HomoFC-to-mCherry (HomoFC/mCherry) ratios were calculated from the mean intensity per cell (n ≈ 20 to 33 cells from three independent experiments). To enable direct comparison of oligomerization levels among NSCLC-related EGFR mutants, cells with mCherry intensities within 10,000 to 30,000 a.u. were analyzed, as this range minimizes low-expression background, avoids GFP saturation at high expression, and ensures consistent quantification across figures and experiments.

Drug treatment
To evaluate the effects of EGFR inhibitors on receptor oligomerization and downstream signaling, HEK293A cells were pretreated with 0.05 μM orthosteric (e.g., osimertinib), allosteric (e.g., EAI045), or combination drugs for 30 min at 37°C without washout. Cells were subsequently stimulated with EGF (10 ng/ml) (Abcam, ab9697), and changes in oligomerization and ERK activity were monitored in real time using confocal microscopy (LSM980, Zeiss) over a 120-min time course. HomoFC-to-mCherry fluorescence ratios and cytoplasm-to-nucleus ERK-KTR localization were quantified to assess drug-mediated effects. No apparent cytotoxicity was observed under these treatment conditions.

TEV protease cleavage assay
For experiments using the BICD2N-TEV cleavage system with EGFR mutants (Fig. 5), HEK293A cells were transiently transfected with mCherry-BICD2N-TEVcs-EGFR or BICD2N-TEVcs-EGFR variants. Twenty-four hours post-transfection, baseline oligomerization and ERK-KTR signals were recorded using an LSM 980 confocal microscope (Zeiss) equipped with a 63× oil immersion objective in a 37°C, 5% CO2 live-cell chamber. Recombinant TEV protease (20 U/ml; New England Biolabs, P8112S) was added directly to the culture medium, and time-lapse imaging was performed every 10 min for 120 min. HomoFC-to-mCherry (HomoFC/mCherry) fluorescence ratios and cytoplasm-to-nucleus (C/N) ERK-KTR signals were quantified to track dynamic changes in EGFR oligomerization and downstream ERK activation.

Cell proliferation assay (MTT)
Cell proliferation was assessed using a tetrazolium-based colorimetric assay (MTT). EGFR-HomoFC–expressing cells were treated with or without EGF (10 ng/ml), allosteric EGFR inhibitors, and ATP-competitive inhibitors and incubated for 48 hours. MTT reagent (0.5 mg/ml final concentration) was added to each well and incubated at 37°C for 3 hours. After incubation, formazan crystals were dissolved in dimethyl sulfoxide (DMSO), and absorbance was measured at 570 nm using a Varioskan LUX Multimode Microplate Reader (Thermo Fisher Scientific).

In vitro DARPin protein preparation
DARPin (E01) and DARPin (E69) fusion proteins (Fig. 6) were expressed in Escherichia coli BL21 (DE3) cells. Transformed cells were grown at 37°C until the optical density at 600 nm (OD600) reached 0.7, at which point protein expression was induced by adding 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Induction was carried out at 20°C with constant shaking at 220 rpm for 18 hours. Cells were harvested by centrifugation at 4°C and lysed by sonication in His-binding buffer (300 mM NaCl, 50 mM tris, pH 8.0). His6-tagged fusion proteins were purified from the soluble fraction of the lysates using HisPur Cobalt Resin (Thermo Fisher Scientific) via affinity chromatography. Protein concentrations were determined by Bradford assay and absorbance at 280 nm, using the Beer-Lambert equation.

eMA and tetrameric streptavidin preparation
eMA and tetrameric streptavidin were expressed in E. coli BL21 (DE3) cells. Cultures were grown at 37°C until the OD600 reached 0.9, at which point protein expression was induced with 1 mM IPTG for 4 hours at 37°C. Cells were harvested by centrifugation at 4°C, and the resulting pellets were resuspended in phosphate-buffered saline (PBS) containing 1% Triton X-100. After sonication and centrifugation at 12,000 rpm for 15 min, inclusion body pellets were collected, washed three times with PBS containing 1% Triton X-100, and solubilized in 6 M guanidinium hydrochloride (GuHCl) at pH 1.5. Solubilized proteins were refolded by rapid dilution into PBS at 4°C, and the solution was stirred overnight. Equilibrated Ni-IDA resin (Bioprogen) in buffer containing 300 mM NaCl and 50 mM tris (pH 8.0) was added to the refolding mixture and incubated overnight at 4°C. The resin was then transferred to a Poly-Prep column (Bio-Rad), washed with 300 mM NaCl, 50 mM tris, and 30 mM imidazole (pH 7.8), and eluted with 300 mM NaCl, 50 mM tris, and 200 mM imidazole (pH 7.8). Eluted proteins were dialyzed three times against PBS. Protein concentrations were determined using Bradford assays and absorbance at 280 nm, calculated based on the Beer-Lambert equation. Ferritin-eMA was expressed in E. coli BL21 (DE3) cells. Transformed cells were cultured at 37°C until the OD600 reached 0.8, at which point protein expression was induced with 1 mM IPTG for 4 hours at 37°C. Cells were harvested and resuspended in Buffer A (1% Triton X-100 in 1× PBS), followed by sonication. Inclusion bodies were recovered by centrifugation at 12,000 rpm for 15 min at 4°C, washed three times with Buffer A, and solubilized by overnight incubation at 4°C in Buffer B (8 M urea, 250 mM NaCl, 50 mM tris, pH 8.0). Denatured proteins were purified using Ni–nitrilotriacetic acid affinity chromatography under denaturing conditions. The eluted proteins were mixed with an equal volume of Buffer B containing 20 mM dithiothreitol (DTT) and incubated at room temperature for 4 hours to reduce disulfide bonds. Refolding was carried out at 4°C by sequential dialysis against decreasing concentrations of urea (4, 3, 2, and 0 M) in Buffer C (50 mM tris, pH 8.0, 50 mM NaCl, 10% glycerol, 0.1% polyethylene glycol, 0.2 mM reduced glutathione, 0.1 mM oxidized glutathione), followed by a final dialysis in Buffer D (50 mM tris, pH 8.0, 50 mM NaCl, 10% glycerol).

Protein structure prediction
The structural model of YFP–DARPin (E01)–BICD2N was predicted using AlphaFold3 (DeepMind). The full-length fusion protein sequence was submitted to the AlphaFold server, and the resulting model was visualized and analyzed using PyMOL (Schrödinger LLC).

Statistics and reproducibility
Statistical differences between two groups were assessed using a two-tailed Student’s t test, while comparisons among multiple groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test. Graphs and statistical analyses were performed using Microsoft Excel 2019 (Office Professional Plus; Microsoft, Redmond, WA, USA) and GraphPad Prism 10.4.0 (GraphPad Software, San Diego, CA, USA). The number of replicates (n) and details of statistical analyses are provided in the corresponding figure legends. All cellular and in vitro assays were independently repeated at least three times, with consistent results observed across biological replicates.