Targeting intracellular oncoproteins with dimeric IgA promotes expulsion from the cytoplasm and immune-mediated control of epithelial cancers

SUMMARY

Dimeric IgA (dIgA) can move through cells via the IgA/IgM polymeric immunoglobulin receptor (PIGR), which is expressed mainly on mucosal epithelia. Here, we studied the ability of dimeric IgA (dIgA) to target commonly mutated cytoplasmic oncodrivers. Mutation-specific dIgA, but not IgG, neutralized KRAS^G12D within ovarian carcinoma cells and expelled this oncodriver from tumor cells. dIgA binding changed endosomal trafficking of KRAS^G12D from accumulation in recycling endosomes to aggregation in the early/late endosomes through which dIgA transcytoses. dIgA targeting of KRAS^G12Dabrogated tumor cell proliferation in cell culture assays. In vivo, KRAS^G12D-specific dIgA1 limited growth of KRAS^G12D- mutated ovarian and lung carcinomas in a manner dependent on CD8+ T cells. dIgA specific for IDH1^R132H reduced colon cancer growth, demonstrating effective targeting of a cytoplasmic oncodriver not associated with surface receptors. dIgA targeting of KRAS^G12D restricted tumor growth more effectively than small molecule KRAS^G12D inhibitors, supporting the potential of this approach for the treatment of human cancers.

Graphical Abstract

eTOC

Despite their long half-life, therapeutic antibodies are considered ineffective against intracellular antigens. Biswas et al demonstrate that dimeric IgA undergoes transcytosis through PIGR⁺ epithelial cancer cells, and can be engineered for specific targeting of intracellular mutated oncodrivers. Binding of mutation-specific IgA expels cells oncodrivers from the cytoplasm and promotes anti-tumor immunity.

INTRODUCTION

Limited by their large size and considered unable to penetrate the cytoplasm, development of antibody-based immunotherapies has focused on transmembrane or extracellular targets. However, most oncodrivers underpinning tumor-promoting pathways are intracellular proteins inaccessible to conventional therapeutic antibodies. Accordingly, considerable efforts have been made to develop small molecules that target these intracellular pathways by inhibiting enzymatic activity or serving as allosteric modulators.

KRAS is considered the most common mutated oncogene driving human cancer¹. KRAS mutations are particularly frequent in pancreatic, colorectal, non-small cell lung and endometrial cancer², which total more than 50,000 cases annually³. KRAS mutations primarily comprise single-base missense mutations in codons 12 (80%), 13 or 61¹. Following a 40-year quest to target KRAS mutations, Sotorasib⁴, a specific KRAS^G12C small molecule inhibitor, was approved by the FDA in 2021. Pharmacological inhibitors against KRAS^G12D, the more common mutation in pancreatic and colon cancer, are currently under testing⁵. Small molecule inhibitors that target oncodrivers offer great therapeutic benefits and have made a big impact in oncology. However, the half-life of many small molecules is ~6 hours⁴, which can limit on target interactions. In contrast, the serum half-life of optimized antibodies is >20 days⁶, which offers more sustained neutralization and can thus be less frequently dosed. Furthermore, small molecule inhibitors, including Sotorasib⁷, are inevitably associated with acquired resistance. Some of these mechanisms include locking KRAS in a GTP-bound (activated) form⁷, which is insensitive to the small molecule inhibitor but could be specifically recognized by an antibody. Furthermore, it is becoming increasingly clear that anti-tumor antibody responses are intrinsically linked to tumor-specific T cell responses, in both preclinical models⁸ and human cancers^9-13.

Although monomeric IgA is found in serum, at mucosal surfaces IgA is predominantly found as a dimer, with 2 subunits bound by a J-chain¹⁴. dIgA binds to the IgA/IgM polymeric immunoglobulin receptor (PIGR), a member of the immunoglobulin superfamily that is expressed on the basolateral surface of mucosal epithelia¹⁴. Recognition of polymeric IgA (and IgM) by PIGR triggers a process called transcytosis, whereby the PIGR:antibody complex traffics inside the epithelial cell from the basolateral to the apical side, and is then secreted into mucosal fluids¹⁵. We previously showed that PIGR is quasi-universally expressed in human ovarian^9,16 and endometrial cancer cells^12,17, and allows transcytosis of dIgA through these tumor cells. However, it is unclear whether IgA transcytosis also occurs in non-gynecologic tumors. Furthermore, IgA transcytoses through endosomes¹⁸, which could prevent direct contact with specific antigens inside tumor cells. Even if transcytosing IgA is not “shielded” by endosomes, targeted antigens could not co-localize inside tumor cells. Experimentation with these processes might determine if it is possible for engineered dIgA to target intracellular molecules, with important therapeutic implications: First, the mechanisms of resistance that invariably follow the administration of small molecule inhibitors¹⁹ could be still sensitive to antibody-mediated neutralization. Second, the half-life of unmodified IgAis ~6 days, but can be further optimized to increase stability²⁰. This could lead to a more sustained activity and therefore enhanced effectiveness. For instance, long half-life and occupancy of PD-1 molecules on T cells appear to be important determinant of effectiveness for immune checkpoint inhibitory antibodies⁶. Third, dIgA is particularly abundant at mucosal surfaces, suggesting that exogenous dIgA could be safely administered to cancer patients²¹. For instance, oral IgA-IgG treatments and antibody preparations containing >20% IgA have been safely administered to patients^22,23. In this study, we investigated the feasibility and therapeutic potential of targeting common mutated oncodrivers using recombinant dIgA. Our results provide a rationale for developing dIgA-based therapeutics to neutralize various intracellular antigens in human cancer and other diseases.

RESULTS

Mutation-specific dIgA neutralizes intracellular KRAS^G12D and expels it outside tumor cells, without effects on KRAS^WT cells

To investigate the spectrum of human malignancies that could be susceptible to PIGR-mediated dIgA transcytosis, we first analyzed PIGR expression in tumors in TCGA datasets. We found a clear dichotomy between epithelial malignancies and B cell lymphoma, which express relatively high PIGR mRNA; and non-epithelial tumors such as melanoma or glioblastoma, which express lower amounts of PIGR mRNA (Suppl. Fig.1A). To test whether antigen-specific dIgA, and not non-specific dIgA (Suppl. Fig.1B&C, Data S1), could target specific molecules in the cytoplasm of different carcinomas, we first focused on targeting KRAS^G12D, a mutational hotspot present in >4% of human cancers, and frequently in pancreatic, colorectal, lung and endometrial carcinomas^2,24. Recombinant antibodies generated against GppNHp-bound KRAS^G12D (equivalent to the oncogenic GTP-bound form²⁵) were produced on a dIgA or monomeric IgG4 backbone, as in anti-PD-1 immune checkpoint inhibitors, using identical variable heavy and variable light sequences (Suppl. Fig.2A-C; Data S1 & Data S2). Although we cannot rule out that GppNHp-KRAS^G12D-specific Abs recognize GTP-bound conformations driven by other mutations²⁵, both dIgA and IgG Abs specifically recognized ectopic or endogenous mutant KRAS^G12D, but not other KRAS mutations, in Western-blot analysis (Fig.1A; Suppl. Fig.2D). Notably, only mutation-specific dIgA, but not IgG, penetrated KRAS^G12D-mutated lung cancer cells (Fig.1B, left), where mutation-specific, but not control dIgA, targeted KRAS (Fig.1B, right). Consistently, KRAS^G12D-specific dIgA penetrated OVCAR3 (ovarian cancer) cells transduced with KRAS^WT or KRAS^G12D fused to photoactivatable (PA)m-Cherry fluorescent protein²⁶, but only disrupts cytoplasmic localization and distribution of intracellular KRAS^G12D (Fig.1C). Notably, mutant KRAS was not only neutralized inside tumor cells, but also expelled outside the cell through transcytosis, as evidenced by co-occurrence of KRAS^G12D with secretory IgA in the supernatant of KRAS^G12D-PAmCherry-transduced OVCAR3 cells upon treatment with KRAS^G12D-specific dIgA (Fig.1D), with decreased intracellular KRAS^G12D (Fig.1E & Suppl. Fig2E). Furthermore, PamCherry fragments, fused to KRAS^G12D, were only found in the supernatant of tumor cells treated with KRAS^G12D-specific dIgA in 3 independent experiments (Fig.1F, left). In contrast, KRAS^G12D was not found in supernatants upon treatment with KRAS^G12D-specific IgG4, irrelevant IgA containing dimers⁹, or vehicle (Fig.1D&F, left). As expected, PIGR was identified in the supernatants of ovarian cancer cells treated with antigen-specific or control dIgA (Fig.1F, right). Together, these results indicate that tumor cell-penetrating dIgA can indeed target specific mutations in KRAS inside tumor (e.g., ovarian cancer) cells, resulting in intracellular decreased levels and expelling the oncodriver outside the tumor cell, without obvious effects in KRAS unmutated epithelial cells.

Figure 1. KRAS^G12D mutation-specific dIgA1, but not IgG4, neutralizes KRAS^G12D inside tumor cells and expels the oncodriver outside tumor cells.

(A) Immunoblots showing that recombinant anti-KRAS^G12D-IgA1 and anti-KRAS^G12D-IgG4 antibodies specifically recognize this mutation in KRAS, but not G12C or G13D mutant KRAS, or WT KRAS (n=3). (B) Confocal immunofluorescence microscopy images showing that Alexa Fluor 647-conjugated anti-KRAS^G12D-IgA1, but not with anti-KRAS^G12D-IgG4, penetrates KRAS^G12D-mutant A427 and SK-LU-1 NSCLC cells, while KRAS^G12D-specific, but not control irrelevant dIgA, co-immunoprecipitated with KRAS^G12D; Scale bar, 25 μm. (C) OVCAR3 cells were transduced with KRAS^G12D or WT KRAs fused to PamCherry (Red). Confocal microscopy shows that treatment with Alexa Fluor 488-conjugated anti-KRAS^G12D-IgA1 (middle panel), but not with anti-KRAS^G12D-IgG4 (right panel), disarrays KRAS^G12D intracellular distribution, without effects on KRAS^WT cells (left panel). (n=3). Scale bar, 50 μm (D) The secretory component of human PIGR was immunoprecipitated from media from KRAS^G12D-PamCherry-transduced OVCAR3 cells, treated with anti-KRAS^G12D-IgA1, IgG4, irrelevant IgA or PBS vehicle. Eluted immunocomplexes were ran by SDS-PAGE, stained with Ponceau Red, and immunoblotted for KRAS^G12D, IgA, and pIgR (n=3). I Immunoblots showing KRAS^G12D levels in anti-KRAS^G12D-IgA1/IgG4 or irrelevant IgA-treated KRAS^G12D-OVCAR3 cell lysates. (F) Supernatants from KRAS^G12d-PAmCherry-transduced OVCAR3 cells, treated with anti-KRAS^G12D-IgA1, IgG4, irrelevant IgA or PBS vehicle, were subjected to liquid chromatography with tandem mass spectrometry (LC–MS/MS). Heatmaps showing intensities of mCherry (left panel) and pIgR (right panel) peptides detected (n=3). (G) Left, Transcytosis experiment adding biotinylated anti-KRAS^G12D-IgA1/IgG4 or irrelevant IgA to the upper chamber of a transwell system, where physical access of antibodies to the basal chamber is prevented by WT or KRAS^G12D-OVCAR3 cells grown to confluence in transwell inserts. Right, dot blots showing presence of KRAS^G12D (top panel) or beads (bottom panel) in the streptavidin immunoprecipitates of the basal and upper chamber contents.

KRAS^G12D and mutation-specific dIgA are found in the same endosomal compartments upon transcytosis

To confirm bona fide unidirectional transcytosis, we used a classical transwell system where ovarian or lung tumor cells prevent the access of dIgA delivered to the upper chamber into the basal chamber, unless it transcytoses through PIGR⁺ tumor cells²⁷. As shown in Fig.1G & Suppl. Fig.2F, we found an accumulation of KRAS^G12D antigen in the basal chamber when KRAS^G12D-mutant ovarian and lung cancer cells were treated with antigen-specific dIgA, but not in wild-type ovarian tumor cells or upon PIGR ablation. However, a frail KRAS signal was detected using a pan-KRAS antibody in the basal chamber of tumor cells with other mutations, suggesting weak recognition of the GTP-bound oncogene²⁵ (not shown). Importantly, only traces of antigen were found in the upper chamber, where dIgA was added and found in similar amounts, supporting that KRAS^G12D-specific dIgA undergoes “true” transcytosis through PIGR⁺ ovarian and lung cancer cells. Accordingly, high-resolution confocal microscopy demonstrated that aggregates of KRAS^G12D-specific dIgA co-localize with KRAS^G12D-PamCherry inside ovarian tumor cells, while the same antibody on an IgG4 backbone cannot penetrate ovarian cancer cells or alter the preferential location of mutant KRAS on the cell surface (Fig.2A).

Figure 2. KRAS^G12D-specific dIgA1, but not the same antibody on an IgG4 backbone, captures mutant KRAS during trafficking.

(A) High resolution confocal microscopy showing that aggregates of KRAS^G12D-specific dIgA, but not KRAS^G12D-specific IgG, co-localize with KRAS^G12D-PamCherry after 1 hr of treatment inside KRAS^G12D-OVCAR3 tumor cells (n=3). No-wash images confirm equivalent positive green fluorescence of added KRAS^G12D-specific IgA and IgG antibodies outside of the cells (45 min after treatment), Scale bar, 50 μm or 10 μm, as indicated. (B) Top panel, Non-denatured lysates (6 mg) of KRAS^G12D-OVCAR3 cells treated with anti-KRAS^G12D-IgA1, anti-KRAS^G12D-IgG4, or vehicle (PBS) were immunoprecipitated using anti-Rab5A, Rab7A, Rab8A, or Rab11A antibodies. Immunoprecipitates were then blotted for KRAS^G12D-specific IgA or IgG. Bottom panel, 60 μg lysates (1% of co-IP lysate) of anti-KRAS^G12D-IgA1, anti-KRAS^G12D-IgG4, or vehicle (PBS)-treated KRAS^G12D-OVCAR3 cells were immunoblotted for Rab5A, Rab7A, Rab8A, and Rab11A as input control. (n=3).

To understand the dynamics of interactions between antigen-specific dIgA and mutant KRAS inside KRAS^G12D-transduced ovarian tumor cells, we immunoprecipitated early (RAB5A⁺), late (RAB7A⁺) and recycling (RAB8A⁺/RAB11A⁺) endosomes^28,29 in KRAS^G12D-transduced ovarian cancer cells treated with KRAS^G12D-specific dIgA or IgG. As expected, given the lack of penetration into tumor cells, KRAS^G12D-specific IgG was undetectable in any endosomes, while dIgA transcytosed through early and late endosomes (Fig.2B). Consistent with known recycling pathways that prevent proteasomal degradation³⁰, mutant KRAS was primarily found in RAB8A⁺/RAB11A⁺ recycling endosomes when ovarian tumor cells were treated with vehicle or antigen-specific IgG (Fig.2B). In contrast, treatment with KRAS^G12D-specific dIgA concentrated mutant KRAS on the same endosomes through which dIgA traffics inside tumor cells (Fig.2B), suggest that transcytosing dIgA encounters the antigen on the cell membrane and hauls it in endosomes, eventually expelling the target through secretory IgA. Although dIgA traffics through endosomes, dIgA retains its capacity to specifically target intracellular antigens, at least those that are located near the cell membrane.

Consistent with capture and extracellular disposal of KRAS^G12D, treatment of KRAS^G12D-transduced ovarian cancer cells with mutation-specific dIgA inhibited tumor cell proliferation, without effects on the same cells transduced with KRAS^WT or untransduced ovarian cancer cells (Fig.3A-C). Treatment of endogenously mutated KRAS^G12D lung cancer cells with mutation-specific dIgA also inhibited tumor cell proliferation (Fig.3D), while neither antigen specific or control dIgA impacted the proliferation of non-malignant HEK293T cells (Fig.3E). Of note, antigen-specific dIgA dampens proliferation of KRAS^G12D-transduced ovarian cancer cells without causing apoptosis (Fig.3F). Therefore, antigen-specific dIgA expels oncodrivers outside tumor (e.g., ovarian cancer) cells by preventing endosomal recycling, resulting in KRAS^G12D accumulation in the same endosomal compartments where dIgA traffics.

Figure 3. KRAS^G12D-specific dIgA treatment reduces proliferation, without causing apoptosis, of KRAS^G12D-mutated, but not KRAS^WT cancer cells.

(A) MTT assay-based absorbance at 570 nm of KRAS^G12D-transduced OVCAR3 cells treated with irrelevant IgA or anti-KRAS^G12D IgA1, relative to mean absorbance of vehicle (PBS) treated cells, at different temporal points. (n=8) (B) MTT assay-based absorbance at 570 nm of KRAS^WT-transduced OVCAR3 cells treated with irrelevant IgA or anti-KRAS^G12D IgA1, relative to mean absorbance of vehicle (PBS) treated cells, at different temporal points. (n=8). (C) MTT assay-based absorbance at 570 nm of untransduced OVCAR3 cells treated with irrelevant IgA or anti-KRAS^G12D IgA1, relative to mean absorbance of vehicle (PBS) treated cells, at different temporal points. (n=16). (D) MTT assay-based absorbance at 570 nm of KRAS^G12D SK-LU-1 cells treated with irrelevant IgA or anti-KRAS^G12D IgA1, relative to mean absorbance of vehicle (PBS) treated cells, at different temporal points. (n=16). (E) MTT assay-based absorbance at 570 nm of PIGR⁻ HEK293T cells treated with irrelevant IgA or anti-KRAS^G12D IgA1, relative to mean absorbance of vehicle (PBS) treated cells, at different temporal points. (n=16). (F) Total apoptotic cells, determined by flow cytometry analysis of KRAS^G12D-transduced OVCAR3 cells, treated with vehicle, or irrelevant IgA, or anti-KRAS^G12D IgA1, or anti-KRAS^G12D IgA4 antibodies, and gated on Annexin V⁺ Propidium iodide^+/− cells (n=3). Data are mean ± SEM. ***P ≤ 0.001, NS, not significant; unpaired two-tailed t-test. Data S3 provides details of statistics.

KRAS^G12D-specific dIgA specifically abrogates the progression of mutant tumors in vivo

To demonstrate the effectiveness of KRAS^G12D-specific dIgA in vivo, we first performed intra-tumoral infusions of different treatments in KRAS^G12D-transduced OVCAR3 (ovarian cancer)-bearing Rag1^−/− mice. Irrelevant dIgA dampens the RAS pathway by increasing expression of DUSP phosphatases that dephosphorylate ERK1/2, while activating inflammatory pathways associated with IFN and TNF signaling, as well as pro-apoptotic pathways linked to ER stress^9,12. Accordingly, non-specific dIgA delayed the growth of established ovarian tumors. However, the anti-tumor effects of KRAS^G12D-specific dIgA were greater (Fig.4A&B), with no obvious signs of toxicity in any treated mice. Consistently with its inability to penetrate tumor cells, the same variable heavy and light chains on an IgG4 backbone did not elicit significant ovarian cancer reduction, potentially attributable to neutralization of extracellular vesicles (Fig.4B & Suppl. Fig.3A&B). Supporting the specificity of this intervention, no effects on tumor growth were elicited in tumor cells transduced with KRAS^WT, (Fig.4C & Suppl. Fig.3A&C), beyond the non-antigen-specific effects of IgA transcytosis. As reported⁵, the KRAS^G12D-selective inhibitor MRTX1133, currently under preclinical development⁵, also elicited significant delays in ovarian cancer growth, although daily intraperitoneal administration (20 injections) were required (Suppl. Fig.3A-C).

Figure 4. KRAS^G12D-specific dIgA1, but not the same antibody on an IgG4 backbone, abrogates KRAS^G12D tumor growth in vivo.

(A) Schematic of design of experiment shown in B and C. Antibody (100 μg per 20 g body weight) or equal volume of vehicle (PBS) was intratumorally (IT) injected. (B) Tumor growth curves (left), tumor weight (right) in KRAS^G12D-OVCAR3 tumor-bearing Rag1-deficient mice receiving control IgA, anti-KRAS^G12D-IgA1, anti-KRAS^G12D-IgG4 antibodies, or vehicle. Growth curves and tumor weights were pooled from 2 independent experiments (n=10 mice per group, total). (C) Tumor growth curves (left), tumor weight (right) from KRAS^WT-OVCAR3 tumor-bearing Rag1-deficient mice receiving the same treatments as in B. Growth curves and tumor weights were pooled from 2 independent experiments (n=10 mice per group, total). (D) Schematic of design of experiment. Antibody (100 μg per 20 g body weight), equal volume of vehicle (PBS), or MRTX1133 (200 μg per 20 g body weight) were intraperitoneally (IP) injected every 4 days, or daily, as indicated, into NSG mice growing KRAS^G12D-OVCAR3 tumors. Tumor growth curves (left) and tumor weight (right) were pooled from 3 independent experiments (n=13 mice per group, total). Data are presented as mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, NS, not significant; paired two-tailed t-test for growth curves or unpaired two-tailed t-test for tumor weight (I). (E) Representative immunohistochemical staining of human IgA of KRAS^G12D-OVCAR3 mice tumors treated intraperitoneally with non-antigen specific or KRAS^G12D-specific dIgA. Scale bar, 400 or 60 μm, as indicated. Data S3 provides details of statistics.

Selective effectiveness of antigen-specific dIgA was recapitulated when antibodies were intraperitoneally administered in KRAS^G12D tumor-bearing NSG mice (Fig.4D & Suppl. Fig.4A&B), ruling out the requirement of functional myeloid cells or NK lymphocytes in this model, while supporting systemic distribution of dIgA. Again, 20 (daily) doses of intraperitoneal MRTX1133 or 5 injections of KRAS^G12D-specific dIgA both delayed tumor progression beyond the effects of irrelevant IgA (Fig.4D & Suppl. Fig.4B). Correspondingly, human IgA was identified in resected tumors (Fig.4E), confirming trafficking of antigen-specific dIgA to tumor beds. These experiments support the potential of tumor cell-penetrating dIgA to specifically target mutated oncodrivers inside carcinoma cells in vivo.

Lung tumors quasi-universally express PIGR and are highly sensitive to mutant oncodriver-neutralizing dIgA

To demonstrate the capacity of dIgA to target human tumors that are spontaneously driven by KRAS^G12D, as opposed to ectopic KRAS expression, we first focused on non-small cell lung cancer (NSCLC), which is frequently driven by mutant KRAS, including KRAS^G12D. Analyses of 30 human lung cancer tissues including squamous cell carcinomas and adenocarcinomas, showed PIGR expression in all tumors, along with spontaneous production of IgA and IgG at tumor beds bound to tumor epithelial tissues as well as to stroma (Fig.5A), without differences for histological subtypes (Suppl. Fig.5A). Notably, PIGR and tumor-bound IgA density, IgG coating of tumor cells, or PIGR:IgA interactions were all associated with higher T cell accumulation (Fig.5B). We therefore aimed to demonstrate the effectiveness of mutation-specific dIgA against lung carcinomas endogenously driven by KRAS^G12D. As shown in Fig.5C&D, 5 intraperitoneal injections of KRAS^G12D-specific dIgA, or 20 daily treatments of MRTX1133 at high dose (200 μg)⁵, effectively abrogated the progression of established A427 NSCLC, while the same antibody on an IgG4 backbone had negligible effects (Fig.5C&D; Suppl. Fig.5B&C). Accordingly, treatment with both KRAS^G12D-specific dIgA and MRTX1133 decreased ERK1/2 phosphorylation in A427 cells in a time-dependent manner (Fig.5E). Anti-tumor effects were not limited to this lung cancer model, because established KRAS^G12D-mutant SK-LU-1 lung cancers also responded to KRAS^G12D-specific dIgA with comparable effectiveness, although they were more resistant to MRTX1133 (Fig.5C&F; Suppl. Fig.5B&D). Although established lung tumors were not fully eliminated, different models underwent a significant reduction in tumor growth. These experiments support the potential of dIgA-mediated intracellular targeting of KRAS^G12D to abrogate the progression of spontaneously mutated NSCLCs, in a mutation-specific manner and without off-target effects in unmutated epithelial cells.

Figure 5. KRAS^G12D-specific dIgA1 is more effective than small molecule KRAS^G12D inhibitors at abrogating the progression of KRAS^G12D NSCLC.

(A) Combined staining of pIgR, IgA, IgG, CD3, Pan cytokeratin (panCK) and DAPI, representative of 30 clinical cases of squamous cell carcinoma or adenocarcinoma of the lung. Scale bar, 200 μm (upper panel), 100 μm (magnified lower panels). (B) Accumulation of CD3⁺ T cells in human panCK⁺ NSCLC islets are associated with increased coating of tumor cells with IgA and IgG, expression of pIgR and colocalization of IgA with pIgR. Data are mean ± SEM. **P≤0.01; ***P≤0.001, unpaired two-tailed t-test. (C) Schematic of design of experiment shown in D&E. Antibody (100 μg per 20 g body weight), or equal volume of vehicle (PBS) were administered every 4 days, while MRTX1133 (200 μg per 20 g body weight) was administered daily or every 4 days, as indicated. All treatments were IP injected. (D) Tumor growth curves from KRAS^G12D A427 tumor-bearing NSG mice receiving (IP) anti-KRAS^G12D-IgA1, anti-KRAS^G12D-IgG4 or vehicle every 4 days. MRTX1133 was administered in different mice every 4 days (left) or daily (middle), as indicated. Tumor weight is shown on the right. Growth curves and tumor weights were pooled from 3 independent experiments (n=14 mice per group, toI). (E) Western blot showing phosphorylated and total ERK1/2 in lysates of KRAS^G12D-mutated A427 cells, treated with anti-KRAS^G12D-IgA1 or MRTX1133. (F) Identical treatment as in ‘D’ of advanced KRAS^G12D-mutated SK-LU-1 lung cancers. Growth curves and tumor weights were pooled from 3 independent experiments (n=14 mice per group, total). (G) Left, schematic design of treatments in KPMSH2^KIN lung tumor-bearing immunocompetent mice. Antibody (100 μg per 20 g body weight) or equal volume of vehicle (PBS) was IP injected. When indicated, CD8 T cells were depleted with intraperitoneal anti-CD8 antibodies. Right, Tumor growth curves from mice receiving intraperitoneal vehicle or anti-KRAS^G12D-dIgA1, or anti-IDH1^R132H-dIgA1, with or without CD8 T cell depletion. Growth curves were pooled from 2 independent experiments (n=8-10 mice per group). Data are mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001; paired two-tailed t-test (D&F) and non-parametric Wilcoxon test (G) for growth curves or unpaired two-tailed t-test for tumor weights. Data S3 provides details of statistics.

dIgA transcytosis sensitizes tumor cells to T cell-mediated killing⁹. To define the contribution of this mechanism upon KRAS^G12D-specific dIgA treatment, we utilized a syngeneic, immunogenic KRAS^G12D-mutant KPMSH2^KIN lung tumors, generated by cloning lung cancer cells from C57/BL6 mice with inducible mutations, activated using intra-nasal adenoviral Cre; plus restoring MSH2 through lentiviral transduction. dIgA targeting KRAS^G12D substantially delayed tumor growth (Fig.5G). Anti-tumor effects in this immunogenic system were dependent on CD8⁺ T cells (Fig.5G), highlighting the complexity of a coordinated adaptive immune response. Further supporting the specificity of dIgA for cognate antigen, similar anti-tumor effects were not observed upon administration of dIgA targeting a different mutation (IDH1^R132H) (Fig.5G). Corresponding effects were observed in another syngeneic KRAS^G12D breast cancer model (Brpkp110³¹; Suppl. Fig.5E).

dIgA is effective against mutations located deeper into the cytoplasm of tumor cells, in a PIGR-dependent manner

To demonstrate the feasibility of targeting additional intracellular antigens, located deeper into the cytoplasm of tumor cells; and also the potential of using dIgA to target other epithelial cancers different from ovarian and lung carcinomas, we generated dIgA1 or IgG4 specifically targeting the cytoplasmic^32,33 R132H mutation in isocitrate dehydrogenase 1 (IDH1), a common hotspot in gliomas³⁴ and sporadically found in gastroesophageal³⁵, colorectal³⁶ and lung³⁷ carcinomas (Fig.6A; Suppl. Fig.2A-C; Data S2). To further evaluate the role of PIGR abundance for dIgA effectiveness, we procured isogenic IDH1^WT and IDH1^+/R132H HCT116 colon cancer cells, which express low PIGR levels, and transduced these with PIGR, resulting in expression comparable to those found in multiple carcinoma cell lines (Fig.6A). Low PIGR expression abrogated the non-specific anti-tumor effects of irrelevant IgA intraperitoneally infused into flank colon cancer-bearing mice (Fig.6B-D). In contrast, similar to KRAS^G12D targeting, IDH1^R132H-specific dIgA delayed tumor growth in mutant flank colon tumors, without effect in the absence of IDH1^R132H mutations (Fig.6C&D). Notably, anti-tumor effects were dependent on PIGR expression in colon cancer cells, because the administration of IDH1^R132H-specific dIgA resulted in a 3.7-fold reduction in flank tumor growth, compared to vehicle, when PIGR-transduced IDH1^R132H HCT116 colon cancer cells were treated, compared to a 1.8-fold reduction in low-PIGR-expressing parental IDH1^R132H HCT116 colon tumors (Fig.6B,C&E). Importantly, anti-tumor effects were again mutation-specific, because a comparable reduction in colon cancer growth was observed when mice with IDH1^WT tumors were treated with irrelevant IgA or IDH1^R132H-specifc dIgA (Fig.6F). Dimerization of mutation-specific IgA is also required because the monomeric version of the same IDH1^R132H-specific IgA showed no significant anti-tumor effects (Fig.6G). Together with data from targeting KRAS-mutated lung tumors, these results support the potential of targeting carcinomas of multiple histological origins and different mutated oncodrivers, using antigen-specific dIgA. They also support the feasibility of targeting tumors despite low PIGR expression, albeit with possibly decreased effectiveness.

Figure 6. IDH1^R132H-specific dIgA1 abrogates IDH1-mutated colon cancer growth in vivo in a PIGR expression-dependent manner.

(A) Immunoblots showing that recombinant anti-IDH1^R132H-IgA1 antibodies specifically recognize the mutation in the lysates of IDH1^R132H/+ HCT116 cells, but not in IDH1^+/+ HCT116 cells, with or without PIGR-transduction. Overexpression of PIGR protein in PIGR-transduced cells is also shown (n=3). (B) Schematic of design of experiment shown in C-F. Antibody (100 μg per 20 g body weight) or equal volume of vehicle (PBS) was injected IP. (C) Untransduced (PIGR^low) mutant-IDH1 (IDH1^+/R132H) tumor-bearing NSG mice received IP control IgA, anti-IDH1^G12D-IgA1, or vehicle. Tumor growth curves (left), tumor weight (middle) and tumor volume of one representative experiment (right) are shown. Growth curves and tumor weights were pooled from 3 independent experiments (n=14 mice per group, total). (D) Tumor growth curves (left), tumor weight (middle) and tumor volume of one representative experiment (right) from identically treated untransduced (PIGR^low) WT-IDH1 (IDH1^+/+) tumor-bearing NSG mice. Growth curves and tumor weights were pooled from 3 independent experiments (n=15 mice per group,Ital). (E) Tumor growth curves (left), tumor weight (middle), tumor volume of one representative experiment (right) from PIGR-transduced mutant-IDH1 (IDH1^+/R132H) tumor-bearing NSG mice treated as in C&D. Growth curves and tumor weights were pooled from 3 independent experiments (n=16 mice per group, total). (F) Tumor growth curves (left), tumor weight (middle), tumor volume of one representative experiment (right) from PIGR-transduced WT-IDH1 (IDH1^+/+) tumor-bearing NSG mice treated as in C-E. Growth curves and tumor weights were pooled from 3 independent experiments (n=16 mice per group, total). (G) Schematic of design of experiment. Monomeric or dimeric anti-IDH1^G12D-IgA1, anti-IDH1^G12D-IgG4 antibody or equal volume of vehicle (PBS) were injected IP into mutant-IDH1 (IDH1^+/R132H) tumor-bearing NSG mice. All antibodies were administered at 100 μg per 20 g body weight. Growth curves pooled from 2 independent experiments (n=8-9 mice per group, total). Data are mean ± SEM. *P ≤ 0.05, ***P ≤ 0.001, NS, not significant; paired two-tailed t-test for growth curves or unpaired two-tailed t-test for tumor weights. Data S3 provides details of statistics.

DISCUSSION

We found that PIGR-mediated transcytosis of dIgA occurs in non-gynecological carcinomas⁹, including lung and colorectal cancers. Given that PIGR is expressed in most epithelial malignancies, but only at lower levels in non-epithelial cancers³⁸, this mechanism could be relevant for most human epithelial cancers. This should include recalcitrant tumors such as pancreatic or small cell lung cancer, which our data show quasi-universally express PIGR. PIGR-mediated dIgA transcytosis progresses through endosomal trafficking but their route of trafficking may slightly differ in malignant cells than in non-malignant cells¹⁸. Despite trafficking through endosomes, dIgA retains its capacity to specifically target intracellular antigens. Our results show that dIgA transcytosis elicits loss of recycling endosomes carrying oncogenic KRAS and colocalization of antigen and antibody in early and late endosomes. This is consistent with heterotypic fusion of the early endosomes that internalize dIgA with the recycling endosomes bringing back KRAS to the cell surface. However, this activity is not limited to mutated oncodrivers that are located near the cell membrane, such as KRAS, but also mutated oncogenes such as IDH1, which are located deeply inside the cytoplasm. Future studies should examine whether other commonly mutated oncogenes such as PI3K or AKT, or immunosuppressive intracellular pathways such as IDO, could be more effectively targeted with dIgA than with small molecule inhibitors.

dIgA-mediated targeting, along or in combination with small molecules, could have substantial advantages, compared to small molecules: First, our results underscore the specificity of this approach. Second, although the half-life of IgA (~6 days in primates and >15 hrs in mice)^39,40 is lower than that of IgG³⁹, and there could be important differences in availability of PIGR-binding dimeric IgA vs. the monomeric form, which does not bind to PIGR but could reach the tumor microenvironment, persistence is nevertheless longer than for small molecule inhibitors. Therefore, the anti-tumor effect could be more sustained as the antibody will remain available to target tumor cells for much longer periods. Moreover, modifications in glycosylation or attaching an albumin-binding domain to the heavy or light chain could increase their half-life^20,41. Third, dIgA expels the mutated oncogene outside tumor cells, which is obviously advantageous, compared to temporary neutralization. Nevertheless, our data also show that the effectiveness of intracellular targeting of oncodrivers using dIgA is dependent on the expression levels of PIGR in tumor cells. Therefore, one could envision that resistance to dIgA could arise from PIGR loss by tumor cells. However, PIGR is quasi-universally expressed in most epithelial cancers, suggesting that it provides a survival advantage to tumor cells. The use of dIgA to target KRAs mutations should complement existing and emerging small molecule inhibitors, which inevitably elicit mechanisms of resistance, Antibodies could therefore provide a valid therapy when this occurs, while small molecule inhibitors could be used if PIGR was lost after dIgA targeting. Finally, the prospect of oral secretory IgA treatment⁴² could make this intervention ideal against tumors of the digestive tract. We therefore propose a rationale for using dIgA as a form of immunotherapy to target intracellular oncodrivers, which could open multiple new avenues to treat otherwise undruggable carcinomas.

Although our dIgA treatment did not fully eliminate xenograft tumors, we administered 5 injections and we only treated established (i.e., palpable) tumors. It is possible that earlier interventions, more sustained antibody administration, or higher doses of dIgA could have resulted in complete tumor rejection. We also found a patchy staining for human IgA, which could reflect suboptimal doses of dIgA, but also different transcytosis timing for distinct tumor areas, or mutational heterogeneity. Furthermore, treatment of an immunogenic syngeneic lung tumor in immunocompetent mice was found to be dependent on T cells, because depletion of CD8⁺ T cells abrogated therapeutic benefits. This is consistent with our previous report that dIgA transcytosis sensitizes ovarian cancer cells to antigen-specific T cell-mediated killing⁹. It is therefore possible that combination with immune checkpoint inhibitors synergizes with targeting of mutated intracellular oncodrivers using dIgA, or that our observed effects are maximized in immunocompetent patients. Our reports in ovarian cancer, for instance, support the importance of spontaneous coordinated humoral and cellular adaptive immune responses as predictors of better outcome⁹.

Our data showing therapeutic effectiveness in immunocompetent mice bearing syngeneic immunogenic carcinomas further supports the concept that the interaction between human dIgA and mouse pIgR mirrors the bioavailability of antigen-specific dIgA in vivo, in the presence of PIGR⁺ healthy epithelial cells that could compete for the antibody. Furthermore, unlike humans, rats and hamsters, mice do not express CD89, which also binds IgA and could be partially responsible for enlarged persistence of IgA in primates, compared to mice⁴⁰, and therefore retain dIgA in circulation for much longer.

Finally, an important issue regarding the future translatability of dIgA is potential toxicity. In support of the clinical potential of our approach, oral IgA-IgG treatment has already been safely used to treat children with chronic non-specific diarrhea²², and polyclonal antibody preparations containing >20% IgA can be safely administered in patients with severe pneumonia²³. In addition, mutation-specific dIgA had no effect on wild-type cells, while dimeric IgA is the predominant Ig at mucosal secretions²¹. It is therefore unlikely that IgA administration will cause on-target, off-tumor toxicity. Furthermore, FDA-approved KRAS^G12C small molecule inhibitors already elicit grade 3-4 toxicity in >11% of patients. New studies should evaluate these issues, but our results provide a rationale for developing dIgA-based therapeutics that specifically target intracellular antigens in human cancer patients.

LIMITATIONS OF THE STUDY

Our study demonstrates that KRAS^G12D-specific dIgA transcytosis alters the endosomal traffic of mutant KRAS. However, the intricacies of this mechanism remain unclear. One plausible scenario is heterotypic fusion of the early endosomes that internalize dIgA with the recycling endosomes that bring mutant KRAS back to the cell membrane. However, this does not fully explain how dIgA targets deep cytoplasmic oncogenes such as IDH1.

Future studies should also investigate whether mutant tumors that have become resistant to existing small molecule inhibitors (e.g., sotorasib) could be still sensitive to PIGR-mediated interventions. Antibodies designed to recognize the constitutively active (GTP-bound) KRAS structure elicited by different mutations could be particularly effective in this context. Conversely, loss of PIGR could make tumors resistant to dIgA, while remaining sensitive to drug inhibitors.

Similarly, future studies should dissect the mechanisms whereby CD8⁺ T cells participate in the effects of oncogene-specific dIgA, which may depend on the immunogenicity of each disease (e.g., as a result of different tumor burden).

Finally, generating recombinant dimeric IgA is more challenging than generating IgG, which also has longer half-life. Modifications of the Fc domain of IgG to trigger PIGR-mediated transcytosis, while preserving binding to FcRn to ensure persistence need to be investigated, to pave the way for novel immunotherapies against the most frequent and aggressive human cancers.

STAR METHODS

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jose R. Conejo-Garcia (Jose.Conejo-Garcia@Duke.edu).

Materials availability

Antibody producing plasmids generated in this study will be shared with other research groups only upon receipt of reasonable request on a case-by-case basis with a completed Materials Transfer Agreement (MTA).

Data and code availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD042914 and PXD044426. This paper does not report original code. The datasets generated during the current study are available from the lead contact upon reasonable request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Human samples

A lung cancer tissue microarray of total 30 cases including squamous cell carcinoma, adenocarcinoma, adenosquamous carcinoma and bronchioloalveolar carcinoma with matched metastatic carcinoma and cancer adjacent tissue, with available pathology grade and survival data was procured from US Biomax.

Cell lines

OVCAR3 (RRID:CVCL_0465), A427 (RRID:CVCL_1055), SK-LU-1 (RRID:CVCL_0629), HEK293T (RRID:CVCL_0063), NCI-H23 (RRID:CVCL_1547), NCI-H647 (RRID:CVCL_1574) cell lines were purchased from ATCC (Manassas, VA). IDH1^+/+ and IDH1^R132H/+ HCT116 (RRID:CVCL_0291) cell lines were purchased from Horizon discovery. PIGR-ablated OVCAR3 cells was generated in the lab⁴³. Brpkp110 was established in the lab¹⁸ and KPMSH2^KIN is a new model of lung cancer developed also in our laboratories that carries the same epitope recognized in human G12D-mutated KRAS. Cell lines were routinely cultured in R10 (RPMI-1640, 10% FBS, penicillin (100 IU ml⁻¹), streptomycin (100 μg ml⁻¹), L-glutamine (2 mM), sodium pyruvate (0.5 mM)) media (Thermo). HEK293T cell line was routinely cultured in D10 (DMEM, 10% FBS, penicillin (100 IU ml⁻¹), streptomycin (100 μg ml⁻¹), L-glutamine (2 mM), sodium pyruvate (0.5 mM)) media (Thermo).Cell lines were routinely tested for negative mycoplasma contamination.

Animal models

For mouse models, three different strains were used. Female, 4-6 weeks old Rag1-deficient (Rag1^−/−) mice (RRID:IMSR_JAX:003729) and female, 7 weeks old C57BL/6J mice (RRID:IMSR_JAX:000664) were purchased from The Jackson Laboratory. NOD.Cg-Prkdc^scid Il2rg^tmlWjl/SzJ (NSG) mice (RRID:IMSR_JAX:005557), originally procured from The Jackson Laboratory, were maintained by the Moffitt Cancer Center animal facility. All animals were maintained by the animal facility of the Moffitt Cancer Center and Duke School of Medicine, housed in cages of up to 5 mice per cage, in a temperature controlled (18-23°C), 40-60% humidity, 12 h light/dark cycle facility. Animal studies were performed in accordance with Institutional Animal Care and Use Committee at the University of South Florida Research Integrity and Compliance department (IACUC protocols#IS00006655 and #IS00009457) and at the Duke University and Division of Laboratory Animal Resources (IACUC protocol# A011-23-01)

Method details

Constructs, and transduction

Human PIGR-coding sequence was cloned into pLVX-IRES-ZsGreen1 lentiviral vector (Genscript)⁴⁴. Constructs encoding Wild-type or G12D mutated KRAS fused with PAmCherry1 (pLenti-TO/CMV-PAmCherry1-KRAS, G12D or WT)²⁶ were procured from Dr. Xiaolin Nan Lab, Oregon Health & Science University. The sequence integrity and accuracy of all constructs was assessed and confirmed by sequencing services from GeneWiz.

Virus production and transduction

Lentiviral particles were generated by co-transfecting HEK 293T/17 (ATCC) with pLVX-IRES-PIGR-ZsGreen, or pLenti-TO/CMV-PAmCherry1-KRAS (G12D mutated KRAS or WT) and packaging/envelope vector pMD2.G (RRID:Addgene_12259) and psPAX2 (RRID:Addgene_12260) using Lipofectamine-3000 (Invitrogen) and the viral supernatant was harvested 48h after. 5×10⁵ HCT116 or OVCAR3 cells were seeded in a 6-well plate with 3mL of the corresponding 0.45 μ filtered viral supernatant with added 10 μg/mL polybrene (Millipore). The plates were spun for 90min at 32°C and 1200g. 12h after transduction the medium was changed to R10 and 48h later the expression of the target gene was verified under microscope for KRAS transductions and selected by FACS-sorted for ZsGreen expression (PIGR).

Tumor Models

Flank tumors were initiated by injecting 1×10⁷ KRAS^{WT or G12D} transduced OVCAR3 or A427; or 5×10⁶ IDH1^{WT or Mut} HCT116 cells or 5×10⁵ Brpkp110 cells; or 1 × 10⁶ KPMSH2^KIN cells in PBS. SK-LU-1 tumors were initiated by injecting 1×10⁷ cells in PBS admixed with 1:1 Matrigel into flanks.

Tumor volume was calculated as: 0.5 × (L × W²), where L is length and W is width. Intratumoral or peritumoral injections, as well as intraperitoneal injections of antibodies were done on multiple times at 3-4 days interval starting from day 7 for OVACR3, A427, HCT116, Brpkp110 and KPMSH2^KIN tumors or from day 20 for SK-LU-1 tumors after the tumor challenge at a dose of 100μg/20g. MRTX1133 (MedChemExpress, #HY-134813) was injected daily or multiple times at 3-4 days interval starting from day 7 for OVACR3 and A427 or from day 20 for SK-LU-1 tumors at a dose of 200μg/20g. CD8 T cell depletion has been performed using αCD8 antibodies (BioXCell), injected at same frequency of injected antibodies, at a dose of 200μg/20g.

Recombinant antibody production

Anti-KRAS^G12D, and anti-IDH1^R132H antibody variable heavy (VH) and variable light (VL) sequences were obtained from the patent US10844136B2 and CN106957367A, respectively. VH sequences followed by human IgA1 or IgG4 heavy chain constant region sequences in frame were cloned into PBMN-I-GFP vector (RRID:Addgene_1736). VL sequences along with human kappa light chain constant region sequences in frame were cloned into pVITRO1 (Genscript). Vectors, coding for all the VH and VL, were produced by Genscript. J-chain encoding pcDNA3.0 vector was procured from Addgene (RRID:Addgene_145146).

For recombinant dimeric and monomeric KRAS^G12D-IgA1 and KRAS^G12D-IgG4 antibody production, HEK293T cells were retrovirally transduced with antibody heavy chain encoding PBMN-I-GFP vector and sorted for viable GFP⁺ cells, followed by transfection with antibody light chain encoding pVitro1 vector and selected using Hygromycin, and finally those cells were transfected with J-chain encoding pcDNA3.0 vector and selected using G418. Heavy chain plus light chain plus J-chain expressing HEK293T cells were grown in suspension for 5-7 days in Free-Style 293 expression media (Thermo) for dimeric KRAS^G12D-IgA antibody production. Similarly, for monomeric KRAS^G12D-IgA1 or KRAS^G12D-IgG4, HEK293T cells expressing only light chain and respective heavy chain were used. Supernatants were concentrated through a 100 kDa membrane, to eliminate any possible contaminants of unassembled J-chains, heavy chains or light chains, and antibodies were purified using IgA or IgG purification columns (Ligatrap). Dimerization of IgA antibodies were confirmed by detecting J-chain in the elutes through mass spectrometry. All transfections were performed using Lipofectamine 3000 (Thermo). Cell lines were routinely tested for negative mycoplasma contamination. Dimeric and monomeric IDH1^R132H-IgA1, and monomeric IDH1^R132H-IgG4 antibodies and some additional quantity KRAS^G12D-IgG4 antibodies were procured from GenScript.

Fluorescence-activated cell sorting (FACS)

Sorting of antibody heavy chain-transduced HEK293T cells (GFP expressing) were performed by staining with DAPI (Thermo Scientific) viability dye and gated for DAPI⁻GFP⁺ cells. Similarly, PIGR-transduced HCT116 cells were sorted by gating DAPI⁻GFP⁺ cells. Samples were subsequently fluorescence-activated cell sorting (FACS) sorted using BD FACS ARIA.

One hundred thousand KRAS^G12D transduced OVCAR3 cells were placed in 6-well plates. After 12 h wells were washed, and fresh medium was added, and cells were treated with anti-human KRAS^G12D-IgA1 or KRAS^G12D-IgG4 or non-specific human IgA (Abcam, ab91025) at 0.5 μg/ml final concentration or vehicle (PBS) and incubated for 16 h in a 37-°C incubator. Total apoptotic cells (annexin V⁺ propidium iodide^+/−) were analyzed by flow cytometry.

MTT assay

Five thousand OVCAR3 (wild-type or G12D mutated KRAS-transduced or untraduced) or SK-LU-1 cells, or five hundred HEK293T cells were seeded in each well of 96-well plates. After 12 h cells were washed with PBS, and fresh medium was added, and were treated with anti-human KRAS^G12D-IgA1 or non-specific human IgA (Abcam, ab91025) at 0.5 μg/ml final concentration or vehicle (PBS) and the medium was changed after 16 h incubation in a 37-°C incubator to remove the treatment. Proliferation was measure at 3, 5 or 7 days. MTT reagent solution (10ul of 5mg/ml) was added to the media of cultured cells and incubated for 3 hr. The precipitates were dissolved in DMSO, after remove the medium, and absorbance were measured at 570nm. Fold changes were calculated by dividing with the mean absorbance of the replicates of respective vehicle-treated cell lines.

Immunofluorescence and confocal microscopy

Anti-human KRAS^G12D-IgA1 or KRAS^G12D-IgG4 antibodies were conjugated with Alexa Fluor 488-conjugation kit (Thermo). Fifty thousand OVCAR3 (transduced with wild-type or G12D mutated KRAS, fused with PAmCherry) cells were placed onto a coverslip within 6-well plates and after 12 h treated with the conjugated antibodies for different temporal points. PAmCherry was photoactivated using the Leica TCS SP8 DAPI bandpass filter cube (Excitation BP 450-490 nm) using the widest aperture. Exposure time was dependent on the day of the experiment and varied between 20 to 30 seconds and was optimized using positive and negative controls. The same exposure time was used for all treatments. The gain for the Hoechst channel was adjusted to compensate for photobleaching due the exposure time required for photoactivating PAmCherry. Images were acquired with a 40X/1.3 NA objective in a confocal microscope (Leica SP8) using LAS X (v.3.5.5.19976) software. Hoechst, Alexa fluor 488, PAmCherry were excited with laser lines 405 nm, 488 nm, 552 nm respectively. Emission was captured through band pass settings 415–480 nm, 500–525 nm, 592–651 nm, respectively. For the highest resolution images, we used the same parameters but we increased the pixel resolution to 2048 x 2048 and took the images with speed=400 and z=0.5um.

Exocytosis of KRAS^G12D and LC-MS/MS

Five hundred thousand OVCAR3 (transduced with wild-type or G12D mutated KRAS, fused with PAmCherry) cells were placed in 100mm × 20mm cell culture plates and treated with or without anti-human KRAS^G12D-IgA1 or KRAS^G12D-IgG4 or non-specific human IgA (Abcam, ab91025) at 0.5 μg/ml final concentration or vehicle (PBS) in low serum media. After 12 h, conditioned medium was collected and filtered for contaminant debris removal. Proteins were extracted from the conditioned medium, reduced by DTT, digested by trypsin, and subjected to mass spectrometry analysis by the Moffitt Cancer Center Proteomics Facility. MaxQuant (version 1.5.2.8) was used to analyze the data, identify, and quantify the proteins⁴⁵.

Western blot, Native gel electrophoresis and Co-immunoprecipitation (Co-IP)

Cells were lysed in RIPA buffer (Thermo) with protease-phosphatase inhibitor cocktail (CST, #5872S) and cleared by centrifugation. Proteins were quantified by BCA assay (Thermo). Membranes were blotted with recombinantly produced anti-KRAS^G12D-IgA1 or anti-KRAS^G12D-IgG4 or anti-IDH1^R132H-IgA1 or anti-IDH1^R132H-IgG4 antibodies; or commercial anti-mCherry Rockland, #600-401-P16, RRID:AB_2614470), anti-human immunoglobulin heavy alpha chain (Thermo Fisher Scientific, #A18781, RRID:AB_2535558), or anti-human immunoglobulin heavy gamma (plus light) chain (Thermo Fisher Scientific, #A18805, RRID:AB_2535582), or anti-human immunoglobulin kappa chain (R&D Systems, #MAB10050), or anti-human J-chain (Thermo Fisher Scientific, Mc19-9, #MA1-80527, RRID:AB_934333), or anti-human Rab5A (Proteintech, #11947-1-AP, RRID:AB_2269388), or anti-human Rab7A (Proteintech, #55469-1-AP, RRID:AB_11182831), or anti-human Rab8A (Proteintech, #55296-1-AP, RRID:AB_10858398), or anti-human Rab11A (Proteintech, #20229-1-AP, RRID:AB_10666202), or anti-human PIGR (Abcam, #ab96196, RRID:AB_10677612), or anti-human KRAS^G12D (CST, #14429, D8H7, RRID:AB_2728748) or anti-human β-actin (CST, #5125, 13E5, RRID:AB_1903890), antibodies. Lysates of A427 cells treated with KRAS^G12D-specific IgA or MRTX1133 for different time points were probed with antibodies against total ERK1/2 (CST, #9102, RRID:AB_330744) and phosphorylated ERK1/2 (CST, D13.14.4E, #9102, RRID:AB_330744). Blots were then probed with appropriate horse radish peroxidase (HRP)-conjugated secondary antibodies (room temperature ,1-2 hour), unless the primary antibodies are HRP-conjugated, and immunoreactive bands were developed using Enhanced Chemiluminescence (ECL) substrate (GE HealthCare).

For the native gel electrophoresis, 5μg of dimeric KRAS^G12D-IgA1, dimeric IDH^R132H-IgA1, monomeric IDH^R132H-IgA1 and irrIgA were diluted using Tris-Glycine Native Sample Buffer (ThermoFisher, #LC2673) and ran in a native gel, 12% Tris-Glycine Gel (Invitrogen, # XP00120BOX) for 1 hour at 220V. The gel was stained by overnight shaking with Invitrogen^™ Colloidal Blue Staining Kit (Invitrogen, #LC6025). Then, the staining solution was replaced with deionized water, followed by overnight shaking until the gel had a clear background. The gel was scanned in a Canon TR4700 series HTTP.

For co-immunoprecipitation, cells were lysed using non-reducing non-denaturing lysis reagent provided in the co-immunoprecipitation (co-IP) kit (Pierce, Cat#26149) used. Proteins were immunoprecipitated using anti-human PIGR-secretory component (Abcam, SC-05, #ab3924, RRID:AB_2261963) or anti-human Rab5A, or anti-human Rab7A, or anti-human Rab8A, or anti-human Rab11A, or anti-human IgA (Abcam, EPR5367-76, #ab124716, RRID:AB_10976507). Proteins from concentrated conditioned media from anti-human KRAS^G12D-IgA1 or KRAS^G12D-IgG4 or non-specific human IgA or vehicle treated OVCAR3 cells were immunoprecipitated using anti-human PIGR-secretory component. The elutes were ran for western blotting. For analyzing changes in intracellular KRAS^G12D protein level upon antibodies-treatment, lysates of KRAS^G12D-transduced OVCAR3 or A427 cells, treated every 4 hours for a total of 12 hours, were used for western blotting.

Transwell and transcytosis assay for expelling of KRAS^G12D

KRAS^G12D OVCAR3 cells (2×10⁵) were seeded in the upper chamber of a transwell system (Corning^® 3413 Transwell^® 6.5mm Polycarbonate, membrane Inserts Pre-Loaded in 24-Well Culture Plates, Pore Size: 0.4μm) in 200uL. In addition, 1 mL of fresh medium was added in the basal chamber. The cells were cultured in the incubator at 37°C and 5% of CO₂ during 3 days to create a monolayer. KRAS^G12D IgA1 or KRAS^G12D IgG4 or non-specific human IgA antibodies were biotinylated using a Biotinylation Kit (Fast, Type B) by following the manufacture’s protocol. Then, 1μg of respective biotinylated antibody was added to the upper chamber. After overnight incubation, the supernatant from the upper and basal chambers were collected into separate microcentrifuge tubes and subjected to streptavidin pull-down using Dynabeads MyOne streptavidin T1. Briefly, 10μl of beads per sample were washed according to the manufacture’s recommendation, added to each sample and then incubated overnight at 4°C with head-over rotation. The beads were collected my magnetic separation for 5 min and washed 3 times in PBS containing 0.1% BSA. Finally, the beads were resuspended in 100uL of PBS and used for performing dot blots.

For dot blots, 10μL of beads were blotted onto a nitrocellulose membrane and dried at room temperature 1-2 hours. Then, membranes were blocked using 5% BSA in TBS-T with a gentle agitation for 1 hour at room temperature. Appropriate primary antibodies (αhuman-IgA-HRP, αhuman-IgG-HRP and αKRAS^G12D) were added immediately and incubated 1 hour at room temperature with gentle rotation. After 3 washes, blots were directly chemiluminescent-developed when probed with HRP conjugated antibodies αhuman-IgA/IgG antibodies, or first incubated with a secondary antibody and developed.

Multiplex immunohistochemistry

FFPE tissue microarray was immunostained using the PerkinElmer OPAL TM Automation IHC kit (Waltham, MA) on the BOND RX autostainer (Leica Biosystems, Vista, CA) and the following anti-human antibodies: IgA (Abcam, EPR5367-76, ab124716, RRID:AB_10976507; 1:1000), IgG (Abcam, EPR4421, ab109489, RRID:AB_10863040; 1:500), PIGR (Abcam, ab96196, RRID:AB_10677612; 1:100), CD3 (Thermo Fisher Scientific; Cat# MA5-14524, Clone SP7; RRID: AB_10982026, 1:100), and pan-cytokeratin (PCK, Dako, AE1/AE3, M3515, 1:200). Nuclei were stained with DAPI. Precisely, tissues were baked at 65°C for 2 hours then transferred to the BOND RX (Leica Biosystems) followed by automated deparaffinization, antigen retrieval using OPAL IHC procedure (PerkinElmer). Autofluorescence slides (negative control) were included, which use primary and secondary antibodies omitting the OPAL fluors. Slides were scanned and imaged with the PerkinElmer Vectra^®3 Automated Quantitative Pathology Imaging System. Multi-layer TIFF images were exported from InForm (PerkinElmer) and loaded into HALO (Indica Labs, New Mexico) for quantitative image analysis. Each fluorescent fluorophore is assigned to a dye color and positivity thresholds were determined per marker based on published nuclear or cytoplasmic staining patterns. Tumor islets and stroma were distinguished by PCK staining.

The Cancer Genome Atlas (TCGA) data analysis

Molecular data for PIGR mRNA expression (RNA Seq V2 RSEM) from TCGA, Firehose Legacy for several human cancer types were analyzed and graphically exported from the cBio Cancer Genomics Portal (http://www.cbioportal.org/).

Quantification and statistical analysis

Unless mentioned otherwise all data are presented as mean ± SEM. Two-tailed t-tests (unpaired and paired, as appropriate) were performed between two groups, and one-way ANOVA were performed for comparisons between more than two groups, unless indicated otherwise. A significance threshold 0.05 for P values was used. Analyses were carried out in Graph Pad Prism (v.9.0).

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
non-specific human IgA	Abcam	Abcam Cat#ab91025
anti-β-actin antibody	Cell Signaling Technology	Cell Signaling Technology Cat# 5125, RRID:AB_1903890
anti-Phospho-p44/42 MAPK(ERK1/2) antibody	Cell Signaling Technology	Cell Signaling Technology Cat# 4370, RRID:AB_2315112
anti-p44/42 MAPK(ERK1/2) antibody	Cell Signaling Technology	Cell Signaling Technology Cat# 9102, RRID:AB_330744
anti-RAS (G12D Mutant Specific) (D8H7) antibody	Cell Signaling Technology	Cell Signaling Technology Cat# 14429, RRID:AB_2728748
anti-mCherry antibody	Rockland	Rockland Cat# 600-401-P16, RRID:AB_2614470
Goat anti-Human IgG (H+L) Secondary Antibody, HRP	Thermo Fisher Scientifi	Thermo Fisher Scientific Cat# A18805, RRID:AB_2535582
Goat anti-Human IgA Secondary Antibody, HRP	Thermo Fisher Scientific	Thermo Fisher Scientific Cat# A18781, RRID:AB_2535558
anti-Human Ig Kappa Light Chain Antibody	R&D Systems	R&D Systems Cat#MAB10050
anti-J Chain Antibody	Thermo Fisher Scientific	Thermo Fisher Scientific Cat# MA1-80527, RRID:AB_934333
anti-PIGR antibody	Abcam	Abcam, ab96196, RRID:AB_10677612
InVivoMAb anti-mouse CD8β (Lyt 3.2)	Bio X Cell	Bio X Cell Cat# BE0223, RRID:AB_2687706
InVivoPlus rat IgG1 isotype control (anti-HRP)	Bio X Cell	Bio X Cell Cat# BE0088, RRID:AB_1107775
anti-Secretory Component antibody	Abcam	Abcam Cat# ab3924, RRID:AB_2261963
anti-IgA antibody	Abcam	Abcam Cat# ab124716, RRID:AB_10976507
anti-IgG antibody	Abcam	Abcam Cat# ab109489, RRID:AB_10863040
Mouse anti-human CD3 antibody	ThemoFisher Scientific	Thermo Fisher Scientific;Cat# MA5-14524, Clone SP7; RRID: AB_10982026
Mouse anti-human PCK antibody	DAKO	Agilent Cat# M3515, Clone AE1/AE3; RRID: AB_2132885
anti-RAB8A antibody	Proteintech	Proteintech Cat# 55296-1-AP, RRID:AB_10858398
anti-RAB5A antibody	Proteintech	Proteintech Cat# 11947-1-AP, RRID:AB_2269388
anti-RAB11A antibody	Proteintech	Proteintech Cat# 20229-1-AP, RRID:AB_10666202
anti-RAB7A antibody	Proteintech	Proteintech Cat# 55469-1-AP, RRID:AB_11182831
Bacterial and Virus Strains
One Shot™ TOP10 Chemically Competent E.coli	ThermoFisher Scientific	Cat# C404010
Biological Samples
Lung tumor tissue microarray (TMA)	US BioMax, Inc.	LC953
Chemicals, Peptides, and Recombinant Proteins
RPMI 1640 Medium	Gibco™	Cat# 11875093
Penicillin/Streptomycin	Lonza	Cat# 17602E
L-glutamine	Genesee Scientific	Cat# 25509
Sodium pyruvate	ThermoFisher Scientific	Cat# 11360070
0.25% Trypsin-EDTA	Gibco™	Cat# 25200056
Dulbecco's Phosphate Buffered Saline 1X	VWR Life Science	Cat# 0201190500
DMEM Medium	Thermo Fisher Scientific	Cat# 11965092
POLYBRENE, 10MG/ML AQUEOUS SOLUTION	Thermo Fisher Scientific	Cat# AB01643-00001
Lipofectamine™ 3000 Transfection reagent	Invitrogen	Cat# L3000015
DAPI	Sigma	Cat# 8417
RIPA Lysis and Extraction Buffer	Thermo Fisher Scientific	Cat#89900
Protease/Phosphatase Inhibitor Cocktail (100X)	Cell Signaling	Cat#5872
Thiazolyl Blue Tetrazolium Bromide, MTT	Millipore Sigma	Cat#M2128
Hoechst 33342 Solution	Thermo Fisher Scientific	Cat#62249
MRTX1133	MedChemExpress	Cat#HY-134813
Tris-Glycine Native Sample Buffer	Invitrogen	Cat#LC2673
Novex™ WedgeWell™ 12% Tris-Glycine Gel	Invitrogen	Cat# XP00120BOX
Invitrogen™ Colloidal Blue Staining Kit	Invitrogen	Cat#LC6025
ECL™ Prime Western Blotting System	Millipore Sigma	GERPN2232
Dynabeads™ MyOne™ Streptavidin T1	Invitrogen	Cat#65601
InVivoPure pH 7.0 Dilution Buffer	BioXCell	Cat#IP0070
Critical Commercial Assays
Biotinylation Kit / Biotin Conjugation Kit (Fast, Type B) - Lightning-Link®	abcam	Cat#ab201796
Alexa Fluor® 488 Conjugation Kit (Fast) - Lightning-Link®	abcam	Cat#ab236553
Pierce™ BCA Protein Assay Kit	Thermo Fisher Scientific	Cat#23225
Pierce™ Co-Immunoprecipitation Kit	Thermo Fisher Scientific	Cat#26149
APC Annexin V, 100 test	Biolegend	Cat#640920
LigaTrap® Human IgG Purification Kit Product Instructions	LigaTrap Technologies	Cat#LT-095KIT
LigaTrap® Human IgA Purification Kit, 10 x 0.10 mL	LigaTrap Technologies	Cat#LT-146KIT
Deposited Data
Mass spectrometry proteomics data	This paper	PRIDE:XXX
Experimental Models: Cell Lines
OVCAR3	ATCC	(RRID:CVCL_0465)
A427	ATCC	(RRID:CVCL_1055)
SK-LU-1	ATCC	(RRID:CVCL_0629)
HEK293T	ATCC	(RRID:CVCL_0063)
NCI-H23	ATCC	(RRID:CVCL_1547)
NCI-H647	ATCC	(RRID:CVCL_1574)
IDH1+/+ HCT116	Horizon Discovery	(RRID:CVCL_0291)
IDH1R132H/+ HCT116	Horizon Discovery	(RRID:CVCL_0291)
Brpkp110	PMID: 27803104	N/A
KPMSH2KIN	This manuscript	N/A
PIGR-ablated OVCAR3	PMID: 33536615	N/A
KRASG12D-PamCherryOVCAR3	This manuscript	N/A
KRASwt-PamCherryOVCAR3	This manuscript	N/A
Experimental Models: Organisms/Strains
NOD.129S7(B6)-Rag1tm1Mom/J	The Jackson Laboratory	RRID:IMSR_JAX:003729
C57BL/6J mice	The Jackson Laboratory	RRID:IMSR_JAX:000664
NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ	The Jackson Laboratory	RRID:IMSR_JAX:005557
Recombinant DNA
pLVX-IRES-Zs	GreenTakara Bio	Cat# 632187
pLenti-TO/CMV-PAmCherry1-KRASWT	PMID: 26080442	N/A
pLenti-TO/CMV-PAmCherry1-KRASG12D	PMID: 26080442	N/A
pMD2.G	Addgene	Cat# 12259; RRID: Addgene_12259
psPAX2	Addgene	Cat# 12260; RRID: Addgene_12260
J-chain encoding pcDNA3.0 vector	Addgene	RRID:Addgene_145146
anti-IDH1R132HVH-IgA1 encoding PBMN-I-GFP	Genscript	N/A
anti-IDH1R132HVH-IgG4 encoding PBMN-I-GFP	Genscript	N/A
anti-IDH1R132HVL-Kappa encoding pVitro1	Genscript	N/A
anti-KRASG12DVH-IgA1 encoding PBMN-I-GFP	Genscript	N/A
anti-KRASG12DVH-IgG4 encoding PBMN-I-GFP	Genscript	N/A
anti-KRASG12DVL-Kappa encoding pVitro1	Genscript	N/A
Software and Algorithms
FlowJo v10.7.2	FlowJoLLC	N/A
GraphPad Prism v9	GraphPad Software Inc.	N/A
Adobe Photoshop 2022	Adobe	N/A
Adobe Illustrator 2022	Adobe	N/A

HIGHLIGHTS.

• Dimeric IgA undergoes bona fide transcytosis through PIGR⁺ carcinoma cells.

• Dimeric IgA specifically neutralizes mutated oncogenes inside tumor cells.

• Oncodrivers are expelled outside tumor cells through altered endosomal trafficking.

• Mutation-specific dimeric IgA specifically abrogates tumor growth in vivo.