In vivo Auto-tuning of Antibody-Drug Conjugate Delivery to Maximize Efficacy using High-Avidity, Low-Affinity Antibodies

Abstract

Antibody-drug conjugates (ADCs) have experienced a surge in clinical approvals in the past few years. Despite this success, a major limitation to ADC efficacy in solid tumors is poor tumor penetration, which leaves many cancer cells untargeted. Co-administration of unconjugated antibody can improve tumor penetration and increase efficacy when target receptor expression is high. However, it can also reduce efficacy in low-expression tumors where ADC delivery is limited by insufficient cellular uptake. This creates an intrinsic problem because many patients express different levels of target between and within tumors. Here, we show how unconjugated High-Avidity, Low-Affinity (HALA) antibodies can automatically tune the cellular ADC delivery to match the local expression level. Using HER2 ADCs as a model, the tumor distribution of trastuzumab emtansine and trastuzumab deruxtecan co-administered with the HALA antibody was improved in vivo, translating to equal or greater ADC efficacy across a range of HER2 expression levels. Furthermore, Fc-enhanced HALA antibodies elicited a strong response in an immunocompetent mouse model. These results demonstrate that HALA antibodies can expand treatment ranges beyond high-expression targets and leverage strong immune responses.

Introduction

Antibody-drug conjugates (ADCs) combine potent cytotoxic payloads with monoclonal antibodies that bind tumor-associated cancer antigens, thereby concentrating the payload within the tumor to increase the therapeutic window. ADCs have experienced a recent surge in clinical approvals, with seven drugs approved in the past five years including several for solid tumors. One of the major limitations of this drug class in solid tumors is limited tissue penetration, leaving many cancer cells with no treatment.(1–5) The newest generation of ADCs in solid tumors partly relies on higher antibody dosing enabled by reduced potency, such as Trodelvy (10 mg/kg twice every three weeks) and T-DXd (5.4 or 6.4 mg/kg every three weeks), which can help overcome limited tissue penetration.(2) Due to their large molecular weight, ADCs and other antibody-based biologics slowly diffuse through tissues, and rapid binding to the target antigen immobilizes the ADC near blood vessels, creating the “binding site barrier.”(4,6) The heterogeneous distribution of antibodies and ADCs has been documented extensively in preclinical models (1–3,7) and, more recently, these trends have also been demonstrated in clinical tumors.(8,9)

There are multiple approaches to overcome this perivascular saturation front, including small molecule formats,(10,11) reduced affinity,(12) and paratope blocking.(13) We have explored the use of co-administering an unconjugated “carrier” antibody to partially block ADC binding, thereby driving the ADC deeper into the tissue. One advantage of this approach is that the additional antibody dose can leverage antibody-intrinsic mechanisms of action, such as receptor signaling blockade and/or Fc-effector function.(14,15) The use of a carrier dose results in a greater number of tumor cells receiving treatment, and several high-expression models have illustrated improved efficacy with this approach.(16–18) Importantly, the amount of drug delivered to cells on average should be tailored to the potency of the payload to improve efficacy. Thus, a ratio of carrier dose to ADC that is too high can reduce efficacy by lowering ADC delivery to cells below a lethal threshold.(19) Similarly, the use of a carrier dose in low-expression systems, where there are fewer receptors to internalize the ADC, has shown mixed results on ADC efficacy.(17,18,20) This creates a problem because patients commonly have variable expression between lesions (e.g., the primary tumor and metastases) and within the same lesion.(21,22) Therefore, the optimal carrier dose does not solely rely on the average expression for a given patient.

To overcome this limitation, we aimed to improve the carrier dose approach by using an antibody that selectively competes with ADC binding on high-expression cells, maximizing tissue penetration and efficacy in these regions, but does not compete on low-expression cells, maximizing ADC uptake and efficacy for these cells. The concept of High-Avidity, Low-Affinity (HALA) antibodies to achieve this goal has been explored theoretically(23) and could be used to expand the use of carrier doses to benefit a wide range of expression levels. HALA antibodies rely on differences in avidity (modeled previously(24–27)), allowing the HALA antibody to compete with the high-affinity ADC to drive better tissue penetration and efficacy on high-expression cells (Figure 1A–1B) while preventing blocking and maximizing ADC uptake on low-expression cells (Figure 1C–1D). Therefore, the HALA and ADC mixture ‘auto-tunes’ the delivery of ADC to each cell to achieve maximum tumor efficacy.

Figure 1. Schematic of HALA design approach to improve in vivo ADC distribution and efficacy and in vitro identification of engineered HALA antibody.

(A) In high-expression tumors without a carrier dose, the ADC will reach a subset of cell layers beyond the blood vessel, limiting efficacy. (B) The addition of a carrier dose (either a high affinity Ab or, as shown, a HALA Ab) can partially block binding sites, increasing tissue distribution and efficacy. (C) However, in a low-expression tumor, a high-affinity carrier dose competes with the ADC and prevents uptake into low-expression cells, reducing efficacy. (B, D) The use of a HALA carrier dose in (B) high-expression systems results in competition with the ADC due to high-avidity interactions, but such competition is not observed in (D) low-expression systems due to ‘auto-tuning’ the competition in situ. Therefore, the HALA antibody intrinsically maximizes efficacy in both high- and low-expression systems. (E) Multiple HALA variants were tested for the optimum affinity on both high-expression and low-expression HER2 cell lines. The S12 HALA variant maintained effective competition on high-expression NCI-N87 cells but not on low-expression MDA-MB-231 cells. (F) The HER2 binding of bivalent S12 HALA Ab (in the absence of an ADC) was high on both high-expression HCC-1954 and low-expression MDA-MB-231 cell lines, but the Fab arm has a much lower intrinsic affinity. The figure was prepared using Biorender.

ADC doses are typically limited by the amount of payload administered because the corresponding antibodies are often well tolerated,(28) allowing high HALA doses relative to the ADC that can improve tissue penetration, leverage Fc effector functions(15), and inhibit cancer signaling pathways (e.g. HER2). Given the effect of binding affinity on immune activity,(29–31) we tested the HALA antibodies with two FDA-approved ADCs, Kadcyla (T-DM1) and Enhertu (T-DXd), in immunodeficient mouse models and a trastuzumab-PBD ADC (T-PBD) in an immunocompetent model. Despite differences in payload potency, bystander effects, dosing, expression levels, and mouse models(32)^,(33), we demonstrate in this work that all ADCs had equal or greater efficacy when co-administered with the HALA antibody, highlighting the ability to enhance multiple mechanisms of action.

MATERIALS AND METHODS

Sorting of high avidity, low affinity (HALA) trastuzumab mutants

The trastuzumab antibody fragment was displayed on the surface of yeast along with a library of mutant sequences with several mutations per antibody. For sequence generation, a heavy chain complementarity determining region (HCDR) focused library for trastuzumab was made in single chain Fab (scFab) format as described previously(34) to isolate trastuzumab mimics with high avidity, low affinity. We targeted sites 53 and 56 in HCDR2 and 96, 97, 98, 99, 100, 100A and 100B in HCDR3. The library DNA for scFab was assembled by PCR, transformed with modified pCTCON2 plasmid in yeast strain EBY100 by homologous recombination as described.(35) The designed library size was ~10⁷ and we achieved >10⁸ transformants.

To enrich the library for antibodies with affinities in the desired range, we selected clones that were able to bind a bivalent antigen (HER2-Fc fusion) at a similar level as wild type trastuzumab but had reduced binding to monovalent HER2. The library was sorted against bivalent HER2-Fc antigen immobilized on dyna beads by magnetic activated cell sorting (MACS) in round 1 and 2. In round 1, 10⁹ yeast cells displaying library were incubated with 10⁷ beads coated with bivalent HER2-Fc for 3 hours at room temperature in 1% milk or 1% serum in PBS+0.1% BSA (PBSB). Post incubation the beads and unbound cells were separated by magnetic separation.(35) The beads were washed once with ice-cold PBSB and then re-suspended in 50 mL SDCAA growth media and grown overnight at 30°C. In round 2, MACS was performed as described for round 1 but starting with 10⁷ yeast cells and 10⁷ beads.

In round 3, the libraries were sorted by florescence activated cell sorting (FACS) against monovalent HER2 fused to Fc domain of human IgG1 (HER2-mFc). 10⁷ yeast cells were incubated with 100 nM of HER2-mFc, 1/1000x dilution of mouse anti-myc antibody and incubated in either 1% milk or 1% mouse serum in PBSB for 3 hours at room temperature. The cells were then centrifuged, washed once with ice-cold PBSB followed by incubation with secondary reagents; 1/100x dilution of goat anti-mouse Alexa Fluor 488 and 1/300x dilution of goat anti-human Fc Alexa Fluor 647 on ice for 4 mins. Post incubation, the cells were washed once with ice-cold PBSB and sorted on Beckman Coulter Astrios sorter. Approximately 40 individual clones were tested on yeast (Supplemental Figure S1), and 12 clones were screened for binding competition on mammalian cells. The characterized clones S12, S18, and M11 were then sequenced (Supplemental Figure S1C).

Tuning HALA antibody binding affinity

The clones were reformatted as IgGs and serially diluted in PBS with 0.5% BSA starting from 300nM and added to 96 well u-bottom plates at 100μl/well. HCC1954 and MDAMB231 cells were trypsinized, resuspended in RPMI1640, spun down, washed in PBS-BSA, and added at 50,000 cells/well in 100μl. Plates were incubated with the antibody on ice for 4 hours. After incubation, cells were washed with PBS-BSA, labeled with Alexa anti-human Fc (Jackson Immuno 109-605-190) at a 300x dilution for 4 min on ice, washed in PBS-BSA, and run on flow cytometry (Bio-Rad Ze5). Data was plotted using Prism (GraphPad, RRID:SCR_002798).

Fab fragments were concentrated to ~2mg/ml in 100 μl PBS and labeled with 2 μl of 10 mg/ml Alexa-647 (Fisher Scientific A37573), reacted for 1 hr, and purified with 10% P6 Biogel (Bio-Rad 1504130) in PBS loaded to a costar spin-x tube (Corning 07-200-387). The protein concentration and degree of labeling were then measured on the nanodrop. The affinity of the fragments was measured as above.

Cell culture and animals

NCI-N87 (RRID:CVCL_1603, purchased 2015), HCC1954 (RRID:CVCL_1259, purchased 2017), MDA-MB-453 (RRID:CVCL_0418, purchased 2021), MDA-MB-231 (RRID:CVCL_0062, purchased 2013), CAPAN-1 (RRID:CVCL_0237, purchased 2023), and E0771 (RRID:CVCL_GR23, purchased 2020) cell lines were obtained from ATCC and maintained according to ATCC guidelines. Cell identity was authenticated by the vendor and cells were tested annually for Mycoplasma using Mycoalert Testing Kit (Thermo Fisher Scientific). Cells were passaged 2–3 times per week up to passage 50 using RPMI 1640 media (NCI-N87, HCC1954) or DMEM (MDA-MB-453, MDA-MB-231) each supplemented with 10% (v/v) FBS, 50 U/mL penicillin and 50 μg/mL streptomycin. Cells were incubated at 37 °C in 5% CO₂. 6 to 8 week old female nude mice were obtained from Jackson Labs and all studies were carried out according to and with approval of University of Michigan IACUC and AALAC guidelines.

In vitro HALA and trastuzumab binding competition

Trastuzumab was fluorescently labeled with AlexaFluor 647 with NHS-ester chemistry as described previously.(36) HCC-1954, NCI-N87, MDA-MB-453, and MDA-MB-231 cells were seeded at 300,000 cells/well in a 24 well plate and allowed to adhere overnight. The following day, media was replaced with 500 μL containing 5 nM fluorescent trastuzumab and 40 nM unlabeled HALA antibody, 5 nM trastuzumab and 40 nM unlabeled trastuzumab, or 5 nM fluorescent trastuzumab only for an 8-hour incubation at 37 °C. (These resulted in an 8:1 ratio, where the carrier dose/concentration is always listed first with the ADC or ADC surrogate antibody listed second for all experiments.) Cells were then washed twice with PBS and detached from wells using Trypsin. Cells were washed again in PBS supplemented with 0.5% bovine serum albumin (BSA), then analyzed on an Attune NxT flow cytometer. Each condition was completed in duplicate and biological replicates were performed in triplicate. The percent blocking was calculated by subtracting autofluorescence from negative samples and was calculated by: (signal from trastuzumab only – signal with blocking Ab)/signal from trastuzumab only. Data was plotted using Prism (GraphPad).

Cell viability assays

Cell viability assays were performed using NCI-N87 cells and MDA-MB-453 cells seeded at 6,000 cells/well in a black-wall, clear bottom 96 well plate and allowed to adhere overnight. Media was replaced daily with serial dilutions of T-DM1, S12, or Trastuzumab, or a combination of increasing T-DM1 concentration and carrier antibody (S12 or trastuzumab) maintaining a total antibody concentration of 10 nM. Media with antibody or ADC solutions was replaced daily for 6 days. On the final day, all media was removed and replaced with 10X PrestoBlue viability reagent diluted in complete media. Cells were incubated with PrestoBlue for two hours before scanning the plate on a BioTek Plate reader with 560/590nm excitation/emission. IC50 curves were fit in Prism (GraphPad).

3D tissue penetration in tumor organoids

Tumor organoids were grown as described previously.(10,37) Briefly, HCC1954 or MDA-MB-231 cells were seeded at 3,000 cells/microwell and maintained by replacing media every other day. A week after seeding, spheroids were incubated with a mixture of 50 nM trastuzumab labeled with AlexaFluor 647 (AF647) and 400 nM M10, S12, or S18 HALA antibodies labeled with AlexaFluor AF750 (AF750) for an 8:1 ratio of carrier dose antibody to trastuzumab (ADC surrogate). HCC1954 spheroids were harvested and flash frozen in OCT, then sectioned on a cryostat. Spheroids were stained for 1 minute with Hoechst 33342 to mark all cell nuclei and imaging was performed on an Olympus FV1200 confocal microscope using the 405 nm, 635 nm, and 750 nm lasers. Image analysis was performed in Fiji (RRID:SCR_002285) and Euclidean distance mapping of spheroids was completed in MATLAB (RRID:SCR_001622).

Low HER2 expression MDA-MB-231 spheroids were digested into a single cell suspension by incubating n=20 spheroids of each condition with 0.05% trypsin-EDTA until spheroids were broken into single cells. The single cells were washed twice and passed through a 40 μm filter to remove cells clumps. Cells were analyzed using an Attune NxT flow cytometer to determine the amount of blocking based on reduction of trastuzumab-AF647 signal. Data shown is the result of three biological replicates consisting of n=20 spheroids for each independent experiment.

In vivo distribution studies

Tumor-bearing nude mice were injected with fluorescent ADC and HALA antibody when tumors reached 250 – 300 mm³. Mice were sacrificed 24 hours after injection and tumors were resected for single-cell digests or embedded in OCT and flash frozen in isopentane for histology. Frozen tumors for histology were sectioned using a cryostat into 12 μm slices. Sections were stained with CD31-AF555 (RRID:AB_312897) for 30 minutes in PBS-BSA, then washed with PBS before imaging. Microscopy was performed using a 20x objective on an Olympus FV1200 microscope using 405 nm, 543 nm, and 635 nm lasers and images were analyzed in Fiji.

In vivo efficacy studies

For the efficacy study with T-DM1, nu/nu mice (Jackson Labs) were injected with 5 million cells of either MDA-MB-453 (moderate HER2 expression) or NCI-N87 (high HER2 expression) into the left flank with 50% v/v Matrigel. Tumor volume was measured using calipers and the formula length² × width/2 where length is the shorter dimension of the tumor. Once tumors reached a volume of approximately 250 mm³, tumor bearing mice were administered one of the following treatments via tail vein injection: 1) PBS vehicle control, 2) 3.6 mg/kg T-DM1, 3) 3.6 mg/kg T-DM1 and 28.8 mg/kg S12 HALA Ab (8:1 ratio of carrier antibody to ADC), 4) 3.6 mg/kg T-DM1 and 28.8 mg/kg Trastuzumab, or 5) 28.8 mg/kg S12 HALA Ab. Tumors were monitored twice per week for MDA-MB-453 tumors and three times per week for NCI-N87 tumors until the humane endpoint. Data were plotted using PRISM (GraphPad).

For the efficacy study with T-DXd, 5 million cells of CAPAN-1 (low HER2 expression) or NCI-N87 were injected to each nude mouse, and tumor volumes was measured as previously described. Once tumors reached a volume of approximately 250 mm³, CAPAN-1 and NCI-N87 tumor bearing mice were administered one of the following treatments via tail vein injection: 1) PBS vehicle control, 2) 6.4 mg/kg T-DXd, 3) 6.4 mg/kg T-DXd and 32 mg/kg S12 HALA Ab (1:5 ratio), 4) 6.4 mg/kg T-DXd and 32 mg/kg Trastuzumab, or 5) 32 mg/kg S12 HALA Ab with n=8 mice for each group. Tumors were monitored once every three days until the humane endpoint. Data were plotted using PRISM (GraphPad).

ADCC assays

ADCC cells (Invivogen, jktl-nfat-cd16, purchased 2022) were cultured in IMDM media (Fisher 12440053) with 10% Heat-Inactivated (HI) FBS, 1% P/S and 100 μg/ml Normocin (Invivogen ant-nr-1). Cell identity was authenticated by the vendor. 10 μg/ml of Blasticidin (Invivogen ant-bl-05) and 100 μg/ml of Zeocin (Invivogen ant-zn-05) were added to the growth media every passage. HCC-1954 or MDA-MB-231 cells were seeded to 96 well Tissue culture-treated flat-bottom plates as 50,000 cells/well and incubated at 37°C and 5% CO2 overnight. Antibodies were diluted as a 3-fold series starting from 100 nM, blocking antibodies were added as the required ratio if needed. After adding 110 μl of diluted antibody to each well, cells were incubated at 37°C and 5% CO₂ for 1 hour. ADCC reporter cells were suspended in test medium (IMDM media with 10% HI FBS, 1% P/S), added to the wells (0.2 million cells in 90 μl) and incubated at 37°C and 5% CO₂ for 6 hours. After incubation, 20 μl per well was placed in a white assay plate (Fisher 720091), 50 μl of QUANTI-Luc^™ (Invivogen REPQLC2) was added per well, and luminescence was measured immediately.

Syngeneic mouse model and treatment

Efficacy studies were also conducted in an immunocompetent syngeneic mouse model. C57BL/6 hHER2 transgenic mice were obtained from Jackson Laboratory, and the breeding colony was maintained by the Unit for Laboratory Animal Medicine (ULAM) at the University of Michigan. 6–8 weeks old hHER2 transgenic mice were injected with 5 × 10⁶ cells of human HER2 (hHER2) transfected E0771 (E0771-hHER2) into the 4^th mammary fat pad with 50% v/v Matrigel. Tumor volume was measured as mentioned above. Once the tumor volume reached approximately 250 mm³, E0771-hHER2 tumor bearing mice were administered one of the following treatments via tail vein injection: 1) no treatment, n=10, 2) 3 mg/kg of S12 eFc HALA antibody, n = 10, 3) 1.1 mg/kg LALAPG-T-PBD (to match the payload concentration), n=10, 4) 1 mg/kg T-PBD, n=15, 5) 1 mg/kg T-PBD and 3 mg/kg S12 HALA antibody, n=13, and 6) 1 mg/kg T-PBD and 3 mg/kg S12eFc HALA antibody, n=15. Tumors were measured every other day until the humane endpoint. Tumors that were not measurable and did not regrow by day 60 were considered complete responders. Data were plotted using PRISM (GraphPad).

Data and materials availability:

Sequences for the characterized clones are provided in Supplementary Figure S1. The other data generated in this study are available upon request from the corresponding author.

RESULTS

Generation and In vitro Binding Competition of HALA Antibodies

HALA antibodies were generated using a mutational library of trastuzumab on the surface of yeast, and lead clones were reformatted as IgGs. Binding competition between the HALA antibody panel and trastuzumab was measured in low (MDA-MB-231), medium (MDA-MB-453), and high (NCI-N87, HCC-1954) HER2 expression cell lines (Figure 1E, Supplementary Figure S2). Expression levels of cell lines are provided in Table 1.(38) A ratio of 8:1 for the competitor to ADC was selected based on the known ratio for efficient tissue penetration in vivo with a clinically relevant ADC dose of trastuzumab emtansine.(16,23) From these results, the S12 clone was identified as the lead compound based on blocking in high-expression systems (similar to the 90% blocking with trastuzumab as the carrier dose) with minimal blocking in low-expression cells (only 35% binding despite an 8-fold higher concentration).

Table 1:

HER2 receptor expression measured for CAPAN-1, MDA-MB-231, EO771-hHER2, MDA-MB-453, NCI-N87, and HCC-1954 (Mean and SD).

Cell line	Measured HER2 Expression (receptors/cell)
CAPAN-1	2.36×104 ± 1,900
MDA-MB-231	3.5×104 ± 320
E0771-hHER2	2.5×105
MDA-MB-453	3.76×105 ± 4,200
NCI-N87(38)	2.1×106
HCC-1954	4.4×106 ± 37,000

Binding affinity measurements were acquired with IgG and Fab fragments to measure bivalent and monovalent binding, respectively, on the cell surface (Figure 1F). In a high-expression system, the S12 HALA clone shows a K_D of 0.7 nM but only 46 nM when measured using the monovalent Fab arm, demonstrating high avidity. Notably, the avidity remains high even on a low-expression cell line. The HALA antibody is only outcompeted on the cell surface when the higher affinity competitor is added, which is consistent with theory.(23)

To further confirm that the HALA antibody was able to effectively block the binding of T-DM1 ADC in an expression-dependent manner, cell viability assays were conducted by co-incubated trastuzumab or S12 HALA Ab with T-DM1 or T-DXd (Supplemental Figure S2). S12 HALA Ab with T-DM1 resulted in a similar level of blocking as trastuzumab on a high-expression cell line, while S12 HALA Ab resulted in less blocking and more cell killing on the low-expression MDA-MB-453 cells, as desired. Likewise, the addition of either trastuzumab or S12 antibody lowered the in vitro efficacy of T-DXd when varying the ratio of antibody to ADC at a total constant antibody level on NCI-N87 cells, consistent with competitive binding. For low expression CAPAN-1 cells, the trastuzumab reduced efficacy of T-DXd at a ratio of 3:1 and 5:1 while the efficacy was similar to T-DXd alone with a 3:1 or 5:1 ratio of S12 antibody, as designed. NCI-N87 and MDA-MB-453 cell lines had modest responses to trastuzumab alone (Supplemental Figure S3).

HALA Antibody Improves 3D Tissue Penetration in Tumor Spheroids

To evaluate whether the binding competition on cell monolayers in vitro would translate to improved tissue penetration, we imaged the distribution of trastuzumab (as a surrogate for T-DM1 with no confounding cell killing) with or without 8:1 concentrations of HALA antibodies in high HER2 expression tumor spheroids (HCC-1954). Trastuzumab alone results in peripheral spheroid uptake due to poor tissue penetration, similar to T-DM1 (Figure 2A). In contrast, the addition of an 8:1 ratio of S12 HALA antibody partially blocks binding in the periphery, forcing deeper penetration of trastuzumab into the spheroid (Figure 2B). The M11 clone had even lower competition on the low-expression MDA-MB-231 cell line compared to S12 HALA Ab (Figure 1E), but the competition was modest on the high-expression cell line. When co-administered, trastuzumab remains bound at the periphery while the high concentration of M11 diffuses deeper into the tissue (Figure 2C), indicating the avidity effect is not sufficient to improve tissue penetration below a certain affinity.

Figure 2. S12 HALA antibody ‘autotunes’ binding and distribution of trastuzumab in tumor spheroids.

(A) Trastuzumab (green; 50 nM) shows limited penetration into high HER2-expressing HCC-1954 spheroids (nuclei stained with Hoechst, blue). (B) The addition of the HALA antibody (8:1 S12 HALA Ab:trastuzumab, red) to trastuzumab (50 nM, green) significantly enhanced trastuzumab tissue penetration. (C) In contrast, lower blocking HALA variants (e.g., M11 HALA clone) did not have sufficient affinity to compete with trastuzumab, resulting in only peripheral targeting of trastuzumab (green) despite M11 penetration to the spheroid center (red). (D) With low-expression spheroids, clones with higher blocking than clone S12 (e.g., clone S18) reduce cellular uptake. Beneficially, S12 HALA Ab shows low levels of competition with trastuzumab for low HER2-expression spheroids. Therefore, S12 HALA Ab has sufficient binding to (B) improve tissue penetration in high-expression spheroids while (D) maintaining low blocking in low-expression spheroids. Scale bar = 100 μm. Unpaired, two-sided t-tests were used (** p<0.001).

Due to the low HER2 expression and full antibody tissue penetration for the MDA-MB-231 cell line, MDA-MB-231 spheroids were digested into single-cell suspensions, and the average fluorescent signals were measured to quantify the competition between the carrier dose and ADC surrogate. Compared to several HALA test candidates, S12 HALA Ab showed the lowest binding competition with trastuzumab on this low-expression cell line (Figure 2D). Therefore, the affinity of the HALA antibody must be balanced, because too high of an affinity (like the S18 clone or trastuzumab itself) will increase tissue penetration but block uptake in low-expression cells, while too low of an affinity will not aid in tissue penetration (Supplemental Figure S4). S12 HALA Ab co-incubation results in improved tissue penetration with slight gradients in trastuzumab uptake, indicating it has sufficient affinity to compete at high-expression levels (Supplemental Figure S5). Together, the competition of S12 on the high-expression cells to improve tissue penetration and low competition on the low-expression cells made it the lead candidate for in vivo testing.

HALA Antibody Improves In Vivo Tissue Penetration

We next measured the distribution of ADCs T-DM1 and T-DXd in vivo with and without a HALA carrier dose to quantify the tissue penetration in xenograft tumors. Because tissue penetration is related to the dose/plasma concentration,(39) we used the clinical doses of these ADCs to match the tissue penetration in the mouse model. Mice with high HER2 expression (NCI-N87) tumors were treated with T-DM1 or T-DXd, with and without a carrier dose, to determine how the HALA carrier dose impacted tissue penetration in vivo. For T-DM1, an 8:1 carrier dose (8-fold higher carrier antibody relative to the ADC, see Methods section) of the S12 HALA antibody or trastuzumab improved the tissue penetration in vivo relative to T-DM1 alone (Figure 3A–C). The carrier doses (trastuzumab and the S12 HALA antibody) improved distribution as quantified with a Euclidean distance map (Figure 3D), with statistically significant differences in drug gradients (p < 0.005).

Figure 3. S12 HALA Ab improves in vivo distribution of T-DM1 and T-DXd.

(A-C) Representative histology images from mice (n=3) 24 h post-injection with (A) 3.6 mg/kg fluorescent T-DM1 (green) alone, (B) 8:1 ratio of S12 HALA Ab to T-DM1 or (C) 8:1 ratio of trastuzumab to T-DM1, showing improved ADC distribution with the addition of a carrier dose. Intravenous Hoechst (blue) labels nuclei around functional blood vessels, with all vessels stained with CD31 (red). (D) Euclidean distance mapping illustrates the mean fluorescence intensity versus distance from blood vessels from multiple mice (n=3). (E-G) Representative histology images from mice (n=3) 72 h post-injection with (E) 6.4 mg/kg fluorescent T-DXd (green) alone, (F) 8:1 ratio of S12 HALA Ab to T-DXd or (G) 8:1 ratio of trastuzumab to T-DXd. Intravenous Hoechst (red) labels nuclei around functional blood vessels. (H) Euclidean distance mapping illustrates the mean fluorescence intensity versus distance from blood vessels from multiple mice (n=3).

T-DXd uses a higher clinical dose of 6.4 mg/kg (in gastric cancer; 5.4 mg/kg in breast cancer) compared to T-DM1 (3.6 mg/kg), so carrier dose ratios were reduced to maintain a similar total antibody dose for maximum tissue penetration. A 5:1 ratio was used to provide sufficient tissue penetration but prevent a reduction in total uptake (%ID/g) from oversaturating the tumor at an 8:1 ratio (Supplemental Figure S6). Similar to T-DM1, the distribution was significantly improved when the ADC was co-administered with a 5:1 ratio of trastuzumab or S12 HALA antibody (Figure 3E–H). Comparing the ADC-only treatments, T-DXd penetrates further than T-DM1 due to the higher ADC dose. In both cases, however, the S12 HALA Ab increased the depth of tissue penetration relative to the ADC alone but did not reduce the perivascular uptake as much as the trastuzumab carrier dose. The differences in drug gradients for all three treatments were statistically significant (p < 0.0001).

We further explored how the increased tumor penetration depth with the addition of a carrier dose influenced payload delivery via pharmacodynamic staining for γH2A.X. The γH2A.X marks double-stranded DNA breaks from the DXd payload. While the bystander effect can clearly be visualized in areas devoid of ADC delivery to aid in improving heterogeneous payload distribution, regions with low or negative γH2A.X staining were still apparent in some regions of the tumor. In contrast, the combination of ADC and carrier dose improves payload distribution compared to ADC with a bystander payload and no carrier dose (Supplemental Figure S7).

Co-administration of HALA Antibody with ADC Maximizes Tumor Growth Inhibition

After establishing that the S12 HALA Ab could improve distribution in high HER2 expression tumors in vivo but maintain minimal blocking in low-expression tumors, we tested the impact of our HALA antibody on ADC efficacy. Using high and moderate HER2 expression cell lines, T-DM1 was administered at its clinical dose of 3.6 mg/kg alone, with an 8:1 S12 HALA Ab dose, or with an 8:1 trastuzumab carrier dose ratio. For the high-expression NCI-N87 cell line, we observed the strongest tumor growth inhibition with the HALA carrier dose, more than with a trastuzumab carrier dose or T-DM1 alone (Figure 4A). The ADC plus trastuzumab or S12 HALA carrier doses were more efficacious than T-DM1 alone by day 7 (p < 0.01) and maintained at day 14 (p < 0.0005). By day 21, the T-DM1 plus HALA carrier dose performed better than the ADC plus trastuzumab carrier dose (p = 0.011).

Figure 4. S12 HALA Ab is uniquely effective at inhibiting tumor growth in vivo for low- to moderate- and high-HER2 expression tumors.

(A-B) Tumor volume (mm³) versus time after treatment for the (A) high HER2 expression (NCI-N87, n=8) cell line and (B) low to moderate-expression (MDA-MB-453, n=5) cell line with T-DM1 monotherapy, T-DM1 with trastuzumab, or T-DM1 with S12 HALA antibody. (C-D) Tumor volume measurements were also performed after T-DXd treatment in (C) NCI-N87 (high HER2 expression, n=8) or (D) CAPAN-1 (low HER2 expression, n=7–10) tumors alone, with T-DXd and trastuzumab, or with T-DXd and S12 HALA Ab. In (A, C), the HALA antibody improved ADC efficacy in high-expression tumors, similar to trastuzumab, but (B, D) the HALA administered with ADC maintained similar efficacy to ADC alone in moderate- and low-expressing tumors, unlike the trastuzumab carrier dose. Error bars indicate SEM.

In the low to moderate HER2 expression tumor model, MDA-MB-453, T-DM1 alone showed strong efficacy, as expected due to good tissue penetration in lower-expression tumors. The HALA carrier dose with ADC or ADC alone caused faster tumor regression than the trastuzumab plus ADC by day 7 (p < 0.02), (Figure 4B). The HALA carrier dose remained more effective at day 14, but the difference was no longer statistically significant by day 28 (p = 0.078), which could be due to the sensitivity of MDA-MB-453 cells to trastuzumab. In both cell lines, the HALA antibody on its own had some efficacy (Supplemental Figure S3, Supplemental Figure S8–S9), which is promising from a clinical perspective given that a patient may benefit from multiple mechanisms of action including HER2 signaling blockade. Overall, the addition of the S12 HALA Ab maintained ADC efficacy, and the HALA approach resulted in equal or greater efficacy than the other treatment arms in both tumor models.

Next, we tested the efficacy of T-DXd with a 5:1 ratio of S12 HALA Ab dose based on improved distribution while maintaining total ADC uptake. T-DXd monotherapy showed higher baseline efficacy than T-DM1 monotherapy in the high-expression NCI-N87 cell line due to higher ADC penetration at clinical doses and the benefit of bystander payload penetration.(40,41) The tumor cohorts started to differentiate at later times, with the HALA carrier dose showing smaller tumor sizes than the trastuzumab carrier (p = 0.041) and ADC alone (p = 0.054) at 8 weeks (Figure 4C, Supplemental Figure S10). For the low-expression CAPAN-1 tumors (Table 1), the 5:1 trastuzumab carrier dose significantly lowered efficacy relative to ADC alone (p < 0.0001) due to competition at the low-expression level, reducing ADC uptake and efficacy (Figure 4D, Supplemental Figure S11). However, S12 HALA Ab co-administered at a 5:1 ratio maintains similar efficacy to the ADC alone (p = 0.15 and 0.38 at 4 and 8 weeks, respectively), suggesting negligible competition for binding sites with low HER2 expression. Importantly, a dose of 32 mg/kg S12 HALA carrier dose alone lacks significant efficacy over negative controls, demonstrating that this xenograft model isolates the impact of payload delivery as a mechanism of action with the strongest benefit from the HALA antibody.

Enhanced ADCC Activity with Fc-engineered HALA Antibodies

The xenograft models show how HALA antibodies improve efficacy in high-expression tumors by increasing tissue penetration while preserving efficacy in low-expression models. However, we wanted to engineer improved efficacy in low HER2 expression tumors as well as those with high expression. Therefore, we utilized another mechanism of action that can be leveraged to enhance killing on low-expression cells, namely Fc effector function.

The high dosing and reduced affinity(29,42) of a HALA antibody can benefit Fc-effector functions such as Antibody-Dependent Cellular Cytotoxicity (ADCC) and Antibody-Dependent Cellular Phagocytosis (ADCP). The co-administration of a HALA antibody with an ADC is expected to achieve a similar level of ADCC on targeted cells (i.e., perivascular cells) as ADC alone, but the higher dose from the HALA antibody will increase the tissue penetration and number of targeted cancer cells in vivo, thereby enabling greater ADCC in high-expression tumors (Figure 5A). The HALA antibody would not impact ADCC with low-expression cancer cells due to a lack of binding and competition (Figure 5B). However, if the binding affinity of the HALA Fc domain to Fcγ receptors is increased using an enhanced Fc receptor binding domain (eFc), the heterotypic avidity between HER2 on the cancer cell and FcγR on an immune cell can enable the HALA to outcompete the ADC in the immune synapse, generating greater ADCC than the ADC alone or ADC plus the WT Fc S12 HALA Ab or trastuzumab carrier dose (Figure 5C). Using an ADCC reporter cell line, the WT antibody induces strong ADCC with or without the S12 HALA Ab dose at a 3:1 ratio at high HER2 expression levels (Figure 5D). In a low HER2 expression cell line (~100-fold fewer HER2 receptors/cell), inducing strong ADCC is more challenging with ~7-fold lower induction with WT trastuzumab alone. However, the S12eFc HALA Ab doubles the ADCC activity, demonstrating the ability of the HALA antibody to outcompete the ADC in the immune synapse (Figure 5E). This provides two mechanisms to enhance the killing of low-expression cells: in the absence of an immune cell, the ADC outcompetes the HALA antibody, maximizing payload delivery, while in the presence of an immune cell, the avidity from the Fcγ receptor and HER2 binding allows the HALA antibody to outcompete the ADC to maximize Fc-mediated killing.

Figure 5. HALA antibody with an engineered Fc region improves the overall ADCC activity for both high- and low-HER2 expression cells.

(A) The HALA antibody is expected to have minimal change in ADCC on high-expression cells that are already targeted, such as perivascular cells. However, it can improve the overall ADCC activity at the tissue level by reaching more cells deeper in the tumor. (B) Lower HER2 expression cells naturally have lower ADCC, which is not impacted by the addition of a HALA antibody since the HALA antibody does not compete well on low-expression cells. (C) However, a higher-affinity Fc domain (referred to as enhanced Fc or eFc) for the HALA antibody can boost the avidity between the Fc receptor and HER2 binding to enable binding in the context of immune cell synapses. Therefore, the higher Fc affinity can increase ADCC beyond a high-affinity antibody alone. (D-E) ADCC activity for trastuzumab alone or mixtures thereof was evaluated for (D) high HER2 (HCC-1954) and (E) low HER2 (MDA-MB-231) expression cells after incubation with trastuzumab (WT), a 3:1 ratio of S12 HALA Ab to WT, or a 3:1 ratio of S12eFc HALA Ab. In (D), a similar maximum induction was observed because of the lack of opportunity for improvement in tissue penetration on the cell monolayers, while in (E) the S12eFc HALA Ab:WT mixture shows a large increase in ADCC relative to WT alone (p < 0.0005), which was only ~3-fold lower than the induction with the high-expression cell line despite a ~100-fold lower HER2 expression.

Co-administered HALA Antibodies Induce Strong Responses in a Syngeneic Mouse Model

To test the ability of the eFc HALA antibodies to improve efficacy in an immunocompetent setting, we used E0771 cells transfected with human HER2 (hHER2) at an expression level of 250,000 receptors/cell and injected them into hHER2 transgenic mice. Mouse cells are less sensitive to DM1 or DXd payloads, requiring large doses for efficacy in syngeneic mouse models that can obscure the need for a carrier dose.(43,44) Therefore, we used PBD as the payload for the syngeneic model, given the high potency in both mouse and human cells. A dose of 1 mg/kg of Trastuzumab-PBD (T-PBD, DAR 2) was selected to provide a moderate response for assessing changes in efficacy. Mice treated with 1 mg/kg of T-PBD showed a 67% overall response rate, including 6/15 complete responses and 4/15 partial responses (Figure 6). To test the impact of Fc-effector function in this model, the Fc-effector function was eliminated by mutating the Fc domain on the ADC (Supplemental Figure S12). Mice treated with 1.1 mg/kg of LALAPG-T-PBD (to match the same total payload dose) showed growth delay, but only a 30% overall response rate and 1/10 mice with a complete response. The HALA antibody with enhanced Fc-effector function only showed a modest growth delay and a single partial response, indicating that effector function from unconjugated antibody alone is not very strong in this model. In contrast, mice treated with 1 mg/kg T-PBD (containing a WT Fc domain) co-administered with 3 mg/kg of S12eFc HALA antibody had the highest overall response rate of 87% including 8/15 complete responses and 5/15 partial responses. This was higher than T-PBD with S12 antibody with a wild-type Fc domain, indicating that enhanced Fc domain further improved responses when paired with the ADC in this model.

Figure 6. A mixture of an ADC and an S12 HALA antibody with an enhanced Fc domain shows the strongest tumor growth inhibition in an in vivo syngeneic tumor model.

Using a syngeneic mouse model with 250K receptors/cell, the impact of an enhanced Fc domain that has increased Fc-effector function was tested in vivo. Trastuzumab was conjugated to the DNA alkylating payload PBD (T-PBD) to treat the syngeneic model because T-DM1 and T-DXd have minimal potency in the mouse E0771 cell line. Human HER2-transfected E0771 cells were orthotopically implanted in the mammary fat pads of human HER2 transgenic mice. When tumors reached 250 mm³, mice were treated with a single injection of either 3 mg/kg of S12-HALA antibody containing an enhanced Fc domain (S12eFc only), 1.1 mg/kg of an Fc-null (LALA-PG mutated) trastuzumab-PBD ADC (DAR 2), 1 mg/kg of trastuzumab-PBD (containing a WT Fc domain), 1 mg/kg of trastuzumab-PBD plus 3 mg/kg of S12 HALA antibody, or 1 mg/kg of trastuzumab-PBD plus 3 mg/kg of an enhanced Fc HALA antibody. The Fc-null T-PBD showed a 30% response rate and one complete response, while the T-PBD had much higher efficacy with a 67% response rate and four complete responses. However, the most effective treatment was the T-PBD co-administered with the enhanced Fc HALA antibody, resulting in an 87% response rate including eight complete responses. This was higher than T-PBD with a WT Fc HALA antibody, demonstrating the increased efficacy from the Fc effector function.

DISCUSSION

Tumor penetration is a persistent challenge in ADC development and is critical to achieving clinical responses in solid tumors. Poor tissue penetration has sparked multiple approaches such as bystander payloads,(45) transient anti-drug binding,(13) smaller format protein-drug conjugates,(10) lower potency payloads to allow a larger antibody dose,(46) and co-administration of an antibody with an ADC to improve efficacy.(7,8,16–18) A major hurdle with any of these approaches is the variable expression found with many targets. If target expression were homogeneous, a patient with high expression could receive an antibody carrier dose, while a patient with low expression could receive the ADC alone. However, this is not possible when expression levels differ between tumors and metastases or within the same tumor. Here, we designed a novel HALA carrier dose to co-administer with an ADC that ‘auto-tunes’ the level of ADC binding to match the amount of payload needed for efficient cell death (Figure 1).

We tested the tissue penetration and efficacy with two clinically-approved ADCs to modulate cellular targeting to the expression level and better leverage multiple mechanisms of action. It was important to balance the HALA dose and affinity for T-DXd and T-DM1 because a low HALA dose would not enhance penetration, but a saturating dose, where ADC uptake becomes limited by available binding sites,(47) would decrease ADC tumor delivery and potentially limit efficacy (Supplemental Figure S6). The HALA:ADC ratio of 8:1 was selected for T-DM1 based on achieving a saturating dose of total antibody in high expressing tumors using the clinical T-DM1 dose (3.6 mg/kg), and the 8:1 ratio is consistent with theoretical predictions in our previous work.(23) Because T-DXd uses a lower potency payload and can be delivered at a higher clinical dose (6.4 mg/kg), a 5:1 HALA:ADC ratio was selected to prevent oversaturation of the tumor, which reduces ADC delivery. One consideration with the ratio selection is that the HALA antibody and ADC may clear at different rates, leading to a different ratio than intended at the site of action. While varying pharmacokinetics could have an impact and is another tool for controlling the ADC and carrier dose ratio, the majority of ADC uptake occurs within the first few days of ADC administration, before a large discrepancy between the ADC and carrier dose concentration can occur, leading to minimal impact on the results presented here.

Overall, HALA co-administration resulted in the greater than or equal efficacy compared to ADC alone or co-administration with trastuzumab. Notably, we observe more durable responses to T-DXd (with and without carrier doses) after a single dose in the high expression NCI-N87 model compared to the low expression CAPAN-1. This difference is likely due to more complete killing of cells with high HER2 expression from higher payload uptake as seen in vitro. Recent clinical results are consistent with this trend, where stronger responses are seen in patients with high HER2 expressing tumors versus HER2 low or ultra-low. In a lower expression system, ADCs tend to be less limited by tissue penetration but rather from insufficient payload uptake into cells. As a result, adding a high affinity carrier dose can decrease efficacy by reducing payload uptake, so our aim with the addition of a HALA antibody was to maintain the efficacy of the ADC alone. For the low to moderate expression cell lines tested, our results show either maintenance of ADC efficacy or improvement with the addition of HALA antibody. Additionally, our results test a single dose of ADC with or without carrier dose, while ADCs in the clinic are administered every 21 days. Under a repeated dosing regimen, we hypothesize the differences in efficacy between ADC alone and ADC with carrier dose would become more pronounced. This approach can be applied to any ADC with high affinity and variable target expression provided the ADC is sufficiently potent to kill low- and moderate-expression cells. Therefore, the approach can broaden the use of carrier doses to clinically relevant scenarios where receptor expression varies within the same tumor or between metastases.(21)

There has been great interest in engineering ADCs for low-to-moderately expressed targets for ADC development,(48) but even moderately expressing tumors can show significant differences in efficacy based on dosing and tissue penetration.(49) Additionally, there has been a shift in ADC design to consider multiple mechanisms of action, such as payload delivery, immunogenic cell death (ICD), antibody signaling blockade, and Fc effector functions that could act synergistically to bolster efficacy.(50) HALA antibody co-administration has the potential to improve all these mechanisms by increasing payload delivery/ICD, blocking receptor signaling (e.g., HER2), and contributing to ADCC (as seen with trastuzumab(14)). We selected the NCI-N87 cell line to investigate HER2-based carrier doses due to its lack of response to trastuzumab alone in order to isolate the impact of improved payload delivery on efficacy. Our previous work and others confirm that NCI-N87 is relatively insensitive to trastuzumab, requiring high, frequent dosing to slow NCI-N87 tumor growth.(16,51) While selecting a trastuzumab-resistant cell line was part of our experimental design, we expect that additional efficacy from antibody dosing is a benefit of using the carrier dose approach.

With the two approved HER2 ADCs, T-DXd monotherapy shows stronger efficacy than T-DM1, which benefits from higher dosing and payload bystander effects(52). Bystander effects provide some improvements against the binding site barrier(20,53), but tissue penetration is important even with bystander payloads due to the lower efficiency of bystander delivery compared to direct cell targeting.(17,41) Therefore, increased penetration via addition of a carrier dose in the high-expression model further improved T-DXd efficacy with statistical significance. Further imaging with pharmacodynamic staining suggests the bystander DXd payload is able to target areas of the tumor that haven’t been reached by the ADC. However, not all regions show payload response. The addition of the HALA carrier dose and improved tissue penetration results in more direct delivery of payload to cells far from blood vessels, consistent with the improvement in efficacy at high expression levels observed with a HALA carrier dose.

However, directly quantifying the improvement of payload delivery in vivo due to increased tissue penetration from the carrier dose versus the bystander effect remains a limitation of this work. We showed qualitatively that T-DXd with and without a carrier dose results in pharmacodynamic signal deeper into tumor tissue, though PD staining is an indirect method of payload detection because it measures DNA damage in response to the payload rather than the payload itself. In previous work, we have demonstrated in spheroids that direct cell killing is more efficient than bystander killing.(54) Therefore, the data are consistent with bystander effects helping tissue penetration, but bystander killing is not as effective as improved ADC distribution. In the heterogeneous tumor microenvironment, quantification to measure payload directly in vivo (e.g. mass spectrometry imaging) are challenging due to the noise of the experiments.

The increased efficacy of an enhanced Fc HALA antibody using the same Fc mutations as margetuximab(55) was demonstrated in a syngeneic mouse model of breast cancer, highlighting the ability of HALA antibodies to further leverage the immune response. Others have shown improved efficacy of ADCs with enhanced Fc regions while also noting that ADCs with enhanced Fc domains have the risk of increasing payload uptake into immune cells, potentially contributing to toxicity (56). In the current results, the eFc-HALA antibody was able to leverage ADCC, even on low expressing cells, without the risks of increased payload uptake in immune cells by utilizing the carrier antibody for enhanced Fc interactions. In contrast, we also tested LALAPG-T-PBD, without effector function, because there has been interest in engineering effectorless ADCs and the role of Fc function in efficacy is not fully understood for ADCs. In this mouse model, we observe lower efficacy for the LALAPG-T-PBD, suggesting that Fc effector function could provide benefits for ADC efficacy particularly for trastuzumab that possesses strong effector function, highlighting potential efficacy/toxicity trade-offs in Fc engineering of ADCs. The HALA antibody is capable of maximizing HER2 signaling blockade as seen with efficacy in the moderate expressing MDA-MB-453 model, increasing payload delivery by the ADC for increased ICD/cell death, and increasing Fc-effector function as seen in the syngeneic mouse model.

Overall, we demonstrate how HALA antibodies can maintain or improve ADC treatment efficacy through multiple mechanisms of action at both high and low HER2 expression levels. We achieved favorable ADC distribution and cellular uptake across different expression levels using HALA carrier doses that ‘auto-tune’ the delivery at the cellular level, addressing a key limitation and expanding the use of carrier doses to a greater number of clinically relevant scenarios. ADCs have been increasingly developed as immunotherapies rather than a payload delivery mechanism, and the HALA approach enhances the contribution of immune effects to provide stronger and more durable ADC responses.