GD2-Targeted Minibody–Drug Conjugates Match the Potency of IgG-Based ADCs in a Mouse Cancer Model

Abstract

Despite the clinical success of antibody–drug conjugates (ADCs), their efficacy in solid tumors remains constrained by limited tumor penetration of the IgG format. Smaller antibody fragment–drug conjugates (FDCs) present a compelling alternative, potentially offering superior intratumoral distribution and a wider therapeutic window driven by rapid systemic clearance. This study compares therapeutic activity of ganglioside GD2-specific minibody–drug conjugates against full-length ch14.18 antibody–drug conjugates, and biodistribution of the respective minibody (scFv-C_H3 homodimer) and IgG formats in the GD2-positive B78-D14 melanoma syngeneic mouse model. We conjugated the minibody and antibody with MMAE or MMAF via a cathepsin-cleavable linker, generating FDCs with drug–antibody ratio (DAR) of 2 and ADCs with DAR of 2 or 4. The biodistribution analysis showed no significant difference in tumor uptake for both formats early in the analysis (2–4 h) and a higher tumor uptake for the IgG at 24 h post-injection. However, the minibody achieved a superior tumor-to-blood ratio (TBR) at all timepoints, reaching a TBR > 1 compared to ~0.2 for the antibody by 24 h. In vitro studies demonstrated higher cytotoxicity for the ADCs regardless of drug load (DAR 2 or 4) compared to the FDCs, although the difference between conjugates with equal DAR was modest in B78-D14 cells. Critically, superior in vitro ADC potency did not translate in vivo. Minibody–MMAF and minibody–MMAE achieved 74% and 55% tumor growth inhibition, respectively, by the study endpoint—demonstrating comparable efficacy to ADCs with twice the drug load when administered to mice at equimass dosing. Stron/g in vivo efficacy of anti-GD2 FDCs, combined with the superior TBR for the minibody format, underscores the potential of minibody–drug conjugates for treating GD2-positive tumors, particularly when ADC-associated toxicity precludes high-dose regimens.

1. Introduction

A wave of global clinical approvals of antibody–drug conjugates (ADCs) over the last decade, including several new entrants to the market in 2025, has solidified their role as a therapeutic class for treating both hematologic and solid malignancies [1,2]. Despite clinical success of ADCs built on the full-length antibody scaffold, their efficacy in solid tumors has been constrained by limited penetration into the dense tumor interstitium, primarily attributed to the large size of the IgG molecule (~150 kDa) [3]. Additionally, high-affinity ADCs are often captured by the first few layers of antigen-expressing tumor cells—a phenomenon known as the “binding site barrier” [4]. This leads to heterogeneous drug distribution, with perivascular regions receiving a high dose while more distal, potentially hypoxic tumor cells remain untreated, fostering an environment for tumor relapse and drug resistance.

Lower molecular mass antibody fragments have been extensively shown [5,6,7] to penetrate tumors faster and to distribute within them more homogeneously compared to antibodies following systemic administration. The trade-off for this is their shorter circulation half-life, driven by rapid renal clearance for fragments below the glomerular filtration cutoff of ~60–70 kDa and absence of FcRn-recirculation for most fragment formats, which often leads to lower accumulation of the fragments in the tumor over long time intervals [8]. However, rapid systemic clearance of antibody fragments often results in their higher tumor-to-blood ratios and reduced exposure to healthy tissues. The hope for antibody fragment–drug conjugates (FDCs) is to offer a wider therapeutic window compared to ADCs, enabling more aggressive dosing regimens—either through dose escalation or more frequent administration—to drive curative efficacy in solid tumors.

Various antibody fragment formats have been employed to generate drug conjugates, including but not limited to Fab-fragments, minibodies, diabodies, scFv-fragments, single-domain antibodies (sdAb), and VHH-Fc fusion constructs [9,10]. Even smaller antibody mimetics such as DARPins, affibodies, or bicyclic peptides have also emerged as promising vehicles for targeted delivery to solid tumors [11,12,13,14]. A key test for determining the therapeutic applicability of each of these platforms probably lies in their head-to-head in vivo comparisons with ADCs that share identical antigen specificity.

An early study by Kim et al. [15] compared an anti-CD30 diabody–MMAF conjugate with an ADC with the same drug–antibody ratio (DAR) of 4: despite a ~30-fold faster serum clearance for the diabody-based format, only a ~3-fold higher dose was required to match the in vivo efficacy of the ADC in a xenograft mouse model, illustrating that dose escalation can compensate for faster clearance. Studies with smaller single-domain antibody formats further highlight the advantages of rapid penetration and optimized antigen binding. Wu et al. [16] found that a 5T4-targeted sdAb-SN38 conjugate (DAR 1), despite a substantially faster systemic clearance, achieved a 12-fold higher tumor uptake at 1 h post-injection and produced greater tumor inhibition than its IgG-based counterpart (DAR ~6). The authors attributed this to the FDC’s smaller size, which enhances intratumoral penetration, and to a faster antigen dissociation rate, which mitigates the binding site barrier. Similarly, Thurber and colleagues [17] demonstrated that a monovalent anti-PSMA sdAb conjugate, despite being less potent in vitro, was more effective in vivo than its rapidly internalizing bivalent version. This superior efficacy was linked to the monovalent FDC’s slower internalization, which allowed it to penetrate deeper into the tumor and deliver a lethal dose to a greater fraction of cancer cells.

The minibody format defined as a dimer of an scFv-fragment fused to the IgG1 C_H3 antibody domain stabilized through two disulfide bonds ((scFv-C_H3)₂) was first described in preclinical studies in 1996 where it was used as an imaging agent [18]. Its intermediate size of 80 kDa provides a longer circulation time than that of the scFv alone while maintaining faster clearance compared to intact antibodies. Minibodies have since become an important platform for molecular imaging and therapy. Recently, an anti-HER2 minibody–MMAE conjugate (DAR 4) achieved excellent tumor inhibition in a xenograft mouse model, outperforming the clinical benchmark T-DM1 in a dose escalation study [19]. A study by Neri and colleagues [20] evaluated a minibody analog of the small immune protein (SIP) format where εC_H4 domain of human IgE was used instead of γ1C_H3, and showed that the EDA domain of fibronectin-targeting SIP-DM1 conjugate produced superior tumor control in mice both in equimass and equimolar dosing compared to its counterpart ADC with equal DAR 2. The authors concluded that rapid drug release combined with early tumor exposure enabled superior efficacy for the fragment format.

A number of gangliosides, primarily the b-series gangliosides GD2 and GD3, and their O-acetylated forms, are overexpressed in multiple tumors while being absent or minimally expressed in healthy tissues, making them attractive targets for immunotherapy [21]. Among them, only ganglioside GD2-targeted therapy has gained therapeutic approval. Two full-length GD2-specific antibodies—dinutuximab (ch14.18) and naxitamab (hu3F8)—received approval for use in high-risk neuroblastoma, demonstrating significant improvements in survival outcomes. Beyond unconjugated antibodies, a range of GD2-targeted modalities are undergoing clinical evaluation, including immunocytokines, radiopharmaceuticals, bispecific antibodies, and chimeric antigen receptor (CAR)-modified T- and NKT-cell therapies [22,23]. Despite this clinical momentum, ADCs targeting GD2 remained comparatively underexplored until recently, and only in 2024 the first-in-human trial of a humanized variant of dinutuximab antibody conjugated to the topoisomerase I inhibitor exatecan payload was initiated in patients with soft-tissue sarcoma and glioblastoma (trial NCT06641908) [24].

Previously, we developed GD2-specific ADCs by conjugating the ch14.18 antibody to the microtubule inhibitor drugs MMAE and MMAF with a drug–antibody ratio (DAR) of 4, and they demonstrated potent antitumor activity in GD2-positive cell lines and in syngeneic mouse models of GD2-positive B78-D14 melanoma and EL-4 lymphoma [25]. We also engineered anti-GD2 FDCs using the minibody (DAR 2) and scFv (DAR 1) formats with the same cytotoxic payloads, which showed substantial and selective cytotoxicity in GD2-expressing cells [26]. The primary aim of this study was to compare biodistribution profiles of the anti-GD2 minibody and full-length ch14.18 antibody and to evaluate the therapeutic efficacy of their respective drug conjugates (ADCs with DAR 4, FDCs with DAR 2) in the B78-D14 melanoma mouse model. Cytotoxicity of antibody–drug conjugates and minibody–drug conjugates was also evaluated in vitro in B78-D14 melanoma cells. To facilitate a more direct comparison, we additionally generated ADCs with a lower DAR of 2 and characterized their cytotoxic effects alongside other conjugates.

2. Results

2.1. Generation of Minibody-Based and IgG-Based Auristatin Drug Conjugates

To address the primary objective of this study—evaluating comparative in vivo efficacy of anti-GD2 minibody-based and ch14.18 antibody-based auristatin conjugates—we employed previously characterized minibody fragment–drug conjugates (FDCs) with an average drug–antibody ratio (DAR) of 2 [26] and antibody–drug conjugates (ADCs) with an average DAR 4 [25]. For a direct comparison with the FDCs, we also generated ADCs with a lower average DAR of 2. The GD2-specific minibody was engineered using the variable light (V_L) and heavy (V_H) chains from the ch14.18 antibody. Both molecules were conjugated with monomethyl auristatin E (MMAE) or F (MMAF) by thiol-maleimide chemistry via a cathepsin-cleavable linker, as illustrated in the reaction schemes in Figure 1A. Representative absorbance spectra for the FDCs and ADCs employed in the study are shown in Figure 1B and Figure 1C, respectively.

Figure 1.

Preparation and antigen-binding properties of anti-GD2 FDCs and ADCs. (A) Reaction schemes for generating minibody–drug conjugates and antibody–drug conjugates with low (DAR2) and high (DAR4) drug ratio. Illustration created with BioRender.com. (B,C) Representative absorption spectra for FDCs (B) and ADCs (C) normalized at 280 nm. (D) Binding of minibody-based conjugates (1 µg/mL) to ganglioside GD2 in direct ELISA. (E) Binding of antibody-based conjugates (0.25 µg/mL) to ganglioside GD2 in direct ELISA. Bars represent mean ± SEM; ns, no significant difference between groups; **, p < 0.01, unpaired Student’s t-test.

The conjugation of maleimide-activated MMAE and MMAF to reduced interchain cysteines did not impair the antigen-binding properties of the parent minibody or ch14.18 antibody, which was evaluated by direct ELISA against immobilized ganglioside GD2. Separately, for biodistribution studies, fluorescently labeled molecules were generated by conjugating the Cy5 dye to protein lysines. Cy5-labeled proteins also demonstrated high affinity to GD2; however, the antigen-binding of minibody-Cy5 was slightly inferior to that of the unconjugated minibody (Figure 1D,E).

2.2. In Vitro Activity of the FDCs and ADCs

Both the minibody-based [26] and the antibody-based (DAR 4) [25] drug conjugates demonstrated direct dependence of the cytotoxic effects on the level of GD2 expression in a panel of cell lines with varying antigen expression in our earlier works, including potent effects in the B78-D14 murine melanoma cell line with high and stable expression of GD2 and no effects in its parent GD2-negative B16 cell line. Here, the GD2-positive B78-D14 cell line was used in order to compare the cytotoxic effects of the anti-GD2 FDCs and ADCs with different drug–antibody ratios.

As shown in Figure 2, GD2-specific conjugates with MMAF exhibited stronger cytotoxicity in B78-D14 cells than the corresponding conjugates with MMAE regardless of the antibody format, which is consistent with our previous findings [25,26]. Half-maximal inhibitory concentration (IC50) values for ADCs with DAR 4 aligned with previously obtained data, comprising 0.34 ± 0.02 nM for ch14.18-MMAE and 0.04 ± 0.01 nM for ch14.18-MMAF. The cytotoxic effects for ADCs with low drug load (DAR 2) were less pronounced, with IC50 values of 3.1 ± 0.7 nM and 0.59 ± 0.08 nM for ch14.18-MMAE and ch14.18-MMAF, respectively. Notably, a two-fold reduction in the drug–antibody ratio resulted in a strong and disproportionate decrease in cytotoxicity in the B78-D14 cell line.

Figure 2.

Viability of the GD2-positive B78-D14 melanoma cell line evaluated by MTT assay following 72 h incubation with (A) ch14.18-MMAE DAR 4, ch14.18-MMAE DAR 2, and minibody-MMAE DAR 2, or (B) ch14.18-MMAF DAR 4, ch14.18-MMAF DAR 2, and minibody-MMAF DAR 2. Data presented as mean ± SEM.

The cytotoxic effects in B78-D14 cells observed for the minibody-based FDCs generated in this work were stronger than those for the conjugates with identical DAR 2 obtained in our earlier work. The IC50 values for minibody–MMAE and minibody–MMAF were 9.5 ± 0.7 nM and 1.9 ± 0.2 nM, respectively. This improvement may be explained by an optimization of the FDC generation protocol, which now ensures minimal aggregation following conjugation with the drugs.

Comparison of the cytotoxic activity of ADCs and FDCs with the same DAR 2 in the B78-D14 cell line indicated that, despite somewhat lower IC50 values for the ADCs, the difference in observed effects was modest. This is consistent with similar internalization mechanisms and efficiency of accumulation in the lysosomes of tumor cells for the anti-GD2 minibody and ch14.18 antibody that were shown in our recent work [27].

2.3. The Minibody Demonstrates Higher Tumor-to-Blood Ratio, Altered Biodistribution Profile Compared to ch14.18 Antibody

For biodistribution analysis, C57BL/6 mice bearing subcutaneous B78-D14 melanoma tumors received intravenous injections of Cy5-labeled minibody or ch14.18 antibody, and their accumulation in plasma and major organs was analyzed at 2, 4, 8, and 24 h post-injection. The results of the analysis are summarized in Figure 3 and also presented in Supplementary Table S1.

Figure 3.

Tissue biodistribution for the minibody and ch14.18 antibody analyzed at 2, 4, 8, and 24 h after intravenous administration of 7.5 mg/kg mass of the fluorescently labeled molecules in the B78-D14 melanoma mouse model. (A) Biodistribution to the tumor and organs. For accumulation in the tumor, individual data points for both molecules are additionally presented in a dot plot (top right); ns, no significant difference between groups; ***, p < 0.001, unpaired Student’s t-test. (B) Biodistribution to the blood plasma and liver. (C) Tumor-to-blood ratio (TBR) for the protein accumulating in the tumor and the protein remaining in the blood at respective timepoints. Data are presented as percentage of injected dose per gram of tissue type or blood plasma (% ID/g), and values are expressed as mean ± SEM calculated for groups of 3 animals.

As can be seen from Figure 3A, the dynamics of accumulation in the tumor differed for the minibody and the antibody. At early timepoints post-injection (2–4 h), the minibody exhibited a moderately more effective tumor uptake relative to the antibody, though the difference did not reach statistical significance. Specifically, the minibody uptake in the tumor at 2 h and 4 h post-injection was 2.1% and 2.8% of injected dose per gram of tumor (% ID/g), respectively, while the ch14.18 uptake constituted 1.6% and 2.0% ID/g. This reversed by 8 h post-injection, when the antibody accumulation in the tumor (2.3% ID/g) slightly exceeded that of the minibody (1.5% ID/g). The trend strengthened over time, and by 24 h post-injection, the ch14.18 content in the tumor (4.3% ID/g) significantly surpassed that of the minibody (1.0% ID/g). It is important to note that when calculated based on molar amounts of the molecules that reached the tumor, the minibody uptake at early timepoints post-injection was significantly higher than that of ch14.18 (e.g., 54 vs. 20 pmol/g at 4 h; p = 0.013), reflecting the difference in the molecular mass of the molecules—80 kDa for the minibody vs. 150 kDa for the full-length IgG.

A significant difference in the concentration of the molecules in the blood was observed as early as 2 h post-injection (Figure 3B), when 23% ID/g of the minibody (19% of the injected dose) and 57% ID/g of the antibody (47% ID) were detected in blood plasma. This gap became even more pronounced over time; by 24 h, minibody levels in the plasma had dropped to 0.9% ID/g (0.7% ID), while antibody levels remained much higher at 22% ID/g (18% ID).

Among the examined mouse organs, most significant differences were observed in the accumulation profiles of the minibody and antibody in the liver and kidneys, indicating the route and intensity of elimination for the two formats from the body. Uptake of the ch14.18 antibody in the kidneys was modest (reaching 2.6% ID/g at 8 h), whereas the minibody uptake was considerably higher (7.7% and 5.4% ID/g at 4 h and 8 h, respectively). Accumulation of both formats in the liver was the highest among all organs, with the minibody uptake being higher compared to the antibody at all analyzed timepoints. Liver uptake was highest in the first hours following administration—specifically, 29% ID/g for the minibody and 15% ID/g for the antibody at the 4 h mark.

Drug content in mouse tissues was evaluated following thorough cardiac perfusion, with the standard endpoint determined by the lightening of the liver tissue to a pale brown color. A significant portion of the drug (sometimes above 50%) in tissue homogenates is attributed to residual blood from the vasculature, which can lead to substantial differences in detected drug concentrations in perfused vs non-perfused tissues [28,29]. We observed similar differences for fluorescently labeled molecules in the liver, kidneys, and other mouse organs in our preliminary experiments.

The tumor-to-blood ratio (TBR)—the ratio of the drug accumulated in the tumor to the drug remaining in the blood—serves as a critical metric for evaluating in vivo tolerability and the feasibility of dose escalation strategies. In our analysis, the TBR for the minibody significantly exceeded that for ch14.18 antibody across all evaluated timepoints (Figure 3C). Specifically, at 24 h post-administration, the minibody reached a TBR greater than 1, whereas the value for ch14.18 was ~0.2. This difference in TBR correlates with lower accumulation of the minibody in healthy tissues such as the lung, spleen, and heart that was observed at the 24 h mark.

2.4. FDCs with DAR2 Show Antitumor Effects In Vivo Comparable to ADCs with DAR4 at Same Dose

We next evaluated the antitumor activity of minibody–drug conjugates compared to antibody–drug conjugates in the B78-D14 melanoma mouse model. For this, the ADCs with DAR 4 were selected to test against the FDCs with DAR 2. The choice was driven by several factors. An average DAR of 4 represents an established benchmark for clinically approved ADCs carrying auristatin payloads since almost every approved ADC carrying MMAE or MMAF utilizes such DAR [1]. Furthermore, anti-GD2 ch14.18-MMAE and ch14.18-MMAF with DAR 4 showed substantial antitumor efficacy in the same model in our earlier study [25], while ch14.18-MMAE DAR 2 resulted only in minimal tumor growth inhibition in a head-to-head analysis against the DAR 4 analog that led to significant growth inhibition compared to the control (Supplementary Figure S1).

Minibody–MMAE and minibody–MMAF were administered to mice via 5 injections of 5 mg/kg body mass, resulting in strong tumor growth inhibition (Figure 4A,B). Their antitumor effects were comparable to those of ch14.18-MMAE and ch14.18-MMAF with DAR 4, respectively, administered in the same dose and regimen (p > 0.05 between pairs, two-way ANOVA with Tukey’s post hoc test). At the study endpoint (day 41), mean tumor growth inhibition (TGI) in experimental groups compared to the control group constituted, in ascending order, 51% for ch14.18-MMAE, 55% for minibody–MMAE, 70% for ch14.18-MMAF, and 74% for minibody–MMAF. Minibody–MMAF demonstrated somewhat more pronounced activity compared to minibody–MMAE, and a similar trend was observed for the ADCs carrying MMAF and MMAE; however, differences in tumor growth inhibition between the conjugates did not reach statistical significance. All conjugates were well-tolerated by the animals and did not cause significant body weight loss (Figure 4C).

Figure 4.

Antitumor activity of the minibody–drug conjugates evaluated next to antibody–drug conjugates (DAR 4) in the B78-D14 melanoma mouse model. Tumor growth curves from the combined experiment are shown separately for groups treated by (A) conjugates with MMAE and (B) conjugates with MMAF. Control groups were treated with PBS solution. All conjugates were administered at 5 mg/kg body mass, and arrows indicate days of administration. ***, p < 0.001 vs. control, two-way ANOVA with Dunnett’s post hoc test. (C) Weight change. Values represent mean ± SEM for groups of 4 animals.

The comparable therapeutic activity of the FDCs and the ADCs in the B78-D14 melanoma mouse model appears unexpected, considering the much higher cytotoxic activity of the ADCs with DAR 4 in vitro. The high in vivo efficacy of minibody-based conjugates is likely attributed to advantages associated with their more efficient delivery to the tumor compared to full-sized ADCs. Notably, mice were not only administered an equal dose of the ADCs and FDCs in the study (5 mg/kg), but also a similar amount of the auristatin payload (~120–130 nmol/kg in each case), which was not deliberately normalized but rather followed from accounting for differences in molecular mass of the conjugates (approx. two-fold higher for ADCs) and the drug load (DAR 2 for the minibody vs. DAR 4 for the antibody).

3. Discussion

Our study provides a direct comparison of the therapeutic efficacy of ganglioside GD2-specific minibody–drug conjugates and full-length antibody–drug conjugates in a syngeneic mouse model, demonstrating that minibody conjugates with MMAE and MMAF can achieve antitumor activity similar to that of ADCs despite substantially faster systemic clearance for the minibody format. The GD2-specific minibody exhibited a significantly higher tumor-to-blood ratio at all evaluated timepoints, reaching a TBR > 1 compared to ~0.2 for the ch14.18 antibody at 24 h post-injection. This favorable pharmacokinetic profile translated into comparable therapeutic efficacy of the FDCs, with minibody–MMAE and minibody–MMAF (DAR 2) demonstrating tumor growth inhibition comparable to ch14.18-based ADCs with twice the drug load (DAR 4) when administered to mice at equimass dose. To our knowledge, this represents the first in vivo analysis of an anti-GD2 antibody fragment–drug conjugate.

A central finding of the pharmacokinetic analysis was the distinct difference in tumor accumulation between the minibody and the antibody. The two molecules showed no significant difference in tumor uptake at early timepoints, specifically 2.8 ID/g for the minibody vs. 2.0 ID/g for the antibody at 4 h post-injection (n = 3, p = 0.15). However, their accumulation profiles diverged markedly over time. By 24 h, the ch14.18 antibody achieved substantially higher tumor uptake compared to the minibody (4.3% vs. 1.0% ID/g, respectively), a result likely driven by the extended blood half-life of the IgG format. At the same time, rapid elimination of the minibody from circulation—0.7% of the injected dose for the minibody vs. 18% for the ch14.18 antibody left in blood plasma at the 24 h mark—led to reduced exposure of healthy tissues to the antibody fragment over prolonged time intervals. Considerably lower uptake of the minibody compared to ch14.18 was observed in the lung, spleen, and heart at 24 h post-injection.

The uptake in the kidneys was significantly higher for the minibody compared to the antibody in the first hours post-injection. Difference in kidney clearance can be at least partially attributed to the size of the molecules relative to the glomerular filtration cutoff of ~60–70 kDa [30,31]: while the minibody’s molecular mass is slightly above this threshold, the full-length antibody substantially exceeds it.

Accumulation of both formats in the liver was the highest among all organs. Full-length antibodies are typically characterized by an increased clearance via the mononuclear phagocytic system compared to small-molecule drugs. This is primarily explained by expression of various types of Fcγ receptors on cells of these organs (Kupffer cells in the liver), that interact with the antibody Fc domain and drive internalization and catabolism of the molecules [32]. However, minibody accumulation in the liver was still higher than that of the IgG at all analyzed timepoints of the analysis.

The minibody’s inability to interact with the neonatal Fc receptor (FcRn), which mediates the recirculation of full-length antibodies from tissues back into the bloodstream, contributes to its higher accumulation in both liver and kidneys. Although the minibody contains the C_H3 constant domain, the functioning of FcRn in both mice and humans requires its specific binding to the interface between the C_H2 and C_H3 domains, which is formed only in IgGs [33]. Studies by the research group that originally developed the (scFv-C_H3)₂ format demonstrated that the introduction of an additional C_H2 domain into the minibody structure resulted in the pharmacokinetic properties of the 105 kDa (scFv-C_H2-C_H3)₂ molecule in mice becoming similar to those of a full-length antibody of the same specificity (increased blood half-life, decreased accumulation in kidneys and liver). However, when two point mutations preventing binding to FcRn were introduced into this molecule, its biodistribution mirrored that of the 80 kDa minibody [34,35]. Thus, for the minibody format, absence of interaction with FcRn plays a role no less important than the size of the molecule in determining its half-life in the blood.

Higher total accumulation of the GD2-specific antibody compared to the minibody in the tumor must be interpreted within the broader context of antibody pharmacology, where total accumulation and spatial distribution represent fundamentally different pharmacokinetic parameters. The primary mechanism of antibody transport within tumor tissue is diffusion, which is inversely related to molecular size. Smaller antibody fragments, including minibodies, typically exhibit enhanced penetration capabilities compared to IgGs, enabling more uniform distribution throughout the tumor volume rather than peripheral accumulation around tumor vasculature. Several independent studies comparing the in vivo antitumor activity of FDCs vs ADCs with identical specificity have demonstrated that fragment-based formats—including an anti-HER2 minibody [19], a fibronectin Extra Domain A-targeting small immune protein (SIP) [20], and an anti-CD30 diabody [15]—can exhibit potent therapeutic activity despite achieving comparable tumor accumulation only at early timepoints post-injection (typically ≤4 h), and inferior tumor accumulation at later timepoints (≥24 h). Among these studies, the non-internalizing SIP-DM1 fragment–drug conjugate with DAR 2 is particularly noteworthy, as it demonstrated superior antitumor efficacy compared to the corresponding ADC with DAR 2 under equimolar dosing conditions. While our biodistribution analysis employed the unconjugated minibody rather than the fluorescently labeled fragment–drug conjugate itself—a methodological constraint imposed by stability challenges during generation of the fluorescently labeled FDC constructs—the observed correlation between the markedly elevated TBR for the minibody format and the antitumor efficacy of the minibody-based FDCs is consistent with the mechanistic advantages reported in the aforementioned studies. At the same time, given that conjugation of hydrophobic payloads can substantially modify the pharmacokinetic profile of antibody fragments, future preclinical characterization of the anti-GD2 minibody-based format will require direct evaluation of FDC biodistribution.

Similar tumor growth inhibition in vivo for the anti-GD2 FDCs (DAR 2) compared to the ADCs (DAR 4) was achieved despite the considerably higher IC50 values for minibody–drug conjugates in vitro. This disparity suggests that factors beyond intrinsic cytotoxicity—specifically penetration of the tumor interstitium and intratumoral distribution—play decisive roles in determining therapeutic outcomes in the B78-D14 melanoma model. However, in order to confirm the mechanistic basis for therapeutic advantage of the anti-GD2 FDCs, future studies will need to directly visualize the intratumoral spatial distribution for both the minibody-based and IgG-based formats. Equal or superior in vivo efficacy despite lower in vitro cytotoxicity has been reported for a number of FDCs when compared to ADCs of the same specificity, specifically when given at higher doses or more frequent intervals [15,16,17]. We did not evaluate dose-escalation for the GD2-specific minibody–drug conjugates; however, the pharmacokinetic data suggest that the minibody format may tolerate higher therapeutic doses than the antibody format. Rapid clearance of the minibody from the circulation creates a favorable therapeutic window, potentially allowing for increased dosing frequency or magnitude without exacerbating systemic toxicity. The improved safety margin is particularly relevant for GD2-targeted therapies, where dose-limiting toxicities in patients are common due to GD2 expression on peripheral nerves [36].

Limited stability of the minibody–drug conjugates with DAR > 2 precluded evaluation of higher loading capacities in our study. FDCs with DAR3 were generated in the work, but suffered from a disproportionately higher aggregation rate compared to the DAR2 variants when analyzed by size-exclusion chromatography (Supplementary Figure S2). Aggregation rates for all ADC variants were minimal regardless of the drug load. This disparity likely reflects the constraints of the particular conjugation chemistry employed. The highly hydrophobic nature of potent cytotoxic payloads such as auristatins, maytansinoids, or PBD dimers can destabilize smaller protein scaffolds that possess less capacity to spatially shield the payloads compared to IgGs. Multiple strategies have emerged to overcome hydrophobicity-driven instability and enable higher DAR in antibody–drug conjugates that include rational design of conjugation sites and incorporation of hydrophilic elements such as polyethylene glycol (PEG) or polysarcosine into the linker [37,38,39]. Branched PEG chains have demonstrated particular success for FDCs in this regard. Douez et al. [19] incorporated a branched PEG₁₂ unit into an anti-HER2 minibody–MMAE, generating a homogeneous conjugate with DAR 4 with improved internalization rate and cytotoxicity. A branched PEG₈ glucuronide-based linker increased the overall hydrophilicity of an anti-HER2 scFv-MMAE conjugate and prevented aggregation even at DAR 6, also leading to a 14.4x increase in half-life of the original scFv [40]. We previously employed a 10 kDa 4-arm PEG-maleimide for conjugating anti-GD2 scFvs with several molecules of maytansinoids DM1 and DM4, yielding stable FDCs that retained high binding and selective cytotoxicity for GD2-positive cells [41]. Incorporating hydrophilicity-enhancing linkers into anti-GD2 minibody–drug conjugates developed in this work could enable higher drug loading while preserving their favorable pharmacokinetic profile.

Despite strong antitumor activity, neither the ADCs nor the FDCs achieved complete tumor eradication in the B78-D14 melanoma model. Lack of curative efficacy may be attributed to the immunologically “cold” nature of the B78-D14 cell line, characterized by a microenvironment that limits the secondary immune response typically required for tumor clearance in syngeneic hosts [42]. This supports the use of combinatorial strategies with immune checkpoint inhibitors—namely those targeting PD-1, CTLA-4, and CD47—that have all been reported to improve outcomes of anti-GD2 antibody-based therapeutics in the preclinical setting [22,43]. Furthermore, engineering FDCs to carry several payloads simultaneously (e.g., both MMAE and MMAF [44]) could enhance efficacy through complementary mechanisms of action and reduced drug resistance. Both efficacy and selectivity of GD2-targeted FDCs could also be refined through bispecific strategies. Co-targeting ganglioside GD2 and the B7–H3 immune checkpoint with a bispecific full-length antibody was shown to enhance selectivity in B78 melanoma murine xenografts [45]. This approach is exemplified by several VHH-Fc format bispecifics that have recently entered clinical trials—among the first FDCs to ever reach this stage—the anti-PD-L1/αvβ6 drug conjugate JSKN022 (trial ChiCTR2500110741) [46] and the anti-EGFR/c-MET drug conjugate RB-601 (trial NCT06928363).

Our study presents anti-GD2 minibody–drug conjugates as a possible alternative to ADCs, offering distinct pharmacokinetic advantages including rapid clearance and reduced healthy tissue accumulation. The comparable antitumor efficacy observed despite lower total drug accumulation, and a favorable therapeutic index support exploration of dose-escalation strategies for this FDC format, particularly in contexts where ADC toxicity in GD2-positive tumors limits high therapeutic dosing [47].

4. Materials and Methods

4.1. Production of Minibody– and Antibody–Drug Conjugates

GD2-specific FDCs were generated by a modification of a previously described protocol [26]. First, minibody interchain cysteines were reduced by incubation with 0.5 mM TCEP in Versene solution (0.2 g Na₄EDTA per 1 L PBS, pH 7.3) for 1.5 h at room temperature (RT) and with agitation. The reducing agent was removed by size-exclusion chromatography in Zeba Spin Desalting Columns (7 kDa MWCO; Thermo Fisher Scientific, Waltham, MA, USA). The minibody (1–2 mg/mL in Versene solution) was then immediately conjugated with 4:1 molar excess (for DAR 2) of maleimide-activated auristatins MC-VC-PABC-MMAE or MC-VC-PABC-MMAF (MedChemExpress, Monmouth Junction, NJ, USA) for 3 h at RT and with agitation; auristatins were added as stock solutions in DMSO, maintaining the DMSO concentration in the reactions below 3%. The FDCs were buffer-exchanged into PBS in centrifugal filters, and disulfide bridges were re-oxidized by 10:1 molar excess of dehydroascorbic acid for 2 h at RT. Final purification was performed in Zeba columns.

ADCs were prepared following a modified protocol from [25]. Briefly, complete reduction of interchain cysteines of the antibody was performed by incubation with 1 mM TCEP in Versene solution (pH 7.3) for 1 h at 37 °C with agitation. The reducing agent was removed using Zeba columns. Thiol-maleimide conjugation of the antibody at concentrations in the range of 1–5 mg/mL was then immediately performed with drug molar excess of 7:1 (for DAR 4) or 3.5:1 (DAR 2) in Versene solution for 3 h at 37 °C and with agitation. The drug molar excess used to generate the conjugates with the average DAR specified above was validated based on at least 5 batches of each conjugate. Reaction products were purified from unreacted auristatins and buffer-exchanged into PBS by two consecutive rounds of size-exclusion chromatography in Zeba columns.

4.2. Protein Labeling for the Biodistribution Study

Labeling of the minibody and ch14.18 antibody was performed via lysine amino groups using the fluorescent dye Sulfo-Cyanine5 NHS ester (Lumiprobe, Moscow, Russia; referred to as Cy5 in this work). Proteins were buffer-exchanged into 0.1 M bicarbonate buffer, pH 8.3, using Amicon Ultra-4 10 kDa centrifugal filters (Thermo Fisher Scientific), following which Cy5 was added at 5:1 molar excess. Reactions were carried out for 3 h in the dark at RT and with agitation, and the unreacted fluorophore was removed using Zeba columns. The degree of labeling of Cy5 to both proteins calculated by UV-VIS spectroscopy was highly reproducible between different batches under the specified reaction conditions, ranging between 1.8 and 2 Cy5 molecules per protein molecule.

4.3. Evaluation of Drug–Antibody Ratio

Average drug–antibody ratios (DAR) for the conjugates were calculated by UV-VIS spectroscopy using the wavelengths of the drug local absorption maxima and 280 nm, the characteristic protein absorption maximum [48], on a BioDrop µLITE spectrophotometer (BioChrom, Cambridge, UK). The degree of labeling of the Cy5 fluorophore was calculated analogously. Extinction coefficients used in the calculations are presented in Supplementary Table S2. All extinction coefficients except for Cy5 were determined experimentally, and the characteristics for Cy5 were provided by its manufacturer.

4.4. Direct ELISA

Nunc MaxiSorp high protein-binding capacity 96 well ELISA plates (Thermo Fisher Scientific) were coated with ganglioside GD2 (purchased from Merck Group, Darmstadt, Germany) at a concentration of 0.25 μg in 100 μL of 96% ethanol per well. Following air drying, wells were blocked with 100 μL 2% BSA in PBS supplemented with 0.1% Tween-20 (PBS-T) per well for 2 h at RT. GD2-binding molecules (100 μL solution in PBS-T per well) were added in triplicates at 1 µg/mL concentration for minibody and minibody-based conjugates and at 0.25 µg/mL for antibody and antibody-based conjugates. Incubation was carried out for 1.5 h. After washing with PBS-T, anti-human Fc-specific HRP-labeled antibodies were added to the wells (1:6000; Santa Cruz Biotechnology, Dallas, TX, USA). Following 40 min incubation and further washing, 1-Step Ultra TMB-ELISA Substrate Solution (Thermo Fisher Scientific) was added, and the color reaction optical density (OD) was measured at 450 nm by DEL-100 microplate reader (MIULAB, Zhejiang, China).

4.5. Cytotoxicity Assay

The GD2-positive B78-D14 mouse melanoma cell line was cultured in RPMI-1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 0.2 mg/mL G418, 0.2 mg/mL hygromycin B, 100 μg/mL penicillin, and 100 U/mL of streptomycin (all from Thermo Fisher Scientific, Waltham, MA, USA). The B78-D14 cell line generated by transfection of the GD2-negative B16 line with genes coding for GD3 and GD2 synthases [49] was a kind gift from David Schrama (University Hospital Wuerzburg, Germany). The cell line was maintained at low passage numbers and routinely checked for Mycoplasma by PCR.

ADC- and FDC-induced decrease in cell viability was analyzed by colorimetric MTT assay (MTT purchased from Merck Group) previously described by Denizot and Lang [50], with modifications specified earlier [51]. For analysis of cell viability, B78-D14 cells were incubated in 96-well flat-bottom tissue culture plates (2 × 10⁴ cells/well; Greiner, Kremsmünster, Austria) with serial dilutions of the conjugates for 72 h under standard culture conditions. Following incubation, the MTT solution in a final concentration of 250 μg/mL was added to each sample for 3 h. Reaction OD was assessed at 570 nm by DEL-100 microplate reader. Cell viability was evaluated by the formula (OD _{treated cells} − OD _blank)/(OD _{control cells} − OD _blank) × 100%, where OD _blank represents OD in control wells with no cells. GraphPad Prism 8 software was used to generate dose–response curves. All MTT experiments were reproduced at least three times.

4.6. Biodistribution Study

Biodistribution of the minibody and ch14.18 was analyzed ex vivo in the syngeneic mouse model of B78-D14 melanoma. All experiments with mice were approved by the IBCh RAS Institutional Animal Committee, protocol #325/20-21, from 14.06.2021, and performed in compliance with AAALAC guidelines. To establish the model, 6–8-week-old female C57BL/6 mice were used, that were purchased from the laboratory animal nursery “Pushchino” (Moscow region, Russia). All animals were housed in a clean barrier facility in microisolator cages and standardized conditions, and had unhindered access to food and water. Melanoma cells were resuspended in incomplete RPMI medium and administered to mice (4 × 10⁶ cells in 150 µL of medium) subcutaneously in the right flank.

Mice were randomized into groups (n = 3 per group) when the mean tumor volume reached 500 mm³, and fluorescently labeled ch14.18 antibody or minibody were administered via the retro-orbital venous sinus at 7.5 mg/kg of body mass (~150 µg). At 2, 4, 8, or 24 h post-injection, mice were anesthetized by isoflurane, and their blood was retro-orbitally collected into tubes containing heparin; blood was centrifuged, and the supernatant (plasma) was stored at 4 °C for simultaneous analysis with the organs. Mice were then immediately euthanized by cervical dislocation.

To analyze drug accumulation in tissues with minimal residual blood, blood was further removed by transcardial perfusion with 10–15 mL of heparinized PBS until the liver tissue turned pale brown. Tumors and organs were excised, homogenized, and incubated overnight at 4 °C in a solution of hydrochloric acid and ethanol (0.3 M HCl, 70% C₂H₅OH in water). Samples were centrifuged, and fluorescence intensity in the supernatants was analyzed using a GloMax-Multi Detection System fluorometer (Promega, Madison, WI, USA); excitation: 625 nm; emission: 660–720 nm. Organs and blood plasma from intact mice were used to subtract autofluorescence, and the results were presented as percentage of injected dose per gram of tissue or plasma (% ID/g). Drug content in blood plasma was estimated assuming a total mouse blood volume of 1.5 mL, plasma volume of 55% of total blood volume [52], and plasma density of 1 g/mL.

4.7. Antitumor Activity In Vivo

Therapeutic activity of GD2-specific FDCs and ADCs was analyzed in the B78-D14 melanoma mouse model. Subcutaneous tumors in mice were induced following the procedure described above. Treatment was initiated when tumors reached a mean volume of 50 mm³. Mice were randomized into experimental and control groups (n = 4 per group), and experimental groups received retro-orbital injections of 150 μL of the conjugates in a regimen of 5 injections at 3-day intervals, while the control group received PBS in the same regimen. For an independent experiment comparing IgG-based drug conjugates with different DAR (Figure S1), groups of n = 3 animals were administered 3 injections every 3 days. Conjugates were sterilized on 0.22 μm filters prior to administration to mice. Tumor volumes were measured with a caliper at least twice weekly using the modified ellipsoid formula V = (length × width²)/2, where length and width represent the largest and second-largest perpendicular linear dimensions of the tumor, respectively. Efficacy was quantified by relative tumor growth inhibition (TGI), calculated according to the formula TGI (%) = (1 − ΔV_treated/ΔV_control) × 100, where ΔV_treated and ΔV_control denote the change in mean tumor volume in the treated and control groups, respectively, relative to the start of treatment. All animals were euthanized when the mean tumor volume in the control group reached ~1500 mm³.

4.8. Statistical Analysis

Graphs were generated using GraphPad Prism software. Data are presented as arithmetic mean ± standard error of the mean (SEM) from at least three independent experiments or as values from one representative experiment of three. Unpaired Student’s t-test was used to compare two independent groups. For multiple comparisons of more than two independent groups, two-way repeated measures analysis of variance (ANOVA) with Dunnett’s or Tukey’s post hoc tests was applied. Statistical significance was set at p < 0.05; the following notations for significance levels were used: ns, no significant difference between groups; *, p < 0.05; **, p < 0.01; ***, p < 0.001.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms27041974/s1.

Author Contributions

Conceptualization, D.V.K. and R.V.K.; Data curation, D.V.K. and M.M.T.; Formal analysis, D.V.K., M.M.T., I.V.K. and E.V.S.; Funding acquisition, S.M.D. and R.V.K.; Investigation, D.V.K., M.M.T., I.V.K., A.V.K., E.V.S. and R.V.K.; Methodology, D.V.K., M.M.T., I.V.K., A.V.K., E.V.S. and R.V.K.; Project administration, S.M.D. and R.V.K.; Resources, S.M.D. and R.V.K.; Software, D.V.K., M.M.T., A.V.K., E.V.S. and R.V.K.; Supervision, R.V.K.; Validation, D.V.K., M.M.T. and E.V.S.; Visualization, D.V.K. and M.M.T.; Writing—original draft, D.V.K. and R.V.K.; Writing—review and editing, D.V.K. and R.V.K. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The animal study protocol was approved by the IBCh RAS Institutional Animal Committee, protocol #325/20-21, approval date: 14.06.2021.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available within the article text and figures.

Conflicts of Interest

The authors declare that they have no competing interests.

Funding Statement

The work was supported by the Ministry of Science and Higher Education of the Russian Federation, agreement # 075-15-2024-536.