Article Text
PDF
Abstract
Background Advanced clear cell renal cell carcinoma (ccRCC) is a prevalent kidney cancer for which long-term survival rates are abysmal, though immunotherapies are showing potential. Not yet clinically vetted are bispecific T cell engagers (BTEs) that activate T cell-mediated cancer killing through intercellular synapsing. Multiple BTE formats exist, however, with limited cross-characterizations to help optimize new drug design. Here, we developed BTEs to treat ccRCC by targeting carbonic anhydrase 9 (CA9) while characterizing the persistent BTE (PBTE) format and comparing it to a new format, the persistent multivalent T cell engager (PMTE). These antibody therapies against ccRCC are developed as both recombinant and synthetic DNA (synDNA) medicines.
Methods Antibody formatting effects on binding kinetics were assessed by flow cytometry and intercellular synaptic strength assays while potency was tested using T-cell activation and cytotoxicity assays. Mouse models were used to study antibody plasma and tumor pharmacokinetics, as well as antitumor efficacy as both recombinant and synDNA medicines. Specifically, three models using ccRCC cell line xenografts and human donor T cells in immunodeficient mice were used to support this study.
Results Compared with a first-generation BTE, we show that the PBTE reduced avidity, intercellular synaptic strength, cytotoxic potency by as much as 33-fold, and ultimately efficacy against ccRCC tumors in vivo. However, compared with the PBTE, we demonstrate that the PMTE improved cell avidity, restored intercellular synapses, augmented cytotoxic potency by 40-fold, improved tumor distribution pharmacokinetics by 2-fold, and recovered synDNA efficacy in mouse tumor models by 20-fold. All the while, the PMTE displayed a desirable half-life of 4 days in mice compared with the conventional BTE’s 2 hours.
Conclusions With impressive efficacy, the CA9-targeted PMTE is a promising new therapy for advanced ccRCC, which can be effectively delivered through synDNA. The highly potent PMTE format itself is a promising new tool for future applications in the multispecific antibody space.
- Renal Cell Carcinoma
- Antibody
- Bispecific T cell engager - BiTE
- Kidney Cancer
Data availability statement
Data are available on reasonable request. All data generated or analyzed during this study are included in this published article or its online supplemental information files.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Advanced and metastatic clear cell renal cell carcinoma (ccRCC) is a challenging cancer from which only 1 in 10 patients will survive.
For other cancers, bispecific T cell engagers (BTEs) are antibody therapies that have gained traction as effective treatment strategies.
Synthetic DNA (synDNA) medicines can bypass recombinant protein manufacturing to deliver durably expressing, efficacious biologics in vivo.
WHAT THIS STUDY ADDS
The clinically relevant persistent BTE (PBTE) format experiences significant functional attenuations due to its Fc domain appendage for half-life improvement.
Equipping the PBTE with an additional tumor-targeted binding domain can recover its functional losses, generating a format that is both highly potent and Fc-equipped for half-life improvements, referred to as the persistent multivalent T cell engager (PMTE).
Mostly, it presents the CA9-PMTE as a promising, new anticancer for the treatment of advanced ccRCC while also advancing the delivery capabilities of synDNA.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
It demonstrates the importance of BTE format cross-characterization and tuning during early development to navigate potentially adverse effects on structure–activity relationship. It also supports similar investigations and optimizations for other existing BTE formats.
Introduction
Renal cell carcinoma (RCC) is a rare and orphan disease responsible for 90% of kidney cancers, which stand as the 9th most common cancer in men and 14th most common cancer in women worldwide.1 The clear cell RCC (ccRCC) comprises approximately 75% of cases, which in the USA has risen to 55,000 annually. Problematically, one-third of patients are already metastatic on diagnosis due to a mild symptomatic nature that slows diagnosis, and unlike earlier grades that are effectively treated with surgical resection, advanced lesions are inoperable.1 2 Intermediate-risk and poor-risk patients go on to acquire resistance to traditional chemotherapies, radiation, and targeted antiangiogenics. With only 10%–15% of these patients surviving 5 years, advanced ccRCC represents a cancer population in significant need of new treatment options.3
Immunotherapy has transformed the landscape of cancer care by empowering the immune system to detect and kill cancers, with particular potential for advanced ccRCC. Checkpoint inhibitors that unleash T cells on tumors serve as a first-line therapy where they offer a relatively high response rate of 50%–60%. However, beyond still leaving nearly half of patients nonresponsive, progression-free survival is only extended to 1–2 years for responders due to discontinuations from acquired resistance and side effects.3 Resistant tumors subsequently acquire a more aggressive and metastatic phenotype and with no treatment options left, approximately 10,000 ccRCC patients die annually in the USA.1 New therapies could leverage and expand this cancer’s immune responsivity by driving alternative mechanisms of action, like focusing immune activity directly on tumors through the select targeting of tumor-associated antigens (TAAs), more effectively eradicate ccRCC.
Bispecific T cell engagers (BTEs) are a powerful, antibody-based immunotherapy designed to simultaneously bind immune and cancer cells to trigger cancer’s selective destruction. To accomplish this, they are equipped with multiple binding domains that recognize distinct epitopes or antigens, with most research directed at binding to the CD3 complex of T cells alongside a relevant TAA.4 The dual specificity forces a proximal relationship between the target and T cells, with the goal of engaging the T cell-mediated cellular cytotoxicity (CMC) of cancer cells through the formation of artificial, MHC-independent synapses. While clinically successful, first-generation BTEs given their simplified, dual single-chain variable fragment (scFv) design suffer a short half-life of ~2 hours (hrs) owing to the absence of an Fc region that would otherwise facilitate an extended circulation time from a higher molecular weight (MW) and neonatal Fc receptor (FcRn)-mediated recycling.5 Dramatic improvements to circulation time have been made by numerous second-generation formats that reintroduce an Fc domain, such as the Persistent BTE (PBTE) that is equipped with a single-chain Fc domain (scFc) linked to the conventional BTE molecule.6 There have since been at least eight PBTEs in clinical development, among them an anti-CD19 PBTE which demonstrated a well-extended half-life of 210 hours.5 7 Our group has also leveraged this format with a recently reported PBTE against CD45, made possible through epitope base editing, to engineer a universal blood cancer therapy.8 However, detailed comparative characterizations between first-generation and second-generation formats that shed light on potential trade-offs for clinically relevant formats like this one have yet to be reported. Nevertheless, the known benefits of PBTEs promoted the application of this format not only for additional targets, but to the creation of new structures that advance the scFc concept. For instance, PBTEs are still limited by their 1:1 valency that detracts from the higher avidity naturally afforded to antibodies. Increasing valency with additional binding domains has endowed other therapies with more functional affinity to wield greater potency and efficacy.9 10 The longer residence times enabled by higher valency can also drive better tumor distribution and help distinguish between high-target-expressing and low-target-expressing cells to better mitigate toxicity.11–13 We, therefore, hypothesize that bispecific designs integrating both scFc and valency adaptations could give rise to new and effective, antibody therapies for advanced ccRCC if provided a highly targetable TAA by the disease.
90% of RCC cases involve notable overexpression of the TAA, carbonic anhydrase 9 (CA9).14 Its high and homogeneous expression is observed at both primary and metastatic sites and its expression patterns correlate with poor prognoses.14–16 This contrasts with its otherwise scarce expression in normal tissue, as its aberrant upregulation arises from the inactivation of a specific regulatory von Hippel-Lindau tumor-suppressor gene.17 Functioning as a modulator of extracellular pH by catalyzing the interconversion of carbon dioxide to protons and bicarbonate, CA9 serves cancer cells by acidifying the tumor microenvironment (TME). In doing so, it drives the cancer growth, metabolic adaptation, invasion, and metastasis associated with poor patient prognoses.17 Eliminating CA9-expressing cells in cancer or inhibiting its enzymatic activity could thus dismantle these effects. Given its consistent expression in advanced ccRCC and the immune responsivity of this disease, CA9 represents an ideal antigen for targeted immunotherapies aimed at the isolated eradication of cancer and pH equilibration of the TME, wherein relief from acidic conditions could further facilitate immune function and tumor clearance.18
Despite the promising alignment of these parameters, there are no approved therapies targeting CA9 and clinical testing is largely confined to radiotherapies and chemotherapies.19 Vaccination strategies have demonstrated limited clinical efficacy for RCC and the phase III failure of girentuximab, an IgG1 monoclonal antibody relying on an antibody-dependent cellular cytotoxicity mechanism of action, emphasizes the need for newer preclinical approaches to better harnessing the power of the immune system in targeting CA9.20 A recent investigation of a CA9-directed BTE supported this technology’s potential for advanced ccRCC therapy after it exhibited tumor control in patient-derived xenograft models; the first-generation format described, however, would face limited translatability in the modern clinical landscape and CA9-targeted BTEs that instead adopt from advancing bioengineering strategies would maximize the clinical impact for ccRCC patients.21 In addition, the simultaneous advancement of the antibody delivery platform could further extend a meaningful reach to ccRCC patients.
Nucleic acid medicines such as synthetic DNA (synDNA) and messenger RNA (mRNA) represent innovative strategies for delivering self-synthesized, durably expressed biologics for cancer. In contrast to the manufacturing and infusion of recombinant medicines, synDNA leverages one’s own cellular machinery in expressing and continually disseminating anticancer biologics.22 Pronounced improvements to pharmacokinetic (PK) parameters and efficacy have been reported with in vivo DNA-launched, BTE (dBTE) targeting HER2, IL13Ra2, follicle-stimulating hormone receptor (FSHR), and EGFRViii in ovarian and brain cancers.23–26 The physiological benefits stand alongside manufacturing ones that include reduced costs and long-term temperature stability, which could eliminate the requirement for cold-chain storage and expand patient access.22 Coupling bispecific antibodies to synDNA through the development of in vivo-produced dBTEs is thus a promising approach for the potential improvement of patient clinical outcomes, as well as the upstream manufacturing and expanded distribution of antibody therapies.
Here, we present two novel, T cell engager molecules targeting CA9 for ccRCC, one of which is designed in a novel format for multivalent or multispecific T cell engagers. The first of these molecules is a PBTE while the second, referred to as a persistent multivalent T cell engager (PMTE), builds off the PBTE with an additional binding domain to generate a multivalent format, providing CA9 bivalency in this iteration. We characterize and compare the functional differences in recombinant form between the prototypical 1:1 BTE, 1:1 PBTE, and 2:1 PMTE formats as it pertains to avidity and potency in vitro using ccRCC cell lines, before studying their PK/pharmacodynamic (PD) relationships in mouse models. We then transform the biologics into nucleic acid medicine for a multidisciplinary examination of their therapeutic potential. Lastly, we test the reproducibility of these formatting trends beyond the CA9-targeted molecules with novel, FSHR-targeted antibodies for ovarian cancer. This study ultimately determines that the PMTE could be a superior agent to existing single-chain formats, warranting its further investigation for medicinal use against ccRCC and other cancers or antigens. Further, it supports the application of synDNA for the sustained delivery of these bispecific antibody therapeutics.
Results
PBTE displays reduced affinity and potency compared with BTE
Early testing began with only the BTE and PBTE formats bispecific for CA9 and CD3. To generate the PBTE format, an scFc was linked to the conventional BTE N-terminus (online supplemental figure 1A). The antibody DNA was codon-optimized and inserted into a pVax1 expression vector for Expi293 cell expression and purification (online supplemental figure 1B). Flow cytometry binding analyses against target cell lines revealed an interesting consequence of linking an scFc to the BTE to generate a PBTE. Doing so attenuated binding affinity to target cells, evidenced by right-shifted median fluorescent intensity (MFI) curves for ACHN-CA9, 293T-CA9, and T cells, with corresponding dissociation constant (KD) values declining approximately twofold for all cell types (online supplemental figure 1C). To next assess the effects of the PBTE’s binding affinity reductions on cytotoxic potency in vitro, we leveraged xCELLigence technology that proxies cell death as a function of electrical impedance by adherent, target cells. Increasing cell index over time is a measure of unprohibited proliferation, unlike low indices resulting from the cytotoxic removal of impeding target cells, indicative of an effective therapy (online supplemental figure 1D). At 30 nanomolar (nM), both the BTE and PBTE engaged T cell killing of the primary RCC cell line 786-O with over 50% cell death by the 24-hour time point. While the BTE maintained this activity at 200 picomolar (pM), PBTE activity declined nearlytwofold (online supplemental figure 1E). Trending reductions in potency for the PBTE were similarly observed against the primary cell line A-498 and accompanying microscopic images were taken at the terminal time point (online supplemental figure 1F).
Supplemental material
PMTE overcompensates for PBTE’s reduced CA9 avidity
To attempt recovering this lost activity, a second CA9 binding domain was then linked to the PBTE to generate the PMTE, whose anti-CD3 domain was placed medially to maintain equidistantly tight synapses for either engagement arm, to render a maximally potent molecule.27 28 The full panel of single-chain formats to undergo further testing in unison for assessment of the functional effects of compounding domain appendages is visually represented (figure 1A, online supplemental figure 2A). Western blotting presented bands at appropriate MWs for each antibody (online supplemental figure 2B). The synDNA therapies are also visually represented by their plasmid map, which describes the insertion of antibody sequences into individual, codon-optimized pVax1 expression vectors, together with Kozak and IgE leader sequences for enhanced expression in humans and mice, as previously described26 (figure 1B).
PMTE restores avidity for target cells from PBTE levels. (A) Graphic of the single-chain BTE, PBTE, and PMTE formats. Gray lines represent GS linkers. (B) Plasmid map for the antibody-encoding pVax1 vectors used for synDNA therapies. (C) Flow cytometry binding curves to ACHN-CA9 (n=3), 293T-CA9 (n=4), and T cells (n=3). An irrelevant IgG1 was used as a negative control. (D) KD calculations from respective cell binding curves were plotted as donor mean±SEM. E) z-Movi workflow depicting insertion of adherent cancer cells, suspended T cells, and compounds to a flow chamber. Ultrasound waves pull T cells from cancer cells toward acoustic nodes. Force required to separate cells is recorded at single cell resolution, and detachment force represents synaptic strength generated by bispecific antibodies. (F) Curves represent the percent of cells remaining bound with increasing detachment force over time. Antibodies were analyzed at 30 nM (n=5), 3 nM (n=3), and 300 pM (n=3). An irrelevant BTE was used as a negative control. (G) Area under curve (AUC) analyses of adjacent force curves at 30 nM (n=5), 3 nM (n=3), and 300 pM (n=3) were plotted as mean±SEM. (H) Single cell resolution of force required to detach T cells from cancer cells for individual antibodies for respective antibody concentrations. Points represent individual T cells and data represent all replications collectively. (I) Representative images of the flow chamber at 1000 pN display localization of detached T cells at acoustic nodes for lower avidity antibodies at 30 nM. BTE, bispecific T cell engager; PBTE, persistent bispecific T cell engager; PMTE, persistent multivalent T cell engager. Significance is assigned as * (p<.05), ** (p<0.005), *** (p<0.001), or **** (p<0.0001).
Binding was first reexamined where the PBTE reproducibly displayed right-shifted MFI curves against CA9-expressing cells compared with the BTE, amounting to significant twofold and fourfold higher KD values for ACHN-CA9 and 293T-CA9, respectively (figure 1C). However, equipped with CA9 bivalency, the PMTE rescued this lost binding strength with additional enhancement over the conventional BTE. This overcompensation is illustrated in the PMTE’s left-most shifted MFI curves for binding both target cells. Significantly lower KD values reflect this binding difference, which compared with the PBTE were 12.9-fold and 17.6-fold for ACHN-CA9 and 293T-CA9, respectively. Compared with the BTE, the PMTE displayed a 4.7-fold lower KD for ACHN-CA9 and a significant 4.4-fold lower KD for 293T-CA9. On testing of T cell binding kinetics, both the PBTE and PMTE displayed right-shifted curves with trending 2.5-fold higher KD values than the BTE (figure 1D). Pronounced binding at 500 nM is represented by 3-log population shifts for all formats, unlike the irrelevant-IgG1, negative control (online supplemental figure 2C). This analysis divulged a dynamic relationship between structural appendage and receptor interactions, notably with the PMTE experiencing conflicting effects on CA9 and CD3 binding. Characterization was, therefore, limited by testing avidities in antigenic isolation, and clarifying the net effect on intercellular, and synaptic strength required integrating both antigen-binding kinetic analyses in a single assay.
zMovi technology is an emerging tool that leverages force-generating, ultrasonic waves to measure the collective, intercellular-synapse strength generated by immunotherapies. It has provided valuable insight into the field of CAR T-cell therapy, yet its application to bispecific antibody research herein is novel.29 30 The assay is initiated by target cell monolayer adhesion within a microfluidic chip prior to the introduction of T cells and the bispecific antibody of interest to then precipitate forceful, immunological synapses. Ultrasonic waves then pull and disconnect T cells, allowing measurement of the force required to overcome bispecific avidity at single-cell resolution. T cells in strong relationships due to higher avidity bispecific antibodies can resist ultrasonic forces without release from target cells. Contrarily, those bound by lower avidity bispecific antibodies succumb to detachment at lower forces to be visualized in clusters at acoustic nodes (figure 1E).
The CA9xCD3 engagers were tested at 30 nM, 3 nM, and 300pM while a FAPxCD3 bispecific-negative control was tested at 30 nM with the consideration that lower concentrations of the control would only have equal or less background signal. Approximately 60% of cells remained bound to target 293T-CA9 cells at peak forces for both the BTE and PMTE at 30 nM while only 20% remained for the PBTE. At 3 nM, approximately 30% of cells remained bound with the BTE and PMTE, with almost no bound cells remaining for the PBTE. Neither the BTE, PBTE, or PMTE resisted detachment forces at 300 pM (figure 1F). Transformed to the area under curve (AUC) for statistical analysis, both the BTE and PMTE signaled significantly higher force resistance compared with the negative control, but not the PBTE. In fact, the conversion of the BTE to a PBTE induced a threefold decline in force resistance at 30 nM. Converting the PBTE to the PMTE, however, restored this synaptic strength by threefold. At 3 nM, the BTE generated significant resistance to detachment compared with 30 nM of negative control. This effect was lost with the PBTE but recovered by the PMTE, which demonstrated a twofold improvement in comparative force resistance. The AUC analysis indicates concentration-dependent reductions in force resistance with no antibodies showing a meaningful effect at 300 pM (figure 1G). These data can be observed at single-cell resolution where points represent each T cell at conceding forces, with violin density expressing trends for each treatment (figure 1H). Images taken at max force during the 30 nM tests illustrate clusters of detached T cells from PBTE and negative control treatments, but a lack thereof with BTE and PMTE treatments (figure 1I). Ultimately, this comprehensive assay examined the dual nature of the engagers to form a holistic readout on bispecific avidity. In doing so, the repercussions of the PBTE’s parallel reductions in flow binding appeared to take shape with weaker synapses while the PMTE’s higher avidity for CA9 lent itself to stronger synapses. The ensuing activation of T cells would offer functional appraisals for the engagement capacities of these formats.
PMTE activates T cells with the highest potency in co-culture with RCC cells
To model T cell activation with translational relevance, we acquired SKRC-52 cells which are primary cells derived from a metastatic lesion in the mediastinum of a ccRCC patient. SKRC-52 cells were co-incubated with donor T cells and antibodies for 24 hours before the T cell staining of CD69 and CD25 activation markers. At 10 nM, all three CA9xCD3 engagers induced pronounced upregulation of CD69 in approximately 75% of CD4+ and CD8+ T cell populations, compared with a FAPxCD3 bispecific-negative control. Potency differences began revealing themselves at lower concentrations, such as at 1 pM, where the PBTE lost all activity unlike its BTE predecessor. However, the PMTE significantly restored CD69 expression on CD4+cells with a 14-fold improvement from the PBTE, and twice the activity of the BTE. It too recovered CD8+T cell activation by measure of its 10-fold higher CD69 upregulation than the PBTE, which itself had no activity. Only the PMTE maintained activity at 100 femtomolar (fM), a 100,000-fold dilution from the initial 10 nM, with meaningful differences to the PBTE and negative control in both populations (figure 2A). Potency differences are similarly represented by CD25 expression, whose significant upregulation in CD4+ and CD8+ populations from control levels is distinctly observed with the PMTE at 1 pM (figure 2B).
PMTE activates T cells with the most potency in vitro. (A) CD69+ and (B) CD25+expression on CD4+ (top row) and CD8+ (bottom row) T cells after 24 hours of co-incubation with antibodies and SKRC-52 cells (n=5). (C) CD69+and D) CD25+expression on CD4+ (top row) and CD8+ (bottom row) T cells after 24 hours of co-incubation with antibodies and 293T-CA9 (n=5). An irrelevant BTE was used as a negative control. Data are expressed as mean±SEM. BTE, bispecific T cell engager; PMTE, persistent multivalent T cell engager. Significance is assigned as * (p<.05), ** (p<0.005), *** (p<0.001), or **** (p<0.0001).
T cell activation driven by their bispecific union to CA9-expressing target cells is recapitulated with 293T-CA9, further magnifying the observable effects of multivalent extension. At 10 pM, only the PMTE induced a significant, 17-fold upregulation of CD69 on CD4+ and CD8+ T cells compared with control; its upregulation on CD8+T cells was also fourfold higher than the PBTE. With further examination, the PMTE drove significant CD69 expression compared with all other treatments at 1 pM. Specifically, fivefold and eightfold higher expression was measured on CD4+T cells than those treated with the BTE and PBTE, respectively. For CD8+T cells, these significant, respective differences were 3.5-fold and 10-fold. Only the PMTE drove CD69 upregulation on CD4+ and CD8+ T cells at the very low 300 fM concentration, with particularly meaningful differences to the BTE and PBTE treatments for the latter population (figure 2C). CD25 expression again co-represented activation states and its measurement on CD4+T cells at 1 pM significantly differentiated the PMTE from all other treatments. A meaningful difference from the PBTE was further maintained at 300 fM. Similar behavior trended for CD8+T cells, with differences at 300 fM instead most apparent between BTE and PMTE formats (figure 2D). 293T absent of transduced CA9 failed to induce CD69 or CD25 upregulation under these conditions, as did the absence of co-cultured target cells, which supports the specificity for CA9 antigen in observable T cell activation (online supplemental figure 3A,B). The gating strategy used to capture activated T cell populations is demonstrated with or without 10 nM BTE treatment in co-culture with 293T-CA9 (online supplemental figure 3C,D).
PMTE induces highly potent killing of RCC cells at femtomolar-range concentrations
CMC is a powerful tool in the immune repertoire and its comparative examination among the panel once again leveraged xCELLigence technology. For ccRCC, an effective therapy would exercise high potency at both primary and metastatic sites, the latter of which is here first represented by SKRC-52. Mirroring the observed T cell activation and avidity, the right-shifted curve from the BTE to PBTE represented a 10-fold decline in potency due to scFc linkage. Engrafting the second CA9 binding domain to form the PMTE, however, led to a 44-fold increase in potency from the PBTE to demonstrate an impressive 340 fM half-maximal effective concentration (EC50) against these metastatic cells at this time point. The primary cell lines A-498 and 786-O experienced 33-fold and 14.5-fold reductions in potency with the PBTE compared with the BTE, to be recovered by the PMTE that went on to maintain 25% cytotoxicity at a low 100 fM in both cases. Milder reductions in PBTE potency from the BTE format were observed with ACHN and 293T transduced to overexpress CA9, ranging from 2.5-fold to 4-fold. However, greater potencies were still attained by the PMTE, which in comparison to the PBTE, improved EC50 values by 41-fold and 23-fold for ACHN-CA9 and 293T-CA9, respectively (figure 3A, online supplemental table 1).
PMTE induces the T cell-mediated cytotoxicity of target cells with the highest potency. (A) Percent of terminal cell population death at increasing concentrations of antibody treatment for the following target cell lines: SKRC-52 (n=4), A-498 (n=3), 786-O (n=2), ACHN-CA9 (n=3), and 293T-CA9 (n=2). Percent cytotoxicity was calculated as (1–(treatment/target cell control))×100) at the terminal experimental time point. (B) Real-time analyses of cell growth over time for 10 nM and 1 pM antibody treatments for the adjacent cell lines illustrate potent cytotoxicity. (C) Representative images taken at the terminal time point of single experiments for each treatment group at 10 nM, illustrating the cytotoxicity of the adjacent target cell lines. PMTE, persistent multivalent T cell engager.
Killing data were acquired in real-time and the suppression of cell growth in units of cell index over time is best represented by the saturating, 10 nM condition. Negative controls do not show activity and include an irrelevant bispecific, treatments without T cells, and T cells without treatments. The 1 pM time analysis showed higher PMTE activity across target cells compared with the BTE and PBTE (figure 3B). Unlike with their CA9-transduced counterparts, naïve ACHN and 293T did not succumb to CMC, nor did the CA9-negative OVCAR3 or DAOY (online supplemental figure 4A). This reinforces the on-target selectivity for CA9-presenting cells only. Visual recording accompanies impedance in this system, allowing images at terminal time points to illustrate the effects of 10 nM treatments on cell viability and density compared with control (figure 3C). In contrast, images of CA9-negative cell lines show no effects on growth or viability from treatments (online supplemental figure 4B). The domain-annexing effects on cytotoxic potency with the PMTE closely followed the trends observed with activation markers and prior CA9 binding kinetics. While the scFc came at a functional consequence to the PBTE’s mechanism of action, the multivalent PMTE restored cytotoxic potency in vitro.
PMTE displays improved tumor distribution and tumor control compared with PBTE
PKs were modeled using Balb/c mice at a relatively high, intravenous antibody dose of 2.5 mg/kg, based on the literature describing a common dose range of 0.1–5 mg/kg for studies of this kind.31 32 The low MW-BTE fell to undetectable levels in just after 30 hours which aligns closely with the known kinetics of this first-generation format. However, the PBTE and PMTE comparably extended circulation times out to nearly 1000 hours before escaping detection, providing 13-fold higher exposure by AUC. These data are presented in both units of μg/mL and nM to correct for MW differences (figure 4A). A one-compartment model helped reduce molecular natures into fundamental, pharmacological parameters including volume of distribution (VD), which was highest for the PBTE. This suggests that the PBTE maintains better tissue distribution than the PMTE, although both had approximately twofold higher VD than the original BTE. The PBTE and PMTE shared similarly low clearance (CL) calculations in relation to the BTE, whose faster clearance and consequent 3.4-hour half-life explain the original format’s clinical requirement of continuous intravenous infusion. Comparatively, the PBTE exhibited an improved half-life of 4.5 days for a 32-fold extension over the BTE while the PMTE experienced a mild reduction of half-life from PMTE levels to 3.9 days. While the decline is in the error range, it is possible that changes in the PMTE’s physicochemical properties and MW alter its permeability and excretion, and additional studies are required to draw firmer conclusions. Ultimately, the PMTE’s half-life amounted to a 27-fold extension beyond the first-generation BTE within an overall PK profile with similarity to the PBTE (online supplemental table 2).
PMTE displays improved tumor distribution and tumor control in vivo. (A) Balb/c mice were administered a single 2.5 mg/kg (intravenous) dose of antibodies and circulating antibody was quantified from sera via His-capture ELISA for pharmacokinetic analysis (n=3). Serum concentrations are displayed in both μm/mL and nM units. Corresponding pharmacokinetic calculations are in online supplemental table S2. (B) The tumor distribution PK analysis workflow began with an ACHN-CA9 xenograft in NSG mice (n=4). Single 0.3 mg/kg (intravenous) doses in PBS were administered when tumors reached 200 mm3. Tumors and serum were harvested after 24 hours, tumors homogenized, and antibody concentrations analyzed by His-capture ELISA. An irrelevant BTE was used as negative control. (C) A tumor distribution summary plot of intratumoral concentrations against serum. The tumor/serum concentration ratio (×10−3) is represented by point size. (D) Serum antibody concentrations at 24 hours are given in both ng/mL and nM units. (E) Intratumoral antibody concentrations at 24 are given in both ng/mL and nM units. (F) Relative distribution of antibodies into the tumor compartment is considered as the tumor/serum concentration ratio (×10−3). (G) Nu/J mice harboring subcutaneous ACHN-CA9 tumors were administered 106 donor T cells (IP) and 3 doses of recombinant antibodies at 1 mg/kg, once every 5 days (n=5). Arrows represent doses and data are presented individually and as grouped mean. Bar graphs display average tumor volumes (mm3) at days 25 and 30. Data were plotted as mean±SEM. BTE, bispecific T cell engager; PMTE, persistent multivalent T cell engager. Significance is assigned as * (p<.05), ** (p<0.005), *** (p<0.001), or **** (p<0.0001).
Changes to tumor distribution were next evaluated with a PK analysis in a mouse tumor model. Using ACHN-CA9, NSG mice were implanted with subcutaneous tumors that would be recollected alongside plasma samples 24 hours after a lower antibody dose of 0.3 mg/kg. This dose was selected for its therapeutic relevance given the common dosing range of 0.01–0.3 mg/kg for T cell engagers in tumor efficacy models.12 33 Tumor dissociation enabled the ELISA-based quantification of antibody concentrations in the original tumors, and their ratio to plasma concentrations would grant additional insight into relative shifts in compartmental distribution with domain attachments (figure 4B). All antibodies were detected in plasma samples after 24 hours with concentrations ranging from 750 to 1000 ng/mL for the PBTE, PMTE, and an Fc-containing, FAPxCD3 negative control. However, in line with the plasma PK, the BTE was measured at a significantly lower 100 ng/mL than the Fc-containing formats by this time point. Concentrations are also provided in molar units (figure 4C,D).
Antibodies were detected in tumors at much lower concentrations, characteristic of limited antibody tumor penetration from poor blood flow due to high cell density, abnormal vasculature, and high interstitial pressure that can dramatically limit exposure.34 35 The negative control established background levels of 0.5 ng/mL compared with which all formats demonstrated significant distribution. The BTE had a tumor concentration of 2 ng/mL that was higher than the plasma, demonstrating a distribution that endured systemic clearance. This could be due to its low MW expediating deeper diffusion, coupled with CA9 adherence that slows clearance. PBTE levels of 1.5 ng/mL were below the BTE to indicate a counterbalance between its higher circulating levels and distribution capacity, and this difference was significantly magnified to 2.5-fold with molar unit normalization. The PMTE measured higher than both the BTE and PBTE in units of ng/mL, with a respective 2-fold and 2.7-fold higher tumor concentration. Molar unit normalization flipped the script to highlight the BTE, which retained significantly higher concentrations than both the PBTE and PMTE to create an interesting dichotomy between plasma and tumor PK characteristic desirability while the PMTE demonstrated a 2-fold higher tumor concentration than the PBTE (figure 4C,E). For the Fc-containing formats of comparable plasma PK behavior, relative differences in tumor distribution were examined using the concentration ratios between tumor and plasma compartments. This proportional analysis calculated the PMTE’s significantly higher tumor distribution than both the PBTE and irrelevant bispecific, with respective 2.8-fold and 6.5-fold differences (figure 4C,F).
Translating these PK insights and functional projections into proof of efficacy required mouse tumor models that introduce effector T cells, a critical test first performed using immunodeficient, Nu/J mice harboring subcutaneous, ACHN-CA9 tumors. These mice were treated with three doses of compound at 1 mg/kg intraperitoneally in a 5-day interval, beginning at the time of T cell administration. Despite its previously observed tumor retention of at least 1 day, BTE treatment had little effect on tumor growth compared with vehicle, given this challenging 3×5 dosing regimen in place of the typical once-daily dosing required to derive meaningful, long-term effects with this format. However, both scFc-containing formats had substantially improved tumor control, represented with day 25 comparisons to the BTE. Importantly, significant differences between the PBTE and PMTE also became evident by day 25, with average tumor sizes, respectively, measuring 700 mm3 and 140 mm3. Day 30 marked 20 days since final treatment, by which point mice in both remaining groups harbored regrowing tumors. However, tumor burden was 4.5-fold lower in the PMTE treatment group, signaling significantly more potent tumor control with this multivalent format.
synDNA-delivered PMTE sustains durable activity in vitro with superior tumor control in mice
synDNA delivery of BTEs (dBTEs) is visually represented, whereby DNA vectors and hyaluronidase (HYA) are intramuscularly administrated and accompanied by electroporation for the employment of host cellular machinery to autosynthesize therapeutics.22 The sustained expression and systemic dissemination of dBTEs that follow continually equips T cells to track and kill cancers like ccRCC. Blood from dosed mice is also sampled to confirm dBTE expression and function in vitro (figure 5A).
synDNA-delivered PMTE demonstrates better durability and tumor control in vivo. (A) DNA-based BTEs administered as plasmid DNA with hyaluronidase (50 U/mouse) and electroporation for enhanced cellular uptake. Myocytes synthesize antibodies in situ for secretion into circulation. In tumor models, donor T cells are administered (IP) and treatment efficacy assessed. Serum can be collected from antibody-treated mice, whether naïve or from tumor models, for in vitro functional assessments of in vivo-synthesized antibodies. (B) Naïve, balb/c mice were administered 200 µg of dBTE, dPBTE, or dPMTE DNA (n=5). Sera collected over time were used to stain ACHN and ACHN-CA9 cells by flow cytometry. Untranduced ACHN cells are represented in gray while colored shifts represent ACHN-CA9 cells up to day 21. (C) Sera were also used to stain donor T cells by flow cytometry from days 0 to 21. (D) Sera from synDNA-treated balb/c mice (n=5) were tested for their ability to induce cytotoxicity of SKRC-52 cells in vitro, using an xCELLigence system. Diminished cell indices represent target cell death. Negative controls included sera from empty pVax1-treated mice while recombinant BTE (10 nM) was used as a positive control. (E) NSG mice harboring subcutaneous A-498 tumors were administered 106 donor T cells (IP) and a single 10 ug dose of synDNA on day 0, represented by arrow (n=5). Mice are line-plotted individually while bar graph displays tumor volumes (mm3) at day 33 as mean±SEM. (F) NSG mice harboring subcutaneous SKRC-52 tumors were administered 106 donor T cells (intraperitoneally) and a single 10 ug dose of synDNA, represented by arrow (n=4–5). Mice are line-plotted individually while bar graph displays tumor volumes (mm3) at day 30 as mean±SEM. BTE, bispecific T cell engager; dBTE, delivery of BTE; dPBTE, delivery of persistent BTE; dPMTE, delivery of persistent multivalent T cell engager. Significance is assigned as * (p<.05), ** (p<0.005), *** (p<0.001), or **** (p<0.0001).
200 µg of expression vectors encoding the CA9-targeted dBTE, dPBTE, and dPMTE were administered to naïve, Balb/c mice whose plasma was sampled over time and target cell staining with pooled-group plasma samples illustrate antibody concentrations by MFI shift. Using samples from days 0 to 21, ACHN-CA9 cells were stained alongside their non-transduced counterparts as negative controls. MFI increases were observed for the ACHN-CA9 population compared with ACHN for all formats and given the nature of the dBTE’s half-life in particular, its consistent shift through day 21 represents the persistent in vivo expression achievable with synDNA. dPBTE expression levels peaked more acutely on day 4 before returning to lower maintenance expression levels resembling the dBTE while the dPMTE induced a later and more prominent shift that peaked on day 14 (figure 5B). T cells were next stained to confirm functional CD3 binding, and the patterns of MFI among formats were comparable to those of ACHN-CA9 (figure 5C).
The in vivo-produced T cell engagers were next tested for their induction of SKRC-52 CMC through xCELLigence analyses. Plasma from conventional dBTE-treated mice produced maximal cytotoxicity through day 14 post-treatment, with half-maximal effects from days 28 to 48. Its effects were observed on day 90 before falling by day 180. The dPBTE instead saw earlier, gradual reductions from maximum killing beginning at day 14, with effects also persisting out to day 90 to demonstrate similar activity to the dBTE, most likely due to differences in enduring-synDNA expression profiles. The dPMTE appeared to maintain maximum activity through day 90 without losing its effects until day 180 to outperform both the dBTE and dPBTE in synDNA form (figure 5D).
The reintroduction of mouse tumor modeling next enabled translationally relevant comparisons of efficacy. Subcutaneous xenografts of A-498 and IP human T cells in NSG mice first modeled a primary RCC condition, and a single 100 µg dose of dBTE expression vector elicited complete tumor control (online supplemental figure 5). At a low dose of 10 µg for distinguishment, both the dBTEs and dPBTEs displayed limited tumor control with respective measurements of approximately 1100 mm3 and 1800 mm3 tumors by day 33, with a significant difference between the two formats. dPMTE treatment appeared to induce the early elimination of tumors and enforce their ongoing suppression with a terminal 80 mm3 mean measurement. Conversely, control mice did not survive beyond day 30. Meaningful differences in terminal mean volumes were ultimately calculated for the dPMTE compared with both the dBTE and dPBTE (figure 5E).
To model metastatic ccRCC, SKRC-52 xenografts were implanted in NSG mice, after which the mice were infused with human T cells and administered a single, low dose of 10 µg dBTE or empty vector control. Control mice experienced rapid tumor growth with mice reaching the terminal volumes between days 15 and 27. The dPBTE produced noticeable yet marginal tumor control with a 2000 mm3 mean tumor volume by this day, with only one mouse surviving to day 33. The conventional dBTE trended toward higher activity with mouse tumors measuring 1250 mm3. In contrast, dPMTE treatment eliminated SKRC-52 tumors soon after administration, with three of the five mice remaining completely tumor-free by day 33 and a terminal mean volume of 190 mm3. Day 30 statistics, powered by at least an n=2 per group, demonstrate the dPMTE’s significant tumor suppression compared with both the dBTE and dPBTE (figure 5F).
Follicle-stimulating hormone receptor-targeted PBTE and PMTE display similar formatting effects on binding and potency in vitro
We were next interested in reaffirming if the observed format-dependent effects on function were a singular occurrence with these CA9xCD3 molecules, or if they would once again present in the development of an alternatively targeted PBTE and PMTE. Our group has recently published a first-generation BTE targeting follicle-stimulating hormone receptor (FSHR) for the treatment of ovarian cancer,24 which we here selected for reformatting into a PBTE and PMTE for this comparative analysis. The scFc and CD3 binding domains were unchanged in these designs (figure 6A).
FSHR-targeted BTEs follow similar formatting trends for binding and potency in vitro. (A) Anti-CA9 scFvs were exchanged for anti-FSHR scFvs in BTE, PBTE, and PMTE formats. Anti-CD3 scFvs were unchanged. (B) Flow cytometry binding curves to OVCAR3-FSHR and T cells. An irrelevant IgG1 was used as negative control. Data are presented as mean±SEM. (C) The concentration-dependent degree of OVCAR3-FSHR cytotoxicity is expressed as the percent of healthy cell control after 5 days. An irrelevant BTE was used as negative control. Data are presented as mean of duplicates. (D) KD values from flow cytometry binding analyses for OVCAR-FSHR and T cells. (E) EC50 calculations from the OVCAR-FSHR killing assay. BTEs, bispecific T cell engagers; FSHR, follicle-stimulating hormone receptor.
A prominent reduction in cancer cell affinity was observed from PBTE reformatting, tested by flow cytometry using FSHR-transduced OVCAR3. This was visualized by its 2-log right-shifted MFI curve compared with the BTE and quantified by a >120 fold increase in KD. Effects on T cell affinity were also observed, greater here than for the CA9-PBTE, by measure of a 4.5-fold higher KD compared with the BTE (figure 6A,D). This PBTE was then linked with a second anti-FSHR scFv to form the PMTE, which to OVCAR3-FSHR led to an 11-fold recovery in binding strength with the PMTE compared with the PBTE. PMTE formatting also affected T cell binding kinetics, where a threefold recovery in KD was observed compared with the PBTE. This left the PMTE with a minimal, 1.6-fold higher T cell KD than the BTE while the PBTEs were higher by >4 fold (figure 6A,D).
The combined effects of these binding alterations on cytotoxic potency were subsequently tested on the monitored co-incubation of treated OVCAR3-FSHR and T cells. After a 5-day incubation, the PBTE displayed a sizeable, 150-fold reduction in potency compared with its predecessor format. However, with an ~11 nM EC50 compared with the PBTE’s 710 nM, the PMTE format generated an almost full recovery of cytotoxic potency (figure 6C,E). The end result is an FSHR-targeted molecule that can benefit from half-life extension without sacrificing potency.
Discussion
Reported here are three bispecific antibodies targeting CA9 for the treatment of advanced ccRCC, among them a comparative, multidisciplinary characterization to better understand the functional impact of advancing formatting modifications across both fields of recombinant biologics and nucleic acid medicines. Specifically, we developed a novel, CA9xCD3 PBTE whose scFc enabled prolonged exposure and improved tumor control compared with the first-generation BTE, after three recombinant doses. Through a new application of intercellular-force sensors did we find, however, that there existed a previously unknown toll for its scFc-driven half-life extension, taken in the form of avidity that would consequently impact potency and synDNA efficacy. Also reported is the PMTE, a novel, multivalent format in antibody therapeutics containing a third scFv in tandem, here providing CA9 bivalency although capable of trispecificity in future iterations. This format was found to not only recover potency losses incurred with the PBTE but to overcompensate through outperformance of the BTE. To test the reproducibility of these formatting effects across other targets, an FSHR-targeted BTE was reconstructed into additionally novel PBTE and PMTE molecules. Here, the PBTE experienced an even greater loss of function in exchange for its expected half-life extension. While PMTE remodeling did not gain more function than the BTE in this iteration, it still drove a nearly full potency recovery from the PBTE, in support of our previous findings. Thus, the PMTE format could dual-wield half-life extension and higher potency, free from the dilemma faced by the other formats.
Although potent, first-generation BTEs are limited by a short half-life on account of their small size. In recent years, an array of more IgG-like formats with improved PKs have undertaken extensive clinical investigations for diverse indications.6 While the half-life extending benefits of these redesigns are understood, there has thus far been little to no insight into the potential negative consequences thereof, through side-by-side characterization. In this study, the PBTE was selected for such an analysis due to its relatively simple scFc annexation to the conventional BTE, a single change that made for a clear evaluation of the domain’s effects on the structure–activity relationship. We found that dramatic plasma PK improvements were indeed realized by a CA9-targeted PBTE, which sported a 32-fold improved half-life over its single-chain BTE relative. Costs came to bear, however, in the form of reduced affinity for target cells. It is possible that the scFc imposes a steric effect, alters protein folding, or obstructs some portion of random molecular collisions from direct, productive contact between ligand and receptor, despite encouraging a more favorable hydrophobic effect on binding entropy. In future studies, optimization of the linker configuration connecting the scFc and variable domain orientations may also help reduce such effects. Unifying the antibody’s bilateral KD changes to capture the net effect on synaptic strength prompted our novel application of zMovi technology for bispecific antibodies, which quantitatively yielded the PBTE’s threefold weaker force resistance than the BTE. Its less-stable synapses best explain its lower potency, observed with T cell activation cell surface markers before in cytotoxicity, as seen by its 33-fold lower potency against A-498 than the BTE. Moving in vivo, a PK analysis of tumor distribution informed its lower intratumoral, molar concentration than the BTE. It is possible that its reduced CA9 affinity contributed to lower tumor retention but differences in MW were not controlled for in this study. Be that as it may, in mouse tumor modeling with ACHN-CA9 and a 5-day recombinant dosing interval, the enduring plasma PK profile of the PBTE drove its far-improved efficacy to the BTE which showed little difference from control. This is in spite of its tumor-distribution PK profile that details an intratumoral concentration lower than that which is required to induce ACHN-CA9 cytotoxicity, informed in vitro. However, the efficacy model benefitted from twice the dose and the compounded accumulation of three treatments. The PK analysis also quantified a homogeneous tumor concentration while it is more natural for bispecific antibodies to generate a penetrative concentration gradient, with the highest peripheral concentrations driving outward-in activity.35 Ultimately, by delivering efficacy through a more translationally suitable dosing scheme, the novel PBTE could potentially be considered a more advanced bispecific format and recombinant therapeutic for CA9+ccRCC than the BTE, notwithstanding its described functional attenuations. Not more so than the PMTE, however, which appeared capable of recovering such activity.
The PMTE was chain-linked with a third-binding domain to afford it greater avidity than its monovalent relatives. This was influenced by previous studies demonstrating improved binding, potency, and tumor distribution with higher valency for TAAs.11 13 Indeed, the PMTE realized an impressive binding recalibration with upwards of 17.6-fold higher binding to CA9+cells than the PBTE, and 4.7-fold higher than the BTEs. The net effect on intercellular synapses was one of augmentation compared with the PBTE, with the PMTE resisting three times as much breakage force when bound T cells were stripped from CA9+cells within zMovi. At comparable strength to the BTE, this represented a functional recovery from the reductive effects of the scFc appendage observed with the PBTE. Yet, despite BTE-equivalent synaptic fortitude, the PMTE commanded more potency. Only it significantly activated T cells at femtomolar-range concentrations by measure of CD69 expression and it displayed unmatched cytotoxic activity in the femtomolar range by killing up to 25% cytotoxicity of RCC cells at 100 fM, demonstrating an uncommon degree of potency. Additionally, the FSHR-PMTE made a more significant recovery of potency than it did for OVCAR3-FSHR binding. While this can be partially explained by an additional T cell binding recovery in this iteration, it is possible that the PMTE format is less prone to lysosomal degradation than the BTE due to FcRn-mediated recycling, which would leave higher remaining concentrations over time for enduring activity.36 In mice, a PK analysis of the anti-CA9 panel generated a profile for the PMTE not unlike the PBTE, although with a mildly shorter, 3.9-day half-life. Regardless, the PMTE maintained the benefit of a 27-fold longer half-life than the BTE with which it could potentially ease clinical dosing demands. Of note, these effects are likely driven primarily through higher MW, as these mice express nonhuman FcRn with no cross-reactivity to human FcRn.37 It is probable that future modeling in FcRn-humanized mice would grant the FcRn-mediated recycling benefit of a prolonged circulation.36 The PMTE also saw improvements to PK tumor distribution, modeled with ACHN-CA9. It accumulated to double the molar concentration than that of the PBTE to signify another impressive recovery from BTE concentrations, something that could make or break activity at later time points and lower doses, or compound with additional doses. The PMTE’s 2×1 design does more than just increase absolute CA9 binding strength, but it does so in relativity to CD3. To clarify, the ratio between the molecule’s CA9:CD3 binding strength has been shown to drive preferential compartmental distribution due to the greater influence from either the tumor or T cell binding arm.33 With this consideration, the PMTE’s antiparallel increase in CA9 avidity and decrease in CD3 avidity together generate an advantageous intrinsic binding strength ratio that could more firmly commit the antibody to tumor retention, relative to the other formats. From here, PK/PD relationships can begin to form by intertwining PK profiles with respective potencies against ACHN-CA9 in vitro, against which the PMTE maintained activity at lower concentrations than the PBTE. Maximum tumor concentrations and potencies, respectively, establish upper and lower limits on biological activity for any given dose, thus indicating that the PMTE expands the dynamic range of half-lives spent at effective concentrations in tumors, signifying a more lasting PK/PD relationship. It also generated a nearly threefold higher tumor:plasma ratio than the PBTE, arbitrarily used to compare relative compartmental distribution. Interestingly, the BTE maintained higher intratumoral concentrations than the PMTE, likely attributed to more thorough penetration on account of its smaller size and Fick’s Law, which governs rates of diffusion in inverse, exponential proportionality to molecular radii.38 Counterintuitively, this may not be optimal as more homogeneity mitigates localized, high effective concentrations and trades them for a uniform yet lower concentration that will work submaximally and last more fleetingly above the effective threshold. This is in contrast to larger molecules like the PMTE whose slower diffusion will maintain higher, peripherally localized concentrations that could, therefore, maximize enduring activity.35 In any case, the experimentally dissected features of the PMTE were functionally reintegrated through mouse-efficacy modeling. Where the BTE had no effect and the PBTE had moderate activity with only three total doses, the PMTE demonstrated the most pronounced tumor control after leaving a 4.5-fold lower tumor burden than the latter format. Taken together, these findings indicate that the PMTE could be the superior recombinant biologic among the panel. Yet, demand remained for both routine dosing and the laborious bottleneck of recombinant antibody purification that limited dosing, mirroring the great challenges faced with large-scale manufacturing. Finding clear solutions thereof meant focusing further formatting inspection through the lens of the drug delivery platform.
synDNA provides genetically encoded instructions to patient cells for durably self-expressing their own medicines built to weaponize the immune system against cancers. We reasoned that coupling second-generation bispecific antibodies to this delivery platform could, therefore, have an additive effect on exposure, and in turn, efficacy. Counterintuitively, no meaningful differences existed between dBTE and dPBTE killing activity in vitro from mouse plasma, although both still actively demonstrated impressive endurance of 90 days from a single dose in the latter assay. This, together with the PBTE’s earlier decline from peak expression levels by flow cytometry, could suggest a potentially less-enduring synDNA expression profile. The dPMTE plasma did, however, display the hypothesized additive effect, both in flow staining with 2-log population shifts beyond the dBTE that speak best for its relative plasma accrual and in the concomitant lethal suppression of SKRC-52 in vitro from a likely combination of greater exposure and potency. It is possible that protein stability differences could underpin its more prominent expression profile compared with the PBTE and further studies are required to elucidate this behavior. Tumor modeling began with a preliminary test of administering one dose of 100 µg dBTE to A-498 tumor-bearing mice; this single dose impressively elicited complete tumor control. Subsequent modeling instead relied on 10 µg doses to avoid signal saturation and better distinguish activity between treatment groups in a full study. Nonetheless, the dPMTE did well to swiftly eliminate and largely suppress both primary A-498 and metastatic SKRC-52 tumors with up to 20-fold better tumor control than the less efficacious dBTE and dPBTE, between which the former performed better in both models, possibly due to its relatively higher potency and tumor distribution. This is likely not attributed to distinct transfection efficiency differences among the plasmids, as small changes to the sequences of supercoiled plasmids have not been found to significantly affect electroporation-assisted transfection.39 All in all, these data demonstrated the impressive functional benefits and potential of synDNA, to be considered alongside the platform’s advantageous manufacturing and distribution requirements. Principally did they support the dPMTE, whose propensity for synDNA expression could offer a new approach for sustaining drug delivery to a disease in demand of new medicine.
Advanced ccRCC is an aggressive form of cancer for which there are no BTEs in the clinic or its advanced trials. To enable this targeted approach, there is the central TAA in ccRCC, CA9, whose high and homogeneous expression in 90% of cases together with its pro-tumorigenic function make it a strong candidate for targeted therapy.14 17 In healthy tissue, CA9 is expressed at low levels in the biliary tract which caused on-target, off-tumor toxicities in an early CAR-T cell clinical trial.40 However, anti-CA9 monoclonal antibodies were well tolerated and their use for pretreatment in a follow-up study shielded healthy tissue from CAR-T cells and prevented toxicities.41 It is thus possible that BTEs could induce hepatic toxicity, although the same protective pretreatment could be followed. Alternatively, BTEs can be affinity tuned to change the therapeutic index and reduce off-target binding to lowly expressing cells and it is possible that the valency of the PMTE format in particular could permit better distinguishment between high-target-expressing and low-target-expressing cells, to improve clinical safety.13 Lastly, synDNA could be investigated for local delivery at the tumor site to reduce systemic exposure of dBTEs and circumvent potential off-target toxicities. Our selected TAA, therefore, could be highly tumor targetable with a PMTE treatment approach to, in turn, fill a gap in the immunotherapy landscape of this disease.
To conclude, single-chain bispecific antibodies are a potent and promising immunotherapy, here inciting the eradication of ccRCC outfitting CA9. Our PBTE is a novel molecule constructed in a second-generation format to secure immune engagement with less-frequent dosing while the multivalent PMTE is both a novel molecule and format in antibody therapeutics that here elicits greater anticancer immune activity than its single-chain relatives. Through synDNA, all formats sustained durable expression with the dPMTE in particular exhibiting impressive functional longevity and pronounced tumor control, supporting this promising delivery platform for antibody therapies. Their activity is a testament to their compelling target, CA9, an aberrantly upregulated and widely expressed antigen in kidney cancers, which are a public health crisis that lack the clinical application of bispecific antibodies and investigational trials thereof, despite the diseases’ encouraging immune responsivity. Thus, we present this CA9-targeted dPMTE as a new immunotherapy for kidney cancers like ccRCC, alongside the format itself for recommitment to other cancers and their antigens.
Methods
Antibody design and synthesis
To create CA9xCD3 engagers, heavy and light chain sequences were sourced from the CA9-targeting clone G37.42 G37 is a human antibody with high affinity and no observable propensity to induce antigen internalization that could otherwise attenuate a treatment’s potency.43 This clone was also selected for its membrane-proximal epitope, which could optimize potency through tighter cellular synapses.27 CD3 targeting, on the other hand, was enabled by equipping a humanized, high affinity UCHT1 scFv.44 For single-chain construction, human IgG1 CH2 and CH3 domains constitute the N-terminal Fc domain, which are united in duplicate by GS linkers ahead of C-terminal binding domains. His tags were included for detection in early studies. Fc domains were silenced with mutations ablating binding to both FcγR (L234A/L235A) and complement (D270A/K322A) to avoid potential toxicities or reductions in efficacy that could arise from inadvertent Fc-FcγR interactions while leaving FcRn binding intact.33 45 46 Protein sequences for the three BTE designs were codon-optimized in pVax1 expression vectors for expression in humans and mice using GenScript services. Expression in Expi293 cells preceded supernatant purification by way of HPLC and SEC for Nanodrop quantification.
Cell lines and animals
SKRC-52 cells were acquired from Memorial Sloan Kettering Cancer Center. Remaining cell lines were purchased from ATCC. The plasmid vector used to retrovirally transduce 293 T cells with CA9 was acquired from GenScript. Stable expression was achieved following a previously described protocol.26 Primary human T cells were derived from healthy donors by The Human Immunology Core at the University of Pennsylvania. Balb/c and NuJ mice were obtained from The Jackson Laboratory. NSG mice were purchased from The Wistar Institute Animal Facility. All animal experiments were done with approval from the Institutional Animal Care and Use Committee at The Wistar Institute.
Immunoblotting
Denaturation and western blotting were done as previously described.26 Briefly, samples were separated using NuPAGE Bis-Tris gels and transferred to iBlot PVDF membranes (Invitrogen). Membranes were blocked with LI-COR blocking buffer and incubated overnight at 4°C with anti-his primary antibody (Invitrogen). Washing with .1% Tween 20 in PBS preceded a 1-hour incubation with LI-COR detection secondary antibodies. Images were taken using an Odyssey CLx (LI-COR BioSciences).
Cell binding
Cells were stained with Zombie violet Live/Dead (Biolegend) ahead of the primary stain with bispecific antibodies in PBS 2% FBS, whether they be from recombinant purification or mouse plasma. Cells were washed twice with PBS 2% FBS before an anti-his tag secondary stain (Jackson ImmunoResearch). Following an additional three washes, binding was assessed by flow cytometry using a BD FACSymphony (BD Biosciences). The sorting of transduced cells involved staining with a commercial anti-CA9 antibody (Invitrogen) before sorting on a BD FACSAria II (BD Biosciences). Sorting was conducted by The Wistar Institute Flow Cytometry Core.
Avidity force
HEK-293T cells were dissociated from culture flasks using TrypLE (Thermo Scientific, 12604013) for 5 min at 37°C and prepared in a single-cell suspension at a concentration of 6×107 cells/mL. The suspended cells were added to the z-Movi (LUMICKS Inc.) microfluidic chips coated with poly-D-lysine (20 ug/mL; Thermo Scientific, A3890401) for at least 3 hours of attachment. Fresh naïve T cells were stained with CellTrace Far Red Cell Proliferation Kit (Thermo Scientific, C34564) according to the manufacturer’s protocol 1 hour prior to the assay and resuspended at a concentration of 2×107 cells/mL. Treatments were prepared at a twofold higher concentration and added to an aliquot of T cells 5 min prior to the start of the assay. 20 µL of antibody-T cell suspension was introduced into the z-Movi microfluidic chip and incubated with the target HEK-293T cells for 5 min. Following incubation, an acoustic force ramp from 0 to 1000 pN was applied within the microfluidic chip and cell detachment was observed using real-time fluorescence imaging on the z-Movi system. Each microfluidic chip was used for serial escalating doses of a single antibody.
T cell activation
Target cells were plated between 5×103 and 2×104 per well in R10 and incubated overnight. T cells were introduced to the target cells at given E:T ratios along with 10 nM treatments and all conditions assessed in duplicate. Plates were incubated for 24–48 hours at 37°C. T cells were distinguished with CD4 and CD8 antibodies (Biolegend) and activation states assed with CD69, PD1, and CD25 antibodies (Biolegend). Staining was conducted using 2% FBS in PBS buffer and incubation took place at 4°C for 30 min flanked by washes.
Cytotoxicity
3×103–2×104 target cells per well were plated 3 hours before the intended addition of treatment and introduced to the xCELLIGENCE RTCA system. Human donor T cells were added to cancer cells at an effector ratio of 5:1 alongside serially diluted recombinant antibody or dBTE-treated mouse plasma, in duplicate. Control groups received additional R10 in the absence of T cells or antibody. No exogenous cytokines were used as supplementation. Impedance data collected from xCELLIGENCE RTCA eSight (Agilent Technologies) was used as an indirect measurement of cytotoxicity. Units are arbitrary and expressed as cell index. The plate reader was housed in a CO2 incubator allowing measurements for 24–72 hours, during which time cell index was recorded every 15 min and images taken every hour. The experimental duration for individual cell lines was dependent on cell line-distinct onset of T cell activation. Percent cytotoxicity was calculated at the terminal time point accordingly: (1–(treatment/T cell control))×100).
ELISA
ELISA plates (Corning) were coated with anti-his capture antibody (Invitrogen) at 2–10 µg/mL and incubated overnight at 4°C. Blocking relied on Superblock T20 (Thermo Fisher) between washes with 0.05% Tween 20 in PBS. Plasma samples were diluted in 1% FBS/0.05% Tween 20 in PBS and incubated at RT for 2 hours. To accommodate differences in the number of F(ab) domains and MW among treatments, plasma from distinct groups was plated separately, each with a standard curve made from the recombinant counterpart of that antibody treatment. After the primary incubation came two more washes, preceding a 1 hour, secondary incubation with HRP anti-human F(ab) antibody (Jackson) diluted to 1:10 000. Reactions with 1-Step Ultra TMB substrate (Thermo Fisher) were quenched with 4% H2SO4.
Plasma PK
After a baseline blood draw by check bleed, Balb/c mice were anesthesized under 4% isoflurane for an intravenous dose of 2.5 mg/kg recombinant antibody in PBS via tail vein. Subsequent microsample blood samples were consecutively drawn from the open cheek for short-term time points, with long-term time points collected from the adjacent cheek. Plasma was isolated from blood for ELISA detection of circulating antibody. PK parameters were calculated using a single-compartment model. AUC was determined in Graphpad Prism V.9. VD was calculated as VD=dose/Co. The elimination rate constant was calculated as Ke=(ln(Cx)–ln(Cy))/(Ty–Tx). Clearance was determined by Cl=VD×Ke while half-life was inferred by T1/2=0.693/Ke.
Tumor PK
2×105 ACHN-CA9 cells were suspended in a 50% mixture of Matrigel (Corning) in PBS. Cells were injected SC in the right flank of NSG mice under anesthesia with 4% isoflurane. Tumor growth was measured over time using calipers and tumor volume calculated using the formula [volume=(length×width2)/2]. When the average tumor volume reached 200 mm3 after 10–14 days, mice were randomized to groups while normalizing average group tumor volume. Mice were given a 0.3 mg/kg dose of recombinant antibody suspended in 100 µL PBS by intravenous tail injection. After 24 hours, blood was drawn by cheek blood in plasma collection tubes and mice were subsequently euthanized in a CO2 chamber and perfused with PBS. Tumors were dissected from the mice and added to tubes containing RIPA Lysis & Extraction buffer (Thermo) and Pierce protease inhibitor (Thermo) brought to 100 mg/mL. Dissociation was conducted with a Precellys Evolution Homogenizer (Bertin) with 3, 15 s cycles at 8800 RPM. Supernatant was collected and centrifuged at 15 K g for 30 min. Antibody concentrations in plasma and tumor samples were quantified by ELISA.
In vivo expression
Balb/c mice were anesthetized with 4% isoflurane and administered 200 µg of relevant plasmid with 50U HYA in the tibialis anterior (TA). The muscle was then electroporated using a CELLECTRA device (Inovio Pharmaceuticals). Blood was collected via cheek bleed at the given time points and plasma was isolated with MiniCollect Plasma Tubes (Greiner Bio-One). dBTE concentration was calculated using a His Tag ELISA Detection Kit (GenScript) and the manufacturer’s protocol.
Tumor challenges
3.0×105 ACHN-CA9, 2×106 A-498, or 2×106 SKRC-52 cells were suspended in a 50% mixture of Matrigel (Corning) in PBS. Cells were injected SC in the right flank of Nu/J or NSG mice under anesthesia with 4% isoflurane. Tumor growth was measured over time using calipers and tumor volume calculated using the formula [volume=(length×width2)/2]. When the average tumor volume reached 50 mm3 after 10–14 days, mice were randomized to groups while normalizing average group tumor volume. On day 0, mice were injected IP with 10×106 human T cells suspended in PBS and dosed with antibody therapy. For recombinant models, treatments followed a 3×5 dosing regimen wherein they were administered IP in 100 µL PBS on days 0, 5, and 10 for a total of 3 treatments. For DNA models, mice were anesthetized with 4% isoflurane ahead of plasmid vector administration. The expression vectors, alongside 50U HYA, were administered IM in the tibialis anterior before electroporation using a CELLECTRA device (Inovio Pharmaceuticals). All mice were sacrificed when tumors reached 2.5×103 mm3 in accordance with IACUC protocols.
Graphing and statistics
Data were processed and visualized using the Tidyverse package within RStudio or with Graphpad Prism V.9. Flow cytometry data were processed and visualized using FlowJo V.10 while zMovi data analysis was performed using Oceon V.1.2.8 software. Ordinary one-way analysis of variance (ANOVA) tests were used to compare KD values, zMovi avidity AUC, T cell activation markers, and the tumor PK study. Depending on group number, ordinary one-way ANOVA tests or unpaired t-tests were used to compare tumor volumes in mouse tumor models.
Data availability statement
Data are available on reasonable request. All data generated or analyzed during this study are included in this published article or its online supplemental information files.
Ethics statements
Patient consent for publication
Acknowledgments
We would like to thank The Wistar Institute’s Animal Facility for providing care to the animals and Flow Cytometry Core for providingflow cytometers. We would also like to thank the University of Pennsylvania Cell Center for providing donor immune cells and MSK for providing the SKRC-52 cell line.
References
Footnotes
Contributors Conceptualization: RPO'C and DBW. Methodology: RPO'C, KL, CAC, PSB and NS. Investigation: RPO'C, KL, NW, CAC, PSB, DB, DP and NS. Visualization: RPO'C. Funding acquisition: DBW, CHJ and DK. Project administration: DBW. Supervision: DBW, CHJ and DK. Writing–original draft: RPO'C. Writing–review and editing: RPO'C, DBW, NW, PSB and DB. Guarantor: DBW.
Funding This study was supported by T32 CA-115299-15 (to RPO'C), The Jill and Mark Fishman Foundation (to DW) and the WW Smith Charitable Trust (to DW). The Wistar Institute cores were supported by #CA010815. This study received funding in part by Inovio Pharmaceuticals, SRA 21-04 (to DBW), in association with the electroporation device used for animal experiments.
Competing interests Authors (DW and RPO'C) have filed a patent with The Wistar Institute based on this work. DW has received grant funding from industry for sponsored research collaborations, he has received speaking honoraria, and received fees for consulting or serving on scientific review committees. Remunerations received by DW include direct payments and equity/options. DW also discloses the following associations with commercial partners: Geneos consultant/advisory board, AstraZeneca advisory board and speaker, Inovio board of directors and consultant, Sanofi advisory board, BBI advisory board, Pfizer advisory board, Flagship consultant, and Advaccine consultant. CHJ is a scientific co-founder and holds equity in Capstan Therapeutics, Dispatch Biotherapeutics and Bluewhale Bio. CHJ serves on the board of AC Immune and is a scientific advisor to BluesphereBio, Cabaletta, Carisma, Cartography, Cellares, Cellcarta, Celldex, Danaher, Decheng, ImmuneSensor, Kite, Poseida, Verismo, Viracta, and WIRBCopernicus group. The other authors declare that they have no competing interests.
Provenance and peer review Not commissioned; externally peer reviewed.
Author Information DBW is the Executive Vice President and Director of the Vaccine & Immunotherapy Center at The Wistar Institute. He is the WW Smith Distinguished Chair in Cancer Research and Professor Emeritus at University of Pennsylvania Perelman School of Medicine. CC is currently employed by Immudex LLC (Fairfax, VA, USA).
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.