Adaptable antibody Nanoworms designed for non-Hodgkin lymphoma

Abstract

Despite advancements in antibody-based therapies for non-Hodgkin lymphoma (NHL), at least two major therapeutic needs remain unmet: i) heterogenous activation of host immunity towards B cell NHL; and ii) lack of antibody-based therapeutics for T cell NHL. This study explores the molecular characteristics of an adaptable modality called antibody Nanoworms and demonstrates their receptor clustering activity as a means to overcome and address abovementioned needs. To test this, four selected therapeutic receptors of B cell (CD19, CD20, HLA-DR10) and T cell (CD3) NHL were targeted by Nanoworms. Regardless of the target or the cell type, Nanoworms inherently clustered bound receptors on the cell-surface through their multivalency and activated intracellular signaling without any secondary crosslinker. As a sole agent, Nanoworms induced apoptosis by clustering CD20 or HLA-DR10, and arrested the cell cycle upon CD19 clustering. Interestingly, CD3 clustering was particularly advantageous in inducing activation-induced cell death (AICD) in an aggressive form of T cell NHL named Sézary syndrome that is fatal, limited in antibody-based therapeutics, and has poor outcomes to traditional chemotherapy. As Nanoworms can be easily designed to target any receptor for which a scFv is available, they may provide solutions and add therapeutic novelty to underserved diseases.

1. Introduction

Defined by the United States National Cancer Institute, lymphoma arises from malignant lymphocytes that can be divided into two categories, Hodgkin lymphoma and non-Hodgkin lymphoma [1]. Hodgkin lymphoma can be often cured by standard therapy whereas non-Hodgkin lymphoma (NHL) is more progressive and has a 5-year survival rate around 71%. Estimated by the American Cancer Society, the overall chance of a man will develop NHL in his lifetime is about 1 in 41; that for a woman is about 1 in 52.

NHL can be sub-categorized into two types: B cell NHL and T/NK (natural killer) cell NHL [2]. B cell NHL makes up 85–90% of total cases, while T cell NHL makes up less than 15%. Because of the disproportionate incidence rates between the two types, the majority of therapies address B cell NHL. 23 antibody-based immunotherapies have been approved by the USFDA for B cell NHL [3], whereas only two are approved for T cell NHL. These are brentuximab vedotin (anti-CD30 monoclonal antibody drug conjugate with monomethyl auristatin E) and mogamulizumab (anti-CCR4 monoclonal antibody) [4]. The addition of therapeutic antibodies to traditional chemotherapy significantly improved the prognosis of B cell NHL [5,6]; however, the heterogenous involvement of host immune system continues to result in sub-optimal therapeutic responses and fatal adverse events [7-11]. Moreover, T cell NHL deserves better antibody-based therapeutic options.

This report addresses these needs using the modality that shares mechanisms of action with that of existing therapeutic antibodies [12]. The modality called antibody Nanoworm is built using elastin-like polypeptides (ELPs). Recombinant fusions between an ELP and a single chain antibody fragment (scFv) are produced as homogenous fusion proteins. Upon purification, these fusions self-assemble into colloidally stable and biologically active nanomaterials with a distinct worm-like morphology (hence, Nanoworms) that exhibit exceptionally high degrees of multivalency. Upon binding to cell-surface receptors, they promote lateral engagement of multiple cell surface receptors and subsequent activation of apoptosis, without the requirement for any additional crosslinkers or other nanostructures.

This study evaluates four different antibody Nanoworms that target therapeutically relevant cell surface receptors on B cells (CD20, HLA-DR10, and CD19) and T cells (CD3). This panel was chosen to explore three aspects. First, CD20 and HLA-DR10 have been extensively studied as therapeutic targets in the context of Fc-FcR mediated host immune system activation and its therapeutic effect. Because the Nanoworms lack the Fc domain, these are the most suitable candidates to test whether the Nanoworms can achieve therapeutic equipotency only by clustering cell surface receptor in B cell NHL. Second, CD19 clustering was also explored with its therapeutic implications towards B cell NHL. CD19 has become a prominent therapeutic target with the advent of BiTE [13] and CAR-T [14] therapy; however, its therapeutic effect upon clustering had acquired less attention. Third, to determine if molecular traits of Nanoworms can be applied to other NHL models, CD3 clustering Nanoworm was tested in the model of Sézary syndrome, an aggressive cutaneous T cell NHL [15]. As a rare disease, Sézary syndrome lacks reliable diagnostic biomarkers and has a 5-year survival of only 26% [16]. Sézary cells are resistant to activation-induced cell death (AICD) and one of the mechanisms responsible for this resistance is the lack of T cell receptor (TCR) signaling [17]. Therefore, particular attention was given to this disease in the context of CD3 clustering to determine if CD3 clustering Nanoworms induce AICD.

The dynamic clustering of cell-surface receptors induced by antibody Nanoworms was observed in real-time using the experimental method pioneered by our group to detect the rapid and reversible segregation of the cytoplasmic proteins [18] or membrane-bound receptor [19] in live cell cultures and in zebrafish [20] relative to the controlled application of heat. Using this powerful technique, we present the real-time, visual evidence of Nanoworm-mediated clustering of cell surface receptors, the exact temperatures at which that occurs, and simultaneous induction of membrane blebbing, which all directly support the mechanistic studies of activation and apoptosis.

A particularly compelling feature of this study relates to the therapeutic activity observed in T cell NHL. Nanoworms that clustered CD3 on HuT78 cell line (a Sézary syndrome model) induced activation-induced cell death (AICD), which the OKT3 (a widely studied CD3-targeting mAb) treatment was unable to achieve. Nanoworm-mediated AICD was even more potent than CD3 clustered by OKT3 supplemented with a secondary crosslinker (OKT3+2°). Furthermore, co-incubation of OKT3 and CD3 clustering Nanoworms showed an additive effect in T cell activation. The AICD and additive effect on T cell activation were exclusively observed in HuT78 cell line but not on other CD3⁺ T cell lines, such as CEM or Jurkat. These observations suggest that CD3 clustering via Nanoworm may be an effective approach to overcome resistance to AICD in Sézary syndrome.

2. Materials and methods

2.1. Synthesis, expression, and purification of Nanoworms

The pET-25b(+) vector was purchased from Novagen (#69753) and further modified for ELP fusion cloning [21]. For the four scFv-ELP fusions, the DNA encoding scFv was cloned to the N-terminus of ELP A192. The amino acid sequence of scFv that targets CD20 was derived from RTX (rituximab) and optimized for bacterial expression. The amino acid sequence of scFvs that target CD19 and CD3 were derived from bispecific antibody blinatumomab and optimized for bacterial expression. The amino acid sequence of scFv that targets HLA-DR10 was derived from the chimeric Lym-1 (chLym-1) monoclonal antibody [22,23]. For the four anti-CD20-ELP fusions, the DNA encoding anti-CD20 scFv used above was cloned to the N-terminus of ELP V2A64, A96G96, S192 or G192. These four ELPs were chosen because they contain identical number (n = 192) of VPGXG pentameric motif and have similar MW but differ in their solution T_t (Table S2). The cloned constructs were sequenced, transformed into and expressed in Shuffle® T7 Express competent E. Coli (#C3029J, NEB, Ipswich, MA, USA) fermented in terrific broth media for 16–18 h at 30 °C without IPTG induction. After bacterial cell lysis (S-4000 Ultrasonic Disintegrator Sonicator Liquid Processor, Misonix, Inc. NY, USA; Amplitude 9, 18 repeats of 10 s on + 20 s off cycle) and clarification of cell debris by centrifugation at 16,100 rcf for 10 min at 4 °C in a Beckman J2-21 Centrifuge, the supernatant was equilibrated to room temperature and ELP-mediated phase separation was induced by 2 M sodium chloride at room temperature (i.e., dissolve 0.12 g NaCl powder per 1 mL cleared lysates by gently inverting until transparent lysates become opaque). Coacervates were pelleted at 5000 rcf for 10 min at 25 °C using a Sorvall RC-3C Plus Centrifuge immediately after the phase separation was observed (hot-spin). After each hot-spin, soluble impurities (supernatant) were removed, and coacervates (pellet) were resolubilized in ice-cold dPBS (#25–508, Gene-see Scientific, San Diego, CA, USA). Thoroughly resolubilized ELPs were centrifuged at 16,100 rcf for 10 min at 4 °C in an Eppendorf 5415 R Centrifuge (cold-spin). At the end of each cold-spin, insoluble impurities (pellet) were again removed by transferring the supernatant to a clean tube. Cycles of hot-spin followed by cold-spin were repeated 2 times to achieve the necessary purity and yield. Purified materials are processed either for refolding or stored at −20 °C for further use.

The identity and purity of Nanoworms were analyzed using SDS-PAGE. The molar extinction coefficient (ε) of aCD20A, aHLADR10A, aCD19A, and aCD3A Nanoworms was calculated at 60,855, 45,630, 59,570, 60,435 M⁻¹ cm⁻¹, respectively [24]. Serial dilutions of aCD3A Nanoworms in Edelhoc buffer were prepared, measured and averaged to acquire the best estimate of concentration in dPBS using Eq. (1) [24,25]. Serial dilutions of aCD20A, aHLADR10A, and aCD19A Nanoworms in Edelhoc buffer were prepared, measured and averaged to acquire the best estimate of concentration in dPBS using Eq. (2) (0.34 = rhodamine correction factor provided by manufacturer).

$$C_{Nanoworms}=\frac{A280-(A333\times 1.929)}{\epsilon }\times Dilutionfactor$$ (1)

$$C_{Rho-Nanowomrs}=\frac{A280-(A555\times 0.34)}{\epsilon }\times Dilutionfactor$$ (2)

For the optical density profile, absorbance at 350 nm, A, was measured in a DU800 UV–Vis spectrophotometer (Beckman Coulter, CA) under a temperature gradient of 0.5 °C/min.

2.2. Labeling and refolding of Nanoworms

Purified Nanoworms (three scFv-A192 fusions except aCD3A and four anti-CD20-ELP fusions) were labeled with 5-fold molar excess NHS-rhodamine (#46406, Thermo-Fisher, IL, USA) for 4 h at room temperature under constant rotation in a 15 mL tube. After the labeling reaction, Nanoworms were denatured by directly adding equilibration solution (20 nM β-mercaptoethanol, 50 mM Tris-base, 500 mM NaCl, pH 8.0) to the labeling reaction solution at a half of the volume of labeling reaction solution, followed by 12 M urea as a powder (i.e., directly add 1 mL of equilibration solution to 2 mL labeling reaction solution and then add 1.44 g (12 M) urea powder calculated based on 2 mL). The tube was gently inverted to completely dissolve urea. Addition of equilibration solution and urea doubled the volume (2 mL → 4 mL), which resulted in 6 M urea as a final concentration. The entire solution was under constant agitation for 10 min and then subjected to refolding process using a 10 kDa MWCO dialysis tubing (#68100, Thermo Fisher, IL, USA) via stepwise dialysis at 4 °C. The following buffers were prepared to promote refolding of scFv domains. Buffer was changed 5 times at intervals of 24 h.: Buffer 1 (20 mM Tris base, 150 mM NaCl, 3 M urea, 500 mM l-Arginine, 2 mM GSH, 0.4 mM GSSH, 0.5 mM PMSF, pH 8.0); Buffer 2 (20 mM Tris base, 150 mM NaCl, 1 M urea, 250 mM l-Arginine, 2 mM GSH, 0.4 mM GSSH, 0.5 mM PMSF, pH 8.0); Buffer 3 (20 mM Tris base, 150 mM NaCl, 0.5 M urea, 125 mM l-Arginine, 0.5 mM PMSF, pH 8.0); Buffer 4 (20 mM Tris base, 150 mM NaCl, 62.5 mM l-Arginine, 0.5 mM PMSF, pH 8.0); Buffer 5 (1x dPBS). Excess rhodamine was removed during this process.

Refolded material was centrifuged at 16,100 rcf for 10 min at 4 °C (Eppendorf 5415 R Centrifuge, Eppendorf AG, Hamburg, Germany), and the supernatant was syringe filtered (450 nm pore, #PN 4614, Pall Corp., NY, USA). At the end of all steps (labeling, refolding, and final purification) degree of labeling was 300–500% per mole protein, which equals to 3–5 mol rhodamine per mole protein. Empirically, labeling was a prerequisite for aCD20A, aLADR10A, and aCD19A Nanoworms for its activity, whereas aCD3A Nanoworms did not require prelabeling. Therefore, aCD3A Nanoworms were immediately subjected to denaturation and refolding after purification. RTX was labeled with 15 M excess NHS-rhodamine for 2 h at room temperature under constant rotation and dialyzed against dPBS for 3 days in a sink condition.

2.3. Cryo-TEM imaging

For cryo-TEM, 5 μM Nanoworm suspended in dPBS were evaluated. Lacey carbon film copper grids (300 mesh, Electron Microscopy Services, Hatfield, PA, USA) were pretreated with plasma air for 30 s to render the lacey carbon film hydrophilic. A 6 μL sample was applied on the grid using a Vitrobot (FEI, Hillsboro, OR, USA) that was maintained at 95% humidity. Following 1 min incubation, blotting was performed using Vitrobot preset parameters and the grid was immediately plunged into a liquid ethane reservoir precooled by liquid nitrogen. Grids were then transferred to a cryo-holder and cryo-transfer stage that were precooled with liquid nitrogen. A FEI Tecnai 12 Twin Transmission Electron Microscope, operating at 100 kV, was used to perform all imaging. The cryo-holder was maintained below −170 °C with liquid nitrogen to prevent the sublimation of vitreous water during the imaging process. All images were recorded with a 16-bit 2 K × 2 K FEI Eagle bottom mount camera (Hillsboro, OR, USA). The length and the width of Nanoworms were measured using ImageJ (v2.0.0, NIH, MD) based on the reference length presented in Fig. 1D.

Fig. 1.

scFv-A192 fusions form stable worm-like structures upon expression and refolding. (A) Design of scFv-A192 fusions. Recombinant DNAs encoding scFv were fused to the amino terminus of ELP A192. (B) SDS-PAGE was used to evaluate the purity of the rhodamine-labeled scFv-A192 fusions. The identical gel was first imaged for fluorescence and then imaged after Coomassie blue staining. (C) The hydrodynamic radii of Nanoworms remain stable at 37 °C for least 5 days in dPBS reflecting their apparent colloidal stability (10 μM, n = 3, mean ± SD). (D) The morphology of refolded scFv-A192 fusions were visualized under Cryo-TEM, which revealed worm-like structures.

2.4. Cell cultures and time-lapse live cell imaging

All cell lines used in this study (Raji, SU-DHL-7, HuT-78, CEM, and Jurkat) were cultured in RPMI 1640 (Corning, MA, USA) supplemented with 10% FBS at 37 °C without any antibiotics. For time-lapse live cell imaging, 10 μM Nanoworms or 5 μM RTX was incubated with 0.5 × 10⁵ cells for 30 min at 4 ° C under constant agitation. Cells were spun down at 300 rcf, washed 3 times with pre-chilled dPBS, resuspended with a pre-chilled fresh media (without FBS), and mounted on a poly-d-lysine (P7405, Sigma-Aldrich, St. Louis, MO) coated 35 mm glass bottom culture dish (#P35G-0-10-C, MatTek Corp. MA). After 15 min, cells were imaged using a DIAPHOT epifluorescence microscope equipped with a DS digital camera (Nikon Instruments, Minato-Ku, Tokyo, Japan) and a temperature control stage (Linkam Scientific Instruments, Epsom, UK). Temperature of media within the culture dish was measured in real-time with the type K temperature probe (TP870, Extech, NH, USA) connected to a thermocouple thermometer (Model:800005, Scottsdale, AZ, USA) during the temperature increase at a rate of 2 °C/min. Fluorescence images were taken at every 0.5 °C from 15 °C up to 45 °C during the heating. Images were further analyzed to identify the T_c of a cell surface bound Nanoworms using ImageJ (v2.0.0, NIH, MD, USA). The T_c of each Nanoworm was defined as the T_i temperature at which the maximum first derivative of background-corrected fluorescence intensity (dI/dT) was estimated using Eq. (3). The I_i is defined as the background corrected fluorescence intensity measured at T_i temperature. The I_ROI and I_BG are defined as the fluorescence intensity measured within the subregion of the cell where Nanoworm cluster is formed and the fluorescence intensity measured in the region devoid of cells where there is only a baseline fluorescence, respectively, at T_i temperature (Eq. (4)). Brightfield images were taken under the identical condition to observe cell membrane blebbing.

$$\frac{dI}{dT}{\mid }_{{}_i}=\frac{I_{i+1}-I_i}{T_{i+1}-T_i}$$ (3)

$$I_i=(I_{ROI}-I_{BG})\times A$$ (4)

2.5. Target specificity of Nanoworms

To test the specificity of aCD20A and aHLADR10A, 0.3 × 10⁵ Raji cells were incubated with 50 μg RTX or chLym-1 (17.4 μg of anti-CD20 scFv or anti-HLA-DR10 scFv) for 20 min at 4 °C. After 20 min incubation, 6 μg of rhodamine-labeled aCD20A or aHLADR10A (1.5 μg anti-CD20 scFv or anti-HLA-DR10 scFv) was added into the solution and incubated for another 20 min. The total volume did not exceed 100 μL. At the end of second incubation, cells were washed with dPBS twice (300 rcf for 5 min) and resuspended in 20 μL Live Cell Imaging Solution (#A14291DJ, Molecular Probes, Eugene, OR, USA) added with NucBlue™ Live Cell Stain ReadyProbes™ reagent (#R37605, Molecular Probes, Eugene, OR, USA). Cells were mounted on a poly-d-lysine (P7405, Sigma-Aldrich, St. Louis, MO) coated 35 mm glass bottom culture dish (#P35G-0-10-C, MatTek Corp. MA, USA) and imaged under the Zeiss LSM880 Confocal Microscope with Airyscan Fast (Carl Zeiss AG, Oberkochen, Germany). Fluorescence intensities of cells were analyzed using ZEN 2 Blue Edition software (Carl Zeiss AG, Oberkochen, Germany).

To test the specificity of aCD19A and aCD3A, 0.2 × 10⁵ cells (Raji and SU-DHL-7 cells for aCD19A, HuT-78 and Jurkat cells for aCD3A) were incubated with 1.25 μg blinatumomab (0.625 μg anti-CD19 scFv or anti-CD3 scFv) for 20 min at 4 °C. After 20 min incubation, 80 μg of rhodamine-labeled aCD19A or aCD3A (20 μg anti-CD19 scFv or anti-CD3 scFv) was added into the solution and incubated for another 20 min. The total volume did not exceed 100 μL. At the end of second incubation, cells were washed with dPBS twice (300 rcf for 5 min), resuspended in 500 μL dPBS, and analyzed with BD LSRFortessa™ X-20 (Becton Dickinson, Franklin Lakes, NJ, USA). Data were further analyzed with FlowJo™ Software v10.4 (Becton, Dickinson and Company, Ashland, OR, USA).

For the non-treated controls, equal volumes of dPBS was added in place of the antibody and the Nanoworm solutions. For positive controls, equal volume of dPBS was added in place of the antibody and then incubated with respective Nanoworms described above.

2.6. Apoptosis, cell cycle distribution, and cell activation upon Nanoworm treatment

To measure the apoptosis, 0.2 × 10⁵ cells in complete media were incubated with antibody or Nanoworms for 15 min at room temperature in the 1.7 mL tube. A final scFv concentration was 20 μM for both antibody and Nanoworms (10 μM for T cell lines), and the total volume was kept below 90 μL. After 15 min incubation, 80 μL complete media was added and the whole solution was transferred to a well in a 48-well plate. The plate was incubated at 37 °C for 18 h. After 18 h incubation, cells were counted and washed twice with pre-chilled dPBS (300 rcf for 5 min). Cells were resuspended in 100–200 μL hypotonic buffer (20 mM Tris-HCl, pH 8.0) and incubated on ice for 15 min. After 15 min, outer membrane of the cells was removed using a Dounce tissue grinder (#D8938, Millipore Sigma, Burlington, MA, USA). About 80–90% of the cells were devoid of outer membrane after 50 times of grinding with tight pestle. Solutions were spun down (10,000 rcf, 7 min) to remove any outer membrane debris and mitochondria, and only the supernatant (cytosolic fraction that contains cytosolic cytochrome C) was transferred to a clean tube. The level of cytosolic cytochrome C was measured from 100 μL of the supernatant using Human Cytochrome C Quantikine ELISA Kit (#DCTC0, R&D Systems, Minneapolis, MN, USA). Resulting cytosolic cytochrome C levels were normalized to counted cells as a ng/mL and then converted to a fold change. To measure the apoptosis in the HuT78 cell line upon co-treatment of OKT3 and aCD3A, both OKT3 and aCD3A were added to the culture media as a mixture at a final concentration of 5 μM and 10 μM, respectively, and incubated at 37 °C for 18 h. The identical procedure described above was applied hereafter.

To analyze the cell cycle distribution, 0.4 × 10⁵ Raji cells in 200 μL was incubated with 200 μL of aCD19A (0.8 mg/mL) for 15 min at room temperature in the 1.7 mL tube and then incubated at 37 °C for 48 h. A 200 μL of 0.1 μg/mL rapamycin formulated with tween 80 and PEG 400 was used as a positive control [26]. After 48 h, collected cells were washed with dPBS three times (300 rcf for 5 min) and fixed with 70% ethanol for 24 h. After fixation, cells were washed with dPBS twice (900 rcf for 5 min) and resuspended in 100 μL of 100 μg/mL RNAse A (component of Miniprep Kit, #27106, Qiagen, Hilden, Germany). After incubation at room temperature for 1 h, 100 μL of 100 μg/mL propidium iodide (component of Apoptosis Kit, #V13241, Thermo Fisher Scientific, Waltham, MA, USA) was added and incubated for another 30 min. Cells were analyzed using BD LSRFortessa™ X-20 (Becton Dickinson, Franklin Lakes, NJ, USA). Data were further analyzed with FlowJo™ Software v10.4 (Becton, Dickinson and Company, Ashland, OR, USA).

To analyze the T cell activation, cell culture media were collected during the process of abovementioned cytosolic fraction preparation. Collected media were spun down once more (300 rcf for 5 min) to completely remove any cellular components and the transferred supernatant was directly used to measure IL-2 concentration using the Human IL-2 ELISA MAX™ Set Deluxe (#431804, BioLegend, San Diego, CA, USA).

To analyze the co-stimulatory effect in T cell activation, 70 μL of 0.52 mg/mL OKT3 antibody solution was filled into wells in 96-well plate (#423501, BioLegend, San Diego, CA, USA), sealed, and incubated at 4 °C one day prior to the experiment (day 1). On day 2, 0.1 × 10⁵ HuT-78 or Jurkat cells in 70 μL was incubated with 70 μL of aCD3A (serial dilution) or dPBS for 40 min at 4 °C under constant agitation. After this incubation, cells were transferred to OKT3 pre-coated wells in 96-well plate. Another set of cells incubated with aCD3A were transferred to OKT3 uncoated wells in the same 96-well plate to observe the activity of aCD3A without OKT3. The 96-well plate was incubated at 37 °C for 18 h. On day 3, solution from each well was collected, spun down (300 rcf for 5 min), and only supernatant was collected. The samples were stored at −20 °C until further analysis. Cell activation was analyzed with same procedure mentioned above.

3. Results

3.1. scFv-ELP fusions form colloidally stable worm-like nanostructures

Four different scFv-ELP fusions were expressed in E. coli from cloned DNAs which encode the scFv followed by an ELP called A192 (Fig. 1A). These scFvs, which target CD20, HLA-DR10, CD19, or CD3, are derived from monoclonal antibodies RTX (rituximab) [27], chimeric Lym-1 (chLym-1) [22,23] or the bispecific antibody blinatumomab [28], respectively. The resulting four scFv-A192 fusions that target these receptors, will be referred as aCD20A, aHLADR10A, aCD19A, and aCD3A for convenience (Table 1). Purification was performed by induction of ELP (A192)-mediated phase separation, which is a non-chromatographic purification method [21,29]. Two rounds of purification yielded ~30 mg/L of each fusion with >90% purity, as verified by SDS-PAGE (Fig. 1B). Given the thermo-responsive nature of ELPs, optical density of each fusion was scanned at 350 nm (OD 350) over a range of temperatures to determine their transition temperatures (T_t). Purified scFv-A192 fusions remained soluble at physiological temperatures (Fig. S1A). Various concentrations of each fusion were tested for optical density, evaluated for T_t using Eq. (5), and then plotted and fit with Eq. (6) to generate a phase diagram (Fig. S1B) [12]. All fit parameters are reported in Table S3. This fit allows estimation of T_t over a range of concentrations. These scFv-A192 fusions were not expected to phase separate at 37 °C at concentrations used in this study (1–100 μM). Dynamic light scattering (DLS) analysis was employed at days 0, 2, and 5 to determine the hydrodynamic radius (R_h) and colloidal stability of all four scFv-A192 fusions (Fig. 1C). Purified scFv-A192 fusions remained stable at 37 °C for at least 5 days. To explore their physical shape and exact dimensions, scFv-A192 fusions were imaged under cryo-TEM (Fig. 1D). All four fusions oligomerized into worm-like nanoparticles with slight differences in their length of 81–89 nm for aCD20A, aCD19A, aCD3A, while aHLADR10A was shorter with a mean length of 54 nm. All four fusions had a mean width between 5–9 nm (Fig. S1C).

Table 1.

Size and shape factor of Nanoworms.

Proteins (anti-(target)- (ELP))	aMW(kDa)/ monomer	bAverage MW(kDa)/particle Mean (±SEM)	cOligomeric state	dRadius of gyration (Rg, nm) Mean (±SEM)	eHydrodynamic radius (Rh, nm) Mean ± PD	Rg/ Rh
RTX (Rituximab)	143.8	144.0 (±0.006%)	1	6.6 (±0.5%)	8.7 ± 1.5	n/a
aCD20A (anti-CD20-A192)	99.1	51550 (±0.006%)	516	86.0 (±1.2%)	76.1 ± 18.1	1.1
aHLADR10A (anti-HLA-DR10-A192)	98.7	10880 (±0.002%)	109	41.6 (±0.5%)	47.0 ± 10.1	0.9
aCD19A (anti-CD19-A192)	100.0	18560 (±0.004%)	186	65.5 (±0.9%)	65.4 ± 15.0	1.0
aCD3A (anti-CD3-A192)	99.4	34470 (±0.007%)	345	77.5 (±0.6%)	69.8 ± 15.6	1.1
aCD20VA (anti-CD20-V2A64)	101.1	8477 (±0.0016%)	85	41.0 (±0.9%)	42.5 ± 10.1	1.0
aCD20AG (anti-CD20-A96G96)	98.1	21280 (±0.007%)	213	55.2 (±0.5%)	56.1 ± 12.3	1.0
aCD20S (anti-CD20-S192)	102.3	9588 (±0.006%)	95	39.1 (±0.6%)	45.3 ± 6.5	0.9
aCD20G (anti-CD20-G192)	96.4	20350 (±0.054%)	204	46.5 (±0.7%)	32.7 ± 7.4	1.4

^aExpected molecular weight based on the amino acid sequence (Table S1)

^b,dMeasured using SEC-MALS, PD = polydispersity

^cCalculated by dividing average molecular weight per particle over molecular weight of the monomer

^eMeasured using DLS.

Size-exclusion chromatography followed by multi-angle light scattering (SEC-MALS) was used to determine the population heterogeneity and the degree of oligomerization. All scFv-A192 monomers are associated with oligomerized products, as no lower-weight population was retained on the column. The single population of oligomerized scFv-A192 showed a molar mass between 10⁷–10⁸ g/mol (Fig. S1D). The oligomeric state of each fusion was calculated from the absolute molecular weight observed by SEC-MALS and the expected MW of the peptide encoded by the open reading frame (Table 1). The ratio between the radius of gyration (R_g, from SEC-MALS) and R_h (from DLS) was used to estimate a shape factor for each fusion [30]. The theoretically expected R_g/R_h ranges from a hard sphere 0.778 to 2.36 for a rigid rod [31, 32]. The random coil model best-fit the data acquired for scFv-A192 fusions in SEC-MALS. Therefore, this model was applied consistently to all observations. The shape factor (R_g/R_h, Table 1) reconfirmed the observations using cryo-TEM (Fig. 1D) that scFv-A192 self-assembly results in an extended chain-like or worm-like shape. When evaluated with the other frequently used fit models [32-34], the R_g/R_h values reported in Table 1 were slightly higher for all fusions, ranging from 1.0 to 1.5, which also was consistent with random coil or flexible linear chains. Based on analysis of both shape factors and morphological observations (cryo-TEM), all four scFv-ELPs thus appear consistent with a designation as ‘Nanoworms.’

3.2. Nanoworms cluster bound receptors on the cell-surface

Having demonstrated their oligomerization into Nanoworms (Fig. 1, Table 1), their specificity was confirmed through competitive binding. Both aCD20A and aHLADR10A showed target specificity towards CD20 and HLA-DR10 in B cells, respectively (Fig. 2A and B). Pre-blocking these receptors using their respective parent monoclonal antibodies, RTX for aCD20A and chLym-1 for aHLADR10A, substantially decreased the cell surface binding of labeled Nanoworms. B cells incubated with aCD20A or aHLADR10A underwent strong homotypic adhesion, which made flow cytometry on individual cells untenable. In contrast, sufficient numbers of individual cells permitted flow cytometry to confirm the specificity of aCD19A in B cells and aCD3A in T cells (Fig. 2C and D).

Fig. 2. Nanoworms spontaneously cluster cell surface receptors below physiological temperature.

(A) Specificity of rhodamine-labeled aCD20A was evaluated on live cells using confocal microscopy on CD20⁺ Raji cells. RTX was used to pre-block the CD20. (B) Specificity of rhodamine-labeled aHLADR10A was evaluated on HLA-DR10+ Raji cells. chLym-1 was used to pre-block HLA-DR10. For (A) and (B), experiments were performed in triplicate with 10–20 cells imaged per field of view (n = 40–50 cells). (C) Specificity of rhodamine-labeled aCD19A towards CD19⁺ Raji and SU-DHL-7 cells. (D) Specificity of rhodamine-labeled aCD3A towards CD3⁺ HuT-78 and Jurkat cells. For (C) and (D), cells were gated (black bar) at the upper 10% of the fluorescence profile of blinatumomab + aCD19A (or aCD3A) cells (blue) to compare the percentage of non-treated cells (Neg cntl) and cells treated with aCD19A (or aCD3A) only. (E) Time-lapse live cell epifluorescence imaging was used to monitor rhodamine-labeled Nanoworms bound to receptor-positive human cell lines (Raji cells for aCD20A, aHLADR10A and aCD19A; HuT-78 cells for aCD3A). Cells were heated (15–45 °C at 2 °C/min) during imaging. Yellow arrows indicate cluster formation upon temperature increase. Representative images are shown (n = 43–45 cells). (F) Receptor clustering was quantified using integrated fluorescence density within the cell. Fluorescence intensity increases as the cluster forms in a subregion of the cell. Each fluorescence intensity profile was acquired by analyzing individual cells shown in panel E. (G) Receptor clustering temperature (T_c) was defined where the maximum first derivative was observed, as estimated by Eq. (3) (Mean ± SD). Experiments were performed in triplicate with 10–20 cells imaged per experiment. A one-way ANOVA was used for statistical comparison.

To observe Nanoworm-mediated receptor clustering, which is mechanistically similar to Fc-FcR mediated receptor clustering [35], real-time live cell fluorescence imaging was used to track receptor-bound Nanoworms during a gradual temperature increase. Nanoworms were deliberately bound to cells under low temperatures where membrane fluidity is low. Upon heating, all Nanoworms induced substantial clustering on the cell-surface (Fig. 2E, Supporting Information Video S1). The integrated fluorescence density was measured to quantify Nanoworm-mediated receptor clustering temperature (T_c) on individual cells (Fig. 2F). This observation suggests that Nanoworms mediate receptor clustering below physiological temperatures, which was consistent across the selected immune receptors (Fig. 2G). Compared to Nanoworms, the monoclonal antibody RTX did not show any apparent clustering under identical conditions (Fig. 2E, Supporting Information Video S2). Integrated fluorescence density of RTX alone shows only a continuous decrease due to photobleaching (Fig. 2F).

Supplementary data related to this article can be found at https://doi.org/10.1016/j.biomaterials.2020.120338.

3.3. Nanoworm-mediated receptor clustering on the immune cell-surface activates intracellular signaling

Cell membrane blebbing is a distinguished phenomenological feature of cells undergoing cellular locomotion, cytokinesis (non-apoptotic), or apoptosis (apoptotic) [36]. When aCD20A Nanoworms were observed on the cell surface, cells immediately experienced massive reformation of their membranes (Fig. 3A). Live-cell brightfield imaging showed about 50% of cells experienced membrane blebbing below physiological temperatures (Supporting Information Video S3), at similar temperatures where aCD20A-mediated clustering occurred (Fig. 2G). Likewise, membrane blebbing was apparent with aHLADR10A (Fig. 3A) at similar temperatures where aHLADR10A-mediated clustering occurred (Fig. 2G). As these two receptors have well-characterized roles in B cell apoptosis, cytosolic cytochrome C level in Raji cells was measured upon incubation with aCD20A or aHLADR10A, and their respective antibody controls (Fig. 3B). Release of mitochondrial cytochrome C into the cytoplasm is a hallmark of apoptosis often considered a ‘point of no return’ [37]. After 18 h, Raji cells incubated with aCD20A and aHLADR10A showed a significantly higher cytosolic cytochrome C level compared to non-treated cells. RTX supplemented with a secondary anti-human crosslinking antibody (RTX+2°) induced release of mitochondrial cytochrome C into the cytoplasm, which was comparable to aCD20A treated cells. RTX alone did not show signs of apoptosis, which was consistent with prior observations [38]. In contrast, the HLA-DR10 targeting antibody chLym-1, either alone or with a secondary anti-human crosslinking antibody, induced release of mitochondrial cytochrome C into the cytoplasm at levels comparable to that of the aHLADR10A treated cells. This observation is consistent with prior characterization of the chLym-1 antibody, which binds lipid-raft associated HLA-DR10 and induces apoptosis without secondary crosslinkers [39].

Fig. 3. Receptor clustering induces cell membrane reformation (blebbing) and activates lymphoid cells.

(A) Time-lapse live cell brightfield imaging shows Raji cells (aCD20A, aHLADR10A, aCD19A) and HuT78 cells (aCD3A) experience blebbing (white arrows) upon receptor clustering. Cells were heated (15–45 °C, 2 °C/min) during imaging. (B) Cytosolic cytochrome C levels in Raji cells (18 h, 37 °C) treated with 20 μM aCD20A or aHLADR10A activate apoptosis similar to positive control chLym-1 (10 μM). (C) Cytosolic cytochrome C levels in Raji cells (18 h, 37 °C) show aCD19A (20 μM) fail to activate apoptosis relative to a positive control chLym-1 (10 μM). (D) As aCD19A failed to induce apoptosis, their effect on the cell cycle was assessed using flow cytometry in Raji cells (48 h, 37 °C). aCD19A (10 μM) appear to arrest the cell cycle similarly to rapamycin, which was selected as a positive control due to its cytostatic property (n = 3). Propidium iodide was used to quantify DNA content. A one-way ANOVA followed by multiple comparisons was used for statistical comparison. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05. Mean ± SD.

Supplementary data related to this article can be found at https://doi.org/10.1016/j.biomaterials.2020.120338.

CD19 is reported to compartmentalize with IgM and IgD on the B cell surface, which plays an important role in B cell receptor (BCR)-mediated signal transduction [40]. Due to its indispensable role in proliferation and survival, most malignant B cells consistently express CD19. As it is found in all NHL subtypes as well as in acute or chronic lymphocytic leukemia (ALL or CLL), CD19 serves as a reliable therapeutic target for lymphoma and leukemia [41,42]. B cells undergo actin cytoskeleton rearrangement upon BCR-mediated signal transduction, which is directly related to cell locomotion but unrelated to apoptosis [43,44]. Therefore, a possible outcome of CD19 clustering might include non-apoptotic cell cycle arrest [45]. Despite blebbing (Fig. 3A), CD19 clustering only arrested the cell cycle in the G1/G0 transition phase but failed to promote apoptosis (Fig. 3C and D). The cell cycle distribution upon aCD19A treatment was similar to that of the cells incubated with rapamycin, an immunosuppressant with cytostatic properties [26].

Supplementary data related to this article can be found at https://doi.org/10.1016/j.biomaterials.2020.120338.

About 2% of all lymphomas are the cutaneous T cell NHL (CTCL). The transformed T cells that are residing at or homing to the skin is known to cause CTCL [46]. Due to its low incidence, clinical trials are rare and there are no universal treatment guidelines [47]. A recent study that involved 21 centers around the globe reported 24 different approaches with no single treatment approach having more than 15% of the market share [48]. Mycosis fungoides and Sézary syndrome are the most common subtypes of CTCL. More than 70% of the patients with Mycosis fungoides are diagnosed at the early stages and have median survival of 13 years, however, Sézary syndrome is a more progressed disease and the patients’ median survival is less than 3 years [16,49]. The Sézary cells are known to be pathogenic due to their resistance to activation-induced cell death (AICD) rather than uncontrolled proliferation [50] and one of the resistance conferring mechanisms is the insufficient signal from the TCR [17]. Therefore, activation of TCR signaling via Nanoworm-mediated TCR clustering may be an effective therapeutic approach to overcome the resistance.

To explore this, the CD3 clustering Nanoworm (aCD3A) was incubated with HuT78 cells, one of the patient-derived Sézary syndrome cell lines [15], and subjected to the process employed in Figs. 2 and 3. The immediate cellular response that was observed upon Nanoworm-mediated CD3 clustering was membrane blebbing (Fig. 3A, Supporting Information Video S4). To see which cellular pathway is activated, IL-2 levels in cell culture media and cytosolic cytochrome C levels were quantified as downstream indicators of T cell activation and apoptosis, respectively. AICD was pronounced upon aCD3A treatment in HuT78 cells. aCD3A treatment directed HuT78 cells more towards apoptosis with lesser activation compared to cells treated with OKT3 (anti-CD3 mAb) supplemented with a secondary crosslinker (OKT3+2°). In contrast, targeting CD3 without clustering (OKT3 only) induced T cell activation without apoptosis (Fig. 4A and B). The aCD3A-mediated AICD was exclusively observed in HuT78 cells, but not in leukemic T cell lines CEM and Jurkat. In CEM cells, which is a mosaic human leukemia T cell line, aCD3A did not induce activation (Fig. S2A); although it induced apoptosis comparable to OKT3+2° (Fig. S2B). In Jurkat cells, which are an acute human leukemia T cell line, aCD3A induced secretion of IL-2 (Fig. S2C); however, the fold-increase was marginal (3-fold) compared to OKT3 alone (132-fold) or OKT3+2° (104-fold) treated cells. aCD3A did not induce apoptosis in Jurkat cells (Fig. S2D).

Fig. 4. CD3 clustering induces activation-induced cell death (AICD) in the HuT78 cell line.

(A) Quantification of IL-2 in HuT78 cell culture media and (B) quantification of cytosolic cytochrome C level in HuT78 cells after 18 h incubation with 10 μM aCD3A or 5 μM OKT3 (=10 μM scFv) at 37 °C (n = 3). (C) Quantification of IL-2 in HuT78 cell culture media after 18 h of co-incubation with plate-adsorbed OKT3 (36.4 μg) and soluble aCD3A (10 μM) at 37 °C (n = 3). Additive effect of OKT3 and aCD3A in IL-2 secretion was observed. (D) Extrapolated EC₅₀ for aCD3A-mediated activation is 4.3 μM. (E) Quantification of cytosolic cytochrome C level in HuT78 cells after 18 h of incubation with 10 μM aCD3A, 5 μM OKT3 (=10 μM scFv), or both at 37 °C (n = 3). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05. Mean ± SD.

Based on a report that IL-2 exposure sensitizes AICD resistant Sézary cells to apoptosis [17], studies were performed to test whether co-treatment of OKT3 and aCD3A can enhance AICD. HuT78 cells showed enhanced activation under the presence of both (Adsorbed OKT3+aCD3A) compared to single agent treatments (Fig. 4C). The estimated EC₅₀ for aCD3A in T cell activation was 4.3 μM (Fig. 4D). Despite the additive effect on activation, co-treatment did not further increase apoptosis as indicated by cytosolic cytochrome C level. Rather, it showed a slight decrease compared to aCD3A treatment alone (Fig. 4E). Based on this observation, it may be preferable to treat Sézary cells with a therapeutic molecule that can inherently cluster CD3 (such as aCD3A) instead of monoclonal antibody (such as OKT3) because heterogenous involvement of endogenous clustering molecules may leave a fraction of CD3 bound antibodies unclustered, which would interfere with apoptotic activity of clustered antibodies. While observing strong additive effects in activation on HuT78 cells, Jurkat cells were also tested because these cells secreted IL-2 upon aCD3A treatment. The cellular response of Jurkat cells was remarkably different from HuT78. Jurkat cells experienced inhibition of activation with increasing concentration of aCD3A (Fig. S2E). The fit IC₅₀ for aCD3A in inhibition was 1.0 μM (Fig. S2F). Since aCD3A treatment did not induce AICD in Jurkat cells (Fig. S2D), no further assessment was performed.

3.4. Nanoworm multivalency is a primary factor that drives receptor clustering on the cell surface

Comprised of pentameric repeats, (VPGXG)_n, ELPs are high molecular weight polymers that show a reversible phase separation in response to heating. The phase transition temperature (T_t), above which ELPs phase separate from bulk water into amorphous coacervates, can be precisely controlled at the genetic level or at the environmental level depending on hydrophilicity of the guest residue (X), number of repeats (n), ELP concentration, and ionic strength. These factors can be modulated individually or simultaneously to control the T_t of a given ELP. Among these factors, the ELP concentration is inversely correlated to T_t (Fig. S1B). Empirically, ELP T_t trends lower when the ELP concentration becomes higher. Based on this, one might argue that Nanoworm binding to cell surface results in high ELP concentration in a confined area, which leads to a significant drop in their T_t. As a result, receptor clustering may proceed via ELP-mediated phase separation but not necessarily through their multivalency.

To delineate the main factor that affected receptor clustering observed in Fig. 2E, four additional anti-CD20-ELP fusions were designed that have similar MW but differ in their solution T_t. ELPs V2A64, A96G96, S192, and G192, were selected to cover a wide range of T_t, from 34–76 °C (Fig. 5A and B). The resulting four anti-CD20-ELPs will be referred as aCD20VA, CD20AG, aCD20S, and aCD20G for convenience (Table 1). These variants were identical in ELP repeat number and only differed in T_t based on their ELP amino acid composition (Fig. 5C, Table S2). Scanning optical density at 350 nm from various concentrations confirmed differential thermo-sensitivity of four anti-CD20-ELP variants and aCD20A (Fig. S3A, Table S3). Calculated shape factors (R_g/R_h) based on DLS and SEC-MALS analyses indicated that all five anti-CD20-ELP variants are consistent with Nanoworms (Figs. S3B and C and Table 1). When these variants were tested on Raji cells, all variants clustered CD20 on the B cell surface (Fig. 5D). It was interesting that even aCD20G, whose solution T_t is about 76 °C, still clustered bound receptors on the cell-surface below physiological temperatures. Based on the single cell quantification (Fig. 5E), aCD20VA, aCD20A, and aCD20AG showed a positive correlation between T_t and T_c; however, this correlation was not apparent in aCD20AG, aCD20S, and aCD20G (Fig. 5F, Table S2). While there remains a minor effect of ELP identity between T_t and T_c, these observations indicate that the multivalency of Nanoworms is more likely the primary factor governing the cell-surface receptor clustering. This trend was less apparent in CD20⁺ SU-DHL-7 cells (Supporting Information Videos S5 and S6) where only aCD20G showed a statistically significant increase in T_c from other variants (Fig. S3D, Table S2).

Fig. 5. Single cell quantification of receptor clustering consistent with multivalency-mediated Nanoworm clustering.

(A) Recombinant DNAs encoding anti-CD20 scFv were fused to the amino terminus of ELPs V2A64, A96G96, S192, and G192. These four ELPs were chosen because they contain identical number (n = 192) of VPGXG pentameric motif and have similar MW but differ in their solution T_t (Table S2). (B) SDS-PAGE was used to confirm the identity of generated fusions. (C) Optical density increases from baseline upon coacervation. Their transition temperatures ranged from 34 to 76 °C (10 μM in dPBS). (D) Time-lapse live cell fluorescence imaging was used to monitor rhodamine-labeled anti-CD20-ELP variants bound to CD20⁺ Raji cells upon heating (15–45 °C, 2 °C/min). A representative cell is shown with the measured media temperature indicated on each panel. (E) Nanoworm clustering was quantified using integrated fluorescence density within each cell as a function of temperature. Each fluorescence intensity profile was acquired by analyzing individual cells as shown in panel D. (F) Receptor clustering temperature (T_c) on each cell was determined using the peak of Eq. (3) (Mean ± SD). Experiments were performed in triplicate with 8–12 cells imaged per experiment (n = 30–32 cells). A one-way ANOVA was used for statistical comparison. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05.

4. Discussion

Standard first-line therapy for the most of the B cell NHL is R–CHOP, which is a combination of immunotherapy (R: RTX, rituximab) and chemotherapy (CHOP: cyclophosphamide, doxorubicin, vincristine, and prednisone) [51]. The CHOP regimen, developed in the 1970’s, is initially effective in 90% of patients but is responsible for severe side effects and subject to relapse [52]. When RTX was supplemented along with CHOP, it improved drug response in resistant NHL and produced remission in 50% of patients by effectively recruiting host clearance machineries through Fc-FcR interactions [5,6]. Although R–CHOP significantly improved therapeutic outcomes compared to CHOP alone, there were also found to be inconsistencies in therapeutic outcomes that stem from heterogenous Fc-FcR interaction [7,8]. On one hand, germline polymorphisms on the FcγR, especially on FcγRIIa and FcγRIIIa, on the immune effector cells [7,8] or premature internalization of RTX by Fc region interacting with FcγRIIb present on the target B cells surface [53] leads to insufficient involvement of the immune system, which confers resistance to immunotherapy. On the other hand, over-activation of the immune system through Fc-FcγR interaction can cause fatal side effects such as cytokine release syndrome [9-11]. As either enhancing or diminishing Fc-FcR interaction may produce unwanted side effects during immunotherapy, modalities that only induce clustering of the drug:target complex without activating immune system may serve as an alternative strategy to overcome Fc-FcR related heterogeneity [54].

To address this issue, several modalities, such as silver nanoparticles, magnetic nanoparticles, or morpholino oligonucleotide conjugates, have been tested as a proof-of-concept with promising in vitro or in vivo results [55-57]. However, the production process used in these examples require chemical modification followed by another round of purification that may introduce heterogeneity in the degree of functionalization and decrease the production yield. The utility of such modalities may be amplified through direct cellular expression of well-designed fusion platforms that eliminate the need for bioconjugation or subsequent purification. Elastin-like polypeptides (ELPs) are one such well-characterized fusion platform. ELPs are genetically encodable thermo-responsive protein-polymers with a sequence derived from human tropoelastin, a precursor of elastin that can be expressed in cells as fusion proteins [58]. This is an exceptional advantage compared to synthetic polymers since purified ELP fusions are monodisperse, exceptionally stable, biodegradable, and biocompatible therapeutics [12]. Due to ease of engineering and unique biophysical properties, their usage has now been widely tested in the fields of protein sciences [59], tissue engineering [60], medical applications [61], and therapeutic applications [12]. Using heterologous expression, the antibody Nanoworms were homogenously biosynthesized and self-assembled into distinct worm-like nanostructures that display more than 100 scFvs per particle. The dynamics of cell-surface clustering mediated by their multivalency was observed using real-time imaging and quantified at a single-cell level. These imaging-based analyses were further correlated to the apoptosis, cell cycle arrest, or cell activation. These observations indicate that the antibody Nanoworms have an inherent ability to cluster and activate bound receptors, which is equipotent to clinically available antibodies supplemented with crosslinker. Because Nanoworms lack the Fc domain, they may overcome the FcR-dependent host immunity and disadvantages observed in the clinic.

Having observed architectural consistency (worm-like morphology) and favorable molecular characteristics (stability and high-degree multivalency) of scFv-ELP fusions, and their cell surface behavior (spontaneous clustering) in B cell NHL, Nanoworms were further modified to cluster CD3 on T cells and tested on a model of Sézary syndrome, an aggressive form of cutaneous T cell NHL. Current therapeutic regimens for T cell NHL vary by stage [48]. In early stages (stages I and II), the first-line therapy bexarotene (cell cycle inhibitor) is often combined with second-line therapies, such as local radiotherapy, phototherapy, skin electron beam therapy, or gemcitabine. In more advanced stages (stages III and IV), methotrexate (antimetabolite), photopheresis (extracorporeal white blood cell treatment), or poly-chemotherapy are used as first-line therapies with IFNα or chlorambucil as second line agents. These are the most commonly adopted approaches; however, treatment options are not limited to the above-mentioned combinations and oftentimes rely on retrospective approaches by clinicians. There are two antibody-based options approved for T cell NHL, brentuximab vedotin (anti-CD30 antibody-drug conjugate) and mogamulizumab (anti-CCR4 monoclonal antibody). Several others are in the clinical trials as well, including anti-CD158K (IPH4102), anti-CTLA-4 (Ipilimumab), and anti-PD-1 (Nivolumab and Pembrolizumab) antibodies [47]. To develop new antibody-based approaches, we explored the therapeutic impact of CD3 clustering in Sézary cells as a means to overcome the resistance to AICD (Fig. 4). These results are consistent with results reported by Klemke et al., which emphasizes the importance of TCR signaling in AICD [17].

One question remaining is how cells exhibit different responses when treated with agents specific to the same receptor. In all cell lines, OKT3+2° treated group experienced AICD, which was not always the case for aCD3A treatment. Based on the fact that both OKT3 and blinatumomab target CD3ε [62], it was perplexing to see a clear difference between these two treatments (OKT3+2° vs. aCD3A) towards AICD and additive/inhibitory effect in T cell activation (Figs. 4 and S2). Two possible explanations can be explored. The first explanation concerns the degree of clustering regarding differential multivalency. The maximum valency that OKT3 further engaged by 2° (OKT3+2°) is four, while a single aCD3A particle is comprised of more than 300 scFv-ELPs. Therefore, aCD3A (and other Nanoworms) forms a single macrodomain on the cell surface (Fig. 2E) that is different from multiple microdomains formed by OKT3+2° or other lower valency clustering agents [63,64]. This difference on the cell surface may activate different intracellular pathways and be responsible for different additive/inhibitory effect in T cell activation in different cell lines (HuT78 vs. Jurkat). The second explanation is based on different epitopes within the CD3ε between OKT3 and blinatumomab. However, the binding region for blinatumomab or its parent clone L2K is not as well characterized compared to that of the OKT3 [65]. Given the scope of the current study, future studies will explore these questions.

Based on the observed colloidal stability and absence of scFv-ELP monomers in SEC-MALS (Fig. 1C), this manuscript makes the assumption that Nanoworm structure is maintained in cell culture until they engage cell surface receptors. A limitation of this assumption is that it is possible that other components in culture media fractionate Nanoworms into smaller fragments. If this is the case, what has been observed (Figs. 2E,5D) could be Nanoworm monomers or fragments that self-assemble into larger clusters via lateral diffusion on the cell surface. To observe the potential fragmentation of the Nanoworms in a physiologically relevant condition, two Nanoworms (aCD19A and aHLA-DR10A) were incubated with 0.5 mM bovine serum albumin (BSA) [66, 67]. During a 2-day period in the presence of physiologically relevant level of BSA, there was no evidence of fragmentation into smaller particles (Fig. S4). Stability was further assessed by comparing size and oligomeric state of Nanoworms before and after lyophilization-redissolution. This assessment was performed in two different concentrations (10 and 100 μM) to test two aspects: i) how initial scFv-ELP concentration before refolding would affect the size and oligomeric state of the refolded Nanoworms; and ii) what would be the effect of lyophilization-redissolution. A 10-fold difference in scFv-ELP concentration did not affect the R_h of the refolded Nanoworms (Fig. S5A, Table S4). In contrast, lyophilization-redissolution increased oligomeric state, R_h and R_g by 2-fold, 1.3-fold and 1.3-fold, respectively. for both R_h and R_g (Fig. S5B, Table S4). Based on these observations, it is concluded that i) the size (R_h and R_g) of the Nanoworms is not governed by the initial concentration of scFv-ELP monomers; and ii) the Nanoworm assembly is stable during the lyophilization-redissolution process, albeit with a slight increase in size after redissolution.

As can be seen in Fig. 5F and Table S2, Nanoworms with more hydrophobic ELPs cluster at lower temperatures on the cell surface. While the extent of this observation differed between cell lines (Figs. 5F and S3D), on Raji cells there was an ~14 °C increase in cell-surface clustering temperature between aCD20VA and aCD20AG. Despite a broad range of ELP hydrophobicity and solution transition temperature (Table S2), all Nanoworms cluster on the cell surface below 37 °C. Since receptor activation could not be studied at physiological temperature, the effect on receptor activation has not been pursued in this manuscript. Instead, a thermodynamics approach was used to identify biophysical factors affecting Nanoworm phase separation in solution. ELP phase behavior depends on many factors, including temperature, molecular weight, hydrophilicity, concentration, co-solutes, and even molecular assembly [68]. Using the estimated enthalpy and entropy of phase separation of 8 Nanoworms presented in this study and free ELP A192 (Table S3), multiple regression was used to identify the most predictive factors associated with thermodynamics. The enthalpy and entropy of Nanoworm phase separation (coacervation) was first estimated using van’T Hoff approximation (Figs. S6A and B). Next, four independent variables (hydrophilicity of ELP, total molecular weight per particle, R_g, and R_h) were subjected to stepwise multiple regression to predict enthalpy and entropy. Among tested variables, R_h of the Nanoworm best predicted the enthalpic gain (p = 0.020) and entropic cost (p = 0.013) of Nanoworm coacervation (Fig. S6C). As the greater degree of self-assembly (higher molecular weight per particle) leads to larger R_h (Pearson correlation = 0.801, n = 9, p = 0.009) and R_g (Pearson correlation = 0.929, n = 8, p = 0.001) (Fig. S6D), this analysis shows that the greater degree of self-assembly (increased R_h) is associated with increased endothermic enthalpy and entropic cost of coacervation per molecule of scFv-ELP. Curiously, the hydrophilicity of ELP was not significantly correlated to the enthalpy (R² = 0.23, p = 0.19) or entropy (R² = 0.33, p = 0.11) of Nanoworm phase separation (Fig. S6E). The thermodynamics observed in Nanoworm phase separation was consistent with that of the Dox-ELP nanoparticles reported by Zai-Rose et al. [69] and also a GMCSF-ELP nanoparticle reported recently by our group [70]. This analysis may be useful to interpret the phase behavior of nanoparticles decorated by ELPs and possibly predict their relationship with receptor activation in future studies.

5. Conclusion

This work correlates the extracellular behavior of antibody Nanoworms to their intracellular signal activation. This approach may overcome heterogenic therapeutic outcomes for B cell NHL observed in the clinic and resistances in T cell NHL including Sézary Syndrome. As a simple yet adaptable therapeutic modality, the applications for antibody Nanoworms are potentially broad.

Data availability

Selected raw videos obtained from live-cell imaging have been uploaded as supplementary information; furthermore, other raw/processed data required to reproduce these findings will be made available on request.