Patient-Specific Targeting of the T-Cell Receptor Variable Region as a Therapeutic Strategy in Clonal T-Cell Diseases

T cells regulate the adaptive immune response and play key roles in diseases such as autoimmunity and the diverse group of malignancies that comprise non-Hodgkin lymphomas (NHL). Although individual NHL subtypes are rare, T-cell NHL accounts for 10% to 15% of new NHL diagnoses, ranging from indolent to aggressive, all of which are associated with morbidity and healthcare utilization (1). Moreover, outcomes are worse than those with B-cell NHL (2). As T-cell–based cancers are poorly treated with existing therapies, improved targeting of clonal T-cell malignancies is needed (3, 4).

Molecular markers that identify target cells have had a significant impact on the treatment of NHL, and have been explored for autoimmune diseases. Pan-specific B-cell markers such as CD20 (Rituximab) and CD19 have been used to target NHL B-cell cancers, with patients tolerating the associated B-cell aplasia (5, 6). Strategies targeting pan-T-cell antigens, such as the anti-CD52 antibody alemtuzumab, have proven challenging, however, because the resultant T-cell depletion can lead to opportunistic infections (7). In contrast, targeted markers found on both T cell and B cell such as CD30 (Brentuximab) successfully treat Hodgkin lymphoma as well as anaplastic large-cell lymphoma (3, 8). Furthermore, antibodies targeting CD3 have been developed as general immunosuppressive agents to treat transplant rejection (muromonab) and autoimmune diseases such as Type 1 Diabetes (Otelixizumab and Teplizumab) (9). Indeed, the Levy lab first showed the potential of using anti-idiotype antibodies to target the malignant B-cell in NHL, seeing remission responses in a majority patients (66%), with some lasting as long as 10 years (10, 11).

T-cell receptors (TCR) possess tremendous diversity owing to diverse variable beta (Vβ) chain (n = 48) and variable α (Vα) chain (n = 46) gene segments, that recombine with D and J gene segments to form a functional TCR (12). In contrast to B-cell receptors which undergo somatic hypermutation, T cells express a static TCR. T-cell malignancies involve clonal expansion of T cells that bear the same Vβ and Vα VDJ rearranged genes (13–18), and autoimmune/inflammatory diseases frequently result in a predominant expansion of specific T-cell clones (19–23). Multiplexed PCR of the TCR gene has been used to identify the presence of a distinct malignant clone and facilitate diagnosis of T-cell NHL (21, 24). Thus, the Vβ chain represents a cell surface marker that discriminates disease associated T-cell clones from the majority of normal T cells. Recent studies have begun to demonstrate the potential of targeting the TCR using bispecific T-cell engagers (BiTE) and chimeric antigen receptor T-cell (CART) directed approaches (4, 25).

As proof of concept, we identified a 76-year-old male (Patient J) with a history of T-cell granular lymphocytic leukemia (T-LGL). T-LGL is an indolent T-cell clonal lymphoproliferative disorder that can be followed for years without therapy (26). This allowed for ongoing sample acquisition through refinement of testing over several years. We hypothesized that a mAb that is exclusive to a specific Vβ would eliminate the malignant clone while having minimal effects on healthy T cells. We thus sought out to build a panel of anti-Vβ candidates (αVβ) using the sequencing information of Patient J's T-LGL clone. We demonstrate the development of high affinity antibodies that bind with specificity to the patient's T-LGL clone and employ cell killing, supporting the clinical efficacy of this approach; furthermore, we show these therapeutic candidates are capable of targeting and eliminating healthy T cells and cell lines from T-cell acute lymphoblastic leukemia and a murine lymphoblast line. From these findings, we are able to expand the generation of a panel of antibodies capable of treating a majority of patients with other clonal T-cell diseases.

Jurkat (DSMZ), HPB-ALL (DSMZ), and Jurkat J.RT3-T3.5 (ATCC) cells were cultured in RPMI1640 medium (Gibco, 11875–093) supplemented with 15% FBS (HyClone, SH30396.03) and 1% PenStrep (Gibco, 15140–122). EL4 cells (ATCC) were cultured in RPMI1640 medium supplemented with 15% horse serum (Gibco 16050122). NK-92 cells expressing the CD16 receptor (haNK; ATCC) were grown in MEM alpha medium (Gibco, 12561–056) with 0.2 mmol/L inositol (Sigma Aldrich, I7508), 0.02 mmol/L folic acid (Sigma Aldrich, F8758), 0.1 mmol/L beta-mercaptoethanol (Sigma Aldrich, M3148), 12.5% FBS, 12.5% horse serum (Sigma Aldrich, H1138), and 200 units/mL human IL2 (PeproTech, 200–02). The cells did not exceed 20 passages from the time of thaw to the time of use in any given experiment. Cells were tested for mycoplasma once a month throughout use in experimentation using a PCR Mycoplasma Detection Kit (Applied Biological Materials, G238). All patient samples were obtained after oral and written informed consent according to our local Institutional Review Board approval (#4422) following our Hyman Research Protection Program that incorporates the ethical principles described in the Belmont Report and U.S. Common Rule. Healthy control patients were excluded if they had a diagnosis of T-cell lymphoma or myeloproliferative disorders. Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll gradient and plated within 24 hours in IMDM with 20% FBS (Gibco, 12440–053). All cells in all experiments were cultured for incubation in a humidified incubator at 37°C with 5% CO₂ unless otherwise noted.

Patient J's TCRVB was cloned into lentiviral vector pLX304 (Addgene, 25890) and used to transduce Jurkat J.RT3-T3.5 cells to generate cell line Jurkat-V13. Patient J's TRB and TRA chain variable regions were cloned into pMIG II by replacing the murine Vbeta5.2 TRB and TRA chain variable regions in Addgene 52111 (27), and used to transduce EL4 cells to generate cell line EL4-V13. Viral particles were generated at VectorBuilder Inc.

For patient samples, 100 μL of ACK-lysed blood was stained with nonconjugated anti-hIgG (BD Biosciences, 555784) for 20 minutes to saturate nonspecific binding sites, followed by a 20-minute incubation with 30 nmol/L αVβ, and detection with BV786-conjugated anti-hIgG (BD Biosciences, 564230), all on ice. Cells were stained with fluorophore-conjugated lymphocyte markers to identify the malignant T-LGL clone or target T-cell population, including CD2-PE (Beckman Coulter, IM0443U), CD3-APC-AF750 (Beckman Coulter, A66329), CD4-FITC (Beckman Coulter, IM0448U), CD5-PE-Cy7 (Beckman Coulter, A51075), CD8-APC (Beckman Coulter, IM2469U), CD16-ECD (Beckman Coulter, A33098), and CD56-PE-Cy5.5 (Beckman Coulter, A79388). To determine the percentage of T-LGL bound by the αVβ13, the fraction of αVβ-bound CD8⁺ cells detected by anti-hIgG were divided by the percentage of abnormal CD8⁺ cells demonstrating partial loss of CD2.

Cell line binding was determined by incubation with 10 nmol/L or 30 nmol/L αVβ followed by incubation with anti-hIgG-BV786. Competition studies were accomplished by preincubating cell line samples with 10 nmol/L or 30 nmol/L αVβ for 30 minutes, followed by washing and incubating for an additional 30 minutes with a commercially available anti-TCR Vβ8 (BioLegend, 348106) or anti-TCR Vβ5a (Thermo Fisher, TCR2642).

All binding incubations were performed on ice.

Patient J T-LGL CD3⁺ T cells, isolated with a magnetic column (BioLegend, 480022), were plated at 3E4 cells/well in a U-bottom 96-well plate (Falcon, 353077; ref. 28). Cells were opsonized with 30 nmol/L αVβ for 10 minutes at room temperature. haNK cells, prepared in Alpha MEM+ media with 10% FBS, 225 U/mL IL2, were added at effector-to-target ratios of 1:1, 5:1, or 10:1, followed by incubation for 4 hours. Cells were then stained with 7-AAD (Beckman Coulter, A07704). Cell lines were labeled with 0.1 μmol/L calcein and plated at 1.25E4 cells/well. For healthy controls, CD3⁺ T cells were isolated and 1.25E5 cells/well were plated and incubated as above. They were then stained with live dead aqua (ThermoFisher, L34957) and anti-CD2. Cells were fixed and permeabilized, followed by staining with anti-Vβ8.1 and anti-Vβ5a. Surviving target cell lines were identified as calcein⁺7-AAD⁻; surviving T-LGL cells were identified as calcein+7⁻AAD⁻CD8⁺CD2^dimCD5^dim; healthy controls as aqua⁻CD2⁺, and cytosolic Vβ8⁺ or cytosolic Vβ5a⁺. Finally, natural killer (NK) activation was measured by the addition of anti-CD56 and anti-Lamp1a-BV650 (BioLegend, 328638). Trials for T-LGL and cell lines were performed 3 times, in triplicate for each trial. Three healthy controls each underwent one biological trial, and were plated in triplicate.

Jurkat cells were plated in U-bottom 96-well plates at 4E5 cells/well in 15% FBS RPMI media supplemented with 2.5% Human Sera Complement (HS, Millipore Sigma, S1764; ref. 29). Cells were incubated for 4 hours with 30 nmol/L αVβ8, and stained with 7AAD. Similarly, CD3+ T-LGL cells were plated at 3E4 cells/well in 20% FBS IMDM media with 2.5% HS and incubated with 30 nm of αVβ13A, αVβ13E, or αVβ13D. Cells were stained with immunophenotyping antibodies as described above. The surviving T-LGL cells were identified as 7-AAD⁻CD8⁺CD2^dimCD5^dim. Complement deposition was measured following 1-hour incubations at 37°C using anti-C3b/iC3b-APC (BioLegend, 846106). Trials were performed 3 times, in triplicate for each trial.

Patient PBMCs were incubated with 30 nmol/L αVβ for 24 to 96 hours in 20% FBS IMDM at 4E5 cells/well. Activation was measured at 24 hours with anti-CD69 (Beckman Coulter, IM1943U). Cytotoxicity was measured as described above. Trials were performed 3 times, in triplicate for each trial.

For all cytotoxicity assays, CountBright counting beads (Thermo Fisher, C36950) were added and cells were analyzed using a Fortessa Flow Cytometer (BD Biosciences). A total of 8,000 bead events were collected for all the samples with the exception of healthy controls for which 30,000 bead events were captured. Percent cytotoxicity was determined by subtracting the ratio of live target cells incubated with αVβ to live target cells incubated with the isotype from 1, and then multiplying by 100. Data manipulation was done using Graphpad Prism software.

First, 10 nm αVβ8, 30 nm αVβ5, and 30 nm αVβ13A was labeled with Zenon pH-Rodo (ThermoFisher, Z25612). Jurkat and HPB-All were plated at 2.5E5 cells/well in 15% FBS RPMI and labeled αVβ was added at designated timepoints. For T-LGL cells, 4E5 T-LGL cells/well in 20% FBS IMDM media was incubated for 24 hours. All samples were analyzed on a Fortessa flow cytometer.

Jurkat cells in 15% FBS RPMI media were plated at 5E3 cells/well with 10 nmol/L αVβ8 for 72 hours. Cells were prepared using an MTS tetrazolium assay kit (Promega, G358) and results were obtained using a BioTech Synergy 2 Reader (30).

Jurkat cells in 15% FBS RPMI media were plated at 2.5E5 cells/mL with 10 nmol/L αVβ8 for 24 hours. The cells were fixed using 70% cold ethanol and stained in 250 μL propidium iodide/RNase staining solution (50 μg/mL PI, 0.1% Tween-20/PBS, 4X of 100 mg/mL RNase A in PBS) for 30 minutes before analyzing on a Fortessa flow cytometer.

Jurkat cells were plated at 1E4 cells/mL in 15% FBS RPMI in a 96-well U-Bottom plate, incubated with 10 nmol/L αVβ8 for 48 hours and analyzed using the Annexin V-APC Assay kit (Abcam, 236215) on a Guava Easycyte HT plate reader.

5E6 Jurkat cells were serum starved for 18 hours prior to incubation with αVβ8 for 10 minutes at 37°C, then addition of 10 μg/mL anti-CD3 or phosphate buffered saline for 5 minutes. Cells were pelleted and lysed with the RIPA Lysis buffer solution on ice. Supernatants were centrifuged and the protein concentration was quantified with BCA Protein Assay. An equal amount of whole-cell lysate was mixed with 3X SDS sample buffer [75 mmol/L Tris (pH 6.8), 3% SDS, 15% glycerol, and 0.1% bromophenol blue] containing 8% β-mercaptoethanol. The mixed samples were boiled for 5 minutes in a 95°C heat block before loading on a 4% to 15% Tris-HCl gradient gel. The proteins were transferred onto a nitrocellulose membrane and blocked with 5% FBS TBST buffer and incubated with antibodies against phospho-Zap70 (Cell Signaling Technology, #2701T), phospho-AKT (Cell Signaling Technology, #9271), phospho-ERK (Cell Signaling Technology, #9101L). The membrane was then stripped, washed, blocked, and probed with antibodies against ZAP-70 (Thermo Fisher, #PA5–28697), AKT (Cell Signaling Technology, #9272S) and ERK (Cell Signaling Technology, #9101L). Horseradish peroxidase–conjugated secondary antibodies against mouse IgG and rabbit IgG were used.

Animals were cared for under the OHSU Institutional Animal Care and Use Committee in compliance with the Animal Welfare Act regulations and Public Health Services policy. Eight-week-old female C57BL/6J mice (Jackson Laboratory, 20 ± 2 kg) were subcutaneously injected on the left flank with EL4-V13 (3E5 cells) (Day 0) using 28-gauge syringes. On Day 1, half of the specimens received an intraperitoneal injection of 200 μg αVβ13A-mIgG2a; the other half received 200 μg mIgG2a isotype. Two additional doses were given on Day 4 and 8. Tumor volumes were measured (volume = (l * w²/2)) using a caliper. On Day 12, the mice underwent CO₂-induced euthanasia.

The data and protocols that support the findings of this study are available from the corresponding author, P.M. Bowers, upon reasonable request.

Patient J's PBMCs were isolated and RNA sequencing (RNA-seq) analysis was performed to identify dominant T-cell clones. A T-cell clone expressing TRBV6–1, corresponding to Vβ13.3 in Wei nomenclature, was identified and constituted 95% of the CD8 T-cell population (Fig. 1A). Flow cytometric analysis revealed an immunophenotype characterized by CD8⁺CD2^dimCD3⁺CD5^dimCD7^dimCD16^variableCD57⁺ (Supplementary Fig. S1). Genomic analysis revealed mutations in known oncogenes: DNMT3A R882S and E907del, TET2 E1144K, STAT3 Y640F.

We sought to develop an antibody specific to Vβ13.3 (αVβ13) for eliminating the malignant T cells (Supplementary Methods). Protein antigen composed of Patient J's Vβ13.3 fused to mouse IgG1 Fc (mFc) was generated and used to immunize mice (Fig. 1B). A total of 320 single B cells were isolated for antibody cloning and expression (31) and screened by ELISA binding to patient J's Vβ13 chain (Fig. 1C). Binding was also assessed to patient PBMCs, defined as the proportion of αVβ13 bound T-LGL cells to the total population of T-LGL cells (CD3⁺CD8⁺CD16⁺CD2^dim). Seven αVβ13 antibodies binding > 80% of the T-LGL population were identified: αVβ13A-αVβ13G (Fig. 1D; Supplementary Fig. S2A), consistent with ELISA binding. These αVβ13 mAbs can bind to TCRs in the Vβ13 family and have the potential for general treatment of all Vβ13 clones across patients. As each TCR containing Vβ13 has a unique CDR3 sequence created by VDJ recombination, we also assessed binding to a mutant antigen in which the CDR3 was scrambled to distinguish αVβ13 from anti-idiotype (Fig. 1C). Importantly, αVβ13J and αVβ13I do not bind the CDR3 mutated antigen and were identified as anti-idiotype antibodies specific to Patient J (Supplementary Fig. S2B and C).

To our knowledge, there are no commercially available cell lines that express Vβ13.3, the Vβ on Patient J's T-LGL clone. To further characterize the cell binding and functional activity of αVβ13 candidates, we generated cell lines expressing the patient's Vβ. We used a lentiviral transfection to transduce a stable TCR complex (TCRβ chain and CD3) characterized by patient TCR Vβ13.3 expression using the NK murine lymphoma line EL4 and the TCR-null Jurkat T-ALL line T3.5 to express Vβ13.3, referred to as EL4-V13 and Jurkat-V13, respectively (Fig. 1E).

Transduced cell lines were sorted by FACS and confirmed for expression using CD3 and Vβ specific antibodies. We found that αVβ13A, a high affinity binder to primary T-LGL patient cells, bound with specificity to Jurkat-V13, indicating expression of a stable TCR complex (Fig. 1F). We also exchanged the human IgG1 (hIgG1) of αVβ13A with a mIgG2a Fc region (αVβ13A-mIgG2a). This bound with specificity to EL4-V13 and was detected using a secondary against mIgG2a (Fig. 1F). αVβ13A did not bind to the wild-type lines.

Next, we assessed binding to healthy control PBMCs (n = 3). We found that αVβ13A bound 1.89% to 3.12% of the CD3⁺ population, a proportion in line with healthy T-cell repertoire diversity (Fig. 2A; ref. 32). Conversely, the anti-idiotype αVβ13J bound to the patient's T-LGL clone and Jurkat-V13, but not to CD3⁺ lymphocytes from two healthy controls (Supplementary Fig. S2B–D).

Given the heterogeneity of T-cell lymphoma phenotypes, we also analyzed antibodies against Vβ8 (corresponding to TRBV12–3, hitherto αVβ8) and Vβ5.2/5.3 (corresponding to TRBV5–5 and TRBV5–6, hitherto αVβ5; Fig. 2B; refs. 33, 34). We found that αVβ8 bound 3% to 4% of CD3⁺ cells from healthy controls, consistent with published values of 3.5% to 6.5% (Fig. 2A; ref. 32). Although αVβ5 binding was relatively weak, preincubation with αVβ5 reduced the binding of a commercially available αVβ5-FITC antibody by twofold (P = 0.002; Fig. 2A; Supplementary Fig. S2E and F). We also observed strong binding of αVβ8 to Jurkat cells that express Vβ8, and αVβ5 to HPB-ALL cells, characterized by Vβ5.3 expression (Fig. 2C). Preincubation with αVβ8 or αVβ5 also competitively inhibited other αVβ8 or αVβ5 antibodies from binding, demonstrating specificity of αVβ candidates (Supplementary Fig. S2G and H).

Having established a panel of αVβ13 antibodies, we assessed the capability of mediating antibody-dependent cellular cytotoxicity (ADCC) or complement-dependent cytotoxicity (CDC). αVβ antibodies were formatted as hIgG1 and mouse IgG2a isotypes (mIgG2a), both competent for Fc-mediated cell death. At baseline, Patient J demonstrated undetectable NK cells, a known sequela of T-LGL (35). We thus investigated the ability of αVβ to induce ADCC using exogenous NK-92 effector cells engineered to express the Fcγ receptor CD16 (haNK; ref. 36). We mixed target T-LGL patient CD3⁺ cells with effector haNK cells at three different effector-to-target ratios (E:T) of 1:1, 5:1 and 10:1 in a 4-hour cytotoxicity assay (28).

In Patient J's CD3+ T cells, we observed cytotoxicity of the T-LGL clone, identified by CD8⁺CD5^dim, using αVβ13A, αVβ13E, αVβ13D at all E:T ratios, relative to isotype control (Fig. 3A). αVβ13A induced target cell death of 37.1% at 1:1 E:T ratio (P = 0.0010), increasing to 70.5% at 10:1 E:T ratio (P < 0.001). Less than 20% of Patient J's CD8⁺ T cells are normal T cells, characterized by CD8⁺CD5⁺. We found no significant change in this healthy CD8 population relative to isotype controls at all E:T ratios (Supplementary Fig. S3A).

Similarly, we observed cell death of engineered Jurkat-V13 and EL4-V13 cell lines but not wild-type cell lines when using αVβ13A at E:T ratios of 5:1 and 10:1 (Fig. 3B and C). Similar results were seen with αVβ5 and αVβ8 in HPB-ALL and Jurkat cells, respectively (Fig. 3D and E).

We then examined the ability of αVβ antibodies to induce target cell death in healthy controls. However, there is no additional unique epitope on the target population. We found that αVβ8-bound cells contain cytosolic Vβ8 that can be detected by a commercially available αVβ8 antibody (Supplementary Fig. S3B), and thus cytosolic expression could be used to track the target population. We proceeded with a 4-hour ADCC assay with healthy control CD3⁺ T cells and haNK cells. We observed elimination of the target population with both αVβ5 and αVβ8 antibodies at all E:T ratios in three healthy controls (Fig. 3F–H; P < 0.0001; Supplementary Fig. S3C–H). We did not test αVβ13 ADCC activity against healthy controls owing to the lack of a commercially available antibody against the target epitope.

In some cases, we measured the NK activation marker CD107a which correlates to NK cell–mediated lysis of target cells (37). We found that CD107a was elevated in haNK cells incubated with Jurkat-V13 and Jurkat and their corresponding αVβ antibodies, αVβ13 or αVβ8, respectively, relative to isotype control (Fig. 3I).

Patient J's CD3+ T cells were incubated with high affinity αVβ13.3 antibodies for 4 hours in the presence of human complement (29). No T-LGL cell death was induced by αVβ13 mAbs, while alemtuzumab resulted in cell death in both the T-LGL cells (17%; P = 0.144) and normal CD8 T cells (71.8%; P = 0.0001; Supplementary Fig. S4A and B). Only Jurkat, incubated with αVβ8, and human complement had cell death at 24 hours (Supplementary Fig. S4C). Complement deposition was assessed and we did not visualize C3 deposition on treated cells. In contrast, a control mAb against CD3, OKT3, showed effective deposition (Supplementary Fig. S4D).

Next, we evaluated additional potential mechanisms that reduce target T-cell viability. Activation-induced cell death (AICD) may occur when repeated stimulations of the TCR lead to apoptosis.

To test for AICD, Patient J's PBMCs were incubated with αVβ13 antibodies for 24 hours. This resulted in expression of CD69, a marker of activation, in the T-LGL clone (Fig. 4A), but not healthy T-cell subsets (Supplementary Fig. S5A). We then incubated the PBMCs for 96 hours with αVβ13A, in the absence of complement or NK effector cells. The target cell underwent cell death at 48 to 96 hours relative to the isotype control (Fig. 4B). Other αVβ13 antibodies behaved similarly after 96-hour incubations (Fig. 4C).

To ensure cell death was not due to effector function, the hIgG Fc region of αVβ13A and αVβ13J was exchanged with human IgG4 which lacks effector function, called αVβ13A-hIgG4 and αVβ13J-hIgG4, respectively. There was significant cell death at 96 hours relative to the isotype control irrespective of the Fc isotype (Fig. 4D). Interestingly, we found that the presence of complement inhibited cell death. This was accompanied by inhibition of CD69 expression, indicating that a lack of activation was critical for the observed cell death (Fig. 4E). In Jurkat cells, complement does not inhibit αVβ8 mediated cell death (Supplementary Fig. S4C).

To further explore reduced cell viability, we focused on αVβ8 and Jurkat cells. Similar to T-LGL cells, CD69 expression in αVβ8 incubated Jurkat cells increased by 31-fold relative to isotype following a 24-hour incubation (Fig. 4F). Jurkat cells had reduced viability relative to isotype control (81.4%; P > 0.0001) in a 96-hour MTS assay (Fig. 4G). Analysis of cell-cycle progression showed a higher percentage of the αVβ8-treated population in G₁ phase (76.0% vs. 53.0%; P < 0.0001) and less in S phase (16.1% vs. 31.0%; P < 0.0001) or G₂ phase (10.6% vs. 16.4%; P = 0.03; Fig. 4H). They also exhibited elevated annexin expression, a marker of apoptosis (Fig. 4I).

We measured the phosphorylation of ZAP70, a critical component in AICD, and phosphorylation of ERK for MAPK pathway which has been shown to mediate apoptosis (38, 39). As a control, we added soluble anti-CD3 (OKT3), a potent activator and apoptosis inducer, following αVβ incubation (40). We found that 10 minute incubation of αVβ8 induced phosphorylation of ZAP70 and ERK in serum starved Jurkat cells, similar to OKT3 (Fig. 4J). AKT was phosphorylated in all sample, consistent with defective expression of phosphatase and tensin homologue (PTEN) and constitutive activation of the PI3K signaling pathway (41).

To demonstrate effective killing of Vβ13.3 cells by αVβ13A in vivo, we used a syngeneic mouse model that allow for both ADCC and AICD as potential mechanisms of cell death. We formatted αVβ13A in mIgG2a, an ADCC-competent isotype (Fig. 1E and F). We performed subcutaneous injection of EL4-V13 into the flank of C57/BL6 mice and observed decreased tumor growth velocity by αVβ13A-mIgG2A relative to isotype control (P < 0.002; Fig. 4K).

TCR down-modulation following ligation is a pivotal event in T-cell activation (42). We assessed the ability of αVβ to impact TCR surface expression. The T-LGL clone exhibited a drop in TCR detection, as assessed by CD3 and αVβ detection at 1 hour, and progressively decreased over the following 18 hours, while CD2 remained stable (Supplementary Fig. S5B). In Jurkat cells, detectable CD3 and αVβ drops within 5 minutes and is sustained for 24 hours (Supplementary Fig. S5C). We next assessed internalization using fluorophore pHrodo-labeled αVβ and found that Jurkat and HPB-ALL cells internalized αVβ8 and αVβ5, respectively, while primary T-LGL cells did not (Supplementary Fig. S5D and E).

Given target cell death induced by αVβ mAb, we sought to derive an efficient path to assemble a panel of antibodies that could target the majority of patients with T-cell NHL. We performed a meta-study of TCRBV gene usage and homology in NHL patient data sets to prioritize antibody specificity. We identified 9 NHL data sets in which 500 NHL patient samples were analyzed by RNA-seq, PCR and/or FACS methods, of which 314 patients had sufficient data to determine the exact gene (Supplementary Table S1A; refs. 13–16, 18, 43–46). On the basis of TCRVβ gene homology, where 80% identity suggests a high probability of obtaining a cross-reactive antibody, we clustered TCRVβ groups to identify a set of Vβ genes likely to offer optimal coverage of patients with NHL (Fig. 5; Supplementary Table S1B). Five antibodies could theoretically recognize the majority (54%) of NHL malignant clones, and 70% of malignant clones could be targeted with 9 distinct αVβ antibodies.

The development of targeted therapies to treat T-cell–based diseases has been a long-term goal of medicine. Most targeted T-cell therapies, however, discriminate poorly between healthy and diseased tissues. Because T cells express a single rearranged TCR, pursuing Vβ represents a potential route to targeted therapy. Two recent studies using CART and BiTE constructs have shown the ability to kill clonal T cells (4, 25). The Vogelstein lab paired two existing anti-TCR Vβ antibodies formatted as scFvs with an anti-CD19 directed scFv to target effector T cells to kill malignant clones. They showed that constructs specifically lysed malignant T-cell lines patient-derived cells in vitro and resulted in tumor suppression in murine models. The use of BiTEs is appealing and could theoretically provide a fast, patient-specific and selective targeting approach that preserves healthy T cells. BiTEs, however, do possess significant limitations including bidirectional killing and the current need for near continuous infusion due to a short half-life (47). Use is hampered by cytokine release syndrome (CRS) and effector cell–associated neurotoxicity syndrome (48). Finally, the manufacture and development of bispecific antibodies is challenging (49).

While having excellent cytotoxicity, the use of CART presents a technical and logistical burden for T-cell disease. CART strategies require cellular engineering to block target antigen expression by the CART cells to prevent T-cell fratricide and palliative strategies to compensate for loss of T-cell function are challenging. A novel strategy was recently documented that targets the TCR constant region (TRBC1) with a CART approach. This approach destroys about one half of the T-cell repertoire, and while preserving a function T-cell response, would potentially possess other clinical risks such as CRS. The CART targeting format can require autologous CART generation and deployment, a cell population which may be unavailable from immunosuppressed cancer patients. Likewise, the use of allogeneic NK and T-cell lines pose other technical issues (50).

The use of mAbs is an established approach enjoying well-defined clinical and developmental properties. Attributes such as pharmocokinetics, ADCC, CDC, manufacturing, delivery, and storage are well understood, and can be easily modulated as part of therapeutic development (51–53). We found that our antibodies effectively engaged effector cells for target cell cytotoxicity while also inducing downstream apoptosis, partially through T-cell activation. Our murine model demonstrated significant reduction in tumor growth rate using an aggressive T-cell tumor model, however did not completely eliminate the subcutaneous tumors. It is possible that the therapeutic antibody penetration of a solid tumor is impaired. For instance, a clinical trial assessing mogamulizumab in patients with cutaneous T-cell lymphoma showed poorer responses in the skin [overall response rate (ORR), 42.1%] and lymph node (ORR, 25.0%) compartments compared with the blood (ORR, 94.7%; ref. 54). Future studies could assess if tumor growth is controlled with ongoing therapy. Furthermore, effector enhancing Fc formats such as afucosylation and mutations that impact FcGR engagement will be assessed to further optimize cell killing as needed (55). Recent studies suggest that levels of galactosylation of antibodies produced in HEK instead of CHO cells leads to improved C1q engagement and opsonophagocytic killing (56). Future development could also focus on enhancing complement fixation (53). Finally, in aggressive, refractory or bulky T-cell lymphoma, multiagent chemotherapy regimens have clinical activity but are associated with greater toxicity and possibly risk of death when used earlier in treatment (57, 58). Immunochemotherapy approaches are often more efficacious without significant toxicity compared with straight chemotherapy regimens. In this setting, use of a targeted αVβ therapeutic antibody, much like the use of Bentuximab vedotin in CD30 positive peripheral lymphomas or rituximab in diffuse large B-cell lymphoma, could improve survival and be accompanied by a manageable safety profile (8).

The exhausted CD8⁺ T-cell phenotype of T-LGL might influence the clone's susceptibility to AICD. Other T-cell lymphomas with different subsets of T cells might behave differently upon targeted Vβ depletion. Collectively, this underscores the reality that T-cell lymphoma represents a heterogeneous disease characterized by varying levels of aggressiveness, location of the T-cell clone, type of T-cell and equitable access to advanced therapies. Thus, multiple modalities to treat this population of patients is likely necessary. In addition, a lack of effector cells might necessitate administration of allogeneic NK cell lines to support efficacy. Allogenic NK cell therapies are currently in clinical development, and would have potential efficacy advantages relative to cell lines such as haNK cells, which must be irradiated, thus limiting persistence, proliferation and long-term impact on efficacy of coadministered antibody (59).

We observed with interest the lack of αVβ-induced T-LGL activation and cell death in the presence of complement. It has been shown that C5a enhances T-cell proliferation and diminishes effector T-cell apoptosis via upregulation of the antiapoptotic protein BCL2, highlighting the importance of complement in T-cell homeostasis (60). While it is unclear how this might hamper the antibody effector function in vivo, it is reassuring that the syngeneic in vivo murine model showed a significant reduction in tumor velocity growth. We also noted the difference in internalization of the TCR complex upon binding of αVβ. The cell lines had internalization of the antibody while the patient's primary T-LGL samples did not. The loss of TCR expression in the primary patient cells might be due to prevention of TCR complex recycling over internalization (42).

We have demonstrated that mAbs targeting specific Vβ subsets has the potential to be an effective, off-the-shelf strategy to eliminate clonal populations of diseased or healthy T cells while leaving the healthy immune system intact. Moreover, we demonstrate a pipeline for the efficient generation and testing of candidates, and use primary patient data to show that a majority of the patients with T-cell NHL may be treated with a handful of antibodies.

O.M. Lucero reports other support from Sjoberg Award and grants from Medical Research Foundation during the conduct of the study; in addition, O.M. Lucero has a patent for Anti–T-Cell Receptor Antibodies and Uses Thereof pending and is a cofounder of VB Therapeutics. O.M. Lucero does not hold a position in this company and has not received any financial assets from this company. This company, in conjunction with OHSU and UCLA, has filed a provisional patent on the intellectual property described in the manuscript. J.-A. Lee reports other support from Jerry Bidwell during the conduct of the study; in addition, J.-A. Lee has a patent for Anti–T-Cell Receptor Antibodies and Uses Thereof pending. J. Bowman reports grants from Sjoberg and Medical Research Foundation during the conduct of the study. K. Johnson reports grants from Sjoberg Foundation and Medical Research Foundation during the conduct of the study. J.K. Thomas reports other support from Jerry Bidwell during the conduct of the study. K. Thiel-Klare reports grants from Sjoberg during the conduct of the study. C.A. Eide reports other support from OHSU Foundation during the conduct of the study. B.J. Druker reports grants from NIH/NCI and other support from HHMI, Burroughs Wellcome Fund, Beat AML LLC, CureOne, AstraZeneca, and Sjoberg Foundation during the conduct of the study as well as personal fees from Aileron Therapeutics, Therapy Architects LLC, Cepheid, Nemucore Medical Innovations, VB Therapeutics, Vincerx Pharma, and MCED Consortium and personal fees and other support from Amgen, Aptose Biosciences, Iterion Therapeutics, Blueprint Medicines, GRAIL, EnlIven Therapeutics, Novartis, Recludix Pharma, and Adela outside the submitted work; in addition, B.J. Druker has a patent for Treatment of Gastrointestinal Stromal Tumors issued, licensed, and with royalties paid from Novartis. N. Lydon reports other support from VB Therapeutics LLC during the conduct of the study as well as other support from VB Therapeutics LLC outside the submitted work; in addition, N. Lydon has a patent for Anti–T-Cell Receptor Antibodies and Uses Thereof pending and is cofounder and stockholder of VB Therapeutics LLC. P.M. Bowers reports grants from Jerry Bidwell during the conduct of the study; in addition, P.M. Bowers has a patent for Anti–T-Cell Receptor Antibodies and Uses Thereof pending to UCLA and is a cofounder and shareholder of VB Therapeutics. This company, in conjunction with OHSU and UCLA, has filed a provisional patent on the intellectual property described in the manuscript. No disclosures were reported by the other authors.

O.M. Lucero: Conceptualization, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. J.-A. Lee: Conceptualization, formal analysis, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. J. Bowman: Formal analysis, investigation, writing–original draft. K. Johnson: Supervision, investigation, writing–review and editing. G. Sapparapu: Investigation. J.K. Thomas: Investigation. G. Fan: Resources, data curation, validation, writing–review and editing. B.H. Chang: Investigation, writing–review and editing. K. Thiel-Klare: Investigation. C.A. Eide: Investigation, writing–review and editing. C. Okada: Conceptualization, writing–review and editing. M. Palazzolo: Supervision, investigation, methodology, project administration. E. Lind: Investigation, writing–review and editing. Y. Kosaka: Investigation, writing–review and editing. B.J. Druker: Conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing. N. Lydon: Conceptualization, resources, funding acquisition, project administration, writing–review and editing. P.M. Bowers: Conceptualization, resources, supervision, writing–original draft.

We would like to thank the OHSU Flow Cytometry Core (Brianna Garcia, Dorian LaRocha) for technical support. Funding for this project was provided by the following sources: The Sjoberg Award and the Medical Research Foundation (GCNCR1217A) to B.J. Druker, O.M. Lucero, J. Bowman, K. Johnson, B.H. Chang, and K. Thiel-Klare; NIH for E. Lind.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.