All chemicals and proteins were purchased from Sigma-Aldrich (St. Louis, Missouri, USA) unless otherwise noted. All cell culture reagents were purchased from ThermoFisher Scientific (Waltham, Massachusetts, USA) unless otherwise noted. Hydrolysable crosslinker poly(DL-lactide)-b-poly(ethylene glycol)-b-poly(DL-lactide)-diacrylate triblock (PLA–PEG–PLA) was purchased from PolySciTech Akina (West Lafeyette, Indiana, USA). Capture antibody for ELISA against RTX was purchased from Bio-Rad Laboratories (MCA2260, Hercules, California, USA). HRP-conjugated goat antihuman IgG Fc for ELISA assay was purchased from ThermoFisher Scientific. Antihuman CD45, antihuman CD3, antihuman CD56, antihuman CD11b, antihuman CD14, antihuman CD4, and antihuman CD8 were purchased from BioLegend (San Diego, California,USA). RTX (RITUXANTM: Genentech, San Francisco, California,USA) and HER (Herceptin: Genentech) were obtained at the UCLA hospital pharmacy. All NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) and NOD.Cg^−Hc1 Prkdc^scid Il2rgtm1Wjl/SzJ (NSG^−Hc1) mice were purchased from The Jackson Laboratory and housed in specific pathogen-free vivarium.

The n-RTX were synthesized with optimized modification for antibodies based on our previous reports.9 17 Briefly, RTXs were encapsulated via in situ polymerization at room temperature using MPC as the monomer, PLA–PEG–PLA and glycerol dimethacrylate (GDMA) as the crosslinker, and ammonium persulfate and N,N,N',N'-tetramethylethylenediamine (TEMED) as the initiator. The n-RTX with mixed crosslinkers of GDMA and PLA–PEG–PLA at a 1:1 molar ratio (n-RTX) were synthesized with optimized modification for antibodies based on our previous reports. The n-RTX were synthesized using a volume of 5 mg RTX at 1.5 mg/mL, a specific amount of MPC (40% m/v in PBS), PLA–PEG–PLA (10% m/v in PBS) and GDMA (10% m/v in Dimethyl Sulfoxide (DMSO)) in a molar ratio of RTX:MPC:PLA–PEG–PLA:GDMA=1:12 000:500:500. Free-radical polymerization from the protein surface was then initiated by adding ammonium persulfate (10% m/v in Phosphate buffered saline (PBS), molar ratio to RTX=2000:1) and TEMED (molar ratio to RTX=4000:1) to the reaction vial and kept in an ice bath for 2 hours. Finally, dialysis was used to remove extra free monomers and initiators. The free RTX was removed using hydrophobic interaction chromatography (Phenyl Sepharose CL-4Bh) as described previously.

The n-Saporin without anti-CD7 conjugation were synthesized using a volume of 100 µg Saporin at 1 µg/µL, a specific amount of acrylamide (AAm, 10% m/v in PBS), N-(3-aminopropyl)methacrylamide (APm, 10% m/v in PBS), citraconic anhydride N-(3-aminopropyl)methacrylamide (cit-APm,10% m/v in PBS) and GDMA (10% m/v in DMSO) in a molar ratio of Saporin:AAm:APm:cit-APm:GDMA=1:6000:100:100:500. Radical polymerization from the surface of the protein was then initiated by adding ammonium persulfate (10% m/v in PBS, molar ratio to Saporin=2000:1) and TEMED (molar ratio to Saporin=4000:1) to the reaction vial and kept in an ice bath for 2 hours. Finally, dialysis was used to remove extra free monomers and initiators. The free Saporin was removed using hydrophobic interaction chromatography (Butyl-S Sepharose 6 Fast Flow columns; GE Healthcare). Anti-CD7 antibodies were conjugated onto nanocapsules through copper-free click chemistry to form n-Saporin^anti-CD7. Dibenzocyclooctyne-PEG4-N-hydroxysuccinimidyl ester (DBCO-PEG4-NHS) in DMSO (11 mg/mL) was added into the nanocapsule solution at a 10:1 molar ratio. NHS-PEG4-Azide in DMSO (39 mg/mL) was added into the anti-CD7 solution at a 10:1 molar ratio. Both reactions were processed in an ice bath for 1 hour. Free reagents were removed by size-exclusion spin columns. The conjugation between DBCO-modified n-Saporin and azide-modified anti-CD7 was reacted at 4°C for 17 hours at a 1:1 molar ratio to synthesize n-Saporin^anti-CD7.

2F7-BR44 cells were transduced with a lentiviral vector encoding mStrawberry and firefly luciferase (Fluc) together with the bleomycin-resistant gene under elongation factor-1 α promoter and maintained in Iscove’s Modified Dulbecco’s supplemented with 15% fetal bovine serum, 1% GlutaMax, 1% Antibiotic–Antimycotic, and 200 µg/mL of Zeocin.

Antilymphoma activity of RTX on 2F7-BR44 cells was assessed by absolute cell number counting using MACSQuant Analyzer 10 (Miltenyi Biotech, Bergisch Gladbach, Germany). For CDC testing, 2F7-BR44 cells were cultured in 24-well plates (1×10⁶/mL) for 24 hours in the presence of RTX (20 µg/mL) and 20% murine serum from either NSG or NSG^−Hc1 mice. For the PBS control group, 2F7-BR44 were cultured in 24-well plates (1×10⁶/mL) for 24 hours in the presence of RTX (20 µg/mL) and 20% of PBS. The same volumes of cell culture were taken to obtain absolute cell numbers on MACSQuant Analyzer 10. Cell death percentages were calculated as cell death (%)=100 × (1−(cell number after RTX treatment/control without RTX)).

2F7-BR44 cells were washed with 2% fetal bovine serum/PBS. Seven antibodies of cell surface markers—CD45, CD3, CD56, CD11b, CD14, CD4, and CD8 —were added and stained at 4°C for 30 min. Expression levels were assessed by BD LSRFortessa (BD Biosciences, San Jose, California, USA) after fixation with 2% formaldehyde in PBS.

In vivo half-life of native RTX and n-RTX was determined by monitoring the free RTX concentration in plasma from NSG and NSG^−Hc1 mice. Briefly, NSG and NSG^−Hc1 mice were randomly divided into two groups and retro-orbitally injected at a dosage of 20 mg/kg/mouse, a dose which has been reported as effective in an NHL xenograft murine model.9 The concentration of free RTX in plasma samples was detected by ELISA using the VersaMaxTM Tunable microplate reader (Molecular Devices, San Jose, California, USA).

The concentration of RTX in plasma was measured by ELISA using anti-RTX antibody and peroxidase-conjugated antihuman Fc antibody as described previously.9 18

To inactivate complement in NSG^−Hc1 serum for in vitro test, the serum was incubated in a water bath at 56°C for 30 min. The complement activity of NSG^−Hc1 serum was depleted by cobra venom factor (CVF) treatment in vivo. NSG^−Hc1 mice (n=4) were treated with 1 mg/kg of CVF at Day 0, 6, 13, and 21 via intraperitoneal injection. The complement C3 depletion by CVF was confirmed by commercial mouse C3 specific ELISA (Abcam, Cambridge, UK). The 50% hemolytic complement activity (CH50) in serum was tested with by hemolytic assay. In brief, serially diluted serum samples were incubated with red blood cells and hemolysin at 37°C for 30 min. The lysis of red blood cells was detected by reading the absorbance of the samples at 540 nm. The %lysis was calculated using the following formula optical density (OD) : %lysis=(OD₅₄₀test−OD₅₄₀blank)/(OD₅₄₀total lysis−OD₅₄₀blank)×100% and plotted with the serum dilutions. The serum dilution required for 50% hemolysis was read out as the CH50 at each time point. The CH50 was normalized to day 0 (Setpoint) and plotted as normalized CH50 (Test time point/Setpoint). The serum from NSG mice was included as a negative control for the CH50 assay.

2F7-BR44 cells were gene marked with a lentiviral vector expressing firefly luciferase, then luciferase-expressing cells were selected through a week of Zeocin treatment (200 µg/mL). The selected 2F7-BR44 cells (2×10⁶/animal) were injected into 6 week-old male NSG^−Hc1 mice via the tail vein to obtain lymphoma xenograft mice. One or 2 weeks after 2F7-BR44 cell injection, mice were treated with a single course of native RTX or n-RTX^CXCL13 via retro-orbital vein injection (4 mg kg/day for five sequential days). Lymphoma formation was monitored by in vivo bioimaging using the IVIS Lumina II (PerkinElmer, Waltham, Massachusetts, USA). In vivo bioluminescence imaging was performed following subcutaneous injection of 4.5 mg D-luciferin (Pierce, Woodland Hills, California, USA). Mice were imaged at the signal plateau (10 min post-D-luciferin injection) under isoflurane anesthesia.

Humanized mice were prepared as previously described with modifications.9 Human fetal liver and thymus tissues from the identical donors were obtained from Advanced Biosciences Resources or Novogenix without identifying information and did not require Institutional Review Boards (IRB) approval for its use. Six-week-old male NSG or NSG^−Hc1 mice were administered Busulfan (35 mg/kg) intraperitoneally. 24 hours later, the mice were implanted with a portion of human fetal thymus combined with 0.5×10⁶ fetal liver-derived CD34⁺ cells obtained from the same donor after solidification in the Matrigel under the kidney capsule and also injected with 0.5×10⁶ CD34+ cells via retro-orbital vein. After the human blood cell reconstitution in peripheral blood, 2F7-BR44 cells (2×10⁶/animal) were injected into humanized BLT mice via tail vein to establish xenograft humanized BLT mice. One or 2 weeks after 2F7-BR44 cell injection, mice were treated with a single course of native RTX or n-RTX^CXCL13 via retro-orbital vein injection (4 mg kg/day for five sequential days). Lymphoma formation was monitored by in vivo bioimaging using the IVIS Lumina II.

Lymphoma burden in both xenograft and xenograft humanized BLT mice was quantified as the total photon flux per second within a region of interest (ROI) whole body of the mouse; ROIs were identically sized for all measurements.

Results are expressed as mean±SDs. Errors depict SD. Statistical significance is presented with a p value calculated by the GraphPad Prism (La Jolla, California, USA). The significance of survival curve was compared with a log-rank test. All other significance comparisons between groups were calculated by either Wilcoxon matched-paired signed rank test or one-tailed unpaired t-test with Welch’s correction, which clarified for each figure legend. P>0.05 was not considered statistically significant and noted not significant.

We have established a highly metastatic 2F7-BR44 lymphoma cell line obtained from a CNS lymphoma.9 This cell line was subcloned from the human 2F7 Burkitt NHL cell line, which was retrieved from an patient with AIDS-lymphoma19 and selected for its high rate of metastasis (100%) to the CNS.9 The 2F7-BR44 cells were modified to express firefly luciferase (Fluc), which allows us to track both cancer spread and therapeutic efficacy through in vivo bioluminescence imaging. These cells also express mStrawberry, which enables their detection by flow cytometry. Although this cell line has already been tested for xenograft mouse generation using NSG or humanized NSG mice in our previous work,9 the impact of complement on xenografts is unknown. We first generated a 2F7-BR44 cell xenograft using NSG^−Hc1 mice and compared lymphomagenesis to that of NSG mice. There was no distinct difference in the timing of lymphoma formation and tissue distribution between two strains of NSG (figure 1: NSG^−Hc1 vs online supplemental figure S1: NSG). Side-by-side comparison of total bioluminescence intensities further supported the similarity in lymphomagenesis between the two xenograft models (figure 1B) even though a higher survival rate was observed in NSG^−Hc1 mice (figure 1C). These results indicate that the fully functional complement system cannot sufficiently mediate antilymphoma activity.

A functional complement system is beneficial to the mediation of additional antilymphoma activity of RTX.14 To confirm this point, we first tested RTX-mediated cytotoxicity against 2F7-BR44 cells in the presence of serum obtained from both NSG and NSG^−Hc1 mice, as well as heat-inactivated NSG^−Hc1 serum in vitro (online supplemental figure S2). Compared with the presence of serum from NSG, approximately three-fold higher levels of cell death induced by RTX were confirmed with NSG^−Hc1 serum; however, the improvement lost when the complement in NSG^−Hc1 serum was inactive by heat inactivation. This indicates that the presence of functional complement plays a valuable role in enhancing antilymphoma effector activity of RTX. We next verified the impact of the presence of functional complement system on RTX efficacy by comparing lymphoma growth among xenograft NSG, NSG^−Hc1, and NSG^−Hc1 diminished complement activity by the administration of CVF—a glycoprotein cleaves complement C3 to C3a and C3b by its enzyme form complexed with factor B and factor D,20 resulting in impairment of functional MAC (figure 2). To monitor the effect of CVF treatment, complement activities in the treated animals were determined by both mouse complement C3 ELISA and the 50% hemolytic complement (CH50) activity in serum through hemolytic assay.21 As shown in online supplemental figure S3, the complement activity significantly decreased in NSG^−Hc1 mice with each CVF treatment, which was still higher than that in NSG mice.

Antilymphoma activity of RTX was monitored weekly by bioluminescence after the administration of native RTX (100 µg/day/animal) five times at 1 week post xenograft. Native RTX treatment delayed lymphoma formation in both xenograft mouse models relative to no treatment (compare figure 2A with figure 1A and online supplemental figure S1). Although lymphomagenesis tended to slow progressively in NSG^−Hc1 mice than that in NSG, the difference became negligible between NSG mice and NSG^−Hc1 mice with CVF treatment (compare week 4 results in figure 2B). The difference was not statistically significant due to high variations, but it suggested that a functional complement system in the NSG^−Hc1 strain could support enhanced antilymphoma activity of native RTX as reported.14

We previously reported that the encapsulation of RTX within a zwitterionic MPC polymer layered together with surface conjugation of CXCL13—the CXCR5 ligand—for B-cell lymphoma targeting (denoted as n-RTX^CXCL13) leads to sustained release and increases CNS levels of free RTX by approximately 10-fold via enhanced penetration of the BBB with respect to the administration of native RTX in NSG mice. The treatment of xenograft NSG mice with n-RTX^CXCL13 exhibited significantly superior control of lymphoma progression compared with that of native RTX when treatment was initiated at an early time point (1 week post xenograft); however, such effects were not observed when the treatment was started at a later time point (2 weeks post xenograft).9 In that system, the death of 2F7-BR44 cells was considered to be induced mainly by RTX-mediated direct killing (apoptosis22 due to the lack of a functional complement system required for the induction of CDC and immune cells supporting antibody-dependent effector activities such as ADCC and ADCP. To confirm improved antilymphoma activity supporting CDC in NSG^−Hc1 mice, we repeated the experiments using both xenograft NSG and NSG^−Hc1 mice side-by-side (compare figure 3 with online supplemental figure S4). Mice were treated with either native RTX or n-RTX^CXCL13 at both 1 week (group I) and 2 weeks post xenograft (group II). Lower lymphoma burden and slower progression were observed in NSG^−Hc1 mice treated with both RTX and n-RTX^CXCL13 (compare figure 1A with figure 3A) relative to that in NSG mice (compare online supplemental figure S1 with online supplemental figure S4). Although the treatment at week 2 resulted in insufficient control of lymphoma proliferation in xenograft NSG^−Hc1 mice, n-RTX^CXCL13 still outperformed RTX in lymphoma control in group II (figure 3B). The therapeutic improvement elicited by n-RTX^CXCL13 was significant with early treatment in group I; significance decreased with delayed treatment in group II NSG^−Hc1 mouse models. These results further confirmed that the n-RTX^CXCL3 mediated greater antilymphoma efficacy in 2F7-BR44 xenograft mice relative to that of native RTX and indicate that the antilymphoma efficacy of n-RTX^CXCL3 can be further heightened in the presence of functional complement. The improved therapeutic efficacy of n-RTX^CXCL13 in NSG^−Hc1 mice was further confirmed by two repeated sets (online supplemental figure S5).

The humanization of NSG mice enables support of antibody-mediated effector activities such as ADCC and ADCP.23 24 We next evaluated the antilymphoma efficacy of RTX using a BLT (bone marrow/liver/thymus) humanized mouse model, which was used in our previous work.9 In BLT mice, human T, B, NK cells, and macrophages were confirmed to be in circulation (online supplemental figure S6). Notably, both NK cells and macrophages are known to mediate ADCC and ADCP.25 Compared with the limited efficacy exhibited by native RTX in the CNS due to the poor antibody penetration, we have demonstrated clearance of CNS lymphomas by n-RTX^CXCL13 treated at 1 week post xenograft in NSG-BLT mice.9 However, when treatment started at the later time point (2 weeks post xenograft), neither native RTX nor n-RTX^CXCL13 could suppress lymphoma proliferation in NSG-BLT mice, resulting in whole-body metastasis and animal death (figure 4A,D). The average survival time of the native RTX-treated group was 4 weeks post xenograft, while the n-RTX^CXCL13 treated group survival time was approximately 7 weeks post xenograft. In contrast, NSG^−Hc1-BLT mice showed enhanced antilymphoma activity by treatment with both native RTX and n-RTX^CXCL13 even when treatment was initiated at 2 weeks post xenograft (figure 4B). Herceptin (n-HER^CXCL13), used as an irrelevant control antibody, did not show any therapeutic effect in either NSG-BLT or NSG^−Hc1-BLT mice (online supplemental figure S7). Although the BBB penetration of native RTX proved inefficient in NSG-BLT mice, three out of four NSG^−Hc1-BLT mice that received native RTX treatment were able to eliminate most tumors and showed no sign of relapse for at least 9 weeks post xenograft. Conversely, n-RTX^CXCL13 mediated far superior antilymphoma efficacy, even against CNS lymphomas; all mice were able to completely eliminate tumors by week 6, and four out of five mice were able to control lymphomagenesis with no relapse observed by week 9. The enhanced lymphoma control in NSG^−Hc1-BLT mice compared with that in NSG-BLT mice was quantitatively confirmed through comparison of total BLIs (figure 4C) and survival rates (figure 4D). Moreover, the antilymphoma efficacy of RTX was further enhanced by the use of n-RTX^CXCL13, allowing for enrichment of RTX around lymphomas—even those distributed within the CNS.9 These results indicate that CDC mediated by a functional complement system together with ADCC and ADCP could be the mechanism behind enhanced antilymphoma efficacy of RTX observed in the NSG^−Hc1-BLT mouse.

Although levels of free RTX can only be maintained at a detectable concentration in plasma for 4 weeks in either native or nanoencapsulated forms (online supplemental figure S8), lymphomagenesis was well-controlled in most NSG^−Hc1-BLT mice over 9 weeks (figure 5). To understand the contribution of other potential mechanisms for controlling lymphoma, we performed a second transplant with the same number of 2F7-BR44 cells at 10 weeks post xenograft in surviving NSG^−Hc1-BLT mice as in figure 4B. The experimental scheme was summarized in online supplemental figure S9. Lymphoma growth was followed weekly by in vivo bioluminescence imaging for an additional 6 weeks (ie, from 10 weeks to 15 weeks post xenograft). Mice that previously received n-RTX^CXCL13 treatment did not allow additional lymphoma formation following the second xenotransplant (figure 5A, n-RTX^CXCL13); in contrast, mice treated with native RTX were susceptible to the second xenotransplant (figure 5A, RTX). The levels of human B-cells in peripheral blood started to recover approximately 3 weeks post RTX administration (ie, at 5 weeks post-1st xenograft) and maintained at 4%–9% after 9 weeks post-1st xenograft (figure 5B). Contrastingly, levels of circulating 2F7-BR44 cells increased after the second xenograft in mice treated with native RTX but not with n-RTX^CXCL13, indicating that the failure of lymphoma growth after the second xenograft in mice treated with n-RTX^CXCL13 was not due to residual RTX. Importantly, all mice treated with n-RTX^CXCL13 achieved sustained control until the endpoint (16 weeks post-1st xenograft) with significant extension of survival time (figure 5C).

To confirm the potential mechanism responsible for preventing xenograft after the second transplant in n-RTX^CXCL13-treated mice, we eliminated CD7⁺ cells using a CD7-specific immunotoxin that uses saporin toxin as a mediator of cell death.26 CD7 expression is present on most T cells,27 28 NK cells, and early stages of B and myeloid cells.29 As reported previously, nanocapsules can achieve broad systemic delivery of macromolecule cargos to sites including the lymph nodes, CNS, gut, thymus, spleen, liver, and kidneys of mice, rats and non-human primates.9 18 To ensure efficient elimination of CD7⁺ cells in animals, we encapsulated saporin within nanocapsules and conjugated anti-CD7 antibody on the surface (n-Saporin^anti-CD7) to target CD7⁺ cells in animals. Saporin mediates cell death by inhibiting protein synthesis through inactivation of the 60S subunit of ribosomes after uptake by cells.30 However, internalization of saporin is inefficient,31 and our nanocapsule platform designed for RTX delivery using zwitterionic monomer MPC is unable to achieve efficient intracellular delivery.9 17 To achieve effective intracellular delivery of saporin, we nanoencapsulated saporin using a mixture of neutrally charged monomer acrylamide and positively charged monomer N-(3-aminopropyl)methacrylamide. As a crosslinker, we used GDMA, which releases encapsulated saporin in low pH conditions, that is, in endosomes. With these modifications, n-Saporin^anti-CD7 can be specifically internalized in cells expressing CD7 molecules through endocytosis, release saporin in endosomes due to low pH conditions, and thereafter mediate cell death. The specificity and efficiency of cell death induced by n-Saporin^anti-CD7 were assessed in human peripheral blood mononuclear cells ex vivo (online supplemental figure S10). The most dramatic reduction in cell number was seen in CD56⁺/CD3⁻ NK cells (approximately 60%), while amounts of both CD4⁺ T cells and total CD8⁺ T cells (CD8^high and CD8^dim) derived from CD3⁺ T cells decreased by approximately 30%. CD19⁺ B cells in the CD45⁺ population and CD14⁺ in the CD3⁻/CD19⁻ population did not seem to be a major target of n-Saporin^anti-CD7. NSG^Hc1-BLT mice that were successfully protected from the second xenograft were treated with either n-Saporin^anti-CD7 or unliganded n-Saporin as a control and monitored for 2F7-BR44 derived bioluminescence 1 week after n-Saporin^anti-CD7 administration (figure 5D). Mice treated with n-Saporin^anti-CD7, but not control n-Saporin, lost their antilymphoma control. These results strongly suggested that CD7⁺ cells, mainly NK cells and T cells to some capacity, in these mice play an important role in cell-associated antilymphoma response in n-RTX^CXCL13-treated NSG^−Hc1-BLT mice.