An Innovative System to Study Immune Responses in Drug Development

International Immunopharmacology 90 (2021) 107237

Contents lists available at ScienceDirect

International Immunopharmacology

journal homepage: www.elsevier.com/locate/intimp

The blood endothelial cell chamber – An innovative system to study immune responses in drug development

Brina Stanˇciˇc a,b, Bodil Qvarfordt a, Magnus M. Berglund c, Nina Brenden d, Mona Sydow Backman¨ d, Moa Fransson a, Sofia Nordling a, Peetra U. Magnusson a,* a Department of Immunology, Genetics and Pathology, Rudbeck Laboratory, Dag Hammarskjoldsv¨ ag¨ 20, 751 85, Uppsala University, Uppsala, Sweden b Department of Molecular Biology, Universidad Autonoma´ de Madrid, and Department of Molecular Neuropathology, Center of Molecular Biology Severo Ochoa (UAM- CSIC), Nicolas´ Cabrera 1, 28049 Madrid, Spain c BeMa BioConsulting AB, Tornv¨ agen¨ 23, 136 72 Vendelso,¨ Sweden d Swedish Orphan Biovitrum AB, Tomtebodavagen¨ 23A, 112 76 Solna, Sweden

ARTICLE INFO ABSTRACT

Keywords: The risk for adverse immune-mediated reactions, associated with the administration of certain immunothera Endothelial cells peutic agents, should be mitigated early. Infusion reactions to monoclonal antibodies and other bio Whole blood pharmaceuticals, known as cytokine release syndrome, can arise from the release of cytokines via the drug target Cytokine release cell, as well as the recruitment of immune effector cells. While several in vitro cytokine release assays have been TGN1412 proposed up to date, many of them lack important blood components, required for this response to occur. The Alemtuzumab Cetuximab blood endothelial cell chamber model is an in vitro assay, composed of freshly drawn human whole blood and Proinflammatory cytokines cultured human primary endothelial cells. Herein, its potential to study the compatibility of immunotherapeutics with the human immune system was studied by evaluating three commercially available monoclonal antibodies and bacterial endotoxin lipopolysaccharide. We demonstrate that the anti-CD28 antibody TGN1412 displayed an + adaptive cytokine release profile and a distinct IL-2 response, accompanied with increased CD3 cell recruit ment. Alemtuzumab exhibited a clear cytokine response with a mixed adaptive/innate source (IFNγ, TNFα and + IL-6). Its immunosuppressive nature is observed in depleted CD3 cells. Cetuximab, associated with low infusion reactions, showed a very low or absent stimulatory effect on proinflammatory cytokines. In contrast, bacterial endotoxin demonstrated a clear innate cytokine response, definedby TNFα, IL-6 and IL-1β release, accompanied + + with a strong recruitment of CD14 CD16 cells. Therefore, the blood endothelial cell chamber model is pre sented as a valuable in vitro tool to investigate therapeutic monoclonal antibodies with respect to cytokine release and vascular immune cell recruitment.

1. Introduction adjusting the infusion rate and/or the dose, its severity should not be neglected [4–6]. That became evident in a clinical trial, where the Intravenous administration of immunotherapeutics may result in administration of an anti-CD28 antibody (TGN1412) caused a nearly immune-related adverse events. Cytokine release syndrome (CRS) fatal systemic inflammatoryresponse in six healthy volunteers [2]. This comprises a subset of infusion reactions that is associated with the use of incident was neither predicted by the available in vitro preclinical assays, therapeutic monoclonal antibodies (mAbs). CRS results in a systemic nor by the examined animal models [7,8], thus demonstrating the need release of several inflammatory mediators, which set off a cascade of for novel in vitro cytokine release assays for therapeutic mAbs. cytokines. It normally follows a similar profile and timing of events, Since then, a variety of cytokine release assays, using either whole starting with the release of tumour necrosis factor alpha (TNFα) and blood [9–14] or peripheral blood mononuclear cells (PBMCs) interleukin 8 (IL-8), followed by interleukin 6 (IL-6) and interleukin 10 [8,10,14–20] have been developed. While whole blood assays proved to (IL-10) and in some cases, interleukin 2 (IL-2) [1–3]. Even though this be more sensitive for mAbs stimulating FcγRI-mediated cytokine release response can be controlled by administering anti-inflammatorydrugs or (alemtuzumab, anti-CD52, IgG1), PBMC assays showed improved

* Corresponding author. E-mail address: [email protected] (P.U. Magnusson). https://doi.org/10.1016/j.intimp.2020.107237 Received 26 August 2020; Received in revised form 4 November 2020; Accepted 23 November 2020 Available online 11 December 2020 1567-5769/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237 sensitivity for TGN1412 and TGN1412-like antibodies (muromonab, and the endothelium, as well as to accurately predict the cytokine anti-CD3, IgG2a) [21]. Nevertheless, mAb immobilisation [8,16], high release in response to mAbs. density cultures [20,22] or co-culture with endothelial cells [17,19] have been identifiedas pre-requisites for TGN1412 to stimulate human 2. Materials and methods PBMCs or human diluted blood. Namely, binding of the mAb to endo thelial cells has been suggested to facilitate the cross-linking of CD28 2.1. Ethical considerations molecules on lymphocytes, thus leading to their activation [8,16,17]. Alternatively, in undiluted human whole blood mAbs could elicit re Blood was drawn from healthy individuals by open system ven sponses in the absence of endothelial interface, highlighting the epuncture in approval with the regional Ethical Review Board in advantage of whole blood cytokine release assays [10,14]. Uppsala (Permit No. 2017/165). Additionally, the vascular endothelium exhibits antithrombotic properties, maintains the quiescence of the circulating leukocytes and 2.2. Heparinization creates an anti-inflammatory intravascular environment under physio logical conditions. During pathological conditions involving inflamma All the plastic surfaces that were in contact with blood were coated tion, the endothelial cells exhibit increased adhesion receptor with a double layer of immobilized heparin, the Corline heparin surface expression, resulting in leukocyte recruitment; and a reduced ability to (hereafter referred as CHS, Corline Biomedical AB, Uppsala, Sweden), inhibit haemostasis, which triggers the activation of the cascade systems according to the manufacturer’s instructions. and the innate response, leading to thrombosis, platelet activation and cytokine release [23,24]. The majority of preclinical safety assays are 2.3. Endothelial cell culture devoid of endothelial cells, thus unable to address the complex interplay between blood and endothelium in the immune response [25]. Primary human umbilical vein endothelial cells (HUVEC #CRL- Notwithstanding, pre-clinical cytokine release assays should incor 1730; ATCC, Manassas, VA, USA) passage 4 to 5 were thawed (1–2 min porate fresh human whole blood that contains a full range of cell types ◦ in 35 water bath) and seeded in 1% gelatine (#BCBL5862V, Sigma- involved in an early immune response, including the cascade systems. ◦ Aldrich, St. Louis, MO) pre-coated 75-cm2 culture flasks at 37 C in a Their investigation is unfortunately many times prevented due to the 5% CO humidified incubator. Cells were maintained in complete common use of excessively anticoagulated blood in cytokine release 2 endothelial cell growth medium microvascular (ECGM MV, cat#C- assays [11]. 22020, lot#404M176, PromoCell) supplemented with 1% PenStrep In the in vitro model, we applied herein, cultured primary human (Penicillin Streptomycin, ref#15140-122, lot#861671, Gibco, Invi endothelial cells are placed in contact with whole human blood (Fig. 1), trogen) until confluence was reached. Once confluent, cells were de thus incorporating all the components present in a blood vessel lumen tached (2.1 µM EDTA/PBS, 0.25% Trypsin), counted (ScepterTM 2.0 [26,27]. A protective layer of immobilized heparin on the surfaces of the Cell Counter, Merck Millipore) and reseeded at 200,000 cells/well in 5 connected chamber allows blood contact without an artificialactivation µg/mL human fibronectin (#F2006-1MG, lot#SLBR1700V, Sigma- of the coagulation cascade and the use of anticoagulants at a minimal Aldrich, St. Louis, MO) pre-coated 1-well glass chamber slides level [26,28]. This so-called blood endothelial cell chamber model has (Thermo Fisher Scientific, Lab-Tek™) and cultured until confluent. previously been used to study the biomaterial-blood-endothelial cell interactions [28] and their role in thrombo-inflammation [26], as well as the effect of TNFα on the immune cell recruitment and coagulation 2.4. Blood sampling and plasma preparation activation [27]. Until now, its capacity as a cytokine release assay has = not yet been assessed. Human blood was drawn from healthy donors (n 2) by open system In the present study, we have evaluated the efficiency of the blood venepuncture into CHS coated 50 mL tubes containing unfractionated endothelial cell chamber model in studying the compatibility of thera heparin (UFH; 100 IE/KY/ml, lot#A08437, Leo Pharma) at a final peutic mAbs with the human immune system. The immunomodulatory concentration of 0.75 IU/mL or 1 IU/mL, respectively. Thereafter, 1.5 effects of therapeutic mAb associated with severe (TGN1412) [2], mild mL of fresh blood was added to each well of 2-well blood chambers to moderate (alemtuzumab) [29,30] and no (cetuximab) [31–33] CRS (Fig. 1B) coated with immobilized heparin, CHS. The following stimuli: were compared to the response, elicited by a known proinflammatory Phosphate buffered saline (PBS; Basal condition), TGN1412 (1 μg/mL, agent, bacterial endotoxin lipopolysaccharide (LPS). To investigate the Evitria AG, Zurich, Switzerland), LPS (5 ng/mL, #055:B5, Sigma- role of endothelium in whole blood cytokine release assay, whole blood Aldrich, Stockholm, Sweden), alemtuzumab (1 μg/mL, Sanofi AB, (Blood) and whole blood with endothelial cell co-culture (Blood + EC) kindly donated by Swedish Orphan Biovitrum AB [Sobi], Stockholm, setups were compared. We present its potential to delineate the coagu Sweden) or cetuximab (10 μg/mL, Merck AB, kindly donated by Swedish lation system activation, the interactions between human whole blood Orphan Biovitrum AB [Sobi], Stockholm, Sweden) were added to the 2- well chambers just before blood chambers were connected to the PBS-

Fig. 1. The blood endothelial cell chamber model [27]. The chamber is composed of two cylinder-shaped wells allowing a volume of 1.5 mL blood/well (A). The blood containing 2-well chamber is connected to either glass culture slides with endothelial cells or heparinized polyvinyl chloride (PVC) slides. The blood chamber is ◦ tightly sealed using heparinized ethylene propylene o-rings (B). Thereafter the blood endothelial chamber is positioned on a rotating wheel at 37 C (C).

2 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237 washed endothelial cell culture glass slides or heparinized PVC slides, DakoCytomation, Glostrup, Denmark) and mouse anti-human CD16, Fc respectively. To specify, alemtuzumab and cetuximab used in the study Gamma Receptor III-FITC (1:100, #F7011, DakoCytomation, Glostrup, were commercial trademark mAbs. The blood endothelial cell chambers Denmark) on the cells in contact with the second blood chamber for 1 h ◦ ◦ were immediately positioned on a rotating wheel at ~30 rpm in a 37 C at RT or at 4 C over night. Primary antibodies were visualized with heating cabinet and incubated for 2 h or 4 h, respectively. After the secondary goat anti-mouse IgG Alexa Fluor® 488 antibody (1:500, incubation, the blood endothelial cell chambers were dismantled and #A11029, Molecular Probes, Eugene, Oregon, USA) and the cytoskel the cells were fixed with ice cold 1–4% paraformaldehyde (PFA), eton was stained with Texas Red®-X phalloidin (1:200, #T7471, Life respectively, for 30 min at room temperature (RT). One mL of blood was Technologies, Carlsbad, California, USA). Nuclei were counterstained ′ immediately collected in Eppendorf tubes with K3EDTA at a final con with 4 ,6-diamidino-2-phenylindole (DAPI; 10 µg/ml; Invitrogen, centration of 10 mM and immediately used for platelet, white blood cell Carlsbad, California, USA) and the slides were mounted with Fluo and red blood cell counting using the Sysmex XP-300 Automated He romount G (SouthernBiotech, Birmingham, AL, USA). Cell culture glass matology Analyzer (Sysmex Corporation, Kobe, Japan). The same cell slides were scanned using a Leica DMi8 Inverted Microscope (Leica types as above were measured in non-incubated whole blood samples Microsystems Wetzlar GmbH, Wetzlar, Germany). All images, including (Initial condition). To prepare plasma, blood was centrifuged at 2000 rcf the ones in Figs. 4 and 5, were taken at 20× magnification. ◦ for 10 min at 4 C and plasma samples obtained were rapidly frozen and ◦ stored at 80 C until analysed for inflammation and coagulation acti 2.8. Image analysis vation markers. + + + The experiments were performed under two different conditions: 2 h Two pipelines for the quantificationof CD3 and CD14 CD16 cells incubation, 0.75 IU/mL heparin (n = 6) and 4 h incubation, 1 IU/mL present on the surface of endothelial cells were developed in CellProfiler heparin (n = 3–7), using blood from different donors. (version 2.2.0, Broad Institute). The major steps of the pipeline were as + + + follows: 1) Identificationof CD3 and CD14 CD16 cells using Identify 2.5. Thrombin anti-thrombin III complex (TAT) enzyme-linked Primary Objects module, 2) Identification of endothelial cell surface immunosorbent assay (ELISA) using Identify Primary Objects module, 3) Relation of the identified + + + CD3 and CD14 CD16 towards the surface area of endothelial cells by Plasma levels of thrombin anti-thrombin (TAT) complex were applying the Relate Objects module. To obtain a relevant overview, the measured with the commercially available sandwich enzyme-linked quantification was performed on average of 228 individual images per immunosorbent assay (ELISA, Enzygnost® TAT micro, Siemens well. Detailed CellProfiler analysis script in Word format is included in Healthcare, Marburg, Germany) according to the manufacturer’s in the Supplementary Data 4 and 5. structions. Depending on the platelet count, plasma samples were diluted 10x, 30x and 50x corresponding to 15%, 20% or 30% platelet 2.9. Statistical analysis reduction, respectively. Samples were diluted in a non-protein binding 96-well microtiter plate (VWR, United Kingdom) prior to being trans Platelet, white blood cell and red blood cell counts were normalized ferred to the Enzygnost® TAT micro plate. The absorbance was read at against the initial cell counts from respective experiments and are pre 490 nm (1.0 s), the limits of detection were between 2 and 60 µg/L and sented as means ± standard error of mean (SEM). The data from the the values were expressed as µg/L. Samples were assayed in 6 (2 h ex multiplex cytokine assays was compared to unstimulated values from periments) and 5 (4 h experiments) biological replicates. respective experiments, excluding the samples with undetectable values below (referred to as˝ Below˝) and above the fitcurve range (referred to as 2.6. Multiplex cytokine assays ˝Above˝). If not stated differently, the values from stimulated blood chamber samples were normalized against basal levels from respective Plasma expression of proinflammatory cytokines and acute inflam experiments. Quantification of the immune cell recruitment was per mation markers was analysed using V-PLEX Proinflammatory Panel 1 formed by normalizing the values, expressed as the number of immune Human Kit and Vascular Injury Panel I Assay that utilize electro cells present on stimulated HUVECs, towards the numbers of cells chemiluminescence and were developed by Meso Scale Discovery (MSD, recruited by unstimulated HUVECs from respective experiment. Statis 1601 Research Blvd., Rockville, Maryland), following the manufac tical analyses were performed using GraphPad Prism version 7.02 soft turer’s instructions. Plasma samples were diluted 1:4 or 1:5 with Diluent ware (GraphPad software) comparing absolute values from 2, respectively, directly in the MSD plate. PBS with 0.05% Tween-20 unstimulated and stimulated conditions with an unpaired t-test, fol (Sigma–Aldrich Inc, St. Louis, MO, USA) was used as washing buffer. lowed by a Mann-Whitney U Test. *p < 0.05, **p < 0.01, (ns = not Briefly, the antibody-coated 96-well plates were washed three times, significant).In each case, the n values refer to independent experiments followed by the 1:1 addition of serially 7-point diluted calibrators or the using plasma from different blood donors. samples. After, the plates were sealed and incubated for 1 h at RT or over ◦ night at 4 C with shaking. Plates were aspirated, washed three times 3. Results and the mix of SULFO-TAG–conjugated secondary antibodies was added for 2 h at RT with shaking. Finally, the plates were washed three times 3.1. Coagulation activation in the blood endothelial cell chamber model and reading buffer was added before the plates were read in the MSD Sector Imager 2400. The assay was run on 3–5 biological replicates per Whole blood assays, nowadays used for prediction of CRS, require experimental condition. The MSD raw data in Excel format is included in high heparin concentrations to inhibit coagulation, which furthermore the Supplementary Data 2 and 3. inhibits the complement system [22]. To determine if the blood endo thelial cell chamber model (Fig. 1) with a low free heparin content can 2.7. Immunofluorescence microscopy be used as a tool to investigate the coagulation cascade, TAT complex formation was investigated. First, we defined the incubation time The cell culture glass slides with PFA-fixed HUVECs were washed required for the stimuli to elicit coagulation effects. This was done by with phosphate-buffered saline (PBS, Medicago AB, Uppsala, Sweden), analysing blood, which was stimulated for 2 h and 4 h in the presence blocked with 10% BSA/PBS (30 min-1 h at RT) and stained with mouse (Fig. 2A) and absence (Fig. 2B) of endothelial cells. PBS was included as anti-human CD3-FITC (1:100, #FR875, DakoCytomation, Glostrup, an unstimulated condition and LPS as a positive control due to its known Denmark) on cells in contact with one of the two blood chambers and a prothrombotic characteristics. Longer incubation period, 4 h, resulted in mixture of mouse anti-human CD14-FITC (1:100, #F0844, a significantincrease of TAT levels in LPS stimulated plasma compared

3 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 2. Coagulation cascade in human whole blood was activated in a time-dependent manner. Fresh human whole blood was incubated with PBS and LPS (5 ng/mL) in the presence (endothelial cells, Blood + EC) or absence (Blood) of HUVECs for 2 (A, n = 6) or 4 (B, n = 3–4) hours, respec tively. Plasma thrombin-antithrombin (TAT) complex formation was measured. While longer incubation time showed limited effect in unsti mulated condition, a significant (A) increase in coagulation activation was observed in LPS stim ulated blood. Unpaired t test was used to compare TAT levels induced after 2 and 4 h of incubation. **p < 0.01, (ns = not significant). to 2 h incubation, while unstimulated condition (PBS) was similarly low 3.3. Cytokine release in the blood endothelial cell chamber model in both incubation times (Fig. 2). Due to the more prominent response in the presence of LPS, 4 h incubation was selected for future experiments Upon the recruitment and the maturation of inflammatory cells, a in this study. Blood cell counts were analysed after incubation showing a proinflammatory phenotype of the endothelium, characterized by the reduction of white blood cells (Supplementary Data 1, Suppl. Fig. 1A) release of cytokines, is initiated [24]. To study if the blood endothelial while platelets and red blood cell counts were non-significantlychanged cell chamber model can predict an immediate cytokine release after 4 h (Supplementary Data 1, Suppl. Fig. 1B and C, respectively). incubation, multiple cytokine analyses were performed (Figs. 7, 8 and Next, we were interested in the coagulation cascade activation upon Supplementary Data 1, Suppl. Fig. 2A-D). The proinflammatory stim stimulation with therapeutic monoclonal antibodies. TAT analysis ulus, LPS, was used as a positive control. Cetuximab, associated with low showed a significantincrease in coagulation cascade activation after a 4 incidence of immediate CRS, was chosen as a negative control. Levels of h stimulation with TGN1412, LPS or alemtuzumab, but not with the most distinctive cytokines, namely IFNγ, IL-2, TNFα and IL-6, are cetuximab (Fig. 3). No difference in the TAT formation was seen when shown in Fig. 7A–D. Additionally, Fig. 7 represents normalized cytokine comparing TAT levels in the presence (Fig. 3A) and in the absence of values as compared to the unstimulated condition, depicting indicative HUVECs (Fig. 3B). proinflammatory profiles for all the stimuli used in this study. At a concentration of 1 µg/mL, TGN1412 induced a significant cytokine release of IFNγ (Blood + EC: mean = 106.00 pg/mL, p-value = 3.2. Immune cell recruitment to the endothelium in the chamber model 0.0317; Blood: mean = 59.90 pg/mL, p-value = 0.0286), IL-2 (Blood + EC: mean = 12.79 pg/mL, p-value = 0.0159; Blood: mean = 11.72 pg/ Increased expression of cellular adhesion molecules, accompanied by mL, p-value = 0.0286), TNFα (Blood + EC: mean = 21.84 pg/mL, p- the recruitment of the immune cells, is one of the characteristics of the value = 0.0317; Blood: mean = 17.95, p-value = 0.0286) and IL-6 endothelial cell activation [34]. The blood endothelial cell chamber (Blood + EC: mean = 200.08 pg/mL, p-value = 0.4127, Blood: 14.95 model’s ability to evaluate the condition of the endothelium was pg/mL, p-value = 0.0286) (Fig. 7A-D) in collected blood plasma investigated by staining the HUVECs using markers for T lymphocytes compared to corresponding PBS condition. High normalized values of (CD3), monocytes (CD14) and macrophages (CD16) after incubation IFNγ and IL-2 (Fig. 8A) suggested a clear cytokine release with an + with blood. A significant increase of CD3 cells compared to basal adaptive cytokine profile, distinctive for this monoclonal antibody. A conditions (Fig. 4A, PBS) was observed on the endothelial surface of strong cytokine release with significantly increased levels of TNFα TGN1412 (Fig. 4B, quantifiedin 6A) and LPS (Fig. 4C, quantifiedin 6A) (Blood + EC: mean = 804.38 pg/mL, p-value = 0.0159; Blood: mean = stimulated HUVECs, while the immunosuppressant alemtuzumab eli 831.31 pg/mL, p-value = 0.0286), IL-6 (Blood + EC: mean = 2019.14 + cited a significant drop in the number of adhered CD3 lymphocytes pg/mL, p-value = 0.0159 ; Blood: mean = 896.85 pg/mL, p-value = (Fig. 4D, ALM, quantified in 6A). Cetuximab treatment was similar to 0.0286) and IL-1β (Blood + EC: mean = 74.23 pg/mL, p-value = 0.0159 PBS control (Fig. 4E, quantified in 6A). LPS as a known stimulator of ; Blood: mean = 74.46 pg/mL, p-value = 0.0286) (Fig. 7C-E) was monocytes and monocyte-derived macrophages induced a robust induced by LPS at only 5 ng/mL. Coinciding with their highly upregu + + recruitment of CD14 CD16 cells (Fig. 5B, quantified in 6B), with no lated relative values (Fig. 8B), these three cytokines defined an innate effect seen in the other conditions (Fig. 5C-E, quantifiedin 6B) compared cytokine response. Stimulation with 1 µg/mL alemtuzumab induced a to PBS control (Fig. 5A).

Fig. 3. Coagulation cascade activation in the blood endothelial cell chamber model. Freshly drawn whole blood from healthy donors was incubated with PBS, TGN1412, LPS, alemtuzumab (ALM) or cetuximab (CET) in the presence (Blood + EC) or absence (Blood) of HUVECs for 4 h. A significant increase in coagula tion cascade activation was observed upon stimula tion with TGN1412, LPS and alemtuzumab (A, B) compared to PBS. Unpaired t test was used to compare TAT levels in stimulated conditions with respective unstimulated PBS condition (Blood + EC or Blood). *p < 0.05.

4 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 4. Representative immunohistochem + ical detection of recruited CD3 immune cells after 4 h incubation of unstimulated (A, PBS), TGN1412 (B, TGN, far right B panel shows high magnificationarea of the section in the middle panel marked with dotted line), LPS (C), alemtuzumab (D, ALM) or cetuximab (E, CET) stimulated HUVECs incubated with blood chambers. Left: + Recruitment of CD3 lymphocytes (green). + Right: Recruitment of CD3 lymphocytes (green), phalloidin (red) and DAPI (blue). Scale bar = 75 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

strong release of IFNγ (Blood + EC: mean = 7221.06 pg/mL, p-value = 4. Discussion 0.0357; Blood: mean = 7991.11 pg/mL, p-value = 0.0571), TNFα (Blood + EC: mean = 706.00 pg/mL, p-value = 0.0357; Blood: mean = Therapeutic mAbs have become widely used as targeted therapies for 763.75 pg/mL, p-value = 0.0571) and IL-6 (Blood + EC: mean = 534.27, cancer and autoimmune diseases [1]. Despite a number of advantages, p-value = 0.0357 ; Blood: mean = 91.00, p-value = 0.0571) (Fig. 7A, C- severe adverse reactions have been associated with their usage [2,3] D), also shown by their normalized values (Fig. 8C). These cytokines highlighting the necessity for careful safety evaluation. A clinically were therefore proposed as an indicative proinflammatoryprofile with a relevant in vitro model with high predictive value should incorporate all mixed adaptive/innate cytokine profile.At a concentration of 10 µg/mL, in vivo components, including involved cells, intact cascade systems, Cetuximab, an antibody with known low incidence of infusion reactions circulating antibodies, neutrophils and endothelial cells [11,19]. As [32,33], did not induce a cytokine release in the blood endothelial cell presented in a survey by Finco et al. in 2014, cytokine release assays chamber (Figs. 7, 8D). frequently utilised in the pharmaceutical industry, contracted research When elucidating the role of endothelial cells, the presence of organisations or academia, apply therapeutic mAbs either in solution HUVECs showed limited effect on cytokine release compared to blood phase (serum, PBMC or anti-coagulated whole blood) or in solid phase alone (Fig. 7). IL-6 was the only cytokine to display significantlyhigher (i.e. mAbs attached to plastic surface either in dry or wet phase, followed levels in the presence of HUVECs in PBS (p-value = 0.0159), TGN (p- by incubation with PBMC) [25]. Furthermore, high density cultures of T value = 0.0286) and LPS (p-value = 0.0286) conditions compared to cells or PBMCs were proven as an elegant strategy to simulate the blood (Fig. 7D). “lymph node like” state, which could elicit a full T cell response and Altogether, the blood endothelial cell chamber model revealed cytokine release [20]. It should be mentioned that the value of using stimulus-specificfirst infusion reaction profiles.Results are summarized complex model systems composed of unmanipulated human whole in Table 1 and will be further discussed below. blood in combination with endothelial cells is appreciated for its

5 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 5. Representative immunohistochem + + ical detection of recruited CD14 CD16 immune cells after 4 h incubation of unsti mulated (A, PBS), TGN1412 (B. TGN), LPS (C, far right C panel shows high magnifica tion area of section in the middle panel marked with dotted line), alemtuzumab (D, ALM) or cetuximab (E, CET) stimulated HUVECs incubated with blood chambers. + + Left: CD14 CD16 monocytes (green). + + Right: CD14 CD16 monocytes (green), phalloidin (red) and DAPI (blue). Scale bar = 75 μm. (For interpretation of the refer ences to colour in this figure legend, the reader is referred to the web version of this article.)

physiological relevance that more closely simulate the in vivo situation. venepuncture, which allows the use of a self-determined level of anti Hereby, an in vitro whole blood assay [27] that incorporates freshly coagulants, herein set to 1 IU/mL for a 4 h incubation. The model takes drawn human blood and primary human endothelial cells was under advantage of an air bubble that creates a rotating blood flow in the taken to investigate the immunomodulatory effects of mAbs, namely chamber, hence keeping the blood in movement. A protective heparin TGN1412, alemtuzumab (ALM, Lemtrada®) and cetuximab (CET, conjugate layer on all blood contact surfaces prevents the clot formation, Erbitux®). TGN1412 was studied as this anti-CD28 superagonist as well as it allows to elucidate blood activation triggered only by induced a severe systemic inflammatory response in the recipients of a endothelial cells. This setup preserves the integrity of the cascade sys first-in-man Phase I clinical trial [2]. Immunosuppressive anti-CD52 tems and permits the investigation of sensitive interactions between monoclonal antibody alemtuzumab was selected as a clinically used whole blood and endothelial cells [26,27]. As such, we believe it pro biological therapy for lymphocytic leukaemia and lymphoma, which can vides a relevant setup for studying the response to immunomodulatory cause inflammatory side effects, including cytokine-mediated first drugs. infusion reactions [30,35]. Cetuximab, epidermal growth factor recep Previous research in the blood endothelial cell chamber model tor (EGFR) inhibitor for cancer treatment was used as a negative control showed that localised anticoagulant treatment, achieved by applying a due to its high tolerability and low incidence of hypersensitive responses surface bound layer of heparin conjugate, can be exploited to study the [31–33]. Additionally, LPS was included as a known exogenous acti activation of the coagulation system [26]. To evaluate the potential of vator of mammalian and human cells and tissues [36], with a profound mAbs in eliciting a pro-thrombotic environment, TAT analysis was cytokine response [37] similar in some respects to that seen in TGN1412 performed upon 4 h of stimulation. The incubation time was set at 4 h [2]. based on the lifespan of the assay and with the aim of capturing late The sensitivity of this model is ensured by an open system inflammatory responses [38]. Indeed, LPS as the positive control

6 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 6. Immune cell recruitment to the endothelium in response to monoclonal anti bodies and LPS. Fresh whole blood from healthy donors was incubated with PBS, TGN1412 (TGN), LPS, alemtuzumab (ALM) or cetuximab (CET) in the presence of + HUVECs for 4 h. CD3 lymphocytes (A) and + + CD14 CD16 monocytes (B) present on either unstimulated or stimulated HUVECs after blood contact were quantified. Data shows immune cell numbers normalized against the unstimulated PBS condition. TGN1412 and LPS exhibit an increased + recruitment of CD3 cells, while a depletion is observed upon incubation with alemtuzu mab (A). LPS demonstrates strong stimula + + tory effect on CD14 CD16 monocytes (B). Unpaired t test was used to compare the + numbers of recruited CD3 (A) or + + CD14 CD16 (B) cells in stimulated condi tions with respective unstimulated PBS con dition. *p < 0.05, **p < 0.01. exhibited a much stronger coagulation activation after 4 h, as opposed to alternative whole blood cytokine release assays [10,11]. Alemtuzumab only 2 h incubation. Hence, our model successfully reported LPS as a exhibited a clear cytokine response with a mixed adaptive/innate potent inducer of coagulation activation. Alemtuzumab exhibited a cytokine source, characterized by the release of INF-γ, TNFα and IL-6. strong ability of eliciting a pro-thrombotic environment, which corre This response constitutes a characteristic first-dose reaction and corre lated well with low blood platelet counts (Supplementary Data 1, Suppl. lates well with the literature [30,47]. LPS demonstrated a clear cytokine Fig. 1B) and clinical data, which associated abnormal coagulation pa response with innate cytokine source, defined by TNFα, IL-6 and IL-1β rameters with this drug [35]. [48]. As these cytokines are released by monocytes and/or macro Prothrombotic environment stimulates the endothelium to adopt a phages, the profile was in line with the increased recruitment of + + pro-inflammatory status and upregulate adhesion molecules, which fa CD14 CD16 cells upon LPS exposure. Consistent with its low incidence cilitates the activation and recruitment of inflammatory cells [24,39]. of infusion reactions in patients [32] and previous pre-clinical studies This phenomenon was investigated by immunofluorescenceanalyses of [10,11], cetuximab did not induce a CRS profile. HUVECs after termination of the experiments. In the blood endothelial It is important to note that whole blood assays show greater sensi cell chamber model, TGN1412 as a superagonistic CD28-specific anti tivity to alemtuzumab-induced cytokine release, which is supposed to be + body, which leads to the proliferation of human CD4 lymphocytes triggered by the Fc portion of the mAb, while PBMC-based assays are + [40], induced a strong recruitment of CD3 T cells to the endothelial cell more sensitive in predicting the cytokine response to anti-CD28 stimuli + + surface. A strong recruitment of CD14 CD16 monocytes upon LPS (e.g. TGN1412), where CRS is target mediated [8,14]. Accordingly, stimulation correlates well with published data, which shows that it much higher absolute levels of certain cytokines (e.g. IFNγ), were stimulates monocytes by binding to their differentiation antigen, namely detected upon exposure to alemtuzumab, as compared to TGN1412, cell surface protein CD14 [36,41]. Alemtuzumab induced a significant even though they were both significantly upregulated in respective + drop in CD3 cells recruited to the endothelial cells. In fact, this mAb conditions. This discrepancy, attributed to their different mechanisms of binds to CD52-expressing cells (lymphocytes, monocytes, some den action, was successfully discarded by normalizing absolute cytokine dritic cells and at low levels natural killer cells), leading to their lysis and release levels towards respective unstimulated conditions, hence depletion [42]. Rapid depletion of T cells is attributed to their high CD52 obtaining relative cytokine values. This process showed how cytokines, expression levels [42,43], hence high susceptibility to alemtuzumab, released at low concentrations, can have a very informative value when which was also observed in significantly decreased white blood cell the background is removed, hence the effect of the stimulus is extracted counts (Supplementary Data 1, Suppl. Fig. 1A). In line with the litera and has also been suggested elsewhere [10]. Our model showed an effect ture, increased levels of soluble intercellular adhesion molecule (sICAM- in IL-6 expression in the unstimulated PBS condition of the Blood + EC 3) upon alemtuzumab treatment indicated complement dependent setup (Fig. 7D). It should be mentioned that IL-6 expression in cultured cytotoxicity as the mechanism of action, employed in this process endothelial cells is already high compared to circulating levels in (Supplementary Data 1, Suppl. Fig. 2C) [44]. To elaborate, ICAM-3 has humans and an increase is easily triggered by thrombin activation previously been suggested as a phagocytic marker of apoptotic human [49–51]. The IL-6 expression in the unstimulated condition thereby leukocytes, involved in chemotaxis and tethering of macrophages, hence exerted an impact on relative values when comparing blood combined in mediating apoptotic cell clearance by the complement system with endothelial cells with blood only (Fig. 8B). Since increased cyto [45,46]. kine expression in unstimulated condition can undermine the accuracy In the blood endothelial cell chamber, TGN1412 induced a cytokine of the relative values, it is relevant to present the cytokine release pro release with an adaptive cytokine profile (IL-2 and IFNγ), which files both ways. mirrored high IL-2 and IFNγ levels in TGN1412 clinical trial volunteers With respect to the cytokine release, our model exhibited high [2]. Our model could therefore simulate the indicative TGN1412 sensitivity, with comparable relative levels in whole blood and whole proinflammatory cytokine profile that many of the so far developed in blood endothelial cell co-culture setup. Unlike in assays using PBMCs vitro approaches failed to reproduce. More specifically, the severity of and in line with some previous studies, endothelial interface was not the adverse response to TGN1412 was previously correlated with the required for blood to respond to mAbs, thus confirmingthe advantage of level of IL-2 release, indicating IL-2 as TGN1412-differentiating cytokine whole blood pre-clinical assays, which include a full range of cell types, [15]. Indeed, when compared to other mAbs, TGN1412 induced the (auto) antibodies and cascade system factors [10]. Its incorporation, highest levels of IL-2 in our model, with its release being comparable to however, provides valuable data considering immune cell recruitment

7 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 7. Proinflammatory cytokine release in the blood endothelial cell chamber model. Fresh whole blood from healthy donors was incubated with PBS, TGN1412, LPS, alemtuzumab (ALM) or cetuximab (CET) in the presence (Blood + EC) or absence (Blood) of HUVECs for 4 h. Plasma levels of IL-2 (A), IFNγ (B), TNFα (C), IL-6 (D) and IL-1β (E) after 4 h stimulation are shown (n = 3–5). Unpaired t test was used to compare the cytokine levels in stimulated conditions with respective (Blood + EC or Blood) unstimulated PBS, as well as whole blood with endothelial cell co-culture (Blood + EC) and whole blood (Blood) setups within the same stimulated condition. *p < 0.05. and interactions, which underlines the added value of the blood endo create reliable results upon secured unstimulated conditions. Incorpo thelial cell chamber model as a pre-clinical tool for mAbs in ration of endothelial cells increases the complexity in comparison to development. other available whole blood models and enables the visualisation of In conclusion, we have shown that the blood endothelial cell immune cell interactions with the vasculature. The advantageous pres chamber model can, despite its complex biological heterogeneities, ence of the coagulation system increases the relevance of the presented

8 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Fig. 8. Proinflammatorycytokine profilesof TGN1412, LPS, alemtuzumab and cetuximab in human blood plasma. Freshly acquired whole blood was incubated with PBS, TGN1412, LPS, alemtuzumab or cetuximab in the presence (Blood + EC) or absence (Blood) of HUVECs for 4 h. All panels show normalized cytokine levels compared to the unstimulated PBS condition (n = 3–5). Unpaired t test was used to compare the cytokine levels in stimulated conditions with respective (Blood + EC or Blood) unstimulated PBS condition. *p < 0.05.

system, making it an important tool for evaluating the risk for CRS in Table 1 therapeutic mAbs. Principal proinflammatorycytokines, immune cell recruitment and activation of the coagulation system by the investigated antibodies and LPS. CRediT authorship contribution statement TGN1412 Alemtuzumab Cetuximab LPS Cytokines: Cytokines: Cytokines: Cytokines: Brina Stanciˇ c:ˇ Data curation, Formal analysis, Methodology, Vali IL-2, IFNγ IFNγ, IL-6, / IL-1β, TNFα, IL-6 dation, Writing - original draft, Writing - review & editing. Bodil TNFα Qvarfordt: Methodology, Formal analysis. Magnus M. Berglund: IC recruitment: IC recruitment: IC IC recruitment: + + CD3 Absent recruitment: CD3 and Conceptualization. Nina Brenden: Conceptualization. Mona Sydow + + Absent CD14 CD16 Backman:¨ Conceptualization. Moa Fransson: Conceptualization, Coagulation +++ / ++ Methodology. SofiaNordling: Conceptualization, Methodology. Peetra system: + U. Magnusson: Conceptualization, Methodology, Funding acquisition, Investigation, Project administration, Supervision, Writing - original Bold font: stimulus-specific reaction. draft, Writing - review & editing. Scoring system: weak upregulation (+), moderate upregulation (++), strong +++ upregulation ( ), weak reduction ( ), moderate reduction ( ), no effect Declaration of Competing Interest (/). IC = immune cell. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

9 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

Acknowledgements R. Stebbings, J.A. Mitchell, An autologous endothelial cell:peripheral blood mononuclear cell assay that detects cytokine storm responses to biologics, FASEB J. 29 (6) (2015) 2595–2602. The research leading to these results has received funding from the [20] P.S. Romer,¨ S. Berr, E. Avota, S.Y. Na, M. Battaglia, I. ten Berge, H. Einsele, SciLife Innovation Project, Uppsala University, Knut & Ragnvi Jacobs T. Hünig, Preculture of PBMCs at high cell density increases sensitivity of T-cell son’s Foundation, The Swedish Research Council (2013-9279) and the responses, revealing cytokine release by CD28 superagonist TGN1412, Blood 118 (26) (2011) 6772–6782. Åke Wiberg Foundation. [21] S. Vessillier, D. Eastwood, B. Fox, J. Sathish, S. Sethu, T. Dougall, S.J. Thorpe, R. Thorpe, R. Stebbings, Cytokine release assays for the prediction of therapeutic – Appendix A. Supplementary material mAb safety in first-in man trials Whole blood cytokine release assays are poorly predictive for TGN1412 cytokine storm, J. Immunol. Methods 424 (2015) 43–52. [22] K. Hussain, C.E. Hargreaves, A. Roghanian, R.J. Oldham, H.T. Chan, C. Supplementary data to this article can be found online at https://doi. I. Mockridge, F. Chowdhury, B. Frend´eus, K.S. Harper, J.C. Strefford, M.S. Cragg, γ org/10.1016/j.intimp.2020.107237. M.J. Glennie, A.P. Williams, R.R. French, Upregulation of Fc RIIb on monocytes is necessary to promote the superagonist activity of TGN1412, Blood 125 (1) (2015) 102–110. References [23] A.K. Bongoni, E. Salvaris, S. Nordling, N. Klymiuk, E. Wolf, D.L. Ayares, R. Rieben, P.U. Magnusson, P.J. Cowan, Surface modification of pig endothelial cells with a branched heparin conjugate improves their compatibility with human blood, Sci. [1] P.J. Bugelski, R. Achuthanandam, R.J. Capocasale, G. Treacy, E. Bouman-Thio, Rep. 7 (1) (2017) 4450. Monoclonal antibody-induced cytokine-release syndrome, Expert. Rev. Clin. [24] F. Fischetti, F. Tedesco, Cross-talk between the complement system and endothelial Immunol. 5 (5) (2009) 499–521. cells in physiologic conditions and in vascular diseases, Autoimmunity 39 (5) [2] G. Suntharalingam, M.R. Perry, S. Ward, S.J. Brett, A. Castello-Cortes, M. (2006) 417–428. D. Brunner, N. Panoskaltsis, Cytokine storm in a phase 1 trial of the anti-CD28 [25] D. Finco, C. Grimaldi, M. Fort, M. Walker, A. Kiessling, B. Wolf, T. Salcedo, monoclonal antibody TGN1412, N. Engl. J. Med. 355 (10) (2006) 1018–1028. R. Faggioni, A. Schneider, A. Ibraghimov, S. Scesney, D. Serna, R. Prell, [3] U. Winkler, M. Jensen, O. Manzke, H. Schulz, V. Diehl, A. Engert, Cytokine-release R. Stebbings, P.K. Narayanan, Cytokine release assays: current practices and future syndrome in patients with B-cell chronic lymphocytic leukemia and high directions, Cytokine 66 (2) (2014) 143–155. lymphocyte counts after treatment with an anti-CD20 monoclonal antibody [26] S. Nordling, J. Hong, K. Fromell, F. Edin, J. Brannstr¨ om,¨ R. Larsson, B. Nilsson, P. (rituximab, IDEC-C2B8), Blood 94 (7) (1999) 2217–2224. U. Magnusson, Vascular repair utilising immobilised heparin conjugate for [4] P. Godel,¨ A. Shimabukuro-Vornhagen, M. von Bergwelt-Baildon, Understanding protection against early activation of inflammation and coagulation, Thromb. cytokine release syndrome, Intensive Care Med. 44 (3) (2018) 371–373. Haemost. 113 (6) (2015) 1312–1322. [5] D.W. Lee, R. Gardner, D.L. Porter, C.U. Louis, N. Ahmed, M. Jensen, S.A. Grupp, C. [27] S. Nordling, B. Nilsson, P.U. Magnusson, A novel in vitro model for studying the L. Mackall, Current concepts in the diagnosis and management of cytokine release interactions between human whole blood and endothelium, J Vis Exp (93) (2014) syndrome, Blood 124 (2) (2014) 188–195. e52112. [6] A. Osterborg, C. Karlsson, J. Lundin, E. Kimby, H. Mellstedt, Strategies in the [28] J. Hong, K. Nilsson Ekdahl, H. Reynolds, R. Larsson, B. Nilsson, A new in vitro management of alemtuzumab-related side effects, Semin. Oncol. 33 (2 Suppl 5) model to study interaction between whole blood and biomaterials. Studies of (2006) S29–S35. platelet and coagulation activation and the effect of aspirin, Biomaterials 20 (7) [7] T. Expert Scientific Group on Phase One Clinical, G.W. Duff, Expert Scientific (1999) 603–611. Group on Phase One Clinical Trials final report, 2006. [29] T.T. Hansel, H. Kropshofer, T. Singer, J.A. Mitchell, A.J. George, The safety and [8] R. Stebbings, L. Findlay, C. Edwards, D. Eastwood, C. Bird, D. North, Y. Mistry, side effects of monoclonal antibodies, Nat. Rev. Drug Discov. 9 (4) (2010) P. Dilger, E. Liefooghe, I. Cludts, B. Fox, G. Tarrant, J. Robinson, T. Meager, 325–338. C. Dolman, S.J. Thorpe, A. Bristow, M. Wadhwa, R. Thorpe, S. Poole, “Cytokine [30] K. Thomas, J. Eisele, F.A. Rodriguez-Leal, U. Hainke, T. Ziemssen, Acute effects of storm” in the phase I trial of monoclonal antibody TGN1412: better understanding alemtuzumab infusion in patients with active relapsing-remitting MS, Neurol. the causes to improve preclinical testing of immunotherapeutics, J. Immunol. 179 Neuroimmunol. Neuroinflamm. 3 (3) (2016) e228. (5) (2007) 3325–3331. [31] D. Cunningham, Y. Humblet, S. Siena, D. Khayat, H. Bleiberg, A. Santoro, D. Bets, [9] N.S. Alakhras, J. Qiu, G.V. Rocha, D.R. Witcher, A. Koester, J. You, D.A. Schaer, R. M. Mueser, A. Harstrick, C. Verslype, I. Chau, E. Van Cutsem, Cetuximab B. Holmgaard, K. Driscoll, J.A. Willy, L.P. Malherbe, FcγRIIIa-dependent IFN-γ monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic release in whole blood assay is predictive of therapeutic IgG1 antibodies safety, colorectal cancer, N. Engl. J. Med. 351 (4) (2004) 337–345. MAbs 10 (6) (2018) 913–921. [32] T.J. George Jr., K.D. Laplant, E.O. Walden, A.B. Davis, C.E. Riggs, J.L. Close, S. [10] L. Bailey, L. Moreno, T. Manigold, S. Krasniqi, H. Kropshofer, H. Hinton, T. Singer, N. George, J.W. Lynch, Managing cetuximab hypersensitivity-infusion reactions: L. Suter, T.T. Hansel, J.A. Mitchell, A simple whole blood bioassay detects cytokine incidence, risk factors, prevention, and retreatment, J Support Oncol 8 (2) (2010) responses to anti-CD28SA and anti-CD52 antibodies, J. Pharmacol. Toxicol. 72–77. Methods 68 (2) (2013) 231–239. [33] B.H. O’Neil, R. Allen, D.R. Spigel, T.E. Stinchcombe, D.T. Moore, J.D. Berlin, R. [11] E.A.K. Fletcher, M. Eltahir, F. Lindqvist, J. Rieth, G. Tornqvist,¨ J. Leja-Jarblad, S. M. Goldberg, High incidence of cetuximab-related infusion reactions in Tennessee M. Mangsbo, Extracorporeal human whole blood in motion, as a tool to predict and North Carolina and the association with atopic history, J. Clin. Oncol. 25 (24) first-infusion reactions and mechanism-of-action of immunotherapeutics, Int. (2007) 3644–3648. Immunopharmacol. 54 (2018) 1–11. [34] K.K. Wu, P. Thiagarajan, Role of endothelium in thrombosis and hemostasis, Annu. [12] Y. Iwata, A. Harada, T. Hara, C. Kubo, T. Inoue, M. Tabo, C. Ploix, T. Manigold, Rev. Med. 47 (1996) 315–331. H. Hinton, M. Mishima, Is an in vitro whole blood cytokine assay useful to detect [35] M. Obermann, T. Ruck, S. Pfeuffer, J. Baum, H. Wiendl, S.G. Meuth, Simultaneous the potential risk of severe infusion reaction of monoclonal antibody early-onset immune thrombocytopenia and autoimmune thyroid disease following pharmaceuticals? J. Toxicol. Sci. 41 (4) (2016) 523–531. alemtuzumab treatment in relapsing-remitting multiple sclerosis, Mult Scler 22 (9) [13] M.R. Walker, D.A. Makropoulos, R. Achuthanandam, S. Van Arsdell, P.J. Bugelski, (2016) 1235–1241. Development of a human whole blood assay for prediction of cytokine release [36] M. Fujihara, M. Muroi, K. Tanamoto, T. Suzuki, H. Azuma, H. Ikeda, Molecular similar to anti-CD28 superagonists using multiplex cytokine and hierarchical mechanisms of macrophage activation and deactivation by lipopolysaccharide: cluster analysis, Int. Immunopharmacol. 11 (11) (2011) 1697–1705. roles of the receptor complex, Pharmacol. Ther. 100 (2) (2003) 171–194. [14] B. Wolf, H. Morgan, J. Krieg, Z. Gani, A. Milicov, M. Warncke, F. Brennan, S. Jones, [37] S.F. Lowry, Human endotoxemia: a model for mechanistic insight and therapeutic J. Sims, A. Kiessling, A whole blood in vitro cytokine release assay with aqueous targeting, Shock 24 (Suppl 1) (2005) 94–100. monoclonal antibody presentation for the prediction of therapeutic protein [38] S.L. Blok, G.E. Engels, W. van Oeveren, In vitro hemocompatibility testing: The induced cytokine release syndrome in humans, Cytokine 60 (3) (2012) 828–837. importance of fresh blood, Biointerphases 11 (2) (2016), 029802. [15] D. Eastwood, C. Bird, P. Dilger, J. Hockley, L. Findlay, S. Poole, S.J. Thorpe, [39] V.W. van Hinsbergh, Endothelium–role in regulation of coagulation and M. Wadhwa, R. Thorpe, R. Stebbings, Severity of the TGN1412 trial disaster inflammation, Semin Immunopathol 34 (1) (2012) 93–106. cytokine storm correlated with IL-2 release, Br. J. Clin. Pharmacol. 76 (2) (2013) [40] D. Eastwood, L. Findlay, S. Poole, C. Bird, M. Wadhwa, M. Moore, C. Burns, 299–315. R. Thorpe, R. Stebbings, Monoclonal antibody TGN1412 trial failure explained by [16] L. Findlay, D. Eastwood, R. Stebbings, G. Sharp, Y. Mistry, C. Ball, J. Hood, species differences in CD28 expression on CD4+ effector memory T-cells, Br. J. R. Thorpe, S. Poole, Improved in vitro methods to predict the in vivo toxicity in Pharmacol. 161 (3) (2010) 512–526. man of therapeutic monoclonal antibodies including TGN1412, J. Immunol. [41] H. Tsukamoto, S. Takeuchi, K. Kubota, Y. Kobayashi, S. Kozakai, I. Ukai, Methods 352 (1–2) (2010) 1–12. A. Shichiku, M. Okubo, M. Numasaki, Y. Kanemitsu, Y. Matsumoto, T. Nochi, [17] L. Findlay, G. Sharp, B. Fox, C. Ball, C.J. Robinson, C. Bird, R. Stebbings, K. Watanabe, H. Aso, Y. Tomioka, Lipopolysaccharide (LPS)-binding protein D. Eastwood, M. Wadhwa, S. Poole, R. Thorpe, S.J. Thorpe, Endothelial cells co- stimulates CD14-dependent Toll-like receptor 4 internalization and LPS-induced stimulate peripheral blood mononuclear cell responses to monoclonal antibody TBK1-IKK∊-IRF3 axis activation, J. Biol. Chem. 293 (26) (2018) 10186–10201. TGN1412 in culture, Cytokine 55 (1) (2011) 141–151. [42] S.P. Rao, J. Sancho, J. Campos-Rivera, P.M. Boutin, P.B. Severy, T. Weeden, [18] M.K. Joubert, M. Deshpande, J. Yang, H. Reynolds, C. Bryson, M. Fogg, M.P. Baker, S. Shankara, B.L. Roberts, J.M. Kaplan, Human peripheral blood mononuclear cells J. Herskovitz, T.J. Goletz, L. Zhou, M. Moxness, G.C. Flynn, L.O. Narhi, V. Jawa, exhibit heterogeneous CD52 expression levels and show differential sensitivity to Use of in vitro assays to assess immunogenicity risk of antibody-based alemtuzumab mediated cytolysis, PLoS ONE 7 (6) (2012) e39416. biotherapeutics, PLoS ONE 11 (8) (2016) e0159328. [43] L. Ginaldi, M. De Martinis, E. Matutes, N. Farahat, R. Morilla, M.J. Dyer, [19] D.M. Reed, K.E. Paschalaki, R.D. Starke, N.A. Mohamed, G. Sharp, B. Fox, D. Catovsky, Levels of expression of CD52 in normal and leukemic B and T cells: D. Eastwood, A. Bristow, C. Ball, S. Vessillier, T.T. Hansel, S.J. Thorpe, A.M. Randi,

10 B. Stanˇciˇc et al. International Immunopharmacology 90 (2021) 107237

correlation with in vivo therapeutic responses to Campath-1H, Leuk. Res. 22 (2) by CAMPATH 1-H: involvement of CD16 (FcgammaRIII) and CD11a/CD18 (LFA-1) (1998) 185–191. on NK cells, J. Clin. Invest. 98 (12) (1996) 2819–2826. [44] N.A. Baig, R.P. Taylor, M.A. Lindorfer, A.K. Church, B.R. Laplant, E.S. Pavey, G. [48] A. Ngkelo, K. Meja, M. Yeadon, I. Adcock, P.A. Kirkham, LPS induced inflammatory S. Nowakowski, C.S. Zent, Complement dependent cytotoxicity in chronic responses in human peripheral blood mononuclear cells is mediated through NOX4 lymphocytic leukemia: ofatumumab enhances alemtuzumab complement and Giα dependent PI-3kinase signalling, J. Inflamm. (Lond.) 9 (1) (2012) 1. dependent cytotoxicity and reveals cells resistant to activated complement, Leuk. [49] K. Johnson, Y. Choi, E. DeGroot, I. Samuels, A. Creasey, L. Aarden, Potential Lymphoma 53 (11) (2012) 2218–2227. mechanisms for a proinflammatory vascular cytokine response to coagulation [45] O.D. Moffatt, A. Devitt, E.D. Bell, D.L. Simmons, C.D. Gregory, Macrophage activation, J. Immunol. 160 (10) (1998) 5130–5135. recognition of ICAM-3 on apoptotic leukocytes, J. Immunol. 162 (11) (1999) [50] V. Marin, F.A. Montero-Julian, S. Gres,` V. Boulay, P. Bongrand, C. Farnarier, 6800–6810. G. Kaplanski, The IL-6-soluble IL-6Ralpha autocrine loop of endothelial activation [46] E.E. Torr, D.H. Gardner, L. Thomas, D.M. Goodall, A. Bielemeier, R. Willetts, H. as an intermediate between acute and chronic inflammation: an experimental R. Griffiths,L.J. Marshall, A. Devitt, Apoptotic cell-derived ICAM-3 promotes both model involving thrombin, J. Immunol. 167 (6) (2001) 3435–3442. macrophage chemoattraction to and tethering of apoptotic cells, Cell Death Differ. [51] R.B. Sothern, B. Roitman-Johnson, E.L. Kanabrocki, J.G. Yager, M.M. Roodell, J. 19 (4) (2012) 671–679. A. Weatherbee, M.R. Young, B.M. Nenchausky, L.E. Scheving, Circadian [47] M.G. Wing, T. Moreau, J. Greenwood, R.M. Smith, G. Hale, J. Isaacs, H. Waldmann, characteristics of circulating interleukin-6 in men, J. Allergy Clin. Immunol. 95 (5 P.J. Lachmann, A. Compston, Mechanism of first-dose cytokine-release syndrome Pt 1) (1995) 1029–1035.