Siplizumab Induces NK Cell Fratricide Through Antibody-Dependent Cell-Mediated Cytotoxicity

The glycoprotein CD2 is expressed on T and NK cells and contributes to cell-cell conjugation, agonistic signaling and actin cytoskeleton rearrangement. CD2 has previously been shown to have an important function in natural NK cell cytotoxicity but to be expendable in antibody-mediated cytotoxicity. Siplizumab is a monoclonal anti-CD2 IgG1 antibody that is currently undergoing clinical trials in the field of transplantation. This study investigated the effect of CD2 binding and Fc γ receptor binding by siplizumab (Fc-active) and Fc-silent anti-CD2 monoclonal antibodies in allogeneic mixed lymphocyte reaction and autologous lymphocyte culture. Further, induction of NK cell fratricide and inhibition of natural cytotoxicity as well as antibody-dependent cytotoxicity by these agents were assessed. Blockade of CD2 via monoclonal antibodies in the absence of Fc γ receptor binding inhibited NK cell activation in allogeneic mixed lymphocyte reaction. In contrast, siplizumab increased NK cell activation in both mixed lymphocyte reaction and autologous lymphocyte culture due to FcγRIIIA binding. However, experiments using purified NK cells did not show an inhibitory effect of CD2 blockade on natural cytotoxicity or antibody-dependent cytotoxicity. Lastly, it was shown that siplizumab induces NK cell fratricide. Concluding, siplizumab is a promising biopharmaceutical drug candidate for depletion of T and NK cells with minimal off-target effects.


INTRODUCTION
The glycoprotein CD2 (also known as LFA2) is expressed on T and NK cells where it serves as an adhesion and activation receptor (1). In humans, the main binding partner of CD2 is lymphocyte associated antigen 3 (LFA3; also known as CD58) which is broadly expressed, especially on antigenpresenting cells (APCs). CD2-targeting immunotherapy is a promising treatment for organ Abbreviations: ADCC, antibody-dependent cell-mediated cytotoxicity; ANOVA, analysis of variance; APC, antigenpresenting cell; DG, deglycosylated; FcgR, Fragment crystallizable gamma receptor; FcS, Fc-silent; KIR, killer cell immunoglobulin-like receptor; NKG2D, natural killer group 2D; HLA, human leukocyte antigen; mAb, monoclonal antibody; NK, natural killer; NKIS, NK cell immunological synapse; LFA3, lymphocyte-associated antigen 3; PBMC, peripheral blood mononuclear cells; RA, rheumatoid arthritis; VPD450, violet proliferation dye 450. transplant recipients and patients suffering from autoimmune diseases [Reviewed in (1)]. Siplizumab is a monoclonal anti-CD2 IgG1 antibody (anti-CD2 IgG1 mAb) that is currently undergoing clinical trials in the field of transplantation (ClinicalTrials.gov Identifier: NCT04311632). While the potent in vitro immune modulatory effects of siplizumab on T cells have been documented (2,3), no studies published to date address the in vitro effects of siplizumab on NK cells. Previous in vivo evidence showed siplizumab-induced peripheral T and NK cell depletion in primates (4) and human patients (5). Evidence suggests that antibody-dependent cytotoxicity is one depletory mechanism induced by siplizumab; however, the direct effects of siplizumab on NK cells remain incompletely characterized (3,6).
NK cells express activating and inhibitory receptors. The balance of signaling through both sets of receptors is integrated and determines whether mature NK cells remain in a resting state or become activated (7). Thus, NK cell activation can be elicited by a reduction of inhibitory signaling, an increase in activation signaling, or a combination of both.
Among activating NK cell receptors, CD16 is one whose signaling can induce NK cell activation in isolation. Most remaining activating NK cell receptors require activation in conjunction to elicit NK cell activation. CD16a is a low affinity Fc g receptor (FcgR) and is also known as FcgRIIIA. NK cell binding to target-bound IgG antibodies via CD16a promotes antibody-dependent cell-mediated cytotoxicity (ADCC). Other activating NK cell receptors include NKG2D, certain killer cell immunoglobulin-like receptors (KIRs) and NKp46 which bind to tumor antigens, foreign HLA and viral proteins, respectively. Most prominent among inhibitory NK cell receptors are KIRs which recognize self-HLA (8). Antibody-independent target cell killing by NK cells is commonly termed natural or spontaneous NK cell cytotoxicity. In addition to FcgRIIIA, certain NK cell subsets can express two forms of CD32/FcgRII, FcgRIIB, and FcgRIIC (9).
A critical event preceding NK cell cytotoxicity is stable target cell conjugation via cell-cell adhesion molecules (10). Once stable target cell conjugation has occurred and the NK cell immunological synapse (NKIS) has formed, actin cytoskeleton rearrangement proceeds to facilitate transport of lytic vesicles to the NKIS and subsequent release to achieve target cell killing. Examples of adhesion molecules involved in NK-target cell conjugation and NKIS formation are LFA-1 (8) and CD2 (10) binding to ICAM-1 and LFA3 on the target cell, respectively. Even in natural NK cell cytotoxicity CD16 is enriched in the NKIS through interaction with CD2. Abrogation of this interaction between CD2 and CD16 markedly decreases natural NK cell cytotoxicity but does not affect NK-mediated ADCC (11). CD2 has two Ig domains, one membrane distal domain and one membrane proximal domain. The membrane distal domain binds LFA3 but to date the site on CD2 which interacts with CD16 has not been reported. Depending on where CD16 binds to CD2, simultaneous binding of CD2 to both CD16 and LFA3 may be conceivable. The role of CD2 in NK cell biology is incompletely characterized.
Similar to what has been observed during T cell-antigen presenting cell (APC) conjugation, CD2 accumulates in the immunological synapse that NK cells form upon target cell conjugation. While the CD2-CD58 interaction has traditionally been known for its role in cellular adhesion, recent evidence has demonstrated that it also plays an important role in recruiting and organizing activating receptors to the immunological synapse (11,12).
This study aimed to characterize the effects of siplizumab on NK cell activation in mixed lymphocyte reaction (MLR) and in pure NK cell culture. Further, potential effects of CD2 blockade on natural cytotoxicity and ADCC were investigated.

Generation of Fc-Silent Anti-CD2 Antibodies
Siplizumab (humanized anti-CD2 IgG1k) is an investigational drug and was provided by the manufacturer (ITB-Med, Stockholm, Sweden).
Fc-silent (FcS) anti-CD2 mAbs were generated via introduction of amino acid substitutions into the sequence of siplizumab or standard IgG2/IgG4 frameworks. All Fc-silent antibodies had identical light chains and complementaritydetermining regions and were produced by Wuxi Biologics Limited (Wuxi, China). Genetic sequences for each antibody were cloned into plasmids followed by transient transfection and production in Chinese hamster ovary (CHO-K1) cells. Anti-CD2 mAbs were purified using protein A and IEX/SEC columns to at least 95% purity and endotoxin levels below 1 EU/mg. Fc-silent siplizumab (FcS anti-CD2 IgG1) was generated by introducing the commonly known LALA-PG mutations to equivalent sites in siplizumab (13,14). FcS anti-CD2 IgG2 was generated by introducing the following mutations at equivalent sites in a standard IgG2 framework: V234A, G237A P238S, H268A, V309L, A330S, P331S (15). FcS anti-CD2 IgG4 was generated by introducing the following mutations at equivalent sites in a standard IgG4 framework: S228P, L235E, P329G (16,17).

Surface Plasmon Resonance CD2 Binding
Interaction experiments were performed using a Biacore T200 instrument (Cytiva Life Sciences/Biacore, Uppsala, Sweden). IgG was immobilized to a surface density of 200-1300 RU using standard amine coupling and CM5 biosensor chips (Cytiva Life Sciences/Biacore, Uppsala, Sweden) with a 10 mM HEPES buffer at pH 7.4, with 150 mM NaCl, 0.5 mM EDTA, 0.05% Tween20 running buffer. Each antibody was immobilized at least twice.
All interaction experiments were performed at 25°C. Recombinant truncated human CD2 (Sino Biological, Beijing, China) was diluted in two-fold dilution series with the highest concentration being 5 nM and injected for 180 s over immobilized IgG. After 15 min of dissociation, the sensor chip was regenerated by 30 s injection of 10 mM glycine pH 1.7.
Sensorgrams from reference surfaces and blank injections were subtracted from the raw data prior to data analysis. A two-state (induced fit) interaction model was fitted to the data using T200 evaluation software v3.0 (Cytiva Life Sciences/Biacore, Uppsala, Sweden) to determine kinetic rate constants and affinities.

SPR FcgR Binding
Recombinant truncated human CD2 (Sino Biological, Beijing, China) was covalently immobilized by amine coupling to CM5 biosensor chips (Cytiva Life Sciences/Biacore, Uppsala, Sweden). IgG was captured to amine coupled CD2. All experiments were performed using a Biacore T200 ™ instrument (Cytiva Life Sciences/Biacore, Uppsala, Sweden) at 25°C. The characterization of compound interactions were conducted in a 10 mM HEPES buffer at pH 7.4, with 150 mM NaCl, 0.5 mM EDTA, 0.05% Tween20. Fcg receptors (R&D systems, Minneapolis USA) were injected for 120 s in multi-cycle or single-cycle experiments in concentration series up to 0.5 mM over CD2-captured IgGs. The R131 isoform of FcgRIIA and V176 isoform of FcgRIIIA were used for all experiments. CD2 surfaces were regenerated by a 120 s injection of 50 mM NaOH. Sensorgrams were doublereferenced (reference surface, blanks) prior to global analysis using a 1:1 interaction model or a heterogeneous binding model. For weaker interactions with fast dissociation the affinity was determined by steady state analysis using a sum of a Langmuir isotherm and a linear term (compensating for non-specific interaction). Kinetic parameters were determined as average values based on 2-3 replicate experimental series.

Cell-Based FcgR Signaling Assay
Jurkat reporter cell lines (Promega Corp., Madison, Wisconsin USA) transfected with FcgRI, FcgRIIA or FcgRIIIA, respectively, were incubated with serial dilutions of each antibody agent for 23 h. Binding of a target-bound antibody through the respective FcgR induced expression of luciferase in reporter cells. Luciferase expression was detected after 23 h of incubation at 37°C 5% CO 2 via addition of luciferase assay substrate (Promega Corp., Madison, Wisconsin USA) and luminescence measurement using a Synergy HTX multiplate reader (BioTek, Winooski, Vermont USA).

Isolation of Peripheral Blood Mononuclear Cells
Peripheral blood mononuclear cells (PBMC) were isolated via Ficoll ® Paque Plus (Cytiva Life Sciences, Uppsala, Sweden) density gradient centrifugation from buffy coats. Buffy coats were obtained from healthy donors via Uppsala University Hospital blood bank (Uppsala, Sweden) as well as Karolinska University Hospital blood bank (Stockholm, Sweden) and PBMC isolation was carried out within 24 h of blood collection.
PBMC were washed twice in saline solution. Samples were incubated with Fc-receptor binding inhibitor (Invitrogen; Thermo Fisher Scientific Inc., Waltham, USA) and then washed twice in FBS stain buffer (BD BioSciences, San Diego, USA) followed by incubation with 10 µg/ml siplizumab for 20 min. Unbound antibody was removed before staining with secondary anti-human IgG Fc BV421 (BioLegend, San Diego, USA; Clone HP6017). Unbound secondary antibody was removed before surface staining for T cell subpopulations, regulatory T cells or B and NK cells. All antibodies listed below were purchased from BD Biosciences (San Diego, USA) if not indicated otherwise.
Samples were stained in the dark at 4°C and were acquired using a FACSVerse flow cytometer (BD Biosciences, San Diego, USA). Post-acquisition editing and data analysis was conducted using FlowJo ® 10.5.3 software (FlowJo LLC, Ashland, USA).

Autologous Lymphocyte Culture and Mixed Lymphocyte Reaction
For analysis of NK cell activation over time in the presence of siplizumab and DG siplizumab, equal amounts of PBMC from each donor were stained with violet proliferation dye 450 (BD Biosciences, San Diego, USA; VPD450) for another experiment. Cells were resuspended in 10% heat-inactivated FBS (Gibco, Thermo Fisher Scientific Inc., Waltham, USA) in RPMI-1640 ATCC Mod. supplemented with 50 U/ml Streptomycin and Penicillin (Gibco, Thermo Fisher Scientific Inc., Waltham, USA), respectively. Resuspended PBMC were dispensed into round-bottom 96-well cell culture plates and pure medium (no antibody controls) or one of the respective antibody agents diluted in cell culture medium (10 µg/ml) was added to a final concentration of 2x10 6 cells per ml (final volume 200 µl). ALCs and MLRs were then incubated at 37°C 5% CO 2 for 1, 2, 4, and 7 days, respectively. On each day, cells were washed twice in saline solution and incubated with Fc-receptor binding inhibitor (Invitrogen; Thermo Fisher Scientific Inc., Waltham, USA) followed by staining with anti-CD16 FITC (Clone 3G8), anti-CD56 PE (Miltenyi, Bergisch Gladbach, Germany; Clone REA196), anti-CD3 PerCP-Cy5.5 (Clone SP34-2), anti-CD69 APC (Clone FN50) and anti-CD14 APC-H7 (Clone MjP-9). NK cells were identified as CD3 -CD14 -CD56 + and/or CD16 + lymphocytes. All flow cytometry antibodies were purchased from BD Biosciences (San Diego, USA) if not indicated otherwise. Samples were stained in the dark at 4°C and were acquired using a FACSVerse flow cytometer (BD Biosciences, San Diego, USA). Post-acquisition editing and data analysis was conducted using FlowJo ® 10.5.3 software (FlowJo LLC, Ashland, USA).

Natural Cytotoxicity Assay
NK cells were isolated from PBMC via negative MACS selection using NK cell isolation kit (Miltenyi, Bergisch Gladbach, Germany).
For runs with short pre-incubation, NK cells were resuspended in 10% heat-inactivated FBS (Gibco, Thermo Fisher Scientific Inc., Waltham, USA) in AIM-V medium supplemented with 50 µg/ml streptomycin sulfate and 10 µg/ ml gentamicin sulfate (Gibco, Thermo Fisher Scientific Inc., Waltham) and dispensed into round-bottom 96-well cell culture plates followed by resting at 37°C 5% CO 2 overnight. Following, pure medium (no antibody controls) or one of the respective antibody agents diluted in cell culture medium (final concentration 10-0.001 µg/ml) were added followed by 30 min incubation at 37°C 5% CO 2 before addition of HLA class Inegative target cells (SPI-801, DSMZ Cat. ACC 86; NK cell to target cell ratio between 1:1 and 1:2 depending on yield of MACS isolation). Addition of target cells was followed by incubation at 37°C 5% CO 2 for 2 h and subsequent flow cytometry staining as described in NK Fratricide section.
For runs with pre-incubation of NK cells with antibody, NK cells were resuspended in in 10% heat-inactivated FBS (Gibco, Thermo Fisher Scientific Inc., Waltham, USA) in AIM-V medium supplemented with 50 µg/ml streptomycin sulfate and 10 µg/ml gentamicin sulfate (Gibco, Thermo Fisher Scientific Inc., Waltham) as well as 500 IU/ml human IL-2 (Miltenyi, Bergisch Gladbach, Germany). Further, pure medium (No antibody controls) or one of the respective antibody agents diluted in cell culture medium (final concentration 10-0.001 µg/ml) were added followed by incubation at 37°C 5% CO 2 for 2 days. On day 2, HLA class I-negative target cells were added (SPI-801; NK cell to target cell ratio between 1:1 and 1:2 depending on yield of MACS isolation) followed by incubation at 37°C 5% CO 2 for 2 h and subsequent flow cytometry staining as described in NK Fratricide section.

Rituximab-Induced ADCC Assay
NK cells were isolated from PBMC via negative MACS selection using NK cell isolation kit (Miltenyi, Bergisch Gladbach, Germany). resuspended in 10% heat-inactivated FBS (Gibco, Thermo Fisher Scientific Inc., Waltham, USA) in AIM-V medium supplemented with 50 µg/ml streptomycin sulfate and 10 µg/ml gentamicin sulfate (Gibco, Thermo Fisher Scientific Inc., Waltham) and dispensed into roundbottom 96-well cell culture plates. Further, pure medium (No antibody controls) or one of the respective anti-CD2 antibody agents diluted in cell culture medium (final concentration 10 µg/ml) were added followed by overnight incubation at 37°C 5% CO 2 . The next day Daudi target cells (DSMZ Cat. ACC 78; CD20 + ; NK to Daudi ratio 4:1) and Rituximab (final concentration 0.0001-1 µg/ml) were added followed by 2 h of incubation at 37°C 5% CO 2 . Subsequently, cells were washed in saline solution and incubated with Fc-receptor binding inhibitor (Invitrogen; Thermo Fisher Scientific Inc., Waltham, USA) followed by staining with anti- CD56  anti-CD19 APC-H7 (BD Biosciences, San Diego, USA; Clone SJ25C1). NK cells were identified as CD3 -CD56 + and/or CD16 + lymphocytes. Samples were stained in the dark at 4°C and were acquired using a FACSCelesta flow cytometer (BD Biosciences, San Diego, USA). Post-acquisition editing and data analysis was conducted using FlowJo ® 10.6.2 software (FlowJo LLC, Ashland, USA).

Graphs and Statistical Analysis
Visualization of results and statistical analysis of underlying data were carried out using GraphPad Prism 8 software (GraphPad Software, San Diego). Data were analyzed using repeatedmeasure one-way ANOVA followed by Dunnett's multiple comparison test or two-way ANOVA followed by Dunnett's multiple comparison test unless specified otherwise. Untreated controls served as the comparator if not specified otherwise.

CD2 and Fc Gamma Receptor Binding
Anti-CD2 antibody binding to CD2 and FcgR binding was characterized via surface plasmon resonance (SPR). As shown in Table 1, all tested antibodies displayed a similar CD2 binding affinity with an affinity constant (K D ) of circa 0.9-1.5 nM. Siplizumab binds FcgRI, FcgRIIA and FcgRIIIA. In contrast, deglycosylated (DG) siplizumab binds FcgRI but not FcgRIIA and FcgRIIIA. All three Fc-silent anti-CD2 antibodies did not display detectable FcgR affinity.
To confirm that FcgR-binding results in FcgR-induced signaling, each antibody was tested in a cell-based FcgR signaling assay (Supplementary Tables S1-S3). As shown in Figure 1, consistent with SPR results siplizumab induced signaling through FcgRI, FcgRIIA and FcgRIIIA. In contrast, DG siplizumab only induced signaling through FcgRI. None of the Fc-silenced anti-CD2 variants induced FcgRmediated signaling.

CD2 and FcgR Expression on NK Cells
CD2 expression on NK cells was measured via flow cytometry. As shown in Figure 2A, NK cells tend to have a lower CD2 expression than T cells. While CD56 bright NK cells express CD2 at comparable levels as the average T cell, CD56 dim and CD56 neg CD16 + NK cells express CD2 at significantly lower levels (p=0.0074 and p=0.0111, respectively; Repeated-measure Additionally, CD2 expression on T and NK cells was measured before and after activation of PBMC with anti-CD3 and anti-CD28 antibody-conjugated microbeads (Supplementary Table S5). As displayed in Figure 2B, CD2   Figure 3, almost all NK cells (91.0%) expressed FcgRIIIA (CD16) while FcgRII expression was heterogeneous (19.2%, CD32). NK cells did not express FcgRI (data not shown).

Effect of CD2 Blockade on Antibody-Mediated and Natural NK Cell Activation in Autologous and Mixed Lymphocyte Culture
To investigate the effect of siplizumab on NK cell activation, autologous lymphocyte culture (ALC) and MLR were performed with a saturating dose of siplizumab. As shown in Figure 4, siplizumab significantly increased the percentage of CD69+ NK   Figure 4, DG siplizumab significantly reduced % CD69+ NK cells after 4 (p=0.0012) and 7 (p=0.0035) days (Repeated-measure two-way ANOVA followed by Dunnett's multiple comparison test). Addition of DG siplizumab did not meaningfully change the percentage of CD69 + NK cells in ALC.

NK Fratricide
To test whether siplizumab can induce NK cell fratricide, purified NK cells were cultured in the presence of increasing doses of siplizumab or Fc-silent anti-CD2 mAbs. As shown in Figure 6, siplizumab significantly increased NK cell lysis in purified NK cell culture at 0.01-10 µg/ml (p ≤ 0.0018) relative to untreated controls. Furthermore, siplizumab induced a significant dose-dependent depletion of total NK cells (p ≤ 0.0301), CD56dim NK cells (p ≤ 0.0191) and CD56bright NK cells (p ≤ 0.0093) at 0.01-10 µg/ml when compared to untreated controls. In contrast, FcS anti-CD2 mAbs did not induce dosedependent NK cell lysis or depletion in pure NK culture, except FcS anti-CD2 IgG2 which induced a mild but statistically significant depletion of CD56bright NK cells at 0.1-10 µg/ml (p ≤ 0.0132; Repeated-measure one-way ANOVA followed by Dunnett's multiple comparison test; Supplementary Tables  S10-S13).
Additionally, Figure 7 shows that siplizumab induced a significant dose-dependent increase in NK cell degranulation (p ≤ 0.0114; CD107a + NK cells) and a strong decrease of CD16 expression on NK cells (p ≤ 0.0022), indicative of ADCC. No FcS anti-CD2 mAb induced degranulation and only FcS anti-CD2 IgG2 induced a minor but statistically significant decrease in CD16 MFI at 0.1-1 µg/ml (p ≤ 0.0269; Repeated-measure oneway ANOVA followed by Dunnett's multiple comparison test; Supplementary Tables S14 and S15).

Natural NK Cytotoxicity
To test the effect of CD2 blockade on natural cytotoxicity, purified NK cells were incubated with titrated doses of siplizumab or FcS anti-CD2 mAb in the presence of SPI-801 target cells (HLA).
As shown in Figure 8, CD2 blockade with FcS anti-CD2 mAb did not significantly inhibit NK cell degranulation in response to SPI-801 target cells, irrespective of whether pre-incubation of NK cells with FcS anti-CD2 mAb lasted 30 min or 2 days prior to target cell addition. After 30 min of pre-incubation, siplizumab significantly increased NK cell degranulation at 0.1-10 µg/ml (p ≤ 0.0482) but not after 2 days of pre-incubation (Repeatedmeasure one-way ANOVA followed by Dunnett's multiple comparison test; Supplementary Tables S16 and S17).

Antibody-Dependent Cytotoxicity
To test whether CD2 blockade influences ADCC, purified NK cells were pre-incubated with siplizumab or FcS anti-CD2 mAb before addition of titrated doses of Rituximab and CD20 + target cells (Daudi). As shown in Figures 9A, B, CD2 blockade with FcS anti-CD2 mAb did not significantly affect NK cell degranulation or target cell killing in response to increasing doses of Rituximab (Supplementary Tables S18 and S19). As displayed in Figure 9C, Rituximab induced a significant increase of CD2 expression on NK cells at 0.01-1 µg/ml (p ≤ 0.0248; Repeated-measure one-way ANOVA followed by Dunnett's multiple comparison test; Supplementary Table  S20). While anti-CD2 mAbs induced different degrees of CD2 downregulation in this setting, this did not seem to affect ADCC.

DISCUSSION
The present study has demonstrated that siplizumab can induce NK cell fratricide through ADCC and that CD2 binding by siplizumab does not induce NK cell activation. Further, it was shown that blockade of CD2 with Fc-silent anti-CD2 mAb inhibits CD69 expression on NK cells in MLR. While these results suggested a potential inhibitory effect of CD2 blockade on natural cytotoxicity in MLR, no inhibition of natural cytotoxicity was detected in experiments using purified NK cells. Further, while siplizumab induces ADCC through CD16, this study showed that CD2 blockade by siplizumab does not inhibit ADCC. It should be noted that siplizumab was included in assays evaluating the effect of CD2 blockade on Rituximabinduced ADCC and natural cytotoxicity for completeness, however, NK cell fratricide makes results seen with siplizumab in these assays difficult to interpret. Limitations of this study included the exclusive use of peripheral blood NK cells which often display different functional profiles and phenotypes when compared to tissue NK cells (18)(19)(20)(21). Moreover, all anti-CD2 antibodies used in this study bind the same CD2 epitope [situated near T11.2 and T11.3 in the extracellular domain of CD2 (22)]. Thus, it remains to be determined whether different effects may be induced upon blockade of alternative CD2 epitopes, e.g. in the region binding to LFA3. The data presented in this study had a limited sample size, meaning that it may not necessarily be representative of the average across a broader population.  Previous evidence points towards an important role of CD2 in natural NK cell cytotoxicity. NK cells of patients homozygous for a mutation of L66 in CD16 (L66H) display impaired natural cytotoxicity but normal ADCC (11,23). This seemed to result from recruitment of CD16 to the NK cell immunological synapse (NKIS) by CD2. Even though IgG antibodies are not involved in natural cytotoxicity, CD16 is recruited to the NKIS, potentially to achieve an enrichment of agonistic receptors in this region. Interestingly, NK cells expressing L66H CD16 displayed lower CD2 expression levels than NK cells expressing wild-type CD16 (11). Additionally, evidence indicates an important role of CD2 in nanotube formation between NK cells and target cells during natural cytotoxicity (24). Further, similar to T cells CD2 seems to mainly assume a peripheral positioning in the NKIS (24) and induces signaling through similar pathways as those observed in T cells (25)(26)(27). Surprisingly, no inhibition of natural cytotoxicity resulting from CD2 blockade was detected in this study. This study only used one target cell type (SPI-801) for assessment of natural cytotoxicity and future research should investigate whether CD2 blockade affects natural cytotoxicity against target cells requiring co-expression of CD2 and CD16 for induction of natural cytotoxicity like mel1106.
The relevance of observed CD69 downregulation on NK cells in MLR upon addition of FcS anti-CD2 antibodies remains unclear. While CD2 blockade could not be shown to inhibit ADCC or natural cytotoxicity, CD2-LFA3 interaction may be important for other immunological processes like crosstalk between NK cells and dendritic cells (28) or natural cytotoxicity against target cells requiring co-expression of CD16 and CD2 on effector cells for induction of a cytotoxic response.
Previous clinical trials of CD2-targeting therapies were predominantly conducted with two agents: (1) Alefacept, a fusion protein of an IgG1 Fc fragment and the CD2-binding domain of LFA3 and (2) siplizumab. Both agents are depletory, i.e. their mechanism of action is mediated by both CD2 blockade and FcgR-mediated depletion. Previous studies have detected an initial pro-inflammatory effect of Alefacept treatment in psoriasis patients (29). Recently published evidence indicates no activation effect of siplizumab on T cells in MLR (3) and this study showed that CD2 ligation by siplizumab does not have an inherent activating effect on NK cells. Thus, siplizumab may be a more desirable therapeutic agent for CD2-targeting as it does not exert an agonistic effect upon CD2 binding. Nevertheless, even though CD2 ligation by siplizumab does not induce NK cell activation, siplizumab-mediated ADCC still induces NK cell activation and thus may increase pro-inflammatory signaling by NK cells.
A number of transplantation and autoimmune conditions feature undesired NK cell activation and cytotoxicity. In solid organ transplantation, a large number of cells expressing foreign human leukocyte antigens (HLA) are introduced into the transplant recipient. Mouse models have shown that alloreactive NK cells can be a notable contributor to kidney allograft rejection (30). Further, alloreactive NK cells seemed to contribute allograft rejection in human liver transplantation (31). While this study did not show an inhibition of natural NK cell cytotoxicity by CD2 blockade, depletion of activated NK cells using siplizumab may be a viable treatment modality in conditions where NK cells and T cells are disease-mediating. Addition of siplizumab to commonly used immunosuppressants in transplantation may be desirable since standard immunosuppression inhibits NK cell alloreactivity heterogeneously (32). In autoimmune conditions like rheumatoid arthritis (RA), NK cell count in peripheral blood decreases. In contrast, an increased frequency of NK cells was observed in the synovial fluid of RA patients. Similarly, enrichment of NK cells in tissues affected by autoimmune disease has also been observed in diabetes and psoriasis patients (33,34). Notably, NK cells in synovial fluid of RA patients are predominantly CD56 bright , an NK cell subpopulation characterized by high cytokine secretion potential (35) and relatively high CD2 expression levels. Depletion of cells expressing high CD2 levels in autoimmune conditions involving NK cell migration into affected tissues may decrease NK cell tissue infiltration and related cytokine release into tissue microenvironment. Given the high cytokine secretion potential of CD56 bright NK cells, depletion of CD2 high cells may lower inflammation levels in tissues affected by autoimmune disease. In addition to ADCC, siplizumab may also induce depletion via antibody-dependent cell phagocytosis as a result of CD32 and CD64 binding. Indeed, ADCP-mediated depletion by siplizumab A B D C FIGURE 9 | Antibody-dependent cytotoxicity. Purified NK cells were pre-incubated without antibody or with saturating doses of Siplizumab/Fc-silent (FcS) anti-CD2 IgG antibodies before addition of titrated doses of Rituximab and CD20 + target cells (Daudi). Significance testing was conducted via one-way ANOVA with untreated controls (no antibody) serving as the comparison data set (N=6; ( * ) p<0.1, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). may be underestimated when considering MLR results due the absence of mature phagocytes among PBMC. Interestingly, animal models using other CD2 antibodies (36) or depletory anti-CD20 mAbs (37) have shown accumulation of depleted cells in the liver which contains an abundance of FcgR-bearing cells including hepatic phagocytes.
In conclusion, siplizumab induces NK cell activation and NK cell fratricide through FcgRIIIA-mediated induction of ADCC but not through CD2 ligation. Further, this study found no effect of CD2 blockade by siplizumab on natural cytotoxicity or ADCC induction.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
CB, FS, and FC carried out experiments and wrote the first manuscript. All authors contributed to the article and approved the submitted version.