Background
Interest in the development of antibodies and antibody-derived agents such as antibody-drug conjugates (ADCs) and bispecific agents for the treatment of acute myeloid leukemia (AML) has increased over the past years. A prototypical agent, gemtuzumab ozogamycin (GO; Mylotarg
TM), has been effective in the treatment of AML, but was withdrawn from the market because of side toxicities [
1,
2]. The problem causing this withdrawal was not related to the suitability of the myeloid surface antigen CD33 as a target, but rather to the chemical linker connecting the toxin component of this drug to the antibody carrier. CD33 remains a valuable and clinically validated target for the therapy of CD33-positive subtypes of AML with antibody-derived agents [
3‐
5], and consequently, several major drug developers currently study new agents for the treatment of AML with specificity for CD33. One class of such agents relies on potent new toxins coupled with improved linkers to the antibody, such as the agent SGN-CD33a [
6]. Another class recruits effector cells including NKs or T cells for the elimination of AML cells. This class includes the agent AMG-330 [
7] recruiting T cells as effectors, as well as our team’s single chain triplebodies 123-16-33 [
8] and 33-16-33 [
9], recruiting NKs. It further includes CD33-specific antibodies with Fc domains engineered for improved binding to Fc-receptors on effector cells. Also, antibodies specific for CD123, the alpha subunit of the receptor for Interleukin-3, an important growth and differentiation factor for early hematopoietic cells and the myeloid lineage, have produced promising results in preclinical studies [
10]. Antibodies against CD123 have mediated cytolysis of AML cells and in particular of AML-Leukemia Stem Cells (AML-LSCs) [
11,
12].
Our team has developed a new class of antibody-derived fusion proteins called triplebodies, suited for the elimination of AML cells [
8,
9,
13‐
17]. These agents carry two binding sites for surface antigens on the cancer cell and one for a trigger molecule on an effector cell. They can bind two different tumor antigens on the same target cell and through this mode of “dual targeting” address the cancer cell with enhanced selectivity [
15,
16,
18]. In addition, they bind a trigger molecule on an effector cell, such as CD16 on NKs, and thereby activate the effector for cytolytic elimination of the cancer cell. Triplebody SPM-2 employed in the present study carries single chain Fv (scFv) antibody fragments specific for CD33 and CD123 on AML cells linked in a single polypeptide chain to an scFv module specific for CD16, the low affinity Fc gamma RIII receptor on NKs and macrophages and a few other cells of the hematopoietic system [
8]. SPM-2 was designed for the elimination of both bulk AML cells and AML-LSCs, because both are double-positive for this pair of antigens.
LSCs from patients with most subtypes of AML however carry a higher combined cell surface density of this pair of antigens than bulk AML cells, normal hematopoietic cells, and healthy hematopoietic stem cells (HSCs) [
3,
11,
19,
20]. SPM-2 was therefore designed for a preferential elimination of AML-LSCs
in vivo. It is still unknown whether this objective can be reached because the agent has still not been tested in humans. It is important to eliminate not only bulk AML cells, but also the LSCs, because the LSCs are believed to be the relevant subset within the population of minimal residual disease (MRD) cells, which are responsible for relapse and poor disease outcome [
4,
21‐
23]. New agents intended to induce a deeper remission and to prolong the time to relapse must therefore make a deliberate effort to target the MRD cells. Currently available chemotherapeutic agents achieving remission for many AML patients probably also act (at least in part) through the elimination of some LSCs. However, most often they do not achieve long-lasting complete remissions and have not been designed with the intent to specifically eliminate MRD cells but rather bulk AML cells. A well-defined immunophenotype of MRD cells is not available, which would lend itself as a specific address for the design of antibody-derived agents. However, in a retrospective clinical study the MRD cell compartment responsible for early relapse was tentatively equated with the CD34
+ CD38
- CD123
+ subset of AML cells [
24]. If these results can be confirmed by prospective studies, then triplebody SPM-2 could be especially promising for the therapeutic removal of AML-MRD cells. Its ability to reach this objective critically hinges upon the availability of sufficient numbers of active NKs in AML patients at the time of the intended use of the agent.
The compartment of functional NKs is often reduced in AML patients [
25‐
29]. This impairment has been attributed to different causes, among them a down-regulated cell surface expression of the activating natural cytotoxicity receptors (NCRs) NKp30, NKp44 and NKp46 [
25‐
27,
29,
30]. Low-level expression of NCRs (NCR
dim) on patient-derived NKs was correlated with poor prognosis in AML, as patients with NCR
dim NKs had significantly lower 5-year survival rates than matched patients with NCR
bright NKs [
26]. Deficiencies in cytokine release have also been linked to an impairment of NK cell function in AML patients. The capacity of NKs to secrete IFN-gamma was highly impaired in AML patients and was correlated with suppressed immune responses against autologous leukemic cells [
31‐
33]. Finally, in adult acute leukemia, impaired production of cytokines by NKs was associated with early relapse [
34].
Triplebody SPM-2 was not designed as a frontline therapeutic for the initial debulking of AML blasts. This objective is reached in most cases by the initial chemotherapy. Rather, SPM-2 was designed to be used after chemotherapy when the blast titer is greatly reduced for most patients and when the titer of functional NKs is expected to have at least partially recovered towards normal levels [
28]. Therefore, here we studied the effectiveness of SPM-2 to mediate lysis of an AML patient’s BM blasts as targets, in combination with autologous NKs drawn from peripheral blood at diagnosis and in remission. The analysis was performed in comparison with NKs isolated from the patient’s healthy monozygotic twin. This exceptional situation provided us with a unique opportunity to compare the activity of the patient’s NKs with NKs genetically as closely identical as possible from a healthy donor.
Here the patients NKs, drawn in remission, were restored to titers in the blood comparable to those of the healthy twin and were able to achieve effective lysis of autologous AML blasts in conjunction with SPM-2. Notably, they also achieved comparably effective lysis of an established lymphoma-derived target cell line mediated by the reference antibody RituximabTM, and had similar expression profiles of NCRs as the healthy sibling’s cells. The present study provides a precedent case in support of our hypothesis that SPM-2, administered at appropriate stages in the course of a standard therapy for AML, will become useful as an adjuvant drug.
Methods
Expression in mammalian cells, purification and protein-chemical characterization of triplebody SPM-2
The DNA construct coding for SPM-2 was based on the published triplebody 123-16-33 [
8] and was synthesized by a commercial provider (Eurofins/MWG-Operon, Ebersberg; Germany). The CD16-specific scFv was stabilized following published procedures [
8,
35] and humanized. Humanization of the CD123- and CD33-specific scFvs and the stabilization of the CD123-specific scFv followed similar procedures (S. Wildenhain, I. Schubert, A. Honegger et. al., unpublished data). The N- and C-terminal CD33- and CD123-specific scFvs were separated from the central CD16-specific scFv by (G
4S)x4 linkers. For expression purposes, human 293 F cells (DSMZ; German Collection of Microorganisms and Cell Lines; Braunschweig, Germany) were stably transfected with plasmid DNA using the 293fectin
TM transfection reagent (Life Technologies, Darmstadt, Germany) according to the manufacturer’s instructions in a total volume of 30 ml. The cells were then cultured under continuous selection with hygromycin C. The recombinant protein was captured from culture supernatants via its C-terminal hexahistidine tag by retention on a metal-ion affinity matrix and purified by anion- and cation-exchange chromatography. Concentrations of the purified protein were determined by spectrophotometry and calculated using the molar extinction coefficient derived from the amino acid sequence. The resulting final protein meets current regulatory standards and industry norms, and was named SPM-2 to indicate its status as a candidate for clinical development. Equilibrium binding constants (K
D) of each of the individual scFvs of SPM-2 were in the 20–30 nM range, and thus were similar to the values reported for the initial agent [
8].
Preparation of primary cells from blood and bone marrow of human donors
Peripheral blood and bone marrow samples were drawn from subjects into EDTA solution after receiving informed consent. The project was approved by the Ethics Committee of the University of Munich Medical Center. Bone marrow mononuclear cells (BMMC) containing the leukemic blasts and Peripheral Blood Mononuclear Cells (PBMCs) were enriched by ficoll density centrifugation using the Lymphoflot reagent (Biotest, Dreieich, Germany) according to manufacturer’s instructions. Isolated cells were then either suspended in RPMI medium (Life Technologies) containing 10% fetal bovine serum (FBS) with penicillin and streptomycin (PS) at 100 units/ml and 100 μg/ml, respectively, for immediate use, or stored frozen in a solution containing 90% FBS and 10% DMSO for future use. Cell viability was assessed by trypan blue exclusion.
Ex vivo expansion of MNCs from healthy donors in the presence of IL-2
PBMCs were expanded
ex vivo in RPMI medium containing Interleukin-2 (IL-2) plus 5% human serum (Life Technologies) for 20 days as described [
16,
36], and were then frozen in aliquots for subsequent use. Prior to use in cytolysis experiments, the cells were thawed and cultured overnight in RPMI medium containing 5% human serum plus 50 units/ml and 50 μg/ml PS, respectively, but no additional IL-2
.
Flow cytometric analysis
Flow cytometric analysis was performed with an Accuri C6 flow cytometer (BD Biosciences, Heidelberg, Germany). The CD3-, CD16-, and CD56-specific monoclonal antibodies (mAbs) used for the analysis of NK cell content in PBMC-preparations as well as isotype control mAbs were from Immunotech (Marseille, France), while the NKp30-, NKp44-, NKp46-specific and isotype control mAbs used for the analysis of NK cell receptors (NCRs) [
37] on isolated NKs were from eBioscience (Frankfurt, Germany). Cell surface densities of CD33 and CD123 were measured using a calibrated cytofluorimetric assay as described [
8,
35]. For this purpose, a commercial kit of fluorescent beads with known numbers of fluorescent chromophores per bead (QIFIKIT®; DAKO; Hamburg, Germany) was employed, as well as fluorescent-labeled mAbs. This procedure allows the investigator to express the measured fluorescent intensity of mAbs bound to the surface of the patient’s cells in terms of average number of antigen copies per cell [
38].
Antibody Dependent Cellular Cytotoxicity (ADCC) and Redirected Lysis (RDL) assays using Calcein release
In this study we refer to cell-mediated cytolysis assays with whole antibodies as “ADCC” and to tests with antibody-derived agents such as triplebodies as “redirected lysis (RDL)” assays. Non-radioactive cytolysis assays based on the release of calcein from target cells pre-labeled with calcein AM (Life Technologies) were performed as described [
16,
39]. The cytolytic activity of NKs from various sources was calibrated in standard ADCC assays with the commercial CD20-mAb Rituximab
TM[
39,
40] as the mediator of lysis and lymphoma-derived Raji cells [
41] as targets. This calibration allowed us to assess the cytolytic activity of NKs from various sources using a standard mAb and a commonly used target cell line, and thus to make our results comparable to the current benchmarks in the field. For the calibration reaction with Rituximab
TM, untouched NKs were first enriched by the MACS kit (Miltenyi kit; see above) from PBMC samples and then used at a constant effector-to-target (E : T) ratio of 2.5 : 1 against Raji targets. The same NKs were also used in redirected lysis experiments with the patient’s autologous bone marrow AML blasts as targets in conjunction with SPM-2. Specific lysis was measured by quantitating the release of calcein from target cells using a fluorimeter/ELISA plate reader and expressed in relative light units (RLU) at 485/535 nm. Calcein release was measured at 3 and 4 hour time points for ADCC and RDL experiments, respectively. Specific cellular cytotoxicity was expressed as overall lysis minus the background of spontaneous lysis mediated by the NKs alone, in the absence of added antibody-reagents. Specific lysis was evaluated by the formula:
% specific lysis = 100 * (Experimental Release RLU – Background Release RLU)/(Maximal Release RLU – Background RLU).
Enrichment of human NK cells by preparative sorting with immunomagnetic beads
NKs were enriched by negative selection using a commercial NK cell isolation kit (Miltenyi Biotec MACS sorting kit, Bergisch Gladbach, Germany) according to manufacturer’s instructions. The enriched cells are referred to as “untouched” NKs, because as a result of the negative selection, no residual mAb is bound to their surface and they have not been eluted from an affinity matrix with harsh reagents. Starting material were ficoll-hypaque purified PBMCs from blood samples of the donors. The enriched NKs were then either resuspended in RPMI medium containing 10% FBS with PS at 100 units/ml and 100 μg/ml, respectively, for use in ADCC and RDL assays, or placed in PBS solution containing 1% bovine serum albumin (BSA) for flow cytometric analysis.
Measurement of IFN-gamma and TNF-alpha release into peripheral blood samples by ELISA assays
Concentrations of human cytokines IFN-gamma and TNF-alpha were measured with ELISA kits (eBioscience) following manufacturer’s instructions. Triplebody SPM-2 was added to peripheral blood samples at concentrations of 10, 1, or 0.1 nM , and the reaction mixtures were then cultured for 6 h at 37°C in 96 well round bottom Nunc plates in 200 ul volumes. Blood samples were frozen and stored at -20°C and were thawed only immediately before use in cytokine assays. In addition, cytokine release was studied with the same ELISA kits in supernatants from the 4 h RDL and 3 h ADCC assays, in which MACS-purified NKs had been used as effectors against autologous BM blasts as targets.
Statistics
All statistical analysis was performed by STATVIEW 4.5 programs from Abacus Concepts (Berkeley, CA) using Student’s t-test for the final determination of significance (p < 0.05).
Discussion
The most important result of this study is that it helped to dispel concerns raised by critics about the use of triplebodies for the treatment of AML, because they rely on NK cells as effectors. The main argument was that too few NK cells were present in an AML patient for triplebodies to be effective. The critics held that it would be preferable to recruit T cells rather than NK cells as effectors, because T cells were present in greater numbers, because they have higher intrinsic cytolytic activity, are capable of serial lysis of several target cells in a consecutive manner, and can be stimulated to proliferate as a result of exposure to the agent. We cannot address all of these points here in detail, and some of our arguments regarding the relative merits of recruiting NK cells vs. T cells have been reviewed elsewhere [
17]. Suffice it to state that NKs are also capable of serial lysis and that their numbers can be increased in patients after treatment with antibody-derived agents recruiting NKs as effectors [
28,
47]. Therefore, after clinical administration of agents related to SPM-2 such as the tandem diabody AFM-13 specific for CD30 and CD16, NKs can expand in human recipients
in vivo, and sufficient numbers can become available for therapeutic effects, even for malignancies such as Hodgkin Lymphoma that include semi-solid tumor masses [
47]. Our results presented above (Figure
5B) suggest that for the patient of the present study, CD3-positive T cells were also greatly reduced in peripheral blood at diagnosis and only recovered together with other leukocytes in the remission phase. Finally, it is not known how many T cells and how many NKs are needed for an effective therapy with such agents. The absolute numbers of NK cells in remission (Figure
4B) may be lower than those of T cells, although absolute numbers have not been determined, but even lower numbers of NKs may be sufficient for therapeutic success.
In the present study, the patient’s NK cells were greatly reduced in relative abundance. At diagnosis their frequency was at least 6–7 fold reduced compared to the two reference samples obtained from the patient in remission and the healthy twin. In the remission sample, normal numbers of NKs with typical cytolytic activity were found. This observation is important for us because SPM-2 was not designed to debulk the mass of leukemic blasts as a frontline agent. Instead, it was designed to act as an adjuvant to further reduce the MRD pool after an initial induction chemotherapy and thus to achieve deeper and longer-lasting remissions.
In this context, the observation that not only the total population of AML blasts from this patient was efficiently lysed by SPM-2 together with MNC effectors (Figure
2B), but that the CD34
pos subset was also specifically reduced (Figure
2C), is encouraging. This finding opens the possibility that the AML-LSCs contained within the CD34 compartment were also affected by this treatment. If this were true more generally beyond this single patient, then this would be a welcome result, because AML-LSCs are reported to have increased resistance to treatment with standard chemotherapeutic agents [
19,
48]. AML-LSCs are typically contained in the CD34
pos compartment, and if the LSCs of this patient were similarly sensitive towards cytolysis mediated by SPM-2 in conjunction with NKs as the overall population of CD34
pos cells, then this would suggest, that SPM-2 may become a useful new agent for the removal of LSCs. However, these extrapolations are made with due caution, because the AML-LSCs may only account for a minor subset of all CD34
pos blasts in this patient and therefore may have escaped lysis in our experiments.
We were further encouraged by our preliminary finding reported here that SPM-2 mediated specific lysis of the CD34
posCD38
neg CD123
pos subset contained within the patient AML blasts, which presumably more narrowly confines the relevant MRD cells than the broader CD34
pos compartment. Admittedly, this was only a single initial experiment and further confirmation is needed. Yet, the results obtained so far suggest that SPM-2 in conjunction with functional NKs was capable of effectively eliminating the CD34
posCD38
neg CD123
pos subset, reported by others [
24] to include the MRD cells.
No major changes were observed here in the expression profiles of NCRs on NKs between the samples from the patient at diagnosis and in remission, and both profiles were similar to those observed for the healthy sibling. This result was somewhat surprising, because it had been reported by others that differences in NCR expression profiles could be a major cause for impaired functional activity of NKs from AML patients [
25‐
27,
29]. One possible explanation for our result is that the expression profile of this particular patient is unique and differs from those of the majority of patients reported elsewhere. Another possible explanation is that in the cases published by others, NCR expression levels were generally compared between NKs from AML patients and those from unrelated healthy donors. This comparison may be less informative than the more rigorous comparison reported here, because it is not excluded that patients in the published studies expressed lower intrinsic NCR levels than the average healthy donor, and that this reduced expression may even have been a cofactor predisposing them to the development of the disease.
Competing interest
The authors declare that they have no competing interests. UJ and GHF are employees of SpectraMab, Munich, Germany.
Authors’ contributions
TAB participated in the design of the study, performed experiments, analyzed data and helped write the manuscript. SW participated in the design of the study, performed experiments, analyzed data and helped in editing the manuscript. CCR performed experiments, analyzed data and helped in editing the manuscript. IAS contributed to SPM-2 design and performed functional studies. KPH and GHF participated in the design of the study, secured extramural funding and helped in editing the manuscript. FSO managed and coordinated the process of gathering patients, patient material and patient data, helped to secure funding, to plan the study design and to edit the manuscript. All authors read and approved the final manuscript.