Acute lymphoblastic leukemia (ALL), in particular B-cell precursor ALL, is the most frequent hematological malignant neoplasm occurring during childhood, with a peak incidence between 1 and 4 years of age. The standard ALL therapy lasts 24 months and therapeutic protocols comprise different chemotherapeutic and immunosuppressive agents [1]. Although often burdened by severe side effects, fist-line polychemotherapeutic approaches result in clinical remission in most pediatric patients (5-year survival rate approximately 80%) [2, 3, 4]. In contrast, ALL could be fatal in ~50–60% of children with recurrence [5]. In relapsed/refractory ALL, haematopoietic stem-cell transplantation might represent the only curative option; however, this clinical procedure requires bone marrow remission that is not easily achievable in patients who already failed to respond to pharmacological treatments [6]. A relevant ongoing approach in ALL therapeutic protocols is the introduction of new biological drugs, which should be integrated in salvage therapy to further improve rescue outcome. Among these, there is the bispecific T-cell engager (BiTE) antibody construct blinatumomab (Blincyto${}^{TM}$) [7]. Blinatumomab is a single peptide with a molecular weight of ~55 kDa and consists of two antibody single chain variable fragments (one against the surface antigen CD3 and one against CD19) joined by a flexible glycine-serine linker. Blinatumomab acts as a short adapter molecule that forces mature CD3${}^{+}$ cells (T-lymphocytes) and CD19${}^{+}$ tumor B cells into close proximity, creating a tight cell-specific synapse that induces a strong T-cell activation and proliferation. Effector cells (CD3${}^{+}$) are thus able to recognize leukemic blasts and induce their death by apoptosis, through the release of perforins and granzymes [8]. Blinatumomab has been administered with success in relapsed/refractory B-ALL adult cases. However, primary resistance has been observed in 20–30% of adult patients [6]. Recently, two international randomized clinical trials (www.clinicaltrials.gov; identification numbers: NCT02101853 and NCT02393859) clearly demonstrated the benefit of blinatumomab in terms of overall survival in pediatric patients with high-risk first relapse of B-ALL with chemotherapy-responsive disease. Blinatumomab was administered in combination with conventional chemotherapy and compared to conventional chemotherapy alone. Importantly, blinatumomab showed a favorable toxicity profile, with less life-threatening complications [9, 10].

The role of CD19 in therapy failure with blinatumomab has been hypothesized and investigated to clarify whether blinatumomab exposure exerts a potent selective pressure, resulting in a CD19 negative recurrence; controversial conclusions on this issue were reported [11, 12]. CD19 is a co-stimulatory type I transmembrane protein (~95 kDa), exclusively expressed on B-cells. It functions as a critical co-receptor of the B-Cell Antigen-Receptor (BCR) signal transduction pathway. In particular, CD19 mediates the activation of the Src family protein-tyrosine kinases, such as Lyn and Fyn [13], that enhance BCR-induced signaling through recruitment and activation of PI3K and downstream Akt kinases, thus promoting B-cells survival and cell growth [14]. The CD19 gene contains a binding site for the B-cell lineage specific activator protein (BSAP) in its promoter’s region, and the expression of CD19 is reduced in pre-B-cells deficient in BSAP [15]. BSAP is encoded by the PAX5 gene and is the only PAX protein expressed in the hematopoietic system. The protein functions both as a transcriptional activator and as a repressor on different target genes involved in B lymphoid lineage development [16, 17]. Interestingly, genomic and transcriptomic analyses in pediatric and adult B-ALL identified PAX5 as one of the most often mutated genes in leukemic cells, being involved in several leukemia-associated rearrangements, such as gene fusions (2–3% of ALL cases) or intragenic amplifications [18, 19, 20]. Moreover, several somatic point mutations at the level of the PAX5 exonic sequence, influencing the different structural domains and functions of BSAP, have been identified in over 30% of ALL cases [19, 21]. The genetic basis of interpatient variability in blinatumomab response have not been investigated so far. The aim of this study was thus to provide evidences of the hypothesis that a differential CD19 surface density on B-ALL cells, due to a heterogeneous PAX5 genetic background, could affect blinatumomab response. Therefore, co-cultures of T-lymphocytes from healthy donors and lymphoblastic B-ALL cell lines (with a different PAX5 genetic status) were maintained in the presence of blinatumomab for one week. Cytofluorimetric analyses investigating cell morphology, CD19${}^{+/}$CD3${}^{+}$ composition and viability were performed after 3 and 7 days of incubation with the drug. Similar in vitro analyses were performed on primary cells co-cultures derived from bone marrow aspirates of ALL patients.

Blinatumomab is a promising agent for both first-onset and relapsed/refractory ALL, however, primary and secondary resistance occurs, and still needs to be characterized. Current recruiting clinical trials investigate on blinatumomab effectiveness in comparison to standard therapeutic procedures (e.g., AIEOP-BFM ALL 2017 protocol, NCT03643276) and/or on the proportion of patients with blinatumomab poor-response as primary outcomes. Pharmacogenetic traits or B-lymphoblasts genetics could modulate treatment efficacy or predispose to treatment failure. Only few reports on a very limited number of patients suggest that blinatumomab could be a valid therapeutic approach for specific ALL genetic subtypes. Zhao and coworkers [27] showed that the activation of JAK-STAT signaling in leukemic cells of 44 patients harboring high-risk CRLF2 and EPOR-rearrangements is an important determinant of blinatumomab response. Mouttet and collaborators [28] reported durable remissions after blinatumomab treatment in 9 patients with a rare subtype of ALL with a dismal outcome, characterized by high and homogeneous expression of CD19 on blast cells and by the presence of the t(17;19) (q21-q22;p13), leading to the TCF3-HLF fusion gene. Moreover, Zhao and collaborators [29] observed, in 44 adults with relapsed or refractory B-ALL treated with blinatumomab, that multiple mechanisms may contribute to CD19 loss and relapse, including CD19 mutations, CD19 mutant allele-specific expression, CD19 low RNA expression and also mutations in the member of the CD19 signaling complex CD81. This underlines the importance of possible variations in CD19, including PAX5, in relation to treatment response [29]. Identifying patients’ prognostic biomarkers and/or finding a reliable predictive in vitro sensitivity assay could be crucial for an effective proactive management of blinatumomab, especially considering the high costs, the length of treatment (2 cycles of 28 days continuous infusion), its toxicity [30] and current percentages of failure (20–30% patients with relapse/refractory ALL) associated with blinatumomab therapy. Another CD19-targeted immunotherapy, patient-specific chimeric antigen receptor T cell therapy (CAR-T), was recently introduced in the clinics with encouraging and impressive response rates in the cure of relapsed/refractory ALL and other hematological diseases. The choice between CAR-T and BiTE immunotherapy depends on several clinical features that may affect efficacy, including relative disease burden, antigen expression, and T-cell function, as well as patient and disease characteristics. Adverse effects of CD19-targeted immunotherapies, such as cytokine release syndrome and neurotoxicity, still need to be reduced: these events may be more frequent and severe in patients receiving CAR-T [31, 32]. Blinatumomab has the advantage to use the same standard drug for every person treated and therapy-associated costs may be lower. Ongoing attempts in CAR-T aim to develop off-the-shelf CAR T-cell therapies that should be immediately available for use without being manufactured for each patient. First attempts used CRISPR/Cas9-engineered universal CD19/CD22 CAR-T cells to avoid host immune-mediated rejection when infused in patients, showing a manageable safety profile and prominent antileukemic activity [33].

In this study, a cytofluorimetric approach was used to measure in vitro blinatumomab effects on co-cultures of immortalized CD19${}^{+}$ B-ALL cells and primary CD3${}^{+}$ T-lymphocytes, to investigate whether the drug response could be affected by the CD19 surface expression levels determined by the PAX5 genetic status of leukemic cells. The different expression of BSAP and CD19 in the immortalized cell lines NALM6 and REH was confirmed at the protein level. Western blot analysis revealed the presence of the full-length BSAP in NALM6, whereas only a faint signal appeared in REH cells. Analysis of mRNA expression levels confirmed previous data of Best and co-workers, who showed comparable PAX5 mRNA levels between these two cell lines [26]. Thus the lower BSAP levels could be ascribable to the PAX5 frameshift (fs) somatic mutation (P321fs) harbored by REH, affecting mRNA translation or protein stability. PAX5 fs mutations occurred in ~2% of ALL patients, as demonstrated by Gu and co-workers using integrated genomic analysis of 1,988 childhood and adult cases; ~40% of the PAX5 fs mutations observed lead to truncated BSAP proteins, similar to what is observed in REH cells, thus confirming this cell line as a useful cellular model [25]. The lack of the C-terminal transactivation domain functionally impaired BSAP, and the truncated protein failed to fully-activate BSAP-dependent gene transcription [34, 35]. Indeed, a three times lower mean fluorescence intensity for CD19 was observed in REH compared to NALM6 cells, as previously reported by Haso and collaborators [36]. Indeed, these authors calculated the surface antigen density on different B-ALL cell lines, finding that the average number of CD19 molecules per cell in REH was 15,000 versus 55,000 in NALM6 [36]. Preliminary analyses were carried out to evaluate the reliability of the experimental condition used for the in vitro blinatumomab assay. The effector-to-target ratio (1:10), with exceeding B-precursor cells, was chosen to resemble the invasion of leukemic blasts in the bone marrow, a condition where a primary resistance to blinatumomab could result in a dismal outcome for relapsed/refractory CD19${}^{+}$ ALL patients [6, 29, 37, 38, 39]. Besides being administered at relapse, the role of blinatumomab in at risk patients is now under investigation (e.g., AIEOP-BFM ALL 2017 protocol). In this paper, we have investigated only the 1:10 ratio, but other effector-to target ratio should be used in further studies for a complete overview on blinatumomab in vitro effects. Blinatumomab was refreshed on a daily basis, mimicking more closely drug exposure in patients. Indeed, due to its short half-life (mean $\pm$ standard deviation: 2.11 $\pm$ 1.42 hours in adults) [40], the drug is administered by continuous infusion in cycles of 4 weeks. The blinatumomab concentration used (1 ng/mL) was in line with steady state concentration values measured in adult ALL patients during the second cycle of blinatumomab [41]. The cytofluorimetric assay allowed to appreciate both blinatumomab-induced effects, i.e., the decreased percentage of B-ALL immortalized cells combined to the T-lymphocytes increase and activation, at day +7. However, at this time point, the drug effects were already accomplished since CD19${}^{+}$ cell mortality was almost complete ($>$95%), suggesting that the ideal timing for investigating simultaneously both drug effects should fall between days +3 and +7. Interestingly, CD19${}^{+}$ cell death was already observed at day +3 in blinatumomab-treated samples, with a more pronounced effect in REH than NALM6, contrary to what could be expected considering the reduced amount of GzB released in the co-culture supernatants and the lack of CD3${}^{+}$ activation at this early time point. Several hypotheses could be postulated to explain such results. It is known that REH cells are more sensitive than NALM6 to conventional ALL chemotherapy [42], in agreement to what is observed in ETV6-RUNX1 ALL patients who have good prognostic parameters and a favorable outcome and in ETV6-PDGFRB patients who failed to reach remission [43, 44]. In this report, we showed for the first time the different response of these B-ALL cell lines to the novel immunomodulatory agent blinatumomab. One hypothesis is that the lower CD19${}^{+}$ expressing REH cells could be more sensitive to cytotoxic stimuli than NALM6 because, in the balance between cell survival and apoptotic signals, they could be more susceptible to pro-apoptotic signals induced by the drug. Indeed, cell surface dynamic analysis revealed that CD19 and BCR come together on plasma membrane microclusters, and that CD19-defective B-cells show a reduced microcluster formation and initiation of BCR-dependent signaling, thus affecting the survival pathway [45]. As a second hypothesis, it could not be ruled out that NALM6 and REH have a different GzB susceptibility, since GzB-mediated cell death is largely dependent on a pathway that is regulated by the antiapoptotic protein Bcl-2, another target gene of BSAP, which is less expressed in BSAP-deficient cells compared to wild type cells [46, 47]. In our experiments we did not assess whether NALM6 and REH stimulate T-cell subpopulations (CD8${}^{+}$ effector memory T-cells and CD4${}^{+}$ regulatory T cells (Tregs)), differently. CD8${}^{+}$ cells showed a greater and faster blinatumomab-induced cytotoxic capability than other activated cells, with specific production of cytotoxic factors and cytokines [48]. A higher percentage of Tregs were associated with unfavorable outcomes in adult B-ALL patients treated with blinatumomab [49]. This occurred likely because activated Tregs lead to IL-10 production, resulting in suppression of T-cell proliferation and in a reduction of the CD8${}^{+}$-mediated lysis of ALL cells [50, 51]. Duell and collaborators demonstrated that CD19${}^{+}$ B-ALL blasts and NALM6 cells induced Treg activation to the same degree [49]. In a recent analysis at single cell transcriptome resolution, a target cell-dependent mechanism of T-cell activation by blinatumomab, mediated by the cytokine TNFSF4, was observed [52]. Whether CD4${}^{+}$/ CD8${}^{+}$ ratio (and IL-10 or TNFSF4 levels) change between NALM-6 and REH co-cultures was not established in our experiments. Finally, upregulation of inhibitory immune checkpoints, mainly PD-L1, has been recognized as a major mechanism of resistance to BiTE therapy [53], being increased in relapsed ALL patients and after BiTE treatment in ALLs refractory to blinatumomab [54]. Future studies could consider PD-L1 surface expression in NALM6 and REH co-cultures to investigate its contribution to the different reaction to the drug. All together, these observations suggest that an in-depth analysis of other activation/inhibitory markers involved in blinatumomab in vitro susceptibility are needed.

CD19${}^{+}$ loss is one of the most significant escape mechanisms of B-ALL blasts after blinatumomab therapy [55]; however, CD19 negative relapses still represent a minority of cases, being reported only in 1/3 of patients [56, 57, 58, 59]. Blinatumomab is given to CD19${}^{+}$ B-ALL patients, but there are not studies investigating the impact of blasts CD19${}^{+}$ antigen initial load on treatment success or as determinant of CD19-positive versus CD19-negative relapses after drug exposure. In patients treated with tisagenlecleucel (a CD19-specific CAR-T), low tumor burden is a factor associated with CD19-positive relapses, in contrast to the high burden associated to CD19-negative ones. It was hypothesized that high amount of CD19 target (i.e., high tumor burden) could be involved in CAR-T- persistence, with the drawback of an increased risk of CD19 negative clones selection, able to escape CAR-T immunosurveillance [60]. For other well-established biological drugs such as trastuzumab, specifically used for HER2 receptor positive cancers, therapeutic failure could occur in the presence of high HER2 expression, and different response rates were observed among subsets of HER2-positive patients [61]. All together, these evidences suggest that the presence of the molecular target drives the choice of the biological drug to be used, but a complex and non linear correlation between the target expression levels and the effective therapeutic response takes place, particularly for biological therapies that activate immune cells [5]. Indeed, the therapeutic response is a multifactorial event that could involve several pathways.

Measurement of the in vitro effects of blinatumomab on primary mononuclear cells without purification of CD19${}^{+}$ or CD3${}^{+}$ cells were performed to assess the assay feasibility on biological material derived from patients’ bone marrows. Only a single time point measurement was possible due to the scarce number of alive primary cells recovered after diagnostic procedures and mononuclear cells isolation; moreover, survival of untreated cells was compromised over the 7 days of incubation in ~40% of the cases. Besides these technical limitations, the assay showed a variability in the blinatumomab in vitro response, and identified co-cultures of patients’ primary cells non-responsive to the BiTE antibody. The blinatumomab in vitro cytofluorimetric assay could be used to screen the in vivo patients’ drug response, if its predictive power will be proven; however, this purpose goes beyond the aim of this paper. Because none of the patients of this study underwent blinatumomab therapy, a preliminary correlation analysis between cytofluorimetric in vitro results and in vivo pharmacological response could not be performed.

Results presented in this study are preliminary and far from being exhaustive. However, the in vitro cytofluorimetric assay here proposed is suitable for studying the blinatumomab response in CD3${}^{+}$:CD19${}^{+}$ cell co-cultures, simulating the condition of lymphoblasts invasion in the bone marrow. Nonetheless, optimization of the assay is still required, particularly to find out the ideal time of observation. Screening of a larger panel of B-ALL cell line models (particularly focusing on those with hyper-activated JAK/STAT pathway and/or t(17;19) rearrangement) or engineered cells to modulate PAX5 and/or CD19 expression, would help to better understand the hypothesis-driven question relative to PAX5 or other B-cells genetic abnormalities contribution in drug response [62, 63]. The in vitro assay here proposed could represent a model for better understanding the complex biology of blinatumomab response.

ALL, Acute lymphoblastic leukemia; BiTE, bispecific T-cell engager; BSAP, B-cell lineage specific activator protein; GzB, granzyme B; IL-2, recombinant human interleukin-2; PBMC, peripheral blood mononuclear cells; PBS, phosphate-buffered saline; Treg, regulatory T cell.

Conceptualization—RF, MG, GS and GD; Data curation—SB, MG and RF; Formal analysis—SB and RF; Funding acquisition—GD; Investigation—SB, MG, EP; Writing – original draft—SB and RF; Writing – review & editing—MR, AT, GD and GS.

The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of I.R.C.C.S. Burlo Garofolo (Protocol number CE/V 135; March 5th 2012). Informed consent was obtained from all subjects involved in the study.

The authors gratefully acknowledge the support of Institute for Maternal & Child Health (I.R.C.C.S) “Burlo Garofolo”, Trieste.

This work was supported by the Institute for Maternal and Child Health (I.R.C.C.S) “Burlo Garofolo,” Trieste, Italy (grant number RC 05/2012).

The authors declare no conflict of interest.