Key points

  • DPP4 truncated granulocyte–macrophage colony-stimulating factor (T-GM-CSF) and interleukin-3 (T-IL-3) reciprocally bind with higher affinity to both GM-CSF and IL-3 receptors compared with their full-length (FL) forms.

  • DPP4 T-GM-CSF and -IL-3 reciprocally blunt the activities of both FL-GM-CSF and -IL-3 in vitro and in vivo in primary acute myeloid leukemia patient samples and mouse models of normal/malignant hematopoiesis.

Introduction

Dipeptidylpeptidase 4 (DPP4/CD26), a serine protease found as membrane-bound and soluble forms, enzymatically cleaves select penultimate amino acids of proteins that regulate multiple aspects of hematopoiesis.1, 2 We reported that DPP4 truncated granulocyte–macrophage colony-stimulating factor (T-GM-CSF) has diminished colony-stimulating activity, and intracellular signaling compared with full-length (FL)-GM-CSF, and blunted the in vitro colony-forming effects of FL-GM-CSF.3 Although a large number of cytokines, chemokines and growth factors (including GM-CSF and IL-3) have previously unrecognized putative DPP4 truncation sites,1, 2, 3 their potential interactive roles in hematopoiesis (including modifying the function of other molecules and acting as regulatory molecules) are poorly understood. Our objective was to clarify the regulatory effects of DPP4 T-GM-CSF and -IL-3 on in vivo and in vitro modulation of steady- and diseased-state hematopoiesis.

Hematopoietic progenitor (HPCs)/precursor cells from patients with Acute Myeloid Leukemia (AML) respond to cytokines such as GM-CSF and IL-3 for enhanced growth, survival and resistance to therapy.4, 5, 6, 7, 8 GM-CSF and IL-3 have additive to synergistic proliferative effects on normal and malignant cell growth,9, 10, 11, 12, 13, 14 and 15 and patients with AML manifest increased serum GM-CSF and IL-3 before induction therapy compared with normal controls, and a decline of GM-CSF and IL-3 after successful remission.16 GM-CSF and IL-3 share a common receptor β-chain, IL-3 competes for GM-CSF binding on AML blasts,10 and targeting the IL-3 receptor has been suggested as a useful therapeutic in AML.16, 17 Therefore, it is important to understand how T-GM-CSF and -IL-3 may influence normal and malignant hematopoiesis. We now report that T-GM-CSF and -IL-3 reciprocally blunt receptor binding, as well as functional interactions, of both FL-GM-CSF and -IL-3 in vitro in primary human cord blood (CB) HPCs, primary AML HPCs/precursors, and in the human growth factor-dependent TF-1 cell line. Moreover, these blunting effects of T-GM-CSF and -IL-3 were recapitulated in vivo in murine models of normal and malignant hematopoiesis, thus demonstrating an additional, and perhaps crucial, layer of cell regulation for normal and leukemic hematopoiesis.

Materials and methods

Mice, mouse cells, human CB, TF-1 and primary patient AML cells

C57BL6/J mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA) and treated with Diprotin A (DPA; Peptides International, Louisville, KY, USA) or phosphate-buffered saline control as reported.3 dpp4−/−,18 FMS-like tyrosine kinase 3 gene-internal tandem duplication (FLT3-ITD)19 and PtpnE76K/+LysM-Cre+ (Ptpn11E76K)20 mice are all on a C57BL6/J mouse strain background, and have been described previously. Mouse bone marrow (mBM) and human CB (from the Eskenazi (formerly Wishard) Hospital, Indiana University School of Medicine, Indianapolis, IN, USA, and from Cord:Use Cord Blood Bank, Orlando, FL, USA) were used as described.21 AML patient samples were obtained from peripheral blood samples of patients under an approved institutional review board (9402-10 and 9812-11).

The Indiana University Committee on Use and Care of Animals, and the Indiana University Institutional Review Board, respectively, approved mBM, human CB and AML studies. CB received from Eskenazi (formerly Wishard) Hospital and Cord:Use was to be discarded material. TF-1 cells, originally obtained from T Kitamura, Japan, are available from American Type Culture Collection (CRL-2003) and were used as described.3, 22, 23

HPC colony assay

C57BL6/J, dpp4−/−, FLT3/ITD and Ptpn11E76K mice were injected subcutaneously with combinations of 10 μg of recombinant murine (rm) FL- or T-GM-CSF and/or -IL-3. After 24 h, mice were killed, femurs flushed and mBM cells plated at ~5 × 104 cells per ml in 1% methylcellulose culture medium in the presence of hemin 0.1 mM, 30% fetal bovine serum (Hyclone, Logan, UT, USA) with the following growth factors, unless otherwise noted: 1 U/ml recombinant human (rh) erythropoietin (Amgen Corporation, South San Francisco, CA, USA), 50 ng/ml rm stem cell factor (R&D Systems, Minneapolis, MN, USA) and 5% (vol/vol) pokeweed mitogen mouse spleen cell-conditioned medium.18, 21 Percent HPCs in the S phase of the cell cycle was estimated by high specific activity tritiated thymidine kill technique. Colonies were scored after a 7-day incubation, and colony-forming unit-granulocyte–macrophage, burst-forming unit-erythroid and colony-forming unit-granulocyte, erythrocyte, monocyte/macrophage, megakaryocyte progenitors distinguished.21 Human CB cells and primary AML patient samples were separated into a low-density fraction and plated at ~5 × 104 cells per ml, and TF-1 cells were plated at 250–1000 cells per ml in 0.3% semisolid agar medium (Difco/BD, Franklin Lakes, NJ, USA) with 10 ng/ml (or concentrations listed) of rhFL- or T-GM-CSF or -IL-3. Colony formation for CB was counted on day 14, and colony formation for TF-1, or cluster formation of AML patient samples,24, 25, 26 at days 7–10. CD34+ CB cells (>95% CD34+) were purified.27 RmGM-CSF and rhGM-CSF, IL-3 and stem cell factor were from R&D Systems. Cell cultures were incubated at lowered (5%) O2 tension in 5% CO2 in a humidified chamber.3, 21, 27

DPP4 activity assay

CB, AML patient samples, TF-1 or mBM were harvested and enriched (CD34+) or depleted (Lineage−) using Miltenyi (San Diego, CA, USA) commercially available kits.3, 28 Approximately 1–5 × 105 cells (untreated or incubated with 5 mM DPA for 30 min) were plated/well in duplicate (N=at least three individual samples or mice), and left intact in 50 μl of phosphate-buffered saline in a 96-well flat-bottom microtiter plate. The assay was carried out by combining cells with 50 μl of the DPP4 substrate Gly-Pro-aminoluciferin and assay buffer optimized for this assay (DPP4-Glo Protease Assay; Promega, Madison, WI, USA). Plates were incubated at 37 °C for 30 min, and surface DPP4 enzyme activity was measured using an LMAX luminometer (Molecular Devices, Sunnyvale, CA, USA).3

Flow cytometry

Human TF-1 cells were fixed with 1% formaldehyde after stimulation (with 10 ng/ml rhFL, T or FL:T IL-3), permeabilized using BD Phosflow Perm Buffer III (No. 558050, BD) and stained in BD staining buffer with primary anti-P-Stat5 (BD) and anti-P-Jak2 antibodies (Cell Signaling, Danvers, MA, USA) and secondary antibody (Cell Signaling). mBM was washed in phosphate-buffered saline and 2 × 106 cells were stained with the following phenotyping antibodies to determine hematopoietic stem cell and progenitor populations29 before fixation (Lineage cocktail (BioLegend), CD34, SCA, KIT, FLT3, Fcγ receptor (BD). All samples were run on BD machines (LSR 4 or Fortessa) and analyzed using the FlowJo Software (FLOWJO, LLC, Ashland, OR, USA).

Generation of DPP4 T-GM-CSF and -IL-3

Soluble human or porcine DPP4, prepared from human placental tissue or porcine kidney, were purchased from MP Biomedicals, LLC or Sigma-Aldrich (No. D7052), respectively, and used at ~0.25 μg for every 1 μg of FL protein per digestion at 37°C for at least 18 h.3

Equilibrium receptor binding

Receptor-binding analysis of FL- and T-rhGM-CSF and -IL-3 were carried out under equilibrating conditions, and data were analyzed by Scatchard plotting3, 30, 31 using TF-1 cells, CD34+ CB cells and AML primary patient samples. Carrier-free rhGM-CSF and IL-3 (R&D Systems, Minneapolis, MN, USA) were radio-iodinated by chloramine-T method by Phoenix Pharmaceuticals Inc. (Burlingame, CA, USA). The [125I]GM-CSF and IL-3 were repurified to a concentration of ~0.67 μg/μl and had a specific radioactivity of ~60 μCi/mmol. The chloramine-T method iodinates on tyrosine residues in proteins; therefore, truncation by DPP4 does not reduce the specific radioactivity of the protein by the enzyme’s removal of the N-terminal Ala-Pro. Nonspecific binding was assessed by competing rh[125I]GM-CSF or IL-3 with a 1000-fold excess of unlabeled rhGM-CSF or IL-3. Using TF-1 cells, this results in >95% reduction in non-competed counts per minute of bound ligand, which was near background radiation, thus specific binding under these conditions was considered to be 100% of measured 125I, as determined by Beckman Coulter Gamma 5500B gamma counter (Brea, CA, USA). Linear-regression coefficient of correlation (r2) was considered acceptable at a value of 0.90 or greater.

Statistical analyses

For all colony assays, three plates per experimental point were scored. Results of colony assays and animal studies were assessed by two-tailed Student's T-test or analysis of variance. Receptor-binding studies were assessed by Student’s T-test. Numbers of experiments per group are noted in the figure legends, and all animal experiments analyzing effects in vivo contained at least three mice per group.

Results

Interactions of FL- and T-GM-CSF and -IL-3 on cellular proliferation in vitro and in vivo

DPP4 is active on the cell surface of immature cells

DPP4 has been shown to be present as an active, membrane-bound form on specific cell types. We detected active DPP4 on the surface of the human growth factor-dependent TF-1 cell line, human CD34+ CB cells, primary AML patient samples and Lineage− mBM, and this activity was blunted by DPA, an ILE-PRO-ILE DPP4 inhibitor (Figure 1a), on all samples evaluated.3, 18, 28

Figure 1
figure 1

DPP4 is active on the cell surface of immature blood cells and DPP4 truncated molecules (T-GM-CSF and -IL-3) block colony formation across molecules in vitro at less than a 1:1 ratio. (a) A total of 1–5 × 105 cells per well from TF-1, primary AML samples, CD34+ CB and Lineage− C57BL6/J mBM were analyzed for baseline DPP4 activity. DPA, a DPP4 inhibitor, was added as a control 30 min before the start of the assay (for some samples) to show assay specificity for DPP4 activity on the cell surface and ability to inhibit the DPP4 enzyme. N=3 samples per group was carried out in triplicate for all except for the primary AML samples where N=5 and plating for data shown was lower than other samples (~1 × 105/well) due to sample availability. AML samples were from those shown in Table 1. Low-density CB cells (b, c and e) or TF-1 cells (d) were stimulated with 10 ng/ml of FL or DPP4 T-GM-CSF, IL-3 or G-CSF, unless otherwise listed. (f) TF-1 cells were factor starved and treated with 10 ng/ml of FL, T or 1:1 ratio of IL-3 for 5 min, fixed and assessed for induction of JAK/STAT signaling by flow cytometry (n=2 independent experiments). (g) Mouse BM was plated at 5 × 104 cells per ml in the presence of 10 ng/ml rmFL, T or a mixture of FL/T-GM-CSF and/or IL-3 and colony formation and cycling status of the progenitor cells were determined. All experiments contained three mice per group or were performed at least three times (unless noted otherwise) and analysis of variance (ANOVA) was used to determine statistical significance with the exception of 1A (Student's T-test). *P⩽0.05 compared with control (FL-GM, IL-3 or FL-GM/IL-3, FL-G-CSF and so on). For the combined experiments shown as fold change, the average number of control colonies were: FL-GM=45, IL-3=55, FL-GM/IL-3=95 for LDCB and FL-GM=220, IL-3=170, FL-GM/IL-3=310 for TF-1.

Full size image

T-GM-CSF and -IL-3 blunt colony-stimulating effects of both FL-GM-CSF or -IL-3 in vitro on human (CB and TF-1) and mouse cells

To delineate the ability of the FL- and T-GM-CSF and -IL-3 molecules to reciprocally modulate each other’s function, as well as to better understand the practical relevance of their functions, we analyzed the colony-stimulating activity of these molecules using primary human low-density CB HPCs (Figures 1b and c). This reciprocal interaction was specific for GM-CSF and IL-3, and likely involved the common β-chain they share in their receptors, as DPP4 T-G-CSF (whose receptor does not share the common β-chain) was only able to blunt colony formation induced by FL-G-CSF, but not that of FL-GM-CSF or -IL-3 (Figure 1c). Importantly, the T-factors could be serially diluted to ~12% of the FL amount and still significantly blunt stimulation by either FL-GM-CSF or -IL-3 in both TF-1 (Figure 1d) and CB (Figure 1e) cells. Further, as previously demonstrated for GM-CSF,3 T-IL-3 induced less activation of the JAK/STAT pathway than FL-IL-3 and blunted the activation induced by FL-IL-3 (Figure 1f). T-GM-CSF and -IL-3 alone or in combination blunted numbers of mBM colonies stimulated by their FL forms and manifested decreased percentages of HPCs in the S phase of the cell cycle (Figure 1g).

Effects of T- and FL-CSFs on HPCs in vivo mimic in vitro effects

To evaluate if the functional data above would be replicated in an in vivo situation, we studied effects of mFL- and T-GM-CSF and -IL-3 for effects on the absolute numbers and cycling status of mBM HPCs. We used dpp4−/− mice on a C57BL/6 strain background (Figure 2a), C57BL/6 mice that were pretreated with DPA (Figure 2b), or C57BL/6 mice left untreated (Figure 2c). Mice were administered a single subcutaneous dose of rmFL- or T-GM-CSF or -IL-3, or various combinations of FL-, and T-, GM-CSF or -IL-3. FL- vs T-CSFs were injected separately at different sites in the same mouse. In all three animal models, T-GM-CSF and T-IL-3 resulted in less in vivo stimulatory activity than their respective FL-CSFs with respect to HPC numbers (colony-forming unit-granulocyte–macrophage, colony-forming unit-granulocyte, erythrocyte, monocyte/macrophage, megakaryocyte and burst-forming unit-erythroid)/femur and their cell cycling, as assessed by percent of HPCs in the S phase of the cell cycle (Figure 2). Moreover, T-GM-CSF and T-IL-3 blunted effects of both FL-GM-CSF and -IL-3 similar to that seen in vitro. T-GM-CSF or T-IL-3 each resulted in in vivo suppression of the effects of both FL-GM-CSF and -IL-3 (Figure 2). Most importantly, and impressively, combinations of T-CSFs were able to blunt the increased additive potency of combinations of FL-GM-CSF plus -IL-3. This functional blunting, with respect to HPC numbers/femur and cycling in vivo, demonstrates that both T-GM-CSF and T-IL-3 functionally display downmodulating effects on HPC over both FL factors.

Figure 2
figure 2

T-GM-CSF and -IL-3 block colony formation across molecules in vivo. dpp4−/− mice (a), C57BL6/J mice treated with DPA (b) or untreated C57BL6/J mice (c) were injected with 10 μg of either rmFL, T or a combination of FL/T-GM-CSF and/or IL-3 subcutaneously at different sites. Mice were killed 24 h later, femurs were flushed, and colony formation and cycling status of the progenitor cells were determined. *P⩽0.05 or less by analysis of variance (ANOVA). N=3 animals per group with each mouse individually accessed. One of two representative experiments shown for dpp4−/− mice.

Full size image

Primary progenitor/precursor cells from patients with AML respond in vitro to DPP4 T-rhGM-CSF and -IL-3, in a manner similar to normal progenitors

GM-CSF and IL-3 are important in growth and maintenance of AML cells.6, 10, 32, 33, 34, 35 Modifications in cytogenetic and molecular status, such as ITD mutations in the FLT3/ITD, are common and can alter growth and aggressive progression of disease. Therefore, to understand if HPCs/precursors from patients with AML, with various cytogenetic backgrounds and molecular mutations (Table 1), are sensitive to T-factors in a manner similar to that of normal HPC, we assessed AML patient samples with varied cytogenetic and molecular alterations for responsiveness. Patients with newly diagnosed, or relapsed, AML have varied growth patterns compared with normal cells, and usually form clusters (3–40 cells) rather than colonies (>40 cells) or no colonies or clusters in semisolid agar culture medium.24, 25, 26 In the cases shown (Table 1 and Figures 3a–d), cells from these patients with AML did not grow without CSFs, but formed clusters (~3–20 cells per clone) upon stimulation with either rhFL-GM-CSF or -IL-3, and T-GM-CSF and -IL-3 were each less effective stimulating factors on these cluster-forming cells than their FL forms. The FL-GM-CSF- and -IL-3-stimulated cluster formation of the AML patient samples was reciprocally diminished by either T-GM-CSF or -IL-3 in patient samples tested, regardless of cytogenetic or molecular status. Additionally, for two patient samples tested, and as seen in the CB and TF-1 in vitro models (Figure 1), less than a 1:1 ratio of T-CSF to FL-CSF downmodulated maximal stimulation noted with combinations of FL-GM-CSF and -IL-3 (Figures 3c and d, patients 3135 and 3131). Further, cycling analysis showed an identical trend to that seen in the in vivo studies evaluating normal HPCs, where T-molecules induced less stimulatory activity than the FL-CSF counterparts with respect to the number of clusters formed and the percent of cells in the S phase of the cell cycle (Figure 3e). This marked suppressive response of AML patient samples in response to the T-factors, regardless of aggressive cytogenetic or molecular alterations, prompted us to compare the extent of the inhibition by fold change in primary CB vs primary AML patient samples. Figure 3f shows that, in general in this small sampling of patient cells, AML samples were significantly more sensitive to the T-factors than the CB cells, regardless of cytogenetic/molecular status, suggesting a possibility for consideration of future use of DPP4 T-factors to downmodulate growth of AML progenitors/precursors. We then assessed effects of T-CSFs in vivo in two mouse models of leukemia.

Table 1 Characterization of AML primary patient samples based on disease state, cytogenic/molecular phenotype and utilization
Full size table
Figure 3
figure 3

DPP4 truncation of GM-CSF and IL-3 inhibit AML cluster formation stimulated by FL-CSFs. (a–d) A total of 5 × 104 cells from ficolled peripheral blood of AML patients with 90% or greater blast burden (Table 1) were plated in agar with 10 ng/ml (unless otherwise noted) of rhFL- and/or T-GM-CSF and/or IL-3. Cells were plated in triplicate and incubated at 37°C/5% O2 for 7–10 days before enumeration of clusters formed. (e) Cell cycling evaluation of AML patient samples (n=3, 5 × 104) treated with FL- or T-factors. (f) Fold-change comparison of CB response to DPP4 T-factors compared with AML response. N=4 for AML samples and N=7 for LDCB *P⩽0.05 or less by analysis of variance (ANOVA) compared with control (FL-GM, FL3 or FL-GM/FL3).

Full size image

Effects of T- and FL-CSFs in vivo in mouse models of leukemia

AML primary patient samples were sensitive to the FL- and T-factors in vitro, regardless of cytogenetic or molecular status. Therefore, to begin to assess the potential clinical utility of T-GM-CSF and -IL-3 in models of malignant hematopoiesis in vivo, FLT3/ITD mice were first used.19 Similarly, to in vivo experiments with normal mice (Figure 2), FLT3/ITD mice were injected with rmFL- and T-GM-CSF and -IL-3, and effects on absolute numbers and cycling status of BM HPCs were analyzed. Mice showed a leukemia phenotype with significantly enhanced numbers of stem and progenitor cells in both the BM and spleen (Supplementary Figures 1A and B, respectively) compared with control. In vivo treatment of FLT3-ITD mice with rmFL-factors (GM-CSF and IL-3 in combination to maximize stimulation) resulted in functional increases in the number of progenitors (Figures 4a and c) and cycling status (Figures 4b and d), respectively, in the BM and spleen. T-factors (GM-CSF and IL-3 combination) not only reduced numbers and cycling status of HPCs compared with baseline levels of control mice but also significantly blunted the enhanced stimulation obtained with FL-CSFs in both the BM (A and B) and spleen (C and D). Another model of murine leukemia containing an E76K-activating mutation in protein tyrosine phosphatase, non-receptor type 11 (Ptpn11, Ptpn11E76K),20 commonly detected in childhood acute leukemia and myeloproliferative disorders, was also evaluated for response to rmFL- and T-GM-CSF and IL-3. These mice showed a leukemia phenotype with significantly enhanced numbers of progenitors in both the BM and spleen (Supplementary Figures 2A and B, respectively) compared with control. Identical trends to those seen with the FLT3/ITD mice were detected with respect to progenitor cell numbers and cycling in the spleen (Figures 4e and f). The BM of the Ptpn11E76K mice showed marked reductions in multiple phenotypically and functionally defined cell populations, in response to both FL- and T-factors, such that there was >90% reduction in BM short-term-hematopoietic stem cells, multipotent progenitors, common myeloid progenitors and granulocyte–monocyte progenitors with few or no colony-forming unit-granulocyte–macrophage, burst-forming unit-erythroid or colony-forming unit-granulocyte, erythrocyte, monocyte/macrophage, megakaryocyte detected in these mice after FL-, T- or FL+T-GM-CSF plus -IL-3 administration (data not shown). These decreases in HPC numbers in BM likely reflect large mobilization effects, suggesting that these mice may be ultrasensitive to CSF-induced mobilization regardless of treatment with FL- or T-molecules. Data from these two leukemia mouse models suggest that T-CSFs are able to modify growth and cycling of leukemia cells with molecular alterations in vivo, which may represent a heretofore unknown mechanism of leukemia cell regulation.

Figure 4
figure 4

Effects of T- and FL-CSFs on in vivo model of leukemia. FLT3-ITD (a–d) and Ptpn11E76K (e and f) mice were injected with 10 μg of a combination of either rmFL-GM-CSF and FL-IL-3, T-GM-CSF and T-IL-3 or a mixture of both FL- and T- (GM-CSF and IL-3) subcutaneously at different sites. Mice were killed 24 h later, femurs flushed, spleens made into single-cell suspensions and colony formation (a, c and e) and cycling status (b, d and f) of the progenitor cells were determined.

Full size image

T-GM-CSF and -IL-3 have high receptor binding affinity and compete with FL-GM-CSF and -IL-3 for receptor binding in primary CB and AML cells

To further identify potential mechanistic interactions of T- vs FL-GM-CSF and -IL-3 at a receptor level, receptor equilibrium binding studies were performed (Figure 5). We first used the TF-1 cell line stimulated with rhFL- and T-GM-CSF and -IL-3 that was I125 labeled or unlabeled. Both high- and low-affinity binding sites for FL- and T-GM-CSF and -IL-3 were detected by Scatchard analysis (Figure 5). By calculating dissociation constants (Kd), we determined that T-GM-CSF and -IL-3 bind with higher affinity than their FL counterparts, and both truncated forms are better competitors for receptor binding than FL-GM-CSF and -IL-3 (Figure 5a). We also noted, through binding competition studies using pooled CD34+ CB cells (Figure 5b) and multiple primary AML samples (Figure 5c), that the T-GM-CSF and -IL-3 reciprocally competed with and blunted the receptor binding of both their own, as well as the other FL-CSFs.

Figure 5
figure 5

T-IL-3 and -GM-CSF bind with greater affinity than FL-IL-3 or -GM-CSF, and T-IL-3 and -GM-CSF blunt receptor binding of both FL-IL-3 and -GM-CSF. (a) Scatchard analysis and Kd determination of TF-1(n=2) was performed with I125-labeled IL-3. A total of 1 × 106 TF-1 cells were plated per well, and iodinated FL- or T-IL-3 was added for 1 h. Cells were harvested and levels of I125-labeled ligand bound were detected using a Beckman Coulter Gamma 5500B and low- and high-affinity binding sites, as well as Kd were derived as published previously.3 Pooled CD34+ CB (b) or primary AML patient samples (c) were incubated with I125-labeled FL-IL-3 or -GM-CSF alone or in addition to cold T-IL-3 and -GM-CSF in excess for 1 h. Cells were harvested and the counts per minute (CPM) (or levels of I125-labeled ligand bound) were detected using a Beckman Coulter Gamma 5500B to analyze the ability of T-IL-3 and -GM-CSF to compete with FL-IL-3 and -GM-CSF. *P⩽0.05 AML patient samples n=7 distinct samples; CB n=5 independent experiments with CD34+ cells pulled from multiple CBs for each experiment.

Full size image

Discussion

Our studies have now demonstrated that DPP4 exists in an active form on the cell surface of murine as well as immature normal and malignant human cells, and that DPP4 T-GM-CSF and -IL-3 have enhanced receptor binding as well as modified HPC functional regulatory activity compared with their FL counterparts both in vitro and in vivo for both normal and leukemia progenitors/precursor cells. These results allude to the importance of DPP4, via its regulation of normal and malignant cells by acting as a scaffolding molecule and truncating physiologically and pathologically relevant proteins, thus adding another layer of interactions and complexity to growth modulatory factors and their role in the BM microenvironment. Alterations in receptor binding and signaling after DPP4 truncation is likely not only important for the effects seen on normal and leukemic cells in the settings of GM-CSF and IL-3 that we have elucidated here but also suggests the possibility for broader relevance for the many factors we have recently noted to have DPP4 putative or confirmed T-sites1, 2, 3 and potentially for other proteins with DPP4 T-sites that have not yet been identified. This highlights the critical need to understand how DPP4 T alters receptor binding of individual proteins and how it may alter their ability to act as a negative or positive regulator on normal and malignant hematopoiesis, as well as other steady and disease states outside hematopoiesis.1, 2, 3 Both T-GM-CSF and -IL-3 diminished receptor binding and function of both FL-GM-CSF and -IL-3, at less than a 1:1 ratio, likely through their shared common β-chain receptor. As DPP4 may have a more active role in hematopoiesis under stress conditions,3 a role for T-proteins in the regulation of hematopoiesis may be even more apparent under stress.

DPP4-specific regulation of GM-CSF and IL-3 via its truncation, and in general as a modulator of CSF signaling and consequently alterations in hematopoietic function, may have potential clinical application. Patients with leukemia may respond to T-factors regardless of molecular alterations or may have altered sensitivity to DPP4 T-factors based on leukemia type, similar to effects seen in our sampling of AML patient cells and the two in vivo models of leukemia assessed in our present studies. The knowledge we have gained in this research can possibly be used for investigation of the potential use of DPP4 T-proteins for future treatment of leukemia patients. This will require further mechanistic and therapeutic insight into the actions of T- vs FL-proteins in terms of what overlapping and distinct intracellular signals they elicit, and more in-depth analysis of in vivo mouse models of leukemia to determine the effects on disease progression/relapse and animal survival.

Importantly, current assessment of protein levels, for the most part, do not distinguish between the FL- and DPP4 T-forms of these proteins and hence assessments of proteins by conventional methods such as enzyme-linked immunosorbent assay, bioplex and other methods that use antibodies that do not distinguish the FL- from the T-forms of the proteins therefore may not be fully revealing in terms of the specific physiology or biological activities associated with these molecules. To that end, the possibility exists that functional outcomes that have been previously attributed to FL-molecules may, in fact, actually be due to T-molecules or a mixture of activities between the FL- and T-molecules. Thus, efforts to develop antibodies that can distinguish DPP4 T- from FL-proteins may be of practical, scientific and potentially clinical value.