Background
C-X-C chemokine receptor type 4 (CXCR4) receptor overexpression (CXCR4+) associates with dissemination, relapse, minimal residual disease, and poor survival in acute myeloid leukemia (AML) patients [
1‐
4]. Moreover, CXCR4+ leukemic blasts become anchored and quiescent in the bone marrow (BM) through CXCR4-CXCL12 (C-X-C motif chemokine 12 or stromal cell-derived factor 1; SDF1) interaction [
5], displacing normal hematopoietic stem cells (HSCs) to cause patient death [
6]. CXCR4+ AML cells also circulate in the blood [
4] to reach extramedullar CXCL12 producing sites [
4,
7], mainly the liver and spleen [
5,
7‐
11]. All these functions support CXCR4+ AML cells as Leukemic stem cells (LSCs), despite the lack of a formal proof. Similarly, CXCR4+ cancer cells are metastatic stem cells in several solid tumors [
12‐
17]. Consequently, different groups are developing small drug CXCR4 inhibitors for AML therapy [
5,
7,
18] aimed at eliminating CXCR4+ LSCs, responsible for AML aggressiveness and patient death.
Nanotechnological approaches can increase selectivity regarding target cancer cell killing, while reducing systemic toxicity [
19]. Here, we specifically tested if a novel T22-GFP-H6-Auristatin nanoconjugate (NC) selectively eliminates CXCR4+ AML cells, and therefore LSCs, in an AML disseminated model, to improve chemotherapy performance, which reduces the differentiated tumor bulk but does not eliminate LSCs [
20]. The NC incorporates a protein nanoparticle to selectively internalize in CXCR4+ AML cells by exploiting high CXCR4 overexpression in LSCs as compared to HSCs [
6,
20], similarly to a previous approach in colorectal cancer (CRC) [
21,
22]. The NC is loaded with Auristatin E, a microtubule-destabilizing toxin, introduced in clinical onco-hematology [
23,
24]. This drug eliminates quiescent cancer cells [
23,
24], overcoming the limited effectiveness of current AML chemotherapy, unable to kill LSCs in the G0 phase [
6,
20].
We expected the novel T22-GFP-H6-Auristatin NC to selectively deliver Auristatin E to cycling and quiescent CXCR4+ LSCs target cells achieving a potent inhibition of AML dissemination in the absence of toxicity, because of its low uptake in normal tissues, as we reported for a NC loaded with a CRC sensitive drug [
22]. To that aim, we generated a bioluminescent CXCR4+ disseminated AML model, which allows a non-invasive follow-up of the response in BM, liver, and spleen, the clinically relevant sites. We describe CXCR4-dependent internalization and killing of CXCR4+ AML cells for this novel NC in vitro. We also report in this in vivo model, a potent antineoplastic effect on BM and blockade of dissemination to extramedullar sites.
Methods
Drugs, nanoparticle, and nanoconjugate
Monomethyl Auristatin E (MMAE) is a potent tubulin polymerization inhibitor that was acquired by custom systhesis from Levena Biopharma (Levena Biopharma, San Diego, CA, USA) as a maleimide conjugated MMAE (MC-MMAE). T22-GFP-H6 is a CXCR4 targeted protein nanoparticle produced in
E. coli as previously described [
21]. T22-GFP-H6-Auristatin nanoconjugates were synthesized by covalent binding of the targeting vector (T22-GFP-H6) with the therapeutic moiety (MC-MMAE). For that, an excess of MC-MMAE was incubated with T22-GFP-H6 nanoparticles and reacted with amino groups of external lysines in a 1:50 ratio (protein to MC-MMAE) for 4 h at room temperature. T22-GFP-H6-Auristatin nanoconjugates were then again purified by IMAC affinity chromatography using HiTrap Chelating HP 5 mL columns in an ÄKTA pure (GE Healthcare, Chicago, IL, USA) in order to remove non-reacted free MC-MMAE. Finally, re-purified nanoconjugates were dialyzed against sodium carbonate buffer (166 mM NaCO
3H, 333 mM NaCl pH = 8) and conjugation efficiency and presence of free MMAE checked by MALDI-TOF mass spectrometry.
The volume size distribution of T22-GFP-H6 nanoparticles and resulting nanoconjugates (T22-GFP-H6-Auristatin) was determined by dynamic light scattering at 633 nm in a Zetasizer Nano (Malvern Instruments, Malvern, UK). Measurements were performed in triplicate. In addition, ultrastructural morphometry of T22-GFP-H6-Auristatin nanoconjugates (size and shape) was determined at nearly native state with field emission scanning electron microscopy (FESEM). Samples were directly deposited on silicon wafers (Ted Pella Inc., Redding, CA, USA) for 30 s, excess of liquid blotted, air dried, and immediately observed without coating with a FESEM Zeiss Merlin (Zeiss, Oberkochen, Germany) operating at 1 kV and equipped with a high resolution in-lens secondary electron detector. Representative images of a general field were captured at two high magnifications (× 100,000 and × 120,000). In a quantitative approach, nanoconjugates average size from FESEM images were analyzed by Image J software (1.8.0.172, National Institutes of Health, USA) [
25].
The average molar mass of T22-GFP-H6 nanoparticles and T22-GFP-H6-Auristatin nanoconjugates was measured by a size exclusion chromatography coupled to a multi-angle light scattering (SEC-MALS). Samples were injected in a Superdex 200 increase 10/300 GL column (GE Healthcare, Chicago, IL, USA) and run in a degassed sodium carbonate buffer with Nickel (166 mM NaCO3H, 333 mM NaCl, 0.1 mM NiCl2 pH = 8). The eluent was monitored by an in-line UV-Vis detector, a Dawn Heleos MALS detector and an Optilab rEX RI detector (Wyatt Technology Corporation, Santa Barbara, CA, USA). All data were analyzed using Astra 6.0.2.9 software (Wyatt Technology Corporation). Molecular weights were double-checked from MALS with UV and RI signals using ASTRA software and dn/dc (mL/g) values of 0.185 and UV extinction Coefficient (mL/(mg.cm)) values of 1.099 for all proteins. Conjugated MMAE ratio was calculated by subtracting the average molecular weight of T22-GFP-H6 nanoparticles to the average molecular weight of T22-GFP-H6-Auristatin nanoconjugates and dividing by the molecular weight of MMAE molecule.
Fluorescence of T22-GFP-H6 nanoparticles and T22-GFP-H6-Auristatin nanoconjugates was determined in a Cary Eclipse fluorescence spectrophotometer (Varian Inc, Palo Alto, CA, USA) at 510 nm upon excitation at 488 nm.
T22-GFP-H6-Auristatin nanoconjugates proteolytic stability in serum was determined by western blot immunostaining using a monoclonal anti-His antibody (Santa Cruz Biotechnology, Dallas, TX, USA) upon incubation in front of human serum (Sigma-Aldrich, St-Louis, MO, USA) at different time points (0, 2, 5, and 24 h) at 37 °C and a final concentration of 1 mg/mL.
Cell lines
The AML cell lines used in this study were purchased from DSMZ (Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany). THP-1 and SKM-1 were cultured in RPMI-1640 medium supplemented with 10% FBS, 10 mmol/L L-glutamine, 100 U/mL penicillin, 10 mg/mL streptomycin and 0.45 μg/mL fungizone (Gibco, Thermo Fisher Scientific (TFS), Waltham, MA, USA). Cells were kept at 37 °C in a humidified atmosphere of 5% CO2. THP-1 cells were transfected with a plasmid encoding the luciferase gene that confers bioluminescence (BLI) to the cells using Lipofectamine LTX and PLUS reagents (A12621, Invitrogen, TFS) according to the manufacturer’s instructions.
Quantitation of CXCR4 membrane levels in AML cell lines
To detect cell surface expression of CXCR4 in AML cell lines, fluorescence-activated cell sorting (FACS) analysis was performed as described before [
26]. Two technical and three biological replicates were performed. Data acquisition was analyzed by Cell Quest Pro software (BD Biosciences, San Jose, CA, USA) and results were expressed as mean fluorescence intensity (MFI) ± standard error (SE).
Antineoplastic effect and internalization analyses of T22-GFP-H6-Auristatin in CXCR4+ AML cells in vitro
To assess the antineoplastic effect of the nanoparticle T22-GFP-H6 and the nanoconjugate T22-GFP-H6-Auristatin in AML cell lines, XTT cell viability assays were performed. Cells were cultured at 20 × 104 cells/mL for 24 h in 96-well plates. After that, T22-GFP-H6-Auristatin was added in a range of 30–300 nM and T22-GFP-H6 at 500 nM. Forty-eight hours later, cells were incubated for 4 h with the XTT reagents (Cell Proliferation Kit II, Roche Diagnostics, Basel, Switzerland) and then the absorbance was measured at 492 nm in a FLUOstar OPTIMA spectrophotometer (BMG Labtech, Ortenberg, Germany). Five technical and three biological replicates were performed for each condition. The growth inhibitory activity was obtained by subtracting the absorbance of the blanks and expressed as the percentage of cell growth inhibition (± SE), as compared with untreated controls.
To find out the mechanism of cell death mediating T22-GFP-H6-Auristatin antineoplastic activity in AML cells, nuclear staining with DAPI dye in THP-1 and SKM-1 cells exposed to 240 nM T22-GFP-H6-Auristatin or buffer for 24 or 48 h, in 12-well plates (50 × 104 cells/mL), was performed. After incubation, the media was collected and centrifuged to obtain the suspended cells. Cells were rinsed once with PBS. Afterwards, they were centrifuged and fixed (3.7% p-formaldehyde in PBS, pH= 7.4) for 10 min at − 20 °C, washed with PBS, and resuspended in 10 μL of PBS. Finally, cells were mounted on a slide with DAPI dye (ProLong™ Gold Antifade Mountant with DAPI, P36935, Invitrogen, Carlsbad, CA, USA) and observed for the appearance of the apoptotic bodies or mitotic catastrophe (MC), as induced events, under a fluorescence microscope (Olympus BX53, Olympus, Tokyo, Japan). Cell death events such as MC or apoptosis were quantified using the point tool in Image J software (1.8.0.172) and 4 randomly chosen fields for each condition.
To evaluate the internalization capacity of the nanoconjugate T22-GFP-H6-Auristatin in AML cell lines, FACS analyses were performed as described before [
26] exposing cells to the vehicle or with T22-GFP-H6-Auristatin between 30 and 300 nM. Two technical and three biological replicates were performed for each condition. Data acquisition was analyzed by Cell Quest Pro software (BD Biosciences) and results were expressed as MFI ± SE.
Receptor competition assays were performed with the antagonist of CXCR4, AMD3100. To that aim, internalization analyses or XTT cell viability assays were performed as described above, treating cells with 240 nM of T22-GFP-H6-Auristatin with or without a pre-treatment for 1 h of AMD3100 (2000 nM).
Evaluation of the antineoplastic effect of T22-GFP-H6-Auristatin in a disseminated CXCR4+ AML mouse model
NSG (NOD-
scid IL2Rgamma
null) female mice (4 weeks old) were obtained from Charles River Laboratories (Wilmington, MA, USA). Mice were housed in microisolator units with sterile food and water ad libitum. All procedures were conducted in accordance with the guidelines approved by the institutional animal Ethics Committee of Hospital Sant Pau. Three healthy (not having cancer) and untreated mice (not administered with vehicle or nanoconjugate) were assigned to the Normal mouse group (normal,
n = 3). To generate the AML mouse model, 19 mice were intravenously (IV) injected with luciferase-transfected THP-1 cells (THP-1-Luci; 1 × 10
6 cells/200 μL). These mice were divided randomly into two different experimental groups (day 0) (Fig.
5). One group (VEH;
n = 10) was IV injected (Day 2) with the vehicle of the nanoconjugate (Buffer NaCO
3H + NaCl pH = 8) and another group (T22-AUR;
n = 9) with 100 μg of T22-GFP-H6-Auristatin daily for a total of 9 doses. The evolution of AML dissemination in mice was monitored using the IVIS Spectrum Imaging System (PerkinElmer, Waltham, MA, USA) three times per week until the day of the euthanasia. Animal weight was measured the same day as that of BLI analysis. All mice were euthanized the day that the first animal presented relevant signs of disease such as lack of mobility or 10% weight loss. Animals were intraperitoneally injected with luciferin; after 5 min, blood was obtained by intracardiac extraction, and mice were killed by cervical dislocation. Blood samples were collected for BLI analysis in IVIS and CD45-positive cells detection by flow cytometry. Tissues were excised to analyze the BLI levels ex vivo, and preserved in paraffin for further histological and immunohistochemical analyses. BLI measurements were expressed as total flux of BLI (photons/second; radiance photons) ± SE in both in vivo and ex vivo studies.
Detection of AML cells in circulating blood
To detect leukemic cells in the blood, CD45 leukocyte marker detection by FACS was performed. Red blood cells were lysed from the blood samples according to the instructions of the lysis buffer (eBioscience, TFS). After that, cells were incubated with 10 μL of PE-Cy5 Mouse Anti-Human CD45 (ref. 555484, BD Biosciences) for 20 min in agitation and dark at 4 °C. Cells were washed and re-suspended with PBS-BSA. CD45-positive cells were detected by flow cytometry using FACS Calibur cytometer (BD Biosciences). Data acquisition was analyzed by Cell Quest Pro software (BD Biosciences) and results were expressed as the number of CD45-positive events of a total of 10,000 events ± SE.
Immunohistochemical staining of leukemic cells in AML affected organs
Immunohistochemical analysis was performed using paraffin-embedded tissue samples as described before [
26] using anti-human CD45 antibody (ref. IR75161-2, Dako, Agilent, Santa Clara, CA, USA) to detect infiltration of AML cells and anti-human CXCR4 antibody (ref. AB124824, Abcam, Cambridge, UK) to evaluate CXCR4 expression in tissues. AML infiltration by CD45 detection was evaluated with cellSens Dimension 1.9 software (Olympus, Tokyo, Japan) using the Counter and measurement tool. Results were expressed as the percentage of surface stained ± SE. Stain intensity was calculated using Image J software (1.8.0.172) and the Colour Deconvolution Plugin with the H DAB vector to split the brown staining adjusting the threshold to 150. After that, the “Analyze particles” plugin was used to detect all stained areas and the mean gray value was obtained combining all selected black areas and measuring using the ROI Manager. The intensity value of each analyzed slide was calculated subtracting 255 to the mean gray value obtained by the Image J analysis. Six slides for each treatment type and tissue were analyzed and results were expressed as mean ± SE intensity values. Two independent observers evaluated all samples to made consensus decisions.
Toxicity analyses in tissues and blood
Hematoxylin and eosin staining was performed in the organs infiltrated by leukemia cells (bone marrow, liver, and spleen), and also in non-affected organs to evaluate possible off-target toxicities of the nanoconjugate (lung, brain, pancreas, heart, and kidney). In addition, plasma glutamic oxaloacetic transaminase (GOT) and glutamic pyruvic transaminase (GPT) enzyme activities, and creatinine levels were determined using commercial kits (ASTL ref. 20764949 322; ALTL ref. 20764957 322; CREJ2 ref. 04810716 190, respectively. Roche Diagnostics), adapted for a COBAS 6000 autoanalyzer (Roche Diagnostics), to evaluate toxicity in tissues such as the liver, kidneys, or heart after T22-GFP-H6-Auristatin treatment in mice.
Statistical analysis
Statistical analyses were performed in the IBM SPSS Statistics (Release 22.0.0.0, New York, NY, USA). Mann-Whitney U test was used in both in vitro and in vivo assays, and any differences were considered statistically significant when the p value was < 0.05. IC50 were calculated with SigmaPlot (Release 12.0.0.182; Systat Software, Inc., San Jose, CA, USA) using non-linear regression test with Hill-3 parameter adjustment.
Discussion
We developed a novel targeted drug delivery strategy for acute myeloid leukemia (AML) treatment. Repeated administration of the T22-GFP-H6-Auristatin nanoconjugate (NC) in a CXCR4+ THP-1 AML mouse model achieved potent anticancer activity, by significantly reducing the leukemic cell burden in the bone marrow and circulating blood, and by inducing a potent blockade of leukemic cell spread to extramedullar organs. These results were obtained by measuring the bioluminescence emitted by the CXCR4+ luciferase+ AML cells at each affected site, findings that were validated observing a reduction of CD45+ AML cells in the liver and spleen at the end of treatment. This potent blockade of leukemic dissemination was achieved using a NC dosage that did not induce any histological alteration or toxicity on normal organs; therefore, this nanoconjugate displays a high therapeutic index.
Our results show the importance of using a disseminated model for preclinical AML drug development, displaying bone marrow homing, circulating leukemic cells in the blood, and AML cell involvement at extramedullar sites. The luminescence emitted by luciferase+ AML cells allows the non-invasive follow-up of the response of target CXCR4+ leukemic cells to NC treatment in each clinically relevant site. This model outperforms most current models used to evaluate the effectiveness of novel drugs or therapeutic approaches in AML therapy. Thus, many authors evaluate only antitumor activity measuring subcutaneous (SC) tumor size reduction in mouse models [
27‐
30], which cannot evaluate the relevant drug effect on AML cells anchored in the bone marrow [
6]. Other groups complement the SC model assessment with orthotopic or patient-derived PDX models [
31‐
36]; however, the lack of luciferase expression in most models prevents the direct measurement of the drug effect on AML cells affecting the bone marrow, liver, or spleen [
10].
We explain the potent antineoplastic effect obtained in the bone marrow, circulating cells, and extramedullar sites without associated toxicity, in our model, by the achievement of a high selectivity by the T22-GFP-H6-Auristatin NC in CXCR4+ AML cell killing, and the consequent elimination of leukemic stem cells (LSCs). Thus, CXCR4+ AML cells can be considered LSCs based on their association with bone marrow homing, circulation in the blood, and extramedullar dissemination, all associated with poor prognosis [
2,
5,
20]. Two additional reasons also support the notion that this NC achieves selective killing of LSCs, without having an impact on normal cells:
1.
Selectivity in CXCR4+ AML cell killing exploits the high overexpression of CXCR4 in LSCs as compared to normal HSCs, as previously described [
6,
37], leading to the specific internalization and selective delivery of the conjugated drug to CXCR4+ AML cells in all clinically relevant sites, since our THP-1 model highly express the CXCR4 receptor in the AML cell membrane in BM and extramedullar sites. This selective effect is also supported by the in vitro THP-1 and SKM-1 AML cells results, in which the NC showed a strict selectivity and CXCR4 dependence in internalization and cell killing of CXCR4+ AML cells.
2.
Ability of the NC payload, Auristatin E, for killing quiescent cancer cells, as a microtubule destabilizing agent, through the induction of mitotic catastrophe and apoptosis as described [
23,
24], which particularly, in our model, eliminates the LSCs that enter quiescence after homing in the BM [
6,
20]. This effect may overcome the limited effectiveness of current chemotherapy that blocks AML cells in division but induces life-threatening adverse effects [
38], being unable to kill quiescent LSCs. This is likely the reason why Auristatin E is extensively used in antibody-drug conjugates under development, one of which, Brentuximab vedotin, is currently used in the clinic to treat CD30+ Hodgkin’s lymphoma and anaplastic large-cell lymphoma patients [
39].
We also think these arguments, and particularly their cancer-selective biodistribution, underlie our expectation that this nanoconjugate may outperform the efficacy of drugs used in the clinic or being under development as we following explain. Thus, the NC selectivity in CXCR4+ AML cell uptake, and selective elimination by the payload drug, differs from the molecularly targeted small drugs that inhibit proteins involved in maintaining the cancer stem cell phenotype (e.g. Wnt, Hedgehog NFkB). The passive diffusion of small drugs to all cells implies a lack of selectivity in their biodistribution to cancer sites; therefore, the achievement of therapeutic concentrations in cancer cells is limited by associated toxicity in normal tissues [
40,
41].
The direct killing of CXCR4+ quiescent AML blasts in the BM reached by our nanoconjugate differs also from the effect of low molecular weight inhibitors of the CXCR4 receptor, which in AML are used in combination with chemotherapy. CXCR4 inhibitors mobilize AML blasts from BM to the bloodstream, losing their quiescence, and becoming cycling cells that chemotherapy targets. Nevertheless, AML cell mobilization induces hyper-leucocytosis that could, unexpectedly, generate extramedullar dissemination [
5].
Regarding strategies similar to our targeted drug delivery approach, no antibody-drug conjugates (ADCs) targeting CXCR4 have been introduced in the clinic so far. We believe that our protein-based nanoparticle could, however, improve the reported biodistribution described for ADCs, reaching only 0.01–0.1% of the injected dose in cancer cells [
42]. In contrast, we have reported that T22-GFP-H6 nanoparticles and the derivate NCs achieve around 85% of the administered dose in tumor tissues, both in solid tumors [
22,
43] and hematological neoplasias [
26].
Finally, because of the reported overexpression of CXCR4 in AML cells which associates with chemotherapy resistance, LSC development, relapse, and minimal residual disease in AML patients [
1‐
4,
11,
44‐
47], we believe this nanoconjugate could be used to treat resistant AML. Thus, in the near future, we plan to develop disseminated AML models resistant to therapy to evaluate this hypothesis.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.