Background
We proposed to isolate leukemia specific peptides that have the potential to target and deliver toxins to acute myeloid leukemia cells (AML) or to modify the biological behavior of the cells to which they bind using a phage display library. First described by George Smith in the mid 1980s [
1], this technique allows repertoires of antibodies, proteins or peptides displayed on the surface of phage particles to be screened by any chosen target. Single chain Fv phage libraries have been used to isolate antibodies that recognize cell surface antigens for clinical, diagnostic and therapeutic applications [
2‐
6] or for antigen epitope mapping [
6,
7]. An alternative approach was to use peptide phage libraries to identify small molecules that can bind either purified targets or cell surface receptors. Peptides that are specific for surface expressed immunoglobulins isolated from chronic lymphocytic leukemia (CLL) cells [
8] and multiple myeloma cells [
9] have been identified for potential patient specific targeted therapy. Our approach was to use freshly obtained patient cells to isolate leukemia specific peptides from a phage library. In this study we show that multiple myeloid leukemia specific and non-specific peptides can be identified by this method. In addition, we show that these peptides are capable of altering the biological behavior of AML cells while having no effect on normal marrow elements.
Methods
Patients and normal donors
Specimens were obtained from patients with chronic myelogenous leukemia in blastic crisis (CML-BC), AML of different FAB classifications, as well as from normal donors. Informed consent was obtained from all patients prior to study according to the regulations of the Institutional Review Board of Rush Medical Center.
Cell preparation
The peripheral blood (PB) or bone marrow (BM) cells were subjected to density cut centrifugation over Ficoll-Hypaque. The mononuclear fractions were washed twice and used directly. Granulocytes were obtained from the high-density fraction. Immature CD34+ stem cells were isolated by magnetic cell sorting and separation (Miltenyi Biotec, California). The promyelocytic cell line HL60 (ATCC) was maintained in RPMI 1640/20% FBS. Cells were induced to differentiate into the granulocytic lineage by treatment with 1.5% (v/v) of DMSO during 7 days of culture.
Isolation of phage binding to human malignant myeloid cells
This work was performed using the Display PHAGE system library, purchased from Display System Biotech (Vista, California) with a diversity of approximately 3 × 107.
Initial selection of phage specific for myeloid leukemia used cells from patients with CML-BC. CML-BC cells were incubated with the library (1012 cfu) in PBS/0.1% milk at 4°C for 90 minutes. Unbound phage was removed by 5 washes with buffer. Weakly bound phage was acid eluted by glycine (0.1 M, pH2.2) followed by cell lysis (30 mM Tris pH8.0, 1 mM EDTA) to elute the tightly bound phage. This second fraction was amplified by re-infection and growth in E. Coli. Amplified phage was purified by PEG precipitation and used for another round of binding. The "biopanning" step was repeated five times to obtain a population that was highly enriched with phage expressing peptides that bind to leukemia cells. After growth on agar plates using antibiotic selection, individual colonies were randomly picked, amplified, and PEG purified. Binding to leukemia cells was confirmed by ELISA assay.
ELISA Assay
Amplified phage from a single colony (109cfu/ml) were dispensed into a 96-well plate containing either leukemia or normal peripheral or bone marrow mononuclear cells (250,000 cells/well) previously fixed with glutaraldehyde and blocked with PBS/2% milk. After 2 hours, wells were washed and bound phage was detected by a monoclonal anti-M13 antibody HRP conjugate (1:2,000 in PBS/1% milk, Amersham Pharmacia Biotech, NJ). Following addition of HRP substrate, intensity of color was measured by spectrophotometry. Each phage clone was tested in triplicate with appropriate controls. Identical assays were performed using synthesized peptides that had been conjugated with biotin. The bound peptide was detected by streptavidin-HRP diluted 1:1,000 in PBS/1% milk (Amersham Pharmacia Biotech, NJ). A biotin control confirmed that binding to the cells was via the peptide moiety itself and not via the biotin conjugate.
Peptide synthesis
Phage DNA was extracted using the QIAprep M13 kit (Qiagen). The hypervariable oligonucleotide sequence coding for the peptide was PCR amplified using the following primer set: Primer 1 = 5' GGG ATT TTG CTA AAC AAC 3', Primer 2 = 5' GGA GGT CTA GAT AAC GAG 3'. Each clone was amplified, purified and sequenced (Gene Link, NY) in duplicate. The synthetic octapeptides with terminal cysteine residues were commercially synthesized (New England Peptide, Inc., Massachusetts) with more than 95% purity. Peptides were not cyclized by oxidation; therefore percentage of free sulfhydryl groups was evaluated by Ellman's reaction using a cysteine standard (Pierce Biotechnology). This reaction allowed us to determine the percentage of reduced cysteines in the peptide solution used for our experiments. The percentage ranged between 50% and 75% for most peptides except for HP-B6 (100%) and HP-G2 (30%). A biotin molecule was conjugated to the N-terminal of the peptide for cytochemistry studies.
Cytochemistry
Fresh cells, blocked with PBS/4% BSA for one hour at 37°C, were incubated with biotin-conjugated peptide at 1 μM for 30 minutes unless otherwise stated. Free biotin was used to confirm that binding of biotin-conjugated peptide was mediated by the peptide moiety. After washing, cells were fixed with 3.7% paraformaldehyde, permeabilized with methanol and incubated with streptavidin-FITC (DAKO) diluted 1:500 in PBS/4% BSA for 30 minutes at room temperature and washed again. Alternatively, cytospin preparations of cells were fixed in 3.7% paraformaldehyde and kept at -80°C for further study. The fluorescent signal was analyzed using AxioVision 2.05 software.
Liquid culture of HL60, AML and normal cells
To assess the biological effects of peptides on growth and differentiation of HL60 cells, logarithmic growth phase cells were seeded in 3 ml of RPMI 1640/10% FBS. Cultures in the presence or absence of a single peptide (10-6 M -10-4 M) or DMSO (1.5%) were assessed for the level of cell viability, cell number, and differentiation, after 7 days without changing the medium.
AML or normal bone marrow cells were seeded at a concentration of 106cells/ml in 3 ml of RPMI 1640/15% FBS. Peptides were added at a single dose of 10-4 M without addition of fresh medium or peptide for the duration of the cultures. Granulocyte-Macrophage Colony-Stimulating Factor (GM-CSF, 100 U/ml) was used as a positive control. Viability was evaluated at day 14.
Analysis of cell cycle
After 7 days of culture, HL60 cells (1 × 106) were washed in PBS, fixed in 1% paraformaldehyde and permeabilized with 0.5% Triton X100 in acid solution. After neutralization, cells were incubated with RNase A and stained with propidium iodide. The relative DNA content was analyzed by flow cytometry.
CD11b analysis by flowcytometry
After 7 days of culture, HL60 cells (1 × 106) were stained with fluorescence conjugated monoclonal antibodies recognizing the myeloid maturation marker CD11b (Becton Dickinson Immunocytometry system) for 20 minutes at 4°C. The percentage of cells with a fluorescence intensity above the control was measured.
Colony assay
Mononuclear cells (104 cells/ml) were plated in 0.8% methylcellulose (Methocult, StemCell Technologies Inc., Vancouver, Canada) in Iscoves modified Dulbecco medium (IMDM, Gibco Laboratories, Grand Island, NY) with the peptides at a single dose of 10-4 M or GM-CSF at 100 U/ml without addition of fresh medium or peptide. The solvent used for peptide preparation served as a control. Following 14 days of incubation, the number of colonies was assessed and cytospin preparations were made for Giemsa morphology staining. Percentage of differentiated cells was evaluated by counting 200 cells.
Statistical analysis
All analyses were performed using SYSTAT software version 10.0. The relationship between AML patients' clinical data and responses of AML cells to the peptides was analyzed by Fisher's exact test.
Discussion
Therapeutic progress in AML has been painfully slow despite significant improvement in our understanding of the underlying pathology. A major conundrum in designing treatment strategies relates to the issue of targeting only the malignant cells while sparing their normal counterparts. In this work, we have shown that using a phage library, it is possible to isolate peptides that can bind and induce biologic activity only in leukemia cells. Since cell lines are often not representative of the physiological state of malignant cells, we used freshly obtained patient specimens for the initial screening, and in order to avoid isolating patient specific peptides, the screening and profiling were performed with cells from multiple patients with different myeloid leukemia subtypes. It is of interest that only 12 unique peptide sequences were found and that one sequence (HP-B11) was found in 22 clones. The redundancy of the phage sequence may be related to the copy number of each epitope expressed on hematopoietic cells. Ten synthesized peptides were re-profiled on AML and normal cells. As expected, all the peptides were able to bind to the malignant cells, but four peptides also were found to bind to normal BM cells. The peptides were able to bind to all AML-FAB subtypes in our small sample size, but the percentage of cells recognized by the peptide was lower at relapse.
We have found at least 2 peptides, HP-A2 and HP-G7, that bind only to AML cells and not to normal cells including normal CD34+ cells. Cytochemistry studies showed that after surface binding, these two peptides were internalized within minutes. Intracellular localization is a particularly attractive feature since it offers a novel method to deliver drugs and may reduce toxicity by allowing lower concentrations to be administered. In addition, these studies showed that the peptides do not always label all the leukemia cells which is not surprising since AML blasts are not a homogeneous population [
11] and AML marrows frequently contain normal mononuclear cells.
In order to develop clinically useful strategies, we were looking for peptides that, upon binding, could induce some biological change in the leukemia cell growth and differentiation. The peptides that we have identified could clearly affect both these parameters although the biologic response elicited was variable. A given peptide could induce proliferation, differentiation, both or none in the cells of different AML patients. Of the two leukemia specific peptides, HP-A2 stimulated proliferation in 40% and differentiation in 25% of the patients tested while HP-G7 stimulated proliferation in 35% and differentiation in 60% of the patients. This includes some patients in whom the peptides induced proliferation followed by differentiation. To be therapeutically useful in future, the biologic effects of a given peptide would have to be defined for the individual patient. While the peptides were screened to be myeloid specific and not patient specific, the biologic activity elicited by a peptide could depend on multiple variables. The maturation state at which the AML clone is arrested is defined by the FAB subtype and this may be a factor that determines the specific response to the peptide. We have not analyzed sufficient numbers of AML patients to correlate activity with subtype. Additionally, activity could be determined by the number of epitopes on the cells of the AML clone.
Blast analysis of the sequence of the two leukemia specific peptides revealed that 7 of 8 amino acids of HP-A2 matched mostly to proteins of lower organisms. For HP-G7, the best match was 6 amino acids identical to two different proteins; the Toll-like receptor 2 and the neraminidase protein of the streptococcus pneumonia bacteria. Further studies, however, will be needed to clarify the molecular epitope that is recognized by these peptides in order to understand the mechanism(s) responsible for their biological activity.
The high concentration of peptides used for these studies is non-physiological. However, the peptide solution, administered as a single dose at the start of cell culture, was found to contain a mixture of various conformations including linear, cyclic or concatamers. It is possible that the structure exhibiting biological activity represented only a fraction of the total and its stability in the culture medium is unknown. This may explain the need for a high initial concentration to elicit a biological effect. It will be necessary in the future to identify the conformation responsible for biological activity. In addition to conformation, peptide sequence determines the nature of the biological response. For example, HP-C8 and HP-G2, identical for 6 amino acids, elicited different activities despite binding similarities. Finally, a single amino acid substitution in HP-A2 altered the ability of the peptide to induce proliferation. These observations support a specificity of the structure-function relationships of the peptide and its target.
HP-A2 and HP-G7 were also found to bind HL60 cells that have often been used as a model to study myeloid cell growth and differentiation. We have used HL60 cells to study the effect of these two peptides on cell cycle and differentiation. While DMSO was found to increase CD11b expression and arrest the cell cycle, we found that CD11b expression was increased by the peptides without concomitant cell cycle arrest. Studies with other differentiating agents show accumulation of cells during differentiation in G1 by day 3 or day 4 [
12‐
14], we have not found a G1 block or decrease in cell number in peptide treated differentiating cells even after 7 days of culture. Thus the mechanism of differentiation by the peptides appears to be novel and different than that of the known conventional differentiating agents. It may be that the binding of these peptides results in an uncoupling of the expression of the proliferation and differentiation markers. In the absence of growth arrest, it is conceivable that the peptides can also be used concomitantly with chemotherapeutic agents which depend upon cellular synthesis to increase the therapeutic index.
The leukemia specific peptides are expected to be non-toxic based on in vitro results, and theoretical in vivo clinical applications are multiple. Coupled to toxins, the peptides could specifically target and kill tumor cells as first line therapy and/or as therapy targeting residual disease. When coupled with an imaging reagent, the peptides could be used to locate sites of tumor sanctuary. The peptides could also be combined with peptides that selectively target the tumor vasculature [
15]. The challenge now is to identify the molecular target on the surface of leukemia cells that binds these peptides, demonstrate that the in vitro effects are reproducible in vivo by conducting similar experiments in animal models and finally translating these findings into clinical trials for human use; areas that are currently being actively pursued in our laboratory at present.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
NG participated in study design, conducted functional studies, analysis and wrote the manuscript. ED participated in study design, isolated the peptides conducted functional studies and analysis. AR participated in study concept, design and coordination and wrote the manuscript. All authors read and approved the manuscript.