Abstract
Phage display peptide libraries have enabled the discovery of peptides that selectively target specific organs. Selection of organ‐specific peptides is mediated through binding of peptides displayed on phage coat protein to adhesion molecules expressed within targeted organs. Hematopoietic stem cells selectively home to bone marrow, and certain adhesion receptors critical to this function have been demonstrated. Using a phage display library, we identified a specific peptide that trafficked to murine bone marrow in vivo. We independently isolated exactly the same heptapeptide from the entire library by in vitro biopanning on primitive lineage‐depleted, Hoechst 33342dull/rhodamine 123dull murine bone marrow stem cells and confirmed peptide binding to these cells by immunofluorescence studies. We demonstrated bone marrow–specific homing of the peptide by an in vivo assay in which the animals were injected with the phage displaying peptide sequence, and immunofluorescence analysis of multiple organs was performed. We also showed that the peptide significantly decreased the homing of stem cells to the bone marrow but not to the spleen 3 hours after transplantation using fluorescently labeled Lin−Sca+ hematopoietic cells in an in vivo homing assay. The peptide sequence has a partial (5/7) amino acid sequence homology with a region of CD84. This discovery represents the first application of the phage display methodology to the bone marrow and stem cells and led to the identification of a specific heptapeptide that homes to bone marrow, binds to primitive stem cells, and plays a role in stem cell homing.
Introduction
Hematopoietic progenitors are capable of migration and homing, processes fundamental to embryonal development, circulation in the vasculature, and stem cell transplantation. Engraftment after bone marrow transplant may involve several adhesion receptors, each recruited sequentially in the movement from bloodstream through vascular walls to the bone marrow. Studies in α4 integrin–null mice indicated that migration of embryonic hematopoietic precursors to the fetal liver, bone marrow, and spleen was independent of these receptors but that maintenance of hematopoiesis by these organs was entirely dependent on α4‐integrins [1]. The very late antigen (VLA)‐4 receptor plays the most critical role in engraftment after transplant, as demonstrated by the following observations: First, murine colony‐forming unit spleen cells and reconstituting hematopoietic stem cells attached to the connecting segment‐1 (CS‐I) peptide of alternatively spliced fibronectin (Fn), and this binding was blocked by antibody to α4 [2]; second, human long‐term bone marrow culture‐initiating cells (LTBM‐ICs) bound to Fn in the heparin‐binding region (a function of α4), and these Fn peptides could block their adherence to irradiated stroma [3]; third, antibody to VLA‐4 reduced bone marrow engraftment in mice [4]; and, fourth, blocking antibody to human VLA‐4 reduced bone marrow engraftment of human CD34+ cells in fetal sheep [5]. The VLA‐4/vascular cellular adhesion molecule‐1 (VCAM‐1) pathway was found to be critical for homing to the bone marrow, whereas CD44 was involved in the localization of colony‐forming unit–spleen in the bone marrow and spleen [6]. Antibody blocking of VCAM‐1 in vivo in E‐selectin/P‐selectin knockout mice suggested that all three receptors were required for optimal homing to the bone marrow [7]. Moreover, human CD34+ cells were dependent on CXCR4 (receptor for stromal cell–derived factor‐1) (SDF‐1) for optimal bone marrow homing and engraftment in the nonobese diabetic/severe combined immunodeficiency model [8], indicating that chemotaxis to SDF‐1 may be critical [9]. VLA‐4 was then shown to possess the dominant role over both selectins and β2‐integrins for engraftment by using antibody to α4‐integrin with donor cells from selectin‐null or β2‐integrin–null mice [10]. These prior approaches to the study of surface molecules in homing and engraftment examined receptors known to be expressed by the hematopoietic progenitors or the bone marrow stroma.
We sought to discover novel adhesion mechanisms in bone marrow homing and engraftment using a phage display peptide library (PDPL) [11]. The PDPL consists of an assembly of phages, each of which has fusions of a random peptide to a coat protein, with approximately five copies of each peptide per phage [12]. The number of unique peptide sequences depends on peptide length. Because there are 20 coding amino acids, a random heptapeptide display library has 207 or 1.3 ×109 unique sequences (clones). Phages that display specific peptides are selected by binding to selector or target molecules. Amplification of selected clones bearing peptides of interest is possible because the peptide sequence is coded by phage DNA. This process of selection and amplification is called biopanning and can be repeated several times to enrich the phage pool in favor of binding sequences. Individual clones may then be selected and characterized by DNA sequencing.
PDPLs were previously used in a variety of applications [11], including epitope mapping of monoclonal antibodies [13], identification of peptide ligands for α5β1‐integrin, and isolation of structural and functional mimics of the arginine‐glycine‐aspartic acid–binding site of integrins [14]. Peptides identified by biopanning procedures are frequently involved in important functional interactions; for example, peptides mimicking cytokines may exhibit their functional activities [15]. Also, peptide sequences often mimic the sequence of natural ligands such as peptides selected against anti‐platelet antibodies that share sequence homology with glycoprotein (GP) IIb/IIIa and GPIb surface molecules [16]. Biopanning can also be conducted in vivo, allowing selection of peptides that interact with tissues and cells in their native microenvironment. This approach was used to identify peptides that exhibit organ‐specific homing for the kidney and brain [17] and later for other organs and tumor vasculature [18] in animals. Similar studies to select organ‐homing peptides were initiated in humans [19].
For the present study, we used biopanning in vivo to identify a peptide that homes to the bone marrow. Surprisingly, the identical heptapeptide was isolated by biopanning in vitro on hematopoietic stem cells. A functional assay demonstrated its ability to interfere with early homing of hematopoietic progenitors.
Materials and Methods
Animal procedures were conducted according to protocols approved by the University of Massachusetts Medical School Institutional Animal Care and Use Committee.
Biopanning In Vivo
The phage display peptide library kit used for these experiments was purchased from New England Biolabs, Beverly, MA (Ph.D‐7 Phage Display Peptide Library Kit). The kit uses a combinatorial library of random peptide 7‐mers fused to the N‐terminus of a coat protein (p III) of an M13 phage. Random sequence is followed by a spacer (Gly‐Gly‐Gly‐Ser) and then followed by a wild phage sequence. The library consists of approximately 2.8 ×109 random sequences. Therefore, the 2 ×1011 pfu used for the first round of biopanning contained approximately 70 copies of each possible sequence. The diversity of the library was confirmed by sequencing of the native library by the manufacturer. The amplification of the phage was performed according to the manufacturer's instructions using the Escherichia coli strain ER 2738.
BALB/c mice 6–8 weeks of age (Taconic Farms, Germantown, NY) were injected intravenously (tail vein) with 2 ×1011 transducing units of PDPL (Ph.D‐7 Phage Display Peptide Library Kit, New England Biolabs) diluted in 0.25 ml Dulbecco's modified Eagle's medium (DMEM). The animals were euthanized after 10 minutes, and the bone marrow was harvested. The bound phage was eluted with 0.2 M glycine‐HCl and 0.1% bovine serum albumin. The eluted phage was titered, amplified, purified, and injected again at the same concentration. The procedure was repeated five times, yielding bone marrow–homing phage clones (Fig. 1).
Biopanning in vivo. Phages possessing the capability of binding specifically to the bone marrow were isolated by in vivo biopanning. The phage display peptide library was injected into the tail vein, and the mice were euthanized 10 minutes after injection. The bone marrow was isolated, and phage attached to bone marrow cells was eluted, purified, amplified, and then reinjected. Five such rounds of biopanning were performed.
Isolation of Lin− Hoechst 33342dull/rhodamine 123dull and Stem Cells
Bone marrow cells were lineage depleted [20] (Lin−) with antibodies to Ter119, B220, Mac‐1, GR‐1 Lyt‐2, L3T4, and YW25.12.7 and Dynabeads M 450 anti‐rat immunoglobulin G (IgG) (Dynal, Lake Success, NY) and then labeled with rhodamine 123 and Hoechst 33342 [21]. A population (Fig. 2) with low expression of both (first through 13th and first through third percentiles, respectively) was isolated by fluorescence‐activated cell sorting (FACS).
Fluorescence‐activated cell sorting of lineage‐depleted cells exhibiting low expression of both Hoechst 33342 and rhodamine 123. After depletion of cells bearing lineage markers using magnetic beads, a primitive progenitor cell population (depicted by R3 box) with low Hoechst 33342 fluorescence (first through third percentiles) and rhodamine 123 fluorescence (1st through 13th percentiles) was isolated by fluorescence‐activated cell sorting.
Biopanning In Vitro Using FACS Isolation of Hematopoietic Stem Cells
The PDPL was incubated with Lin− cells (109 pfu/106 cells) during the efflux step of the rhodamine 123 staining. Lin−Hoechst 33342dull/rhodamine 123dull cells were isolated by FACS with attached phage. The bound phage was eluted and amplified. Biopanning combined with FACS was repeated three times, yielding a phage clone that attached to primitive Lin−Hoechstdullrhodaminedull stem cells.
Organ Specificity of Specific Bone Marrow Homing Phage
Amplified specific phages (2 ×1010) were injected into two BALB/c mice, and the mice were euthanized at 10 minutes after injection. The organs were frozen sectioned, or cell suspensions were prepared from the lung, spleen, and bone marrow and cytospins prepared. The sections or cytospins were stained with anti‐phage M13 antibody, followed by secondary fluorescein isothiocyanate (FITC)–conjugated goat anti‐mouse antibody, and the fixed tissues were mounted in 4′,6′‐diamidino‐2‐phenylindole (DAPI)–Vectashield. A control mouse was injected with saline.
Bone Marrow Homing Assay Using 5‐ (and 6‐) Carboxyfluorescein Diacetate Succinimidyl Ester–Labeled Purified Lin−Sca+ Marrow Cells with Heptapeptide and Phage Blockade
Lin− bone marrow cells were labeled with phycoerythrin (PE)–conjugated anti‐Sca‐1 (Ly‐6A/E, Pharmingen, San Diego). The Lin−Sca+ cells were isolated by FACS and then labeled with carboxyfluorescein diacetate succinimidyl ester (CFDA‐SE; Molecular Probes, Eugene, OR) [22]. The heptapeptide STFTKSP was synthesized by the UMass Core Peptide Facility. To assess the heptapeptide's ability to interfere with stem cell homing, nonmyeloablated murine recipients underwent preinjection in the tail vein with a synthetic heptapeptide (0.0105 μg/g) 30 minutes before transplant (one animal) or 1010 pfu of phage displaying heptapeptide (four animals). Recipients were then injected with 25 ×104 CFDA‐SE–labeled Lin−Sca‐1+ cells, with either peptide or phage or equal volume of buffer for the control animals. The cells were preincubated with the heptapeptide at a concentration of 10 μM for 20 minutes (for the recipient pretreated with a heptapeptide) or with 108 pfu of heptapeptide‐displaying phage (for recipients pretreated with a phage) on ice, and the peptide with cells or phage with cells were injected intravenously. There was one recipient of heptapeptide‐treated cells and one control animal, and there were four recipients of phage‐treated cells and one control animal. Three hours after transplantation, the animals were euthanized and the bone marrow and spleen cells were harvested. The cells were analyzed by large‐volume flow cytometry of approximately 20 million events, and the number of CFDA‐SE fluorescence–positive cells per million nucleated cells was calculated [22]. Statistical analysis was performed using nonparametric Wilcoxon rank‐sum test.
The bone marrow homing assay using CFDA‐SE–labeled purified Lin−Sca+ marrow cells was validated using multiple control animals, as reported by our group in a separate publication by Cerny et al. [22]. A total of 44 mice were studied for homing by this assay in 17 experiments, with 20 mice studied in eight experiments at the 3‐hour time point. Within a range of 50,000 to nearly 3 million cells infused per animal, there was a linear relationship with the number of CFDA‐SE–labeled murine marrow Lin−Sca+ cells identified in the bone marrow after transplant [22].
Immunofluorescence Analysis of Binding of Bone Marrow Homing Phage to Lin− HoechstdullRhodaminedull Stem Cells
Lin− Hoechst 33342dull/rhodamine 123dull stem cells were labeled with 103 bone marrow homing phages per cell, antibody against the phage M13 protein (Amersham Biosciences, Piscataway, NJ), and then secondary FITC‐labeled anti‐mouse IgG.
Flow Cytometry Analysis of Phage‐Binding Bone Marrow Cells
Bone marrow mononuclear cells were isolated using Animal Step‐1 (Accurate Chemical and Scientific Corp., Westbury, NY). The specific phage was added at 10,000 phages per cell. Cells were analyzed after staining with antibody to phage M13 and secondary FITC‐goat anti‐mouse IgG and then rat monoclonal antibodies to lineage markers or c‐kit, Sca‐1, or CD84 (mouse anti‐human), followed by staining with PE anti‐rat IgG (mouse absorbed) or anti‐mouse for CD34 with appropriate controls.
Western Blot Stained with Phage to Identify a Membrane Receptor
SDS‐solubilized murine bone marrow membrane proteins were resolved on 10% SDS‐polyacrylamide gels and electrophoretically transferred to nitrocellulose. The blot was stained with specific phage and then anti‐M13 monoclonal antibody (Amersham Biosciences), alkaline phosphatase–conjugated anti‐mouse antibody, and 5‐bromo‐4‐chloro‐3‐indolyl phosphate/nitroblue tetrazalium (BCIP/NBT) liquid substrate (Promega, Madison, WI).
Affinity Chromatography of Bone Marrow Membranes
Aminolink coupling gel (Pierce Biotechnology, Rockford, IL) was used to link the synthetic heptapeptide [23]. Triton X‐100 1% was used to solubilize 3.2 million murine bone marrow cells in DMEM with protease inhibitors and then centrifuged. The supernatant was loaded, and the bound protein was eluted with 1 M NaCl, pH 7. The fractions were assayed by bicinchoninic protein assay and analyzed by SDS‐PAGE.
Results
Biopanning In Vivo
The sequencing of single‐phage colonies from the fifth round of biopanning in vivo (diagram of procedure in Fig. 1) revealed two predominant consensus sequences, STFTKSP and NHWASPR, constituting 50% and 28% of sequences, respectively. The remaining sequences were single or incomplete sequences, some of which contained motifs of the two predominant sequences.
Biopanning In Vitro Using FACS on Hematopoietic Stem Cells
Biopanning combined with FACS was repeated three times, yielding phage clones that attached to primitive Lin−Hoechstdullrhodaminedull stem cells. The results of sequencing showed that 85% of the single‐phage colonies had the same sequence as the most predominant peptide isolated by biopanning in vivo, STFTKSP. The remaining sequences were single or incomplete.
Flow Cytometry Analysis of Phage‐Binding Bone Marrow Cells
Flow cytometry analysis of the density‐depleted bone marrow stained with specific phage, antibody to phage M13 protein, and secondary fluorescent‐labeled antibody revealed a distinctive population constituting 1.10% of the density‐depleted bone marrow cells. The following percentages of C57BL6 bone marrow cells exhibited costaining by both lineage marker and specific phage when gated on cells binding phage: Ter‐119, 3%; YW25, 2%; Mac‐1, 3%; CD45R, 2%; CD4, 0%; CD8, 0%; GR‐1, 1%; and GPIIb/IIIa, 0%. These results indicate minimal expression of lineage markers by cells that bind the specific phage. In a separate set of experiments using BALB/c bone marrow, of the cells with attached specific phage, 36.6 ± 17.5% expressed Sca‐1 (1.01 ± 0.14% for the IgG2a isotype control), 49.9 ± 15.5% expressed c‐kit (1.64 ± 0.16% for the IgG2b isotype control), and 27.18 ± 4.05% expressed CD84 (0.99 ± 0.12% for the IgG1 isotype control). Moreover, 86% of the BALB/c lineage‐depleted bone marrow cells that expressed c‐kit also expressed CD84, and 57% of those that expressed Sca‐1 also expressed CD84.
Organ Specificity of Bone Marrow Homing Phage
The organs sections were stained with FITC‐labeled anti‐phage M13 antibody, and staining was observed within the bone marrow and lungs (Fig. 3), as follows: 12.5% of the DAPI‐counterstained cells from the bone marrow, 3% of the cells from the lung sections, 0.5% of the cells from a lymph node, and 0.5% of the cells from the thymus exhibited staining of bound phage. These numbers may represent an overestimation because of the appearance of staining of clusters of cells or adjacent cells. There was no staining for phage within the kidney, heart, brain, spleen, or liver.
BM and other cells were stained by immunofluorescence for phage to assess the specificity of organ homing. The left panels show the dual stain of the cell nuclei with 4′,6′‐diamidino‐2‐phenylindole (blue), indicating the location of cells and the staining with anti‐phage monoclonal antibodies and phycoerythrin (red)–labeled secondary antibodies. As a control, the BM of the animal that was not injected with the phage is shown in the right panels. Representative cells from the BM, lung, and brain are shown. Abbreviation: BM, bone marrow.
Immunofluorescent Staining of Bone Marrow–Homing Phage Binding to Lin−HoechstdullRhodaminedull Stem Cells
There were two patterns of staining: circumferential membrane and more dense, diffuse staining (Fig. 4).
The Lin−Hoechst 33342dull/rhodamine 123dull stem cells were examined by immunofluorescence to demonstrate binding of the specific phage. The upper panels show staining with DAPI. The lower first panel on the left represents a control with FITC‐labeled secondary antibody without phage; the lower middle and right panels show staining with phage, antibody to phage M13 protein, and FITC‐labeled secondary antibody. Abbreviations: Ab, antibody; DAPI, 4′,6′‐diamidino‐2‐phenylindole; FITC, fluorescein isothiocyanate.
Bone Marrow Homing Assay of CFDA‐SE–Labeled Lin−Sca‐1+ Cells with Heptapeptide/Phage Displaying Heptapeptide Blockade
Large‐event flow cytometry analysis demonstrated a 26% reduction in homing 3 hours after transplant in an animal injected with the synthetic heptapeptide compared with the control animal (data not shown). When the phage displaying the bone marrow homing heptapeptide was used instead of a free heptapeptide, there was a 41% reduction in homing 3 hours after transplant, whereas homing to the spleen was fully preserved at 109% of control. The difference was statistically significant for the difference in homing to the bone marrow compared with control (p = .007) but was not statistically significant for the spleen (p = .66; Fig. 5).
BM homing assay of carboxyfluorescein diacetate succinimidyl ester–labeled Lin−Sca‐1+ cells was performed in the presence of the specific phage‐displaying peptide. In the presence of the phage, there was diminished homing to the bone marrow (p = .007) but preserved homing to the spleen (p = .66) compared with the absence of phage in an in vivo homing assay. Abbreviation: BM, bone marrow.
Database Search Results
Although nucleotide database searches were not successful in identifying homologous sequences, a Genbank search using NCBI protein–protein blast BLASTP 2.1.3 revealed several sequences with varying degrees of homology (Table 1).
Genbank search results using NCBI protein–protein blast BLASTP 2.1.3
| Protein | Accession number | Sequence/position |
| NOV/plexin‐A1 protein (Homo sapiens) | X87832 | 1592 STFTKS 1597 |
| Leukocyte differentiation antigen (H. sapiens) | AF081189 | 45 TFTKNP 50 |
| Plexin 1 (Mus musculus) | D86948 | 1612 STFTKS 1617 |
| CD84 isoform CD84a (H. sapiens) | AF054816 | 265 TFTKNP 270 |
| Plexin (Xenopus laevis) | D38175 | 1623 STFTKS 1628 |
| Putative (M. musculus) | AK016085 | 96 FTKSP 100 |
| Hypothetical protein (H. sapiens) | FLJ13162 | 55 FTKSP 59 |
| 40S ribosomal protein S2 | P25444 | 261 FTKSP 265 |
| KRAB‐zinc finger protein SZF1‐1 (H. sapiens) | AF114816 | 415 TFTRSP 420 |
| Hepatocyte growth factor (Xenopus) | S77422 | 177 FTKSP 181 |
| Protein | Accession number | Sequence/position |
| NOV/plexin‐A1 protein (Homo sapiens) | X87832 | 1592 STFTKS 1597 |
| Leukocyte differentiation antigen (H. sapiens) | AF081189 | 45 TFTKNP 50 |
| Plexin 1 (Mus musculus) | D86948 | 1612 STFTKS 1617 |
| CD84 isoform CD84a (H. sapiens) | AF054816 | 265 TFTKNP 270 |
| Plexin (Xenopus laevis) | D38175 | 1623 STFTKS 1628 |
| Putative (M. musculus) | AK016085 | 96 FTKSP 100 |
| Hypothetical protein (H. sapiens) | FLJ13162 | 55 FTKSP 59 |
| 40S ribosomal protein S2 | P25444 | 261 FTKSP 265 |
| KRAB‐zinc finger protein SZF1‐1 (H. sapiens) | AF114816 | 415 TFTRSP 420 |
| Hepatocyte growth factor (Xenopus) | S77422 | 177 FTKSP 181 |
Genbank search results using NCBI protein–protein blast BLASTP 2.1.3
| Protein | Accession number | Sequence/position |
| NOV/plexin‐A1 protein (Homo sapiens) | X87832 | 1592 STFTKS 1597 |
| Leukocyte differentiation antigen (H. sapiens) | AF081189 | 45 TFTKNP 50 |
| Plexin 1 (Mus musculus) | D86948 | 1612 STFTKS 1617 |
| CD84 isoform CD84a (H. sapiens) | AF054816 | 265 TFTKNP 270 |
| Plexin (Xenopus laevis) | D38175 | 1623 STFTKS 1628 |
| Putative (M. musculus) | AK016085 | 96 FTKSP 100 |
| Hypothetical protein (H. sapiens) | FLJ13162 | 55 FTKSP 59 |
| 40S ribosomal protein S2 | P25444 | 261 FTKSP 265 |
| KRAB‐zinc finger protein SZF1‐1 (H. sapiens) | AF114816 | 415 TFTRSP 420 |
| Hepatocyte growth factor (Xenopus) | S77422 | 177 FTKSP 181 |
| Protein | Accession number | Sequence/position |
| NOV/plexin‐A1 protein (Homo sapiens) | X87832 | 1592 STFTKS 1597 |
| Leukocyte differentiation antigen (H. sapiens) | AF081189 | 45 TFTKNP 50 |
| Plexin 1 (Mus musculus) | D86948 | 1612 STFTKS 1617 |
| CD84 isoform CD84a (H. sapiens) | AF054816 | 265 TFTKNP 270 |
| Plexin (Xenopus laevis) | D38175 | 1623 STFTKS 1628 |
| Putative (M. musculus) | AK016085 | 96 FTKSP 100 |
| Hypothetical protein (H. sapiens) | FLJ13162 | 55 FTKSP 59 |
| 40S ribosomal protein S2 | P25444 | 261 FTKSP 265 |
| KRAB‐zinc finger protein SZF1‐1 (H. sapiens) | AF114816 | 415 TFTRSP 420 |
| Hepatocyte growth factor (Xenopus) | S77422 | 177 FTKSP 181 |
Affinity Chromatography with the Heptapeptide of Murine Bone Marrow
Passage of Triton X‐100–solubilized bone marrow membranes over an affinity column bearing attached synthetic heptapeptide resulted in elution of an 82‐kDa polypeptide from the column with 1M NaCl (Fig. 6A).
Affinity chromatography and protein binding on blots of solubilized BM membranes were performed to determine the identity of the receptor for the specific phage. (A):Affinity chromatography of solubilized BM membranes with synthetic heptapeptide bound to the affinity column. The first lane shows the MW standards; the second, solubilized BM membranes; the third, the proteins of the supernatant of the Triton‐extracted BM cells that did not bind to the affinity column (flow thru); the fourth, the proteins of the Triton‐extracted BM cell pellet; and the fifth, the specific 82‐kDa protein bound to the affinity column that eluted with 1M NaCl. (B): Western blot phage protein binding. Note that a doublet of MW 37/33 kDa was stained. Abbreviations: BM, bone marrow; MW, molecular weight.
Western‐Type Blot of Membrane Receptors for Phage
Bone marrow cell‐membrane proteins resolved on SDS‐polyacrylamide gels were electrophoretically transferred to nitrocellulose. After incubation with specific phage and antibody to phage M13 and then development with anti‐mouse IgG‐conjugated alkaline phosphatase and substrates, staining revealed doublet protein bands of 37 and 33 kDa (Fig. 6B). It is unclear whether these bands represent proteolytic degradation products of the above‐described 82‐kDa protein or whether a different protein was obtained by binding of phage to SDS‐solubilized bone marrow membranes rather than native Triton‐solubilized bone marrow membranes.
Discussion
The heptapeptide, STFTKSP, was identified by successive repetitions of in vivo biopanning in BALB/c mice as a peptide that selectively homed to bone marrow. Parallel studies using the same phage display library and Lin−Hoechst 33342dull/rhodamine 123dull stem cells as a target for in vitro biopanning surprisingly yielded the same predominant heptapeptide. Lin−Hoechst 33342dull/rhodamine 123dull cells (which constitute fewer than 0.01% of nuclear cells within the bone marrow) are characterized by the absence of lineage markers and the exclusion of Hoechst and rhodamine dyes, properties mediated by the ability of the multidrug‐resistance transporter ABCG2 to efflux Hoechst [24] and by the ability of the multidrug resistance gene MDR‐1 to efflux rhodamine. Lin−Hoechst 33342dull/rhodamine 123dull cells are considered to be very primitive, and just a few are capable of reconstituting the entire lymphohematopoietic system in lethally irradiated animals [25]. One potential hypothesis for the finding that the identical sequence was obtained for both homing to marrow and binding to stem cells is that the phage's receptor is homophilic and binds to itself as a ligand.
Peptides obtained by biopanning procedures often possess the unique property of binding to molecules involved in important functional interactions [14,15]. Moreover, binding frequently occurs at the interactive sites of these proteins [14], rendering short peptides obtained by biopanning capable of interfering with protein–protein interactions. This property led us to believe that the bone marrow/stem cell homing peptide might have the ability to impede stem cell homing. In a homing assay using highly purified Lin−Sca‐1+ murine marrow cells labeled with a cytoplasmic fluorescent label, CFDA‐SE, a decrease in homing 3 hours after transplant was observed in animals preinjected with the peptide or phage‐bearing peptide. A more pronounced reduction in homing was seen using a phage‐bearing peptide than peptide alone (41% versus 26%, respectively). This could be related to fact that the phage contains multiple ligand binding sites (average of five displayed peptides per phage) and is less likely to be inactivated through internalization of the receptor or by proteolysis. Inactivation of the peptide/phage or a loss of only part of the homing process may explain incomplete interference with bone marrow homing.
We did not observe interference with homing to the spleen, which correlates well with the finding that the heptapeptide does not home to the spleen in organ specificity studies. This strongly suggests the involvement of a heptapeptide receptor in stem cell homing to the bone marrow but not to the spleen.
The ability of the relatively short peptides obtained in biopanning procedures using phage display libraries to interact with active sites of receptors is frequently based on the high degree of homology between the peptide and ligand. This homology allows identification of the receptors' ligands by sequence homology [17,18]. We performed a Genbank protein blast search, which revealed several vertebrate proteins. Of particular interest were proteins known to be surface receptors such as plexins. Plexins were initially recognized as surface molecules involved in axonal guidance during nervous system development and in cell migration during embryonal development [26]. These properties made plexins interesting candidates for heptapeptide receptors; however, the homologous sequences were far within the cytoplasmic domain.
A Genbank search also revealed a motif of the human CD84 isoform A with five of seven identities, either amino acids 259–264 or 265–270, both of which are in a sequence predicted to be cytoplasmic but very proximal to the transmembrane region, and it is not known if the polypeptide could again cross the membrane. CD84 is a highly glycosylated single‐chain cell‐surface glycoprotein of Mr 64‐82 kDa, which is close to the 82 kDa of the receptor identified by affinity chromatography of bone marrow membranes with the heptapeptide attached to the affinity gel. CD84 is a member of the CD2 subgroup of the Ig receptor superfamily [27]. Its expression is restricted to hematopoietic cells, including a high proportion of mobilized human peripheral blood and marrow committed progenitor cells [28]. It functions as a homophilic receptor, serving as its own ligand [29]. We have herein demonstrated that CD84 is expressed by most of cells that express c‐kit or Sca‐1, indicating that there may be a CD84 isoform expressed by primitive hematopoietic progenitors as well. Because there was no coincidental staining pattern for the phage and CD84, it is unlikely that CD84 isoform recognized by the anti‐human CD84 antibody is identical to the specific phage receptor. At least five CD84 isoforms have been described [30], and we propose that there may be an isoform expressed by more primitive stem cells.
Given the organ and cell specificity of the heptapeptide, there are several potential clinical applications such as drug or gene delivery. For example, peptides that home specifically to tumor vasculature were identified using biopanning and, when coupled with doxorubicin, have been shown to improve the response to chemotherapy of human breast cancer xenografts in mice [18]. Short peptides comprised of two functional domains, one a tumor–blood vessel homing motif and the other a programmed cell death–inducing sequence, were shown to have anticancer activity in mice [31]. Building the peptide into the viral coat protein may be a way of stem cell targeting in gene therapy [32].
In summary, our present work identifies a novel peptide that selectively homes to bone marrow and additionally binds to primitive hematopoietic progenitor cells. This peptide's receptor may have a role in bone marrow engraftment in the multistep process of homing of hematopoietic stem cells. Phage display in vivo biopanning represents a very attractive method for identifying new tissue‐homing proteins. Thus, we have described the application of phage display to ascertain critical homing proteins for stem cell trafficking to marrow; however, the approach also holds unique potential for the study of nonmarrow tissue targeting.
Acknowledgements
This publication was made possible by NIH grants P01 DK50222, R01 DK27424, and R01 DK60084 from the National Institute for Diabetes and Digestive and Kidney Diseases, P01 HL56920, R01 HL63184, and R01 HL073749 from the National Heart Lung and Blood Institute, and P20 RR018757 from the National Center for Research Resources. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
