HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 2005-0484v1 24/4/908    most recent Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints/Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Nagasawa, Y. Articles by Blau, C. A. Search for Related Content PubMed PubMed Citation Articles by Nagasawa, Y. Articles by Blau, C. A.

TECHNOLOGY DEVELOPMENT

Anatomical Compartments Modify the Response of Human Hematopoietic Cells to a Mitogenic Signal

^a Department of Medicine,
^b Department of Laboratory Medicine, University of Washington, Seattle, Washington, USA;
^c Division of Transplantation Biology, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA

Key Words. Gene therapy • In vivo selection • Regulation • Hematopoiesis

Correspondence: C. Anthony Blau, M.D., K260 Health Sciences Building, Mailstop 357710, University of Washington, Seattle, Washington 98195, USA. Telephone: 206-685-6873; Fax: 206-543-3560; e-mail: tblau{at}u.washington.edu

Received September 30, 2005; accepted for publication December 14, 2005.

	ABSTRACT

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Methods for specifically regulating transplanted cells have many applications in gene and cell therapy. We examined the response of human cord blood CD34⁺ cells to a specific mitotic signal in vivo. Using a conditional signaling molecule (F36VMpl) that is specifically activated by an artificial ligand called a chemical inducer of dimerization (CID), human hematopoietic cells transplanted into immune deficient mice were induced to proliferate. Only differentiating erythroid precursors and multipotential and erythroid progenitors (colony-forming unit [CFU]-mix and burst forming unitserythroid [BFUe]) responded; however, the nature of the response differed markedly between bone marrow and spleen. In the marrow, F36VMpl induced a 12- to 17-fold expansion of differentiated erythroid precursors and a loss of CFU-mix and BFUe. In the spleen, F36VMpl induced a marked rise in BFUe and CFU-mix and, relative to marrow, a much less prominent rise in more mature red cells. Clonal analysis was most consistent with the interpretation that the spleen and bone marrow differentially regulate the response of human progenitors to a mitotic signal, possibly influencing progenitor expansion versus differentiation. These findings establish CIDs as in vivo growth factors for human hematopoietic cells.

	INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
A challenge for gene and cell therapy is to develop safe and effective methods for controlling transplanted cells, allowing their number or function to be titrated to a desired therapeutic endpoint. One method for controlling cells uses conditional signaling molecules that can, upon activation, elicit death, proliferation, differentiation, or other desired functions [1]. The activation state of many engineered signaling proteins can be regulated by controlling their self-association, using small-molecule drugs called chemical inducers of dimerization (CIDs) [1, 2]. One application of this technology is in gene therapy for blood disorders, where inefficient gene transfer into hematopoietic stem cells poses a significant roadblock and CID-dependent activation of mitotic signaling molecules can bring the frequency of genetically modified cells into a therapeutic range [3, 4]. The therapeutic potential of this approach, known as "in vivo selection," has been established in gene therapy trials for diseases in which the therapeutic gene itself confers a growth or survival advantage, such as X-linked severe combined immune deficiency and adenosine deaminase deficiency [5, 6].

Methods for regulating cell proliferation most commonly rely on growth factor receptors, modified to substitute their binding site for endogenous ligand with the binding site for a CID. Insensate to endogenous ligand, these modified receptors are specifically activated by the CID, which acts as a growth factor for genetically modified cells. Studies of CID-regulated in vivo selection have identified the signaling domain of the thrombopoietin receptor, mpl, as a potent inducer of CID-dependent hematopoietic cell growth. Results from three different animal species (mice, dogs, and baboons) have been reported [3, 4, 7]. The retroviral vector used in each of these studies incorporated a murine stem cell leukemia virus (MSCV) long terminal repeat (LTR), resulting in expression of the CID-dependent murine mpl derivative (F36VMpl) in all hematopoietic lineages [8, 9]. However, in contrast to this panhematopoietic expression pattern, the effects of F36VMpl differed markedly between different lineages. In both mice and dogs, red cells expanded markedly, platelets moderately, and granulocytes only very minimally [3, 4]. Peculiar to dogs was the effect of F36VMpl on B cells, which expanded even more dramatically than red cells and trafficked to lymphoid tissues [4]. In all cases, the effects of F36VMpl remained completely CID-dependent, with the frequency of genetically modified cells falling upon CID withdrawal. Contrary to the prominent effects of F36VMpl in mice and dogs, the hematopoietic cells of baboons responded negligibly to F36VMpl activation. Most striking in baboons was the absence of a red cell response [7]. These disparate findings between different animal models pose a quandary when attempting to predict how human hematopoietic cells might respond to a F36VMpl signal in vivo. Relevant data are available from studies of human cord blood CD34⁺ cells in culture. Transduced cord blood CD34⁺ cells expanded up to 186-fold when cultured in the presence of CID alone, without additional growth factors [10]. In contrast to mouse marrow cells, which could be expanded exponentially for up to a year of culture [11], the expansion of human hematopoietic cells was transient, peaking at 2 weeks of culture, and resulted almost entirely from the differentiation (but not self-renewal) of transduced burst forming unitserythroid (BFUe). Transduced granulocyte macrophage colony-forming cells (CFU-GMs) appeared impervious to F36VMpl signaling alone; however, they could be rendered F36VMpl-responsive upon the provision of a cocktail of growth factors [12]. Although these findings establish the ability of human hematopoietic cells to expand in response to F36VMpl signaling, it is noteworthy that neither in mice nor dogs were qualitative observations in cell culture predictive of the cell types most affected by F36VMpl signaling in vivo. In mice, the dominant effect of F36VMpl on red cells in vivo contrasts with a predominance of megakaryocytes in cell culture [3, 11], whereas in dogs, macrophages were the predominant cell type to emerge in culture [4] (C.A. Blau, unpublished results). These findings indicate that environmental signals that are absent in cell culture can significantly modulate the response to F36VMpl signaling in vivo. Therefore, prior to a clinical trial, it would be useful to evaluate the response of human hematopoietic cells to F36VMpl signaling in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Construction of Lentiviral Transfer Vectors and Preparation of Viral Vector Stocks
The lentivirus vector encoding F36VMpl was assembled using a lentivirus construct pLHGAFMIG (W. Guo and C.A. Blau, unpublished results; details available upon request), and a construct containing hPGK-GFP (kindly provided by L. Naldini, Università Vita-Salute School of Medicine, Milan, Italy) [13]. This vector contained an MSCV promoter [8, 9] directing expression of F36VMpl and a phosphoglycerate kinase promoter directing expression of green fluorescent protein (GFP). Vector stocks were prepared and concentrated as described [14, 15]. Titer, determined in HT1080 cells, was 4–9 x 10⁸ IU/ml.

Cells
Umbilical cord blood was collected under an Institutional Review Board-approved protocol from the obstetrical unit of the University of Washington Medical Center. CD34⁺ cell selection and cryopreservation were performed as described previously.

Mice
NOD/SCID/ß2M^null mice [16], provided by the Core Center of Excellence in Molecular Hematology at the Fred Hutchinson Cancer Research Center, were bred in a specific pathogen-free facility. All procedures were performed according to protocols approved by the Animal Care and Use Committees of the University of Washington and the Fred Hutchinson Cancer Research Center.

Gene Transfer and Transplantation
Transductions were performed essentially as described [14]. CD34⁺ cells were thawed and pooled, then cultured in serum-free medium (Iscove’s modified Dulbecco’s medium [IMDM] containing 1% bovine serum albumin, 10 µg/ml bovine pancreatic insulin, 200 µg/ml human transferrin, 10^–4 M 2-mercaptoethanol, and 4 mM L-glutamine) in the presence of cytokines (20 ng/ml recombinant human [rh] interleukin 6, 100 ng/ml rh stem cell factor, 100 ng/ml rh FLT-3 ligand, and 20 ng/ml rh thrombopoietin; Peprotech, Rocky Hill, NJ, http://www.peprotech.com) and vector at a concentration of 5 x 10⁵ cells/ml (multiplicity of infection: 10) for 20 hours in a 37°C, 5% CO₂ incubator. Following transduction, cells were injected via the tail vein into sublethally irradiated (3.5 Gy) NOD/SCIDß2M^null mice at a cell dose of 2 x 10⁵ cells per mouse. To determine the efficiency of gene transfer, transduced cells were washed with IMDM and plated in IMDM with 10% fetal bovine serum (FBS) with 20 ng/ml rhIL3, 20 ng/ml rhIL6, and 100 ng/ml recombinant human stem cell factor (rhSCF). After 5 days, the percentage of GFP⁺ cells was assessed by flow cytometry.

Ex Vivo Expansion
Following lentiviral transduction, CD34⁺ cord blood cells were cultured in IMDM containing 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin either in the presence or in the absence of AP20187 (100 nM; ARIAD Pharmaceuticals, Inc., Cambridge, MA, www.ariad.com/regulationkits) in a 37°C, 5% CO₂ incubator.

CID Administration
Lyophilized AP20187 was solubilized in 100% ethanol to produce a 5 mg/ml stock that was stored at –20°C. AP20187 was diluted fresh on the day of injection in sterile water containing 10% PEG 400 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) and 1.44% Tween 80 (Sigma-Aldrich). AP20187 was administered via daily intraperitoneal injection. (0.2 mg/mouse, approximately 10 mg/kg).

Flow Cytometry
Transduced cells were washed and resuspended in phosphate-buffered saline containing 1% FBS. The presence of human cells in mouse tissues was determined using flow cytometric analysis of cells harvested from the bone marrow, spleen, and blood after labeling with monoclonal antibodies against human CD45, CD71, CD33, CD19, CD34, and CD41a (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) and analyzed on a Becton Dickinson LSRII. Syto41 (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com) was used to identify nucleated cells. Nucleated erythroid cells were defined as positive for syto41 and bright CD71 without CD45. Myelomonocytic cells were defined as positive for CD45 and CD33, and B cells were positive for CD45 and CD19 without CD33.

Colony Assays
Bone marrow and spleen cells were plated in methylcellulose medium (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) containing 0.92% methylcellulose in IMDM with 30% FBS, 2 mM L-glutamine, 10^–4 M 2-mercaptoethanol, and 1% bovine serum albumin with 3 U/ml rh erythropoietin (Amgen, Thousand Oaks, CA, http://www.amgen.com), 10 ng/ml rh-granulocyte macrophage colony-stimulating factor, 10 ng/ml rhIL3, and 50 ng/ml rhSCF (Peprotech), at concentrations of either 5 x 10⁴ marrow cells or 1 x 10⁶ spleen cells per ml. Cultures were plated in triplicate and incubated in a highly humidified 37°C, 5% CO₂ incubator. Colonies were scored by light and fluorescence microscopy 14 days later (MZ16FA; Leica, Heerbrugg, Switzerland, http://www.leica.com). The GFP plus filter set contained a 480/40-nm exciter filter and a 510 LP barrier filter.

Single Colony Linear Amplification-Mediated Polymerase Chain Reaction
Colony DNA was prepared by transferring a colony, in minimal volume, from the methylcellulose matrix to 100 µl of lysis buffer containing 10 mM Tris, pH 7.5, 0.1% Triton X-100, and 500 µg/ml Proteinase K. The lysate was digested at 55°C for 2 hours and heat-denatured at 99°C for 10 minutes. Genomic DNA was purified from sorted and unsorted bone marrow and spleen cells using a QiaAmp DNA Blood Mini Kit (Qiagen, Valencia, CA, http://www1.qiagen.com).

Insertion site analysis was performed using a three-arm modification of linear amplification-mediated polymerase chain reaction (LAM-PCR) (M.A. Harkey and C.A. Blau, manuscript submitted). Reactions were conducted with either 10 µl of colony DNA (100–1,000 cells), or 50–100 ng of column-purified DNA. All amplifications were done with Advantage II Polymerase Mix and buffer system (BD Biosciences, San Diego, http://www.bdbiosciences.com) using the following cycling conditions: melt at 95°C for 20 s, anneal at 60°C for 45 s, elongate at 68°C for 90 s. First, single-stranded copies of the viral 3'LTR-host junction region were made by 100 cycles of linear amplification, using a biotinylated LTR primer (biotin-5'AAGCCTCAATAAAGCTTGCC). The product was bound to streptavidin-coated magnetic M280 Dynabeads (Dynal Biotech, Oslo, Norway, http://www.invitrogen.com/dynal). Subsequent reactions and washes (3 x 200 µl after every reaction) took place on this matrix. Matrix-bound DNA was resuspended in 20 µl of random-priming reaction containing 500 nM dNTPs, 100 ng/µl random hexamers, 2–5 units Klenow enzyme (New England Biolabs, Beverly, MA, http://www.neb.com), and manufacturer-supplied buffer at 1x and incubated at 37°C for 60 minutes. Products were divided into three equal arms, and digested with Tsp509I, HaeIII, or RsaI (New England Biolabs). DNA that remained bound to the matrix was then ligated to a double-stranded anchor primer using Fast Link Ligase (Epicenter, Biotechnologies, Madison, WI, http://www.epibio.com). For the Tsp509I arm, component primers GACCCGGGAGATCTGAATTCAGTGGCACAGCAGTTAGG and AATTCCTAACTGCTGTGCCACTGAATTCAGATC were used, yielding a Tsp509I-compatible end. For the HaeIII and RsaI arms, a blunt-ended anchor primer was used, with CCTAACTGCTGTGCCACTGAATTCAGATC as the second component. The double-stranded junction fragments, containing host DNA flanked by LTR and anchor-primer (AP) sequences, were eluted from the matrix in 5 µl of 0.1 N NaOH and amplified by two rounds of nested PCR. Two microliters of the eluate was amplified in a 50-µl PCR with LTR primer 1 (TGAGTGCTTCAAGTAGTGTGTGC) and AP 1 (GACCCGGGAGATCTGAATTC). This reaction was diluted 100-fold in water, and 2 µl of template was amplified by 30 cycles of nested PCR, using LTR primer 2 (ACTCTGGTAACTAGAGATCC) and AP 2 (GATCTGAATTCAGTGGCACAG). The LAM-PCR products were separated by electrophoresis on Novex 4–20% polyacrylamide TBEgels (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), stained with ethidium bromide, and visualized by ultraviolet light fluorescence.

Clonal Tracking by Clone-Specific PCR
In order to track specific lentivirus-containing clones, PCR primers were designed to amplify the viral host junction of the inserted provirus. LAM-PCR products from spleen colonies were separated on 3% agarose gels, and individual bands excised and cloned into the TOPO-TA vector, pCR4 (Invitrogen). The cloned inserts were sequenced from flanking M13-F and M13-R primer sites in the vector using Big Dye fluorescent dideoxynucleotide chemistry (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) and an ABI Prism 310 Genetic Analyzer (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com). Sequences were screened for the presence of LTR and anchor-primer motifs, and the intervening sequence was compared to the human genome using the University of California-Santa Cruz BLAT service (http://genome.ucsc.edu/cgi-bin/hgBlat). Primers were designed from the host sequence to amplify the host-LTR junction in combination with our LTR primers. The insertion site for clone 7 is human chromosome 2: 68,683,670. The insertion-site–specific tracking primers are ATGTTATTATGATAATATGTAAAGGG and TATAATTTCTAATAGGGTAAATATCC, corresponding to chromosome 2: 68,683,306 – 68,683,329, and 68,683,381–68,683,406, respectively. The insertion site for clone 9 is human chromosome 17: 31,324,273. The insertion site-specific tracking primers are CCTGAGTTACTGGGACTATAGG and AAGCACATGTATGCTCTCATGGG, corresponding to chromosome 17: 31,323,801–31,323,822 and 31,323,936–31,323,958, respectively. The presence of clones 7 and 9 in colony or bulk DNAs was determined by nested PCR (with the cycling conditions described above) using LTR primers 1 and 2 and the appropriate clone-specific primers. The nested priming approach was necessary to eliminate nonspecific amplification observed with single primer sets. Round 1 amplification was performed at 15 cycles for the clone 9 insertion and 30 cycles for clone 7. Round 2 amplification was performed at 25, 30, 35, and 40 cycles for both clones. A parallel PCR for an intron of the single-copy mouse GAB2 gene was used to confirm the presence of host DNA. The GAB2 primers were GAGACAGTGAGGAGAACTATGTCC and CACACTTGGTAATGCACAGCTTCC. Cycling conditions for GAB2 were as follows: melt at 94°C for 10 seconds, anneal/elongate at 68°C for 1 minute.

Splenectomies
NOD/SCID mice were anesthetized with a 500-µl intraperitoneal injection of Avertin made by dissolving 2.5 g of 2,2,2-tribromoethanol in 5 ml of 2-methyl-2-butanol (both Sigma-Aldrich), then adding distilled water to a final volume of 200 ml, and then sterile filtering. The mouse was placed in a supine position, the abdomen was shaved and disinfected, and a 2-cm incision was made just below the left costal margin to expose the spleen. The spleen was exteriorized and excised with blunt forceps and scissors. The wound was then closed with Super Glue.

Statistical Analyses
All comparisons were made using Student’s t test.

Globin Immunofluoresence
Sorted CD71⁺ cells from mice transplanted with F36VMpl-transduced cord blood CD34⁺ cells and treated with a 2-week course of CID were resuspended in fetal calf serum and used to produce cytocentrifuge smears. Air-dried slides were methanol-fixed, washed, and incubated with an anti- monoclonal antibody [17] in a humidified chamber at 37°C for 1 h. Slides were washed and developed by incubating with a fluorescein isothiocyanate-conjugated goat anti-mouse secondary antibody. The slides were subsequently examined by fluorescence microscopy.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Lentivirus-Mediated Gene Transfer Confers CID Responsiveness In Vitro
To improve gene transfer into cells capable of repopulating immune-deficient mice [18], we used a lentivirus vector in which an MSCV promoter regulates expression of F36VMpl and a phosphoglycerate kinase promoter regulates expression of GFP. Human cord blood CD34⁺ cells transduced with the F36VMpl vector expanded 28.5-fold by day 14 of culture in the presence of AP20187 (without added cytokines), while falling 2.5-fold in cultures lacking CID (Fig. 1A). More than 80% of cells expanded in response to CID were GFP⁺, CD45^–, and CD71⁺, consistent with their identity as erythroid cells (data not shown). In contrast to the expansion of GFP positive erythroid cells, GFP⁺ BFUe and CFU-mix fell (Fig. 1A). These findings are consistent with our previous work using an oncoretrovirus vector, in which F36VMpl specifically induced the differentiation of human erythroid progenitors in culture [10].

View larger version (43K): [in this window] [in a new window]   Figure 1. F36VMpl-mediated selection of transduced human erythroid cells. (A): Cord blood CD34+ cells were transduced with the F36VMpl lentivirus vector and cultured for 14 days in the presence or absence of AP20187 (100 nM). Numbers of total cells (closed circles, x10–4), GFP+ burst forming units-erythroid (triangles, x 10–1), and GFP+ CFU-mix (diamonds, x10–) in the presence of AP20187 and total cells in the absence of AP20187 (open circles) are depicted over time. (B–I): representative experiment showing that the F36VMpl response is restricted to human erythroid cells. (B–F): The percentages presented are relative to all nucleated cells in the femur. (B): Mean percentages of human non-erythroid (CD45+) and erythroid (CD71+) cells relative to total nucleated cells in the femurs of chemical inducer of dimerization (CID)-treated (+) and non-CID-treated (–) mice either 2 weeks or 3 weeks post-transplant. Mice evaluated at week 2 post-transplant had received CID for 7 days, whereas mice evaluated at week 3 post-transplant had received CID for 14 days. The femurs of CID-treated mice (+) contained a significantly higher percentage of human hemopoietic cells at week 3 compared to non-CID-treated controls (–) due to a preferential expansion of CD71+ erythroid cells (p = 0.04). This expansion in CD71+ cells was entirely attributable to an increase in erythroid cells expressing GFP (C) (p =.04). Responses were not observed in myeloid (CD33+) (D) or B lymphoid (CD19+) lineages (E) or among CD34+ cells (F). Week 2 results reflect mean values from four CID-treated (+) and four non-CID-treated (–) mice; week 3 results indicate mean values from four CID-treated (+) and five non-CID-treated (–) mice. Error bars (B–F) denote standard errors of mean. p values were obtained using a t test. Results at week 3 were confirmed in three independent experiments. (G): Wright Giemsa stain of GFP+ CD71+ cells sorted from the marrow of a CID-treated mouse at week 3 (1,000x). (H): Fluorescent histograms of human erythroid cells from four CID-treated NOD-SCIDß2mnull mice (blue lines, right) and three untreated controls (red lines, left). The mean fluorescent intensity of the GFP+ population is more than two decades higher in CID-treated mice compared to controls. (I): Staining of a marrow section (top, x100 objective; bottom, x400) from a CID-treated mouse evaluated at week 3 shows localization of GFP expression in an island of normoblasts (i) and absence of GFP expression in a cluster of myeloid cells (ii). Staining was performed using a rabbit polyclonal antibody directed against GFP and a secondary antibody conjugated to horseradish peroxidase. Brownish stain indicates GFP positivity. (J): Immunostaining with an anti- globin antibody demonstrates that erythroid cells responding to CID express fetal hemoglobin. Green fluorescence denotes the presence of globin. Abbreviation: GFP, green fluorescent protein.

An initial study using transduced CD34⁺ cells confirmed previous reports showing that human erythroid engraftment declines precipitously beyond week 3 post-transplantation (data not shown) [16, 19], making it necessary to evaluate the effects of CID treatment within this time frame. Mice transplanted with transduced CD34⁺ cells were treated, beginning 1 week later, with daily injections of AP20187. CID-treated and control vehicle-treated mice were evaluated after 7 or 14 days of AP20187 (or control vehicle), 2–3 weeks post-transplant, respectively. Although no effect of CID exposure was evident after 7 days (week 2), the 14-day course of AP20187 (week 3) produced a significant response. The fraction of human erythroid (CD71⁺) cells relative to all nucleated cells in the femurs of CID-treated mice was 24.0%, compared to 6.6% in non-CID-treated controls, a 3.6-fold increase (Fig. 1B) (p < .04), whereas non-erythroid (CD45⁺ CD71^–) cells remained unchanged. This increase in erythroid (CD71⁺) cells was entirely attributable to an expansion of GFP⁺ cells, which were present at levels 12.7-fold higher than in non-CID-treated controls (p < .04) (Fig. 1C, 1G). In contrast, CID-treated mice exhibited no significant changes in GFP⁺ cells within the CD33⁺ (Fig. 1D), CD19⁺ (Fig. 1E), or CD34⁺ (Fig. 1F) subsets. Similar results were obtained in a second experiment, where GFP⁺ erythroid cells were present at levels 16.7-fold higher in four CID-treated mice than in four non-CID-treated controls (p = .03; data not shown). The red cells (CD45^–, CD71⁺) of CID-treated mice expressed GFP much more intensely than controls (Fig. 1H), arose in clusters of GFP⁺ erythroblasts in the marrow (Fig. 1I), and stained strongly for fetal globin (Fig. 1J). GFP⁺ human erythroid cells also increased in the spleens of CID-treated mice, although their frequency was much lower than in the marrow, ranging from 0.16% to 0.54% (data not shown).

CID Treatment Induces the Emergence of BFUe and CFU-Mix in the Spleen
Contrasting the expansion of morphologically identifiable erythroid cells in the marrow at week 3, marrow progenitors fell (Fig. 2A; experiments not shown), mirroring the cell culture results (Fig. 1A). In contrast, GFP⁺ BFUe and CFU-mix were consistently higher in the spleens of CID-treated mice compared with those of controls (Fig. 2A). The effect of CID treatment was especially dramatic among BFUe and CFU-mix capable of colony formation in the presence of erythropoietin (epo) alone (Fig. 2A). Epo-dependent BFUe or CFU-mix was never observed in the spleens of non-CID-treated mice. The vast majority of epo-dependent BFUe and CFU-mix were intensely GFP positive (96% in each of two experiments). The strict correlation between CID exposure and the appearance of epo-responsive BFUe and CFU-mix in the spleen defines these colony-formers as specific indicators of the CID response.

View larger version (25K): [in this window] [in a new window]   Figure 2. Differential effects of F36VMpl signaling on progenitors in the femur and spleen. Unfractionated mononuclear cells from the femurs and spleens of CID-treated (+) and non-CID-treated (–) mice were plated, and GFP-positive colonies were scored. Mice were analyzed 3 weeks post-transplant, following a 14-day course of AP20187 (or vehicle control). Photomicrographs show representative morphologies of GFP+ BFUe (top) and CFU-mix (bottom). Note spherical morphology of BFUe. (A): Total numbers of GFP+ BFUe and CFU-mix per femur in four CID-treated (+) and four non-CID-treated (–) mice. Results reflect mean values. Error bars depict standard errors of mean. Colony assays were performed in IL-3, SCF, GM-CSF, and erythropoietin (Epo) or Epo alone. Results using Epo, GM, IL-3, and SCF were confirmed in three independent experiments. Results using Epo alone were confirmed in two independent experiments. Results from only a single representative experiment are shown. (B): Linear amplification-mediated polymerase chain reaction (LAM-PCR) profiles of three epo-responsive spleen colonies (colonies 1–3) in each of four CID-treated mice. LAM-PCR was performed using three different restriction enzymes, Tsp509I (top panel), HaeIII (middle panel), and RsaI (bottom panel) as described in Materials and Methods. Arrows designate internal control bands (see Materials and Methods). Colonies 2 and 3 from mouse 10 show an identical pattern with all three enzymes, indicating common ancestry. Numbers in each lane show the estimate number of transgene insertions based on results from all three enzymes. Abbreviations: BM, unfractionated bone marrow cells from the same mice; Epo, erythropoietin; GM-CSF, granulocyte macrophage colony-stimulating factor; IL-3, interleukin 3; MW, molecular weight; SCF, stem cell factor.

More extensive clonal tracking was performed in a single CID-treated mouse (Fig. 3). A total of 55 spleen-derived colonies were examined, 24 from cultures containing IL-3, SCF, GM-CSF, and epo and 31 from cultures containing epo alone. Of 55 total colonies, 36 yielded evaluable LAM-PCR patterns.

View larger version (40K): [in this window] [in a new window]   Figure 3. Clonal expansion of progenitors in the spleen and apparent lack of relationship to the marrow red cell response. (A): Schematic depiction of clonal relationships among CID responsive progenitors in the spleen. Linker amplification-mediated-polymerase chain reaction (LAM-PCR) using Tsp509I was performed on 55 colonies from the spleen of one CID-treated mouse (Fig. 1, week 3) that was evaluated 21 days post-transplantation, immediately after completing a 14-day course of CID. Results show evaluable insertion patterns shared by two or more colonies. Red circles depict burst-forming units-erythroid (BFUe). Purple circles depict CFU-mix. Circles enclosed with solid lines portray progenitors identified in this experiment. Circles enclosed with dashed lines depict parental colony forming cells whose existence was not demonstrated, but is inferred. Top, LAM-PCR patterns shared by either three (clone 12) or two (clones 5 and 7) CFU-mix. Middle, LAM-PCR patterns shared by BFUe and CFU-mix. Three colonies derived from clone 9 were identified in cultures containing erythropoietin (epo) only (solid downward arrows), and two colonies were identified from cultures containing IL-3, GM-CSF, SCF, and epo (dashed arrows pointing to right). Bottom: LAM-PCR pattern shared by 2 BFUe. The inferred colony could be either a CFU-mix or a BFUe. (B): Tracking of insertions 9 (top panel) and 7 (middle panel) in selected colonies from the same mouse depicted in A and in sorted GFP+ erythroid (CD71+) and non-erythroid (CD71–) cells in the marrow. Numbers indicate the assigned LAM-PCR pattern from A and data not shown. X, colony that yielded a nonevaluable pattern by LAM-PCR. Bottom panel: GAB2 control primers show abundant DNA in wells from sorted marrow cells. (C): The spleen is not required for the CID-induced expansion of transduced human erythroid cells. Left panel: F36VMpl-transduced human CD34+ cells were transplanted into splenectomized or control NOD-SCID mice, followed 1 week later by a 5-day course of either CID (+) or vehicle only (–). The duration of CID treatment in this experiment was curtailed due to excessive deaths in both CID and vehicle-treated mice in the splenectomized group. After 9 days (3 weeks post-transplant), mice were sacrificed, and the frequency of GFP-positive cells among the CD71+ population was compared between the two groups. Responses obtained following the 5-day course of CID were similar to those of nonsplenectomized NOD-SCID mice treated with a 13-day course of CID (right panel). Each column represents mean values from three mice. Error bars depict standard errors of mean. Abbreviation: GFP, green fluorescent protein.

Distinct Ancestry of CID-Responsive Erythroid Cells in the Marrow and CID-Responsive BFUe and CFU-Mix in the Spleen
To gain further insight into a possible ancestral relationship between CFU-mix and BFUe generated in the spleens of CID-treated mice and differentiating erythroid cells in their marrows, we compared LAM-PCR banding patterns. A shared LAM-PCR pattern between spleen progenitors and differentiating marrow erythroid cells would have two potential explanations. First, BFUe and CFU-mix in the spleen might give rise to differentiated red cell progeny that migrate to the marrow. Second, if BFUe and CFU-mix appearing in the spleens of CID-treated mice had originated in the marrow, then traces of their ancestors or progeny may have been left behind.

We first looked in the bone marrows of the four mice depicted in Figure 2B. Although no shared LAM-PCR patterns were observed, marrow mononuclear cells (Fig. 2B, lanes labeled BM) gave rise to much less distinct patterns compared with the single colonies, consistent with the inability of LAM-PCR to accurately reflect the contributions of individual clones within a complex mixture (M.A. Harkey and C.A. Blau, manuscript submitted). To more accurately establish whether a progenitor/progeny relationship existed between CID-responsive progenitors in the spleen and erythroblasts in the marrow, we sequenced insertions from two clones (clones 7 and 9) from the mouse examined in Figure 3A. Specific primers to detect these insertions were used both to verify the accuracy of the LAM-PCR-designated clonal assignments and to determine whether clones 7 and 9 contributed to red cell progeny in the marrow of the same mouse. PCR products of the predicted size were detected in all colonies previously assigned to clones 7 and 9 (Fig. 3B). Clone 9 primers also detected minor bands of appropriate size in colonies previously assigned to clones 5 and 11; however, these low-intensity bands were likely the result of cross contamination during the colony picking procedure. Neither clone 7 nor clone 9 insertions were detectable in sorted GFP⁺ CD71⁺ cells from the marrow of the same mouse. These findings strongly suggest the absence of a clonal relationship between erythroid and mixed progenitors in the spleens and erythroblasts in the marrows of CID-treated mice. Unexpectedly, a faint band of the appropriate size was obtained from sorted nonerythroid (GFP⁺ CD71^–) cells using the clone 7-specific primers (Fig. 3B). Since both of the clone 7 progenitors identified in the spleen were CFU-mix (Fig. 3A), it is possible that nonerythroid progeny of these progenitors were able to escape the spleen and migrate to the marrow, whereas the erythroid progeny appeared not to traffic from the spleen to the marrow.

The Spleen Is Dispensable for the CID Response
To confirm that the CID-induced appearance of BFUe and CFU-mix in the spleen is not required for the emergence of differentiated erythroid cells in the marrow, we evaluated splenectomized mice. Due to the frailty of NOD-SCIDß2^null mice, we tested the effect of splenectomy on the F36VMpl response in NOD-SCID mice. Splenectomized mice tolerated the transplantation procedure much less well than their nonsplenectomized counterparts. Among 20 splenectomized mice, 10 mice (5 in each group) died between 9 and 16 days post-transplant. By comparison, 18 of 20 nonsplenectomized mice survived, with both deaths occurring among control-injected animals. To minimize further loss of mice, injections were curtailed to 5 days. Both splenectomized and nonsplenectomized mice were analyzed 3 weeks post-transplant, 9 days after completing CID or control injections. To test whether the shorted duration of CID treatment might reduce the magnitude of the response, a cohort of mice was treated with a 13-day course of CID. The percentage of CD71⁺ cells that expressed GFP trended higher in the marrows of CID-treated mice, and was not diminished by splenectomy (Fig. 3C). Among nonsplenectomized animals, responses to the 5- and 13-day courses of CID treatment were similar (Fig. 3C, right).

Evaluating the Persistence of the Response to F36VMpl Signaling
Mice treated with CID between weeks 1 and 3 showed a dramatic decline in GFP⁺ erythroid cells by week 8, falling from up to 50% of all nucleated cells in the femur at week 3 to <1% by week 8 (Fig. 4A). Nonetheless, the fraction of erythroid cells expressing GFP remained significantly higher at week 8 in the marrows (but not the spleens) of mice that had received CID between weeks 1 and 3 (p < .001) (Fig. 4B). Additionally, an effect of CID exposure on progenitors in the spleen remained discernible by week 8, with the persistence of a small population of GFP⁺ epo-dependent BFUe and CFU-mix (Fig. 4C). Not only were these progenitors present at levels significantly lower than at week 3, but the relative proportion of BFUe to CFU-mix was inverted, from 1:3 at week 3 (Fig. 2A) to 9:1 by week 8 (Fig. 4C), suggesting that the CID-responsive CFU-mix may have differentiated into BFUe. Persistent effects of CID treatment were also not evident in secondary recipients transplanted either immediately following CID exposure (at week 3) or at week 8 (not shown).

View larger version (31K): [in this window] [in a new window]   Figure 4. Persistence of the F36VMpl response. (A): GFP+71+ cells as a percentage of total mononuclear cells in the femurs of CID-treated (+) and non-CID-treated (–) mice at week 3 and in CID-treated mice at week 8. Despite a marked expansion of GFP+ erythroid cells in CID-treated mice at week 3, these cells are no longer evident in the marrow by week 8. (B–D): Persistence of CID effects at week 8. A 14-day course of CID (+) or vehicle (–) had been given between weeks 1 and 3. (B): The percentage of CD71+ cells expressing GFP at week 8 post-transplantation was higher in the femurs (white bars) but not in the spleens (gray bars) of mice that had received CID. (C): Erythroid and mixed progenitors capable of forming colonies in the presence of epo alone remained higher in the spleens of mice that had received CID. Hatched boxes indicate GFP+ colonies. (D): The percentage of GFP+ BFUe was higher in the spleens of mice that received a 14-day course of CID between weeks 1 and 3 post-transplant (±) or a 3-day course of CID given at week 7 (–/+) compared to controls (–/–). Differences in the percentages of GFP-positive BFUe in the spleens of CID-treated mice in either the +/– or the –/+ groups are significantly different from the –/– group. (E): From the same experiment depicted D, CID treatment either between weeks 1 and 3 or at week 7 had no effect on the percentage of GFP+ CD19+ B cells at week 8. (D, E): Error bars denote standard deviations for +/– group and show individual values for –/– and –/+ groups. Numbers indicate mice per group. Abbreviations: Epo, erythropoietin; GFP, green fluorescent protein.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
Our results demonstrate that human hematopoiesis can be regulated in vivo using a conditionally active growth factor receptor. Similar approaches may be useful in any situation where it is desirable to retain control over transplanted cells [1]. Despite its panhematopoietic expression pattern, F36VMpl signaling in human hematopoietic cells induces proliferation that appears restricted to transduced erythroblasts and their progenitors. Flow histograms document that CID treatment induces the preferential expansion of cells expressing high levels of the linked GFP marker, a potentially advantageous feature in the context of a linked therapeutic gene.

In a previous study, we found that baboons exhibited a very modest hematopoietic response to F36VMpl signaling both in vitro and in vivo [7]. By comparison, human hematopoietic cells exhibit a significant response to F36VMpl signaling both in vitro [7, 10, 12] and, as shown here, in vivo. The narrow time window permissive for human erythroid engraftment in the NOD-SCID-ß2^null model necessitated that we focus on the first 3 weeks post-transplant. The dramatic decline in GFP⁺ erythroid cells by week 8 is highly likely to be an artifact of this model. These cells did not enter the circulation (data not shown), and thus their disappearance can be explained only by accelerated destruction. Whether this destruction is attributable to residual immune competency in the NOD/SCID/ß2M^null mice or is related to the apparent inability to complete erythroid differentiation is not known. In this regard, it is noteworthy that ex vivo-expanded human cord blood cells have been shown to terminally differentiate upon tail vein injection into NOD-SCID mice [20]. We therefore hypothesize that human erythroblasts are unable to transit from the marrow or spleen to the blood, and this inability may be causally related to their shortened survival. This interpretation is supported by the inability of transduced human erythroblasts, generated in the spleens of CID-treated mice, to appear in the marrow (Figs. 2B, 3B). Of note, CID-responsive erythroid cells persist for more than 60 days in murine models [3, 21, 22] and 90 days in canine models [4] of F36VMpl-mediated in vivo selection, time courses commensurate with red cell survivals in these species. Additionally, we anticipate that in a clinical setting, repeated cycles of CID administration would lead to ongoing red cell responsiveness, as suggested by the persistence of GFP⁺ colony-forming cells at week 8 in mice treated with CID at weeks 2 and 3 post-transplantation (Fig. 4C, 4D) and by the presence of ongoing CID responsiveness in our previously published studies in mice [3] and dogs [4]. Although it remains possible that human hematopoietic cells may behave differently than described here in humans, this can only be definitively ascertained in a clinical trial.

The responses of hematopoietic stem cells (HSCs) and progenitors to mitotic signaling can be significantly modulated by environmental factors. A case in point is chronic myelogenous leukemia (CML), where bcr-abl-expressing HSCs dominate hematopoiesis in vivo but are rapidly out-competed by normal progenitors in suspension culture [23]. A second example occurs with c-kit ligand, which alone is unable to support HSC expansion in cell culture [24] but can support a modest HSC expansion upon in vivo administration [25].

Controversy exists as to whether the anatomical compartment in which a HSC or progenitor resides can influence its response to a mitogenic stimulus. It is not known whether adult human peripheral blood HSCs and progenitors inherently differ from bone marrow HSCs in their capacity for engraftment and repopulation. This question has been especially difficult to address in humans due to the confounding influence of growth factors, which are uniformly administered prior to collection of peripheral blood HSCs and are rarely used prior to marrow HSC collection [reviewed in ref. 26]. One study found that progenitors collected from the blood of patients with CML following "chemo-mobilization" were less likely to contain the Philadelphia (Ph) chromosome than progenitors from the bone marrow [27], suggesting an inherent biological difference between progenitors in these compartments.

Results presented here suggest that the compartment in which a human hematopoietic progenitor resides can influence its response to a mitotic stimulus. F36VMpl signaling induced a consistent decline in marrow BFUe and CFU-mix and a consistent rise in spleen BFUe and CFU-mix. Although we cannot exclude the possibility that BFUe and CFU-mix trafficked from the marrow to the spleen in response to F36VMpl signaling, this appears highly unlikely based on 1) our consistent inability to detect circulating progenitors in the blood in response to CID treatment (data not shown); 2) tracking studies that demonstrate clonal relationships among CFU-mix, BFUe, and between CFU-mix and BFUe in the spleen (Fig. 3A); 3) the inability of tracking studies to find any clonal relationships between spleen CFU-mix/BFUe and marrow erythroid cells. Had a progenitor been induced to travel from the marrow to the spleen in response to CID treatment, it seems reasonable to expect some trace of its former presence in the marrow.

Taken together, our findings are most consistent with the interpretation that the response of human hematopoietic progenitors to a mitotic signal can be influenced by the anatomical compartment in which they reside. Although we have not excluded the possibility of differences in AP20187 uptake, retention, or clearance between marrow and spleen, an environmental influence over the F36VMpl response is in keeping with our previous findings that growth factors can modulate the F36VMpl response in cultured human cord blood CD34⁺ cells [12] and that responses in vitro can differ markedly from those observed in vivo [3, 4, 11].

Although the apparent epo-dependence of the CID-responsive spleen progenitors provided a means to examine the clonality of the CID response, this growth pattern is not characteristic of BFUe or CFU-mix. We have previously shown that F36VMpl-transduced human CD34⁺ cells can form small BFUe in the presence of CID alone and that CID-dependent colony formation is augmented, both in colony size and colony number, by the addition of growth factors, including epo [10]. Furthermore, we have shown that cell growth induced by CID treatment can last for several days following CID withdrawal, presumably due to intracellular retention of the drug [28]. We therefore conclude that this unusual pattern of colony growth likely reflects the combined effect of CID that is retained within the colony-forming cell at the time of plating and epo added to the culture medium.

Our results show that, analogous to the effect of epo on red blood cells, CIDs can act as growth factors for F36VMpl-transduced red blood cells. Although the need for repeated CID treatments poses an encumbrance and entails an additional risk of toxicity due to the need for chronic intermittent drug exposure, reversibility also has considerable conceptual appeal in view of the risk of leukemia associated with gene insertion [29].

	ACKNOWLEDGMENTS

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
We acknowledge the assistance of Sue Chacona with cord blood collections, and expert technical assistance of Amanda Melbye, Betty Mastropaolo, Marilyn Skelly, Allen Gown, and Greg Levin. This work was supported by NIH grants DK52997, DK61844, and HL53750.

	DISCLOSURES

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 
C.A.B. is named as an inventor on a patent application related to the work that is described in the present study that is owned by the University of Washington and is licensed to ARIAD Pharmaceuticals.

	REFERENCES

Top Abstract Introduction Materials and Methods Results Discussion Disclosures References

 

1. Neff T, Blau CA. Pharmacologically regulated cell therapy. Blood 2001; 97:2535–2540.[Free Full Text]

2. Spencer DM, Wandless TJ, Schreiber SL et al. Controlling signal transduction with synthetic ligands. Science 1993;262:1019–1024.[Abstract/Free Full Text]

3. Jin L, Zeng H, Chien S et al. In vivo selection using a cell growth switch. Nat Genet 2000;26:64–66.[CrossRef][Medline]

4. Neff T, Horn PA, Valli VE et al. Pharmacologically regulated in vivo selection in a large animal. Blood 2002;100:2026–2031.[Abstract/Free Full Text]

5. Hacein-Bey-Abina S, Le Deist F, Carlier F et al. Sustained correction of X-linked severe combined immunodeficiency by ex vivo gene therapy. N Engl J Med 2002;346:1185–1193.[Abstract/Free Full Text]

6. Aiuti A, Slavin S, Aker M et al. Correction of ADA-SCID by stem cell gene therapy combined with nonmyeloablative conditioning. Science 2002;296:2410–2413.[Abstract/Free Full Text]

7. Richard RE, De Claro RA, Yan J et al. Differences in F36VMpl-based in vivo selection among large animal models. Mol Ther 2004;10:730–740.[CrossRef][Medline]

8. Gao Z, Golob J, Tanavde VM et al. High levels of transgene expression following transduction of long-term NOD/SCID-repopulating human cells with a modified lentiviral vector. STEM CELLS 2001; 19:247–259.[Abstract/Free Full Text]

9. Kaneko S, Nagasawa T, Nakauchi H et al. An in vivo assay for retrovirally transduced human peripheral T lymphocytes using nonobese diabetic/severe combined immunodeficiency mice. Exp Hematol 2005;33: 35–41.[CrossRef][Medline]

10. Richard RE, Wood B, Zeng H et al. Expansion of genetically modified primary human hemopoietic cells using chemical inducers of dimerization. Blood 2000;95:430–436.[Abstract/Free Full Text]

11. Jin L, Siritanaratkul N, Emery DW et al. Targeted expansion of genetically modified bone marrow cells. Proc Natl Acad Sci U S A 1998;95: 8093–8097.[Abstract/Free Full Text]

12. Richard RE, Blau CA. Small-molecule-directed mpl signaling can complement growth factors to selectively expand genetically modified cord blood cells. STEM CELLS 2003;21:71–78.[Abstract/Free Full Text]

13. Follenzi A, Ailles LE, Bakovic S et al. Gene transfer by lentiviral vectors is limited by nuclear translocation and rescued by HIV-1 pol sequences. Nat Genet 2000;25:217–222.[CrossRef][Medline]

14. Ailles L, Schmidt M, Santoni de Sio FR et al. Molecular evidence of lentiviral vector-mediated gene transfer into human self-renewing, multi-potent, long-term NOD/SCID repopulating hematopoietic cells. Mol Ther 2002;6:615–626.[CrossRef][Medline]

15. Dull T, Zufferey R, Kelly M et al. A third-generation lentivirus vector with a conditional packaging system. J Virol 1998;72:8463–8471.[Abstract/Free Full Text]

16. Glimm H, Eisterer W, Lee K et al. Previously undetected human hematopoietic cell populations with short-term repopulating activity selectively engraft NOD/SCID-beta2 microglobulin-null mice. J Clin Invest 2001;107:199–206.[Medline]

17. Stamatoyannopoulos G, Farquhar M, Lindsley D et al. Monoclonal antibodies specific for globin chains. Blood 1983;61(3):530–539.[Abstract/Free Full Text]

18. Mazurier F, Gan OI, McKenzie JL et al. Lentivector-mediated clonal tracking reveals intrinsic heterogeneity in the human hematopoietic stem cell compartment and culture-induced stem cell impairment. Blood 2004; 103:545–552.[Abstract/Free Full Text]

19. Imren S, Fabry ME, Westerman KA et al. High-level beta-globin expression and preferred intragenic integration after lentiviral transduction of human cord blood stem cells. J Clin Invest 2004;114:953–962.[CrossRef][Medline]

20. Neildez-Nguyen TM, Wajcman H, Marden MC et al. Human erythroid cells produced ex vivo at large scale differentiate into red blood cells in vivo. Nat Biotechnol 2002;20:467–472.[CrossRef][Medline]

21. Richard RE, Weinreich M, Chang KH et al. Modulating erythrocyte chimerism in a mouse model of pyruvate kinase deficiency. Blood 2004;103:4432–4439.[Abstract/Free Full Text]

22. Emery DW, Tubb J, Nishino Y et al. Selection with a regulated cell growth switch increases the likelihood of expression for a linked gamma-globin gene. Blood Cells Mol Dis 2005;34:235–247.[CrossRef][Medline]

23. Petzer AL, Eaves CJ, Barnett MJ et al. Selective expansion of primitive normal hematopoietic cells in cytokine-supplemented cultures of purified cells from patients with chronic myeloid leukemia. Blood 1997;90:64–69.[Abstract/Free Full Text]

24. Broudy VC. Stem cell factor and hematopoiesis. Blood 1997;90:1345–1364.[Free Full Text]

25. Bodine DM, Seidel NE, Zsebo KM et al. In vivo administration of stem cell factor to mice increases the absolute number of pluripotent hematopoietic stem cells. Blood 1993;82:445–455.[Abstract/Free Full Text]

26. Elfenbein GJ, Sackstein R. Primed marrow for autologous and allogeneic transplantation: A review comparing primed marrow to mobilized blood and steady-state marrow. Exp Hematol 2004;32:327–339.[CrossRef][Medline]

27. Verfaillie CM, Bhatia R, Steinbuch M et al. Comparative analysis of autografting in chronic myelogenous leukemia: Effects of priming regimen and marrow or blood origin of stem cells. Blood 1998;92:1820–1831.[Abstract/Free Full Text]

28. Blau CA, Peterson KR, Drachman JG et al. A proliferation switch for genetically modified cells. Proc Natl Acad Sci U S A 1997;94:3076–3081.[Abstract/Free Full Text]

29. Hacein-Bey-Abina S, von Kalle C, Schmidt M et al. LMO2-associated clonal T cell proliferation in two patients after gene therapy for SCID-X1. Science 2003;302:415–419.[Abstract/Free Full Text]

This article has been cited by other articles:

				H. Abdel-Azim, Y. Zhu, R. Hollis, X. Wang, S. Ge, Q.-L. Hao, G. Smbatyan, D. B. Kohn, M. Rosol, and G. M. Crooks Expansion of multipotent and lymphoid-committed human progenitors through intracellular dimerization of Mpl Blood, April 15, 2008; 111(8): 4064 - 4074. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 2005-0484v1 24/4/908    most recent Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints/Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Nagasawa, Y. Articles by Blau, C. A. Search for Related Content PubMed PubMed Citation Articles by Nagasawa, Y. Articles by Blau, C. A.