Hematopoietic stem cell (HSC) therapy for myocardial repair is limited by the number of stem cells that migrate to, engraft in, and proliferate at sites of injured myocardium. To alleviate this limitation, we studied whether a strategy using a bispecific antibody (BiAb) could target human stem cells specifically to injured myocardium and preserve myocardial function. Using a xenogeneic rat model whereby ischemic injury was induced by transient ligation of the left anterior descending artery (LAD), we determined the ability of a bispecific antibody to target human CD34+ cells to specific antigens expressed in ischemic injured myocardium. A bispecific antibody comprising an anti-CD45 antibody recognizing the common leukocyte antigen found on HSCs and an antibody recognizing myosin light chain, an organ-specific injury antigen expressed by infarcted myocardium, was prepared by chemical conjugation. CD34+ cells armed and unarmed with this BiAb were injected intravenously in rats 2 days postmyocardial injury. Immunohistochemistry studies showed that the armed CD34+ cells specifically localized to the infarcted region of the heart, colocalized with troponin T-stained cells, and colocalization with vascular structures. Compared to unarmed CD34+ cells, the bispecific antibody improved delivery of the stem cells to injured myocardium, and such targeted delivery was correlated with improved myocardial function 5 weeks after infarction (p < .01). Bispecific antibody targeting offers a unique means to improve the delivery of stem cells to facilitate organ repair and a tool to study stem cell biology.
Recent advances in the field of cellular cardiomyoplasty have generated enthusiasm for the prospects of stem cell therapy for myocardial regeneration. The plasticity of hematopoietic stem cells (HSCs) and their ability to regenerate nonhematopoietic tissues [1, , , , , , , , –10] provides a means to repair injured cardiac tissue. However, the use of HSCs for myocardial repair remains controversial [9, 11, , , , , , –18]. This suggests that the rate and type of cellular regeneration in the injured myocardium may depend not only on the ability of HSC to transdifferentiate into myocytes or endothelial cells, but also on environmental variables that play essential roles in creating conditions conductive to HSC-mediated repair. Even when the local milieu provides appropriate signals for transdifferentiation, proliferation, or fusion, successful HSC-mediated cardiac regenerative therapy may be limited by inadequate delivery and/or lack of persistence of high numbers of HSCs at the site of injury. If HSC transplantation is to provide a viable clinical approach for the treatment of myocardial injury, new approaches are needed to direct large numbers of HSCs to the infarct site, while avoiding invasive procedures that increase risks for morbidity or mortality.
We report here the results of a new strategy using a bispecific antibody (BiAb) to target systemically administered purified human CD34+ cells to infarcted regions of nude rat myocardium. BiAbs have been used for several decades in oncology to redirect cytotoxic effector cells to kill tumor cells. BiAbs are produced by chemically cross-linking the F(ab′) fragments of two monoclonal antibodies (mAbs) and used to form a bidirectional antibody bridge between the cytotoxic effector cells and tumor cells. The construct contains one mAb directed at effector cells and another mAb directed at the tumor-associated antigen. Retargeting effector cells with BiAbs would, therefore, not only redirect the nonspecific cytotoxicity of effector cells of the immune system, but would also improve trafficking to the tumor. We have adapted a similar approach to target HSCs to injured myocardium. In a mouse model of myocardial infarction (MI), lin-sca+ hematopoietic stem cells were successfully directed to the ischemic injured myocardium . To further explore the utility of BiAb targeting of HSCs for myocardial repair, a BiAb combination of antihuman common leukocyte antigen (CD45) and antirat myosin light chain (MLC) mAbs was constructed. MLC was selected because it would target myocardial infarctions in an organ-specific manner [20, –22]. Arming CD34+ cells with the engineered BiAb not only increased the number of CD34+ cells that homed to infarcted tissue, but also improved the function of the affected left ventricular tissue.
Production of BiAbs
Antihuman CD45 (BD Pharmingen, San Diego, http://www.bdbiosciences.com/index_us.shtml) was cross-linked with Traut's reagent (2-iminothiolane HCl; Pierce, Rockford, IL, http://www.piercenet.com) and anti-MLC (myosin light chain 1 antibody [MLM508]; Abcam, Cambridge, MA, http://www.abcam.com) was cross-linked with SulfoSMCC [sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate; Pierce] as described . Anti-human CD45 was chosen to bind to not only CD34+ cells, but potentially other important progenitor cells contained in white blood cells. MLCs are expressed in myocardial tissue, but become accessible to intravascular anti-MLC Ab only during myocardial injury, when the surface membrane is no longer intact . Therefore, MLC is an injury-specific antigen of the myocardium. The cross-linked antibodies were mixed and allowed to heteroconjugate overnight to produce anti-CD45 × anti-MLC (CD45 × MLC). The proportion of dimers, multimers, and monomers were determined by nonreducing SDS-polyacrylamide gel electrophoresis (PAGE) and the concentration of the entire mix determined. The arming dose of CD45 × MLC was 50 ng per million purified CD34+ cells for 1 hour at 4°C. The mix of monomers, dimers, or multimers was not purified because, in other studies using T cells armed with the same dose of purified dimers or multimers of BiAb did not differ from the mix in cytotoxicity directed at target cells expressing the specific antigen (unpublished data). Cytotoxicity mediated by BiAb-targeted T cells is specific to the targets and not to targets that do not express the antigen. For example, anti-CD3 × anti-CD20 armed T cells kill CD20+ targets and not HER2+ breast cancer cell lines; and anti-CD3 × anti-Her2/neu+ armed T cells kill HER2/neu+ targets and not CD20+ B-cell lines [23, 24].
Unused, excess granulocyte colony-stimulating factor (G-CSF)-primed peripheral-blood mononuclear cells (PBMCs) were obtained by leukapheresing normal human stem cell donations. All blood collection and use of human blood products for research were conducted under internal review board-approved protocols at Roger Williams Hospital (Providence, RI), and signed consents were obtained from the donors. CD34+ purified cells were obtained by positive selection over a clinical Isolex 300i column and the CD34+ selection kit (Baxter, Deerfield, IL, http://www.baxter.com). The purified CD34+ cells were frozen in convenient aliquots and thawed, washed, and armed with BiAb CD45 × MLC and infused within 24 hours of arming.
Impact of BiAb CD45 × MLC Arming on the Growth and Differentiation of CD34+ Cells
Isolex-purified CD34+ cells were armed with titrating doses of CD45 × MLC (0–500 ng per 106 CD34+ cells) by incubating cells with CD45 × MLC for 15 minutes at room temperature. Cells were washed free of unbound BiAb, resuspended in Iscove's modified Dulbecco's medium (IMDM)/2% fetal calf serum (FCS), added to Complete Methocult medium (Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) containing recombinant human stem cell factor, recombinant human granulocyte macrophage colony-stimulating factor, recombinant human interleukin-3, and recombinant human erythropoietin, and seeded into 35-mm culture dishes (2,000 cells per plate). Cultures were incubated for 14 days at 37°C in a humidified atmosphere of 5% CO2, 95% air. Colonies were classified and enumerated on the basis of the morphological recognition of one or more types of hematopoietic lineage cells within the colony.
Myocardial Injury Model
The study protocol was approved by the Committee for Animal Research of the University of California San Francisco (San Francisco, CA) and was performed in accordance with the recommendations of the American Association for Accreditation of Laboratory Animal Care. A previously described ischemia reperfusion model was used in this study [25, 26]. Nude rats (225–250 g) were endotracheally intubated, ventilated with a rodent ventilator (Harvard Apparatus, Holliston, MA, http://www.harvardapparatus.com), and anesthetized with inhalational isofluorane. A 7-0 Ticron suture was placed around the left anterior descending portion (LAD) of the left coronary artery. The suture was tightened to occlude the LAD for 17 minutes, and then removed to allow for reperfusion. The sternotomy was then closed, and the animal was allowed to recover. Experience with this model has previously demonstrated that this technique results in an acute infarct size of at least 30% of the left ventricle (LV) with reperfusion [27, –29] and the negative remodeling of the LV results in a continual decline in LV function for at least 5 weeks .
Intravenous Administration of Hematopoietic Stem Cells
Two days after myocardial infarction, either CD45 × MLC-armed CD34+ cells (2 × 106; n = 9) or unarmed CD34+ cells (2 × 106; n = 8) were injected intravenously via the right internal jugular vein. Animals receiving armed and unarmed CD34+ cells were sacrificed 5 weeks after the injection of cells. In a separate control study, either CD45 × MLC alone in phosphate-buffered saline (PBS; n = 9) or PBS (n = 8) was administered intravenously two days after myocardial infarction.
Transthoracic echocardiography was performed on all animals receiving CD34+ cells in the conscious state before MI; and 12 and 35 days after the MI. The study was concluded 5 weeks after infarction, at which point the remodeling process in the rat is essentially complete . The echocardiographer was blinded to the treatment groups during the acquisition of the images and the data analysis. The methodology of echocardiography used in this study has been previously described , and other reports have demonstrated the accuracy and reproducibility of transthoracic echocardiography in rats with myocardial infarcts [32, –34].
Immunofluorescence staining and infarct size was performed as previously described [35, 36], 5 weeks after the intravenous (i.v.) administration of cells. Five sections equally distributed through the infarct area were double labeled with mouse anti-human human leukocyte antigen (HLA) class I (BD Biosciences, San Diego, http://www.bdbiosciences.com/; 1:200 dilution) and either rabbit anti-smooth muscle α-actin (Lab Vision, Fremont, CA, http://www.labvision.com; 1:300 dilution) or rabbit anti-troponin T (Abcam; 1:300 dilution). To visualize labeled cells, slides were incubated with secondary antibodies anti-mouse Alexa 488 (Molecular Probes, Eugene, OR, http://probes.invitrogen.com; 1:500 dilution) and anti-rabbit rhodamine (Molecular Probes, 1:100 dilution). Nuclei were stained using TO-PRO-3 iodide (Molecular Probes, 1:2000). Confocal microscopy (Leica TCS SL; Leica, http://www.leica.com/) and immunofluorescent microscopy (Nikon Eclipse E800; Nikon, Tokyo, http://www.nikon.com) were used to evaluate the immunohistochemical staining. Total number of nucleated HLA class I positively stained cells and total number of HLA and troponin T double-labeled cells were determined by counting 10 high-powered fields within each section. Planimetry was used to trace the infarct and LV. Infarct scar size was derived by dividing the infarct scar area by the total LV area as measured with SPOT 3.5.1 software (Diagnostic Instruments, Sterling Heights, MI, http://www.diaginc.com). Infarct scar size was recorded as a percentage of the LV.
Statistical analyses were performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, http://www.graphpad.com/). Results from cell density measurements were compared using the unpaired t test. Data are presented as mean ± SD. Echocardiography measurements and infarct size were compared using Student's t test.
Production of Anti-CD45 × Anti-MLC BiAb
The BiAb anti-CD45 × anti-MLC was produced by chemical heteroconjugation, as shown in Figure 1A and as described in the methods . The preparation consisted of 12% conjugated dimers, 66% unconjugated monomers, and 22% multimers, as shown by the SDS-PAGE (Fig. 1B, inset). Our prior experience with BiAbs in preclinical and clinical studies has shown no significant difference in binding and specificity between the purified and unpurified heteroconjugates. Therefore, we used the unpurified heteroconjugated material for arming the CD34+ cells. Binding of the BiAb to CD34+ cells via its anti-CD45 antigen was demonstrated using a goat anti-mouse IgG2a phycoerythrin (PE)-conjugated antibody that recognized the mouse IgG2a anti-MLC portion of the CD45 × MLC (Fig. 1B).
Binding of CD45 × MLC to Human CD34+ Cells Does Not Alter In Vitro Hematopoietic Colony-Forming Assays
To determine whether binding of CD45 × MLC induced functional or altered differentiation of human CD34+, purified CD34+ cells were armed or left unarmed with CD45 × MLC. No significant difference in proliferation (p = .16) or differentiation (p = .43 for blast-forming unit [BFU]-E; p = .09 for colony-forming unit [CFU]-GM) was observed between unarmed CD34+ cells and cells armed with CD45 × MLC (0–500 ng/106 cells). Other colony types, CFU-GEMM and CFU-E, on average, represented 5% or less of total colonies. There were no significant changes induced by binding of the bispecific antibody to CD45 on the stem cells.
CD45 × MLC-Armed CD34+ Cells Persist in MIs and Coexpress Human HLA Class I and Muscle-Specific Antigens 5 Weeks After Infusion
Enriched CD34+ cells from G-CSF primed normal peripheral blood mononuclear cells were used to test whether CD34+ cells armed with CD45 × MLC could be targeted to MIs and preserve left-ventricular function. By flow cytometric phenotyping, CD34+ cells comprised 0.5% of the PBMC population. Isolex 300i-purified CD34+ cells contained 99% CD34+, 99% CD45+, 99% CD38+, 96% CD117+, 87% CD133+, and 70% CD33+ cells by flow cytometry. Flow cytometry also showed that the purified CD34+ cells were negative for CD4, CD7, CD10, CD16, CD19, CD20, and CD23 antigens. Two days after myocardial infarctions, rats received either 2 × 106 CD45 × MLC-armed CD34+ cells or unarmed CD34+ cells intravenously.
Five weeks after the infusion of CD45 × MLC-armed CD34+ cells and unarmed cells, HLA class I positive armed cells significantly outnumbered unarmed CD34+ HLA class I positive cells in the infarcted region (171.8 ± 52.7 HLA class I + cells/hpf vs. < 1.0 cells/hpf, p < .001). Human cells bearing HLA class I markers were distributed throughout the infarct region. Two-color staining for human HLA class I and troponin T (Fig. 2) detected double-stained cells in infarcts of rats treated with the CD45 × MLC-armed CD34+ cells. The HLA class I and troponin T double-stained cells accounted for less than 2.5% of the total HLA class I positive cells. The infarcts of rats that received unarmed CD34+ cells showed only rare human HLA class I and troponin T double-positive cells (Fig. 2H). Colocalization of staining for α-smooth muscle actin and HLA class I was also observed in MIs of rats 5 weeks after infusion of CD45 × MLC-armed CD34+ cells (Fig. 3).
Improved Left Ventricular Function in Rats Treated with CD45 × MLC-Armed CD34+ Cells
Sequential echocardiograms of rats that received CD45 × MLC-armed CD34+ cells showed significantly better cardiac function compared to rats that received unarmed CD34+ cells (Fig. 4). By 5 weeks, the fractional shortening (FS) of the unarmed group (0.23 ± 0.08; n = 8) decreased significantly (p < .01) compared to the CD45 × MLC-armed group (0.34 ± 0.06; n = 9), whereas LV systolic diameter (SD) of the unarmed group (0.60 ± 0.04) was significantly (p < .05) dilated compared to the CD45 × MLC-armed group (0.49 ± 0.12). Systolic anterior wall thickness (AWTs) of the unarmed group (0.10 ± 0.03) was significantly (p < .05) thinner than the AWTs of the CD45 × MLC-armed group (0.15 ± 0.06), whereas the posterior wall had a compensatory increase in the diastolic posterior wall diameter thickness (PWDd) in the unarmed group as compared to the PWDd of the CD45 × MLC-armed group (0.20 ± 0.02 vs. 0.17 ± 0.02; p < .05). There was no significant difference in fractional shortening, wall thickness, or chamber size in the control study comparing i.v. administration of the bispecific antibody alone versus a saline control.
There was no statistical difference in infarct size between the armed and unarmed groups, indicating that the preservation of cardiac function resulted from the treatment with armed CD34+ cells rather than a discrepancy in infarct size.
This investigation reports a new approach to target human HSC to areas of myocardial injury. A BiAb that targets a cardiac-specific injury antigen enhances the homing of intravenously administered CD34+ cells specifically to MIs. The BiAb strategy creates an “artificial” but organ-specific injury homing “molecule” on the surface of CD34+ cells. By arming the CD34+ cell population with a BiAb, we combined immunologic, hematopoietic, and cardiac strategies to bypass ligand-receptor interaction requirements for HSC homing. We reasoned that, by delivering as many CD34+ cells to the site of injury as possible, we could optimize the conditions for CD34+ cell engraftment and repair. The use of xenogeneic human cells in immunodeficient rats facilitated the tracking of human HSCs without the need to genetically modify the cells. Genetic modification of cells with markers or reporter genes has been observed to increase immunogenicity leading to rejection of stem cells in recipients .
The mechanism for HSC homing to injured organs is unknown and presents a challenge for achieving tissue repair. The inability of intravenously administered CD34+ cells to traffic to infarcted myocardial tissue  is likely dependent upon the expression of injury factors as stromal cell-derived factor-1 (SDF-1) , which binds HSC via the CXCR receptor expressed on HSCs . The expression of HSC homing factors or binding of the HSC receptor to the tissue ligand during myocardial injury is insufficient for providing clinically significant HSC homing for myocardial repair. In order to overcome the lack of homing by i.v.-administered HSC, investigators have used direct intramyocardial injection, and intracoronary injection of unseparated bone marrow mononuclear cells or purified stem cells and G-CSF to improve stem cell mobilization and trafficking [40, , , , , , –47]. Despite these direct-injection strategies, inconsistent responses have been reported for the fate and function of transplanted HSC [13, 14, 48], possibly resulting from the nonspecific distribution, which could potentially exaggerate lack of engraftment of stem cells . Retention of the directly injected cells may still be problematic because of washout of the cells in the absence of specific binding or targeting.
The rare event of unarmed human cells expressing muscle antigens corroborates a previous study involving the i.v. infusion of human CD34+ cells into severe combined immunodeficiency (SCID) mice  and demonstrates the increased homing of BiAb armed CD34+cells to the injured myocardium. Rats infused with armed purified human CD34+ cells were detected at a significantly greater frequency than in rats infused with unarmed cells. These results demonstrate that arming of the CD34+ cells with the BiAb CD45 × MLC significantly enhances targeting of the CD34+ cells to the injured myocardium. The targeted CD34+ cells engraft at the site of injury and persist, and some of the human CD34+ cells adopt myocyte-like morphology, whereas others localize to vascular structures and may contribute to arteriole formation. These observations are consistent with previous studies in which human CD34+ cells injected into the myocardial infarction in SCID mice contribute to the formation of myocardium and vascular structures in the heart by both transdifferentiation and fusion . The double-labeled cells in the armed CD34+ animals were distributed throughout the infarct region, as expected due to the surface membrane of the injured myocardial tissues' no longer being intact, allowing MLCs to become accessible to intravascular anti-MLC BiAb [21, 22].
Cardiac function was significantly better in rats that received BiAb-armed CD34+ cells than in rats that received unarmed CD34+ cells. Since the percentage of HLA class I positive cells demonstrating HLA class I and troponin T staining was low, the improvement of cardiac function cannot be attributed to myogenesis. The functional benefits seen in the armed CD34+ group could result from the prevention of the negative remodeling associated with ischemic myocardial injury . By specifically targeting the injured myocardium, the increased numbers of transplanted cells may have retarded negative remodeling by preventing the slippage of myocytes after an MI [51, 52]. The transplanted HSC may provide LV wall support by increasing stiffness, or may affect remodeling by increasing wall thickness. Additionally, the beneficial effects seen in this study and published clinical studies may also result from angiogenesis or paracrine effects of the transplanted cells, inducing myocardial regeneration [53, 54].
This study focused on the ability of the BiAb CD45 × MLC to target CD34+ cells to infracted tissue and augment myocardial function. However, upon injury, MLC is released from the cytoplasm of myocardial cells into the circulation . Therefore, accumulation of MLC in organs other than the heart is possible. Future studies are needed to assess CD34+ cell trafficking, homing efficacy, and fate.
In summary, our results show that antibody targeting of CD34+ cells to ischemic-injured myocardium can be achieved with high specificity, resulting in increased HSC homing and increased numbers of cells that develop into myocyte-appearing cells and localization to vascular structures. In addition to providing a potentially nontoxic, noninvasive systemic delivery system for the repair of cardiac or other organ-related injuries, BiAb targeting of stem cells offers a novel tool for investigating stem cell biology.
The authors indicate no potential conflicts of interest.
This work was supported in part by funds from R01 CA 092,334 (L.G.L.), National Institutes of Health (NIH) New Stem Cell Biology (COBRE) 5P20RR18757 (L.G.L.), and NIH Perinatal Biology (COBRE) 1P20RR018728-01 (J.F.P.).