From www.bloodjournal.org by guest on February 6, 2015. For personal use only. Coblockade of the LFA1:ICAM and CD28/CTLA4:B7 Pathways Is a Highly Effective Means of Preventing Acute Lethal Graft-Versus-Host Disease Induced by Fully Major Histocompatibility Complex-Disparate Donor Grafts By Bruce R. Blazar, Patricia A. Taylor, Angela Panoskaltsis-Mortari, Gary S. Gray, and Daniel A. Vallera We have developed an in vitro system in which C57BLl6 donor splenocytes are exposed t o B1O.BR host alloantigens in the context of deficient CD28:B7 signaling as a means of preventing graft-versus-host disease (GVHD).Although 54% t o 82% of MLR alloresponse was inhibited by cytotoxic Tlymphocyte antigen 4 (CTLA4I-lg treatment of host stimulator cells, treated splenocytes were still capable of causing GVHD when infused in vivo. By adding anti-leukocyte function antigen 1 (anti-LFAl) antibody t o hCTLA4-lg in vitro t o coblock the LFA1:intercellular adhesion molecule (ICAM) signaling, splenic alloresponse was inhibited by 28996, yet GVHD induction capabilities were retained. Because antigen-primed cells might be more susceptible t o CD28:B7 blockade, we investigated whether hCTLA4-lg alone, antiLFAl antibody alone, or the combination of both added t o donor-antihost in vitro primed cells could reduce GVHD. To facilitate hyporesponsiveness induction and t o block B7 and ICAM ligands that are upregulated during GVHD, these reagents were also administered t o recipients post-BMT. We have shown thathCTLA4-lg plus anti-LFAl antibodyis highly effective in preventing GVHD-inducedl e t h a l i i (88% t o 100% of treated mice surviving versus 0% t o 28% of controls surviving). For optimal prevention, both hCTLAClg and antiLFAl must beused in vitro in the context of donor-antihost primed splenocytes and continued in vivo. This in vitro-in vivo combined approach was associated with donor engraftment, and recipients were not globally immunosuppressed. We conclude that blocking both theCD281B7 and the LFA1:ICAM pathways are critical t o effective GVHD prevention and may offer advantages t o in vitro donor T-cell removal. 0 1995 by The American Society of Hematology. T marrow transplant (BMT), the donor splenocytes would be exposed to CTLACIg before these cells would have been primed to host alloantigens. Studies by Turka et al16.’’would suggest that primed cells are more susceptible to CTLA4-Ig infusions than nonprimed cells. Many of the reports of anergy induction have involved the testing of antigen-specific (ie, antigen-primed) T-cell clones. Whereas in vivo antigenspecific hyporesponsiveness has been described, a number of these studies have used model systems that may not be applicable to the transfer of allogeneic lymphohematopoietic cells such as transgenic mice analyzed in a resting state’.’8 or murine recipients of xenografted islets.” In contrast with these examples, GVHD induced in lethally irradiated recipients generates a fulminant systemic reaction characterized by massive tissue destruction, release of inflammatory cytokines and T-cell responses to alloantigens present on multiple tis- CELLS ARE typically activated in vivo by the binding of antigendpeptides to the T-cell receptor (TCR). For an efficient and sustained T-cell response to occur, a costimulatory signal, usually provided by professional antigenpresenting cells (APCs), must be transduced to the T cell.’ Using antigen-specific Thl clones, Jenkins and Schwartz’ first showed that T cells exposed to antigerdpeptide in the presence of paraformaldehyde-fixed APCs may become incapable of responding when reexposed to the antigen/peptide, a state known as anergy. This approach has been shown to inhibit the ability of resting APCs to costimulate T cells by preventing the upregulation of CD2WCTLA4 ligands that occur during the immune response and by precluding efficient interleukin-2 (IL-2) production via the CD2WCTLA4 p a t h ~ a y .A~ .soluble ~ fusion protein, consisting of the extracellular domain of CTLA4 and an Ig moiety, which specifically blocks the CD28/CTLA4:B7 costimulatory pathway, similarly renders T cells anergic to antigen/peptides present during the blockade.””’ Because CD28/CTLA4:B7 blockade reduces the primary mixed lymphocyte reaction (MLR),’’.’3strategies that inhibit this pathway could be used to make tolerant the small subset of donor T cells with host alloreactive capacity and be exploited as a means of preventing lethal GVHD. Our group has shown that in vivo blockade of one such T-cell costimulatory pathway, CD28/CTLA4:B7, was effective in reducing, but not eliminating lethal GVHD in irradiated recipients of fully allogeneic or minor histocompatibility-only disparate donor grafts.I4.l5In those experiments, mice were given a fusion protein consisting of the extracellular domain of human CTLA4 fused with a murine (m) or human (h) Ig heavy chain (termed CTLA4-Ig) to extend circulating half-life of the recombinant protein. There are several reasons why incomplete anti-GVHD responses to CTLA4-Ig infusions might have been observed in these studies. One possibility is that the timing of CTLA4-Ig infusions wasnot optimal for tolerance induction in vivo. Because hCTLA4-Ig infusions were initiated on days 0 or - 1 of bone Blood, Vol 85, No 9 (May l ) , 1995: pp 2607-2618 From the Department of Pediatrics, Division of Bone Marrow Transplantation, and the Department of Therapeutic Radiology, Section on Experimental Cancer Immunology, University of Minnesota Hospital and Clinic, Minneapolis, MN; and RepliGen Corp, Cambridge, MA. Submitted August 26, 1994; accepted December 13, 1994. Supported in part by US Public Health Service Grants No. ROIA134495, ROI-CA 31618, and Pol-AI35296 awarded the byNational Cancer Institute and the National Institute of Allergy and Infectious Diseases, Department of Health and Human Services. B.R.B. is a recipient of the Edward Mallinckrodt Jr Foundation Scholar Award. This is no. 33 in a series on murine bone marrow transplantation across the mixed lymphocyte reaction. Address reprint requests to Bruce R. Blazar, MD, Box 109, University of Minnesota Hospital and Clinic, 420SE Delaware St, Minneapolis, MN 55455. The publication costsof this article were defrayedin part by page chargepayment. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. section 1734 solely to indicate this fact. 0 I995 by The American Society of Hematology. 0006-4971/95/8509-05$3.00/0 2607 From www.bloodjournal.org by guest on February 6, 2015. For personal use only. BLAZAR ET AL 2608 sue targets.’” The graft-versus-host reaction may be less susceptible to anergy induction by CTLA4-Ig by virtue of the systemic inflammatory processes that accompany irradiation and alloreactivity. Another possibility is that other APC:T-cell pathways capable of providing this costimulation might override the beneficial anti-GVHD effect of CTLA4-Ig infusions. Some of the candidates include LFAl :intercellular adhesion molecule (ICAM),’3.2’CD2:CLM8,” CD40 ligand:CD40,23.24 and the VLA4:VCAMl,” and interactions. In vivo blockade of the first two receptor:counter-receptor pairs has been shown to facilitate the acceptance of solid organ grafts,”.” whereas blockade of the last three receptor:ligand pairs23-26 has resulted in diminished antigenic responses, including GVHD, in vivo. The present study was undertaken for two reasons: (1) to determine if donor cells exposed to host alloantigens in vitro before BMT in the presence of a CD28/CTLA4:B7 blockade could be made tolerant of host alloantigens in vivo resulting in a diminished GVHD capacity and, if not, to determine whether continuing the blockade strategy in vivo would be advantageous; (2) to determine if the coblockade of an alternative T-cell signaling pathway, such as LFAl:ICAM, would be additive or synergistic to the blockade of the CD28/ CTLA4:B7 pathway in preventing GVHD in lethally irradiated recipients of fully allogeneic donor grafts, and to determine whether long-term surviving recipients would become severely immunosuppressed. Toward that end, we have examined mice during the graft-versus-host reaction in vivo for upregulation of B7and ICAM ligands to decide if LFA1:ICAM might be an appropriate determinant for coblockade along with CD28:B7 and to decide whether in vivo infusion of hCTLA4-Ig in conjunction with anti-LFAl antibody would be a reasonable approach to augment an in vitro procedure to reduce GVHD alloreactivity in vivo. MATERIALS AND METHODS Mice. B IO.BR/SgSnJ (H-2k)recipient mice were purchased from Jackson Laboratory (Bar Harbor, ME). C57BLl6 (H-2b) donor mice were purchased from the National Institutes of Health (Bethesda, MD). Donors and recipients were female. Donors were 4 to 6 weeks old and recipients were 8 to I O weeks old at the time of BMT. In addition, C.H-2h”’2(designated bm12) congeneic mice, which differ from C57BW6 by virtue of a 3-amino acid mutation in the MHC class I1 locus, were obtained from Jackson Laboratory and used as a source of stimulator cells in some in vitro assays. BMT. Our transplant protocol has been described in detaiLZ7 BIO.BR recipients were conditioned with 8.0 Gy total body irradiation administered from a Philips RT 250 Orthovoltage Therapy Unit (Philips Medical Systems, Brookfield, W)filtered through 0.35-mm Cu at a final absorbed dose rate of 0.41 Gray (Gy)/min at 225 kV and 17 mA. Donor BM was collected into RPMI 1640 medium by flushing it from the shafts of femurs and tibias. Recipients (8 mice per group per experiment) received 25 X lo6 BM cells from C57BLl 6 donors that had been T-cell depleted with anti-Thyl.2 monoclonal antibody (MoAb) (hybridoma 30-H-12, rat IgG2b, provided by Dr David Sachs, Massachusetts General Hospital, Cambridge) plus C‘ as previously Single cell suspensions of splenocytes were obtained (as a source of GVHD-causing effector cells) by passing minced spleens through a wiremeshand collecting them into RPMI 1640. Noncultured or in vitro cultured splenocytes (as indicated) were suspended with anti-Thy1.2 plus C’-treated BM, and 0.5 mL volume containing 25 X 10‘ anti-Thy1.2 plus C‘-treated BM plus 25 X IOh splenocytes was infused. This dose of splenocytes was chosen to provide GVHD-induced mortality in at least 70% of recipients (see Results). The final composition of T cells in this donor graft ranged from 10%to 20% of theentire inocula, simulating the lower spectrum of the content of T cells contained with human BM aspirates used for human BMT. In this latter setting, peripheral blood T cells typically contaminate donor BM aspirates.3” Asan additional control, some mice received splenocytes that were depleted of T cells with anti-Thyl.2 plus C’ as described above. hCTLA4-Ig protein and anti-LFAI MoAb preparation and usage. hCTLA4-Ig waspurified from supernatants ofprotein A Chinese hamster ovary cells electroporated with an expression vector containing the extracellular portionof hCTLA4 joined tothe CHIHinge-CH2-CH3 domains derived from a human genomic IgGl gene with immobilized protein A.” Purified hCTLA4-Ig consisted entirely of the dimeric form. Anti-LFAl (anti-CD1la) (hybridoma FD441 .S, rat IgG2b.kindly provided by Dr Frank Fitch, University of Chicago, Chicago IL) MoAb was used as purified (greater than 85% pure as assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel) ascites fluid obtained after ammonium sulfate precipitation and dialysis as previously described.” For in vivo administration, mice were injected intraperitoneally with phosphate-buffered saline (PBS), hCTLA4-Ig (250 pgldose on days - I and 0, and then 100 &dose thrice weekly until day 28 post-BMT according to our previous dose optimization studiesI4),and/or anti-LFAl MoAb (300 pgl dose beginning on day - 1 and continuing twice weekly through day 29 post-BMT to ensure adequate in vivo blockade of LFA1:ICAM binding). In vitro incubation of splenocytes with hCTLA4-Ig a d o r antiLFAl MoAb. Donor C57BL/6 splenocytes and irradiated (30 Gy) hostBIO.BR splenocytes were concentrated to 8 X 10‘ cells/mL each in Dulbecco’s minimal essential medium (Bio Whitttaker, Walkersville, MD), 10% fetal calf sera (Hyclone, Logan, UT), 2 mom) (Sigma Chemical CO, St Louis, mercaptoethanol ( 5 X MO), 10 mmoVL HEPES buffer, 1 mmol/L sodium pyruvate (GIBCO BRL, Grand Island, NY), and amino acid supplements (1.5 mmol/L. L-glutamine, L-arginine, and L-asparagine) (Sigma), antibiotics (penicillin: 1 0 0 UlrnL, streptomycin: 100 pg/mL) (Sigma), and placed at 37°C and 10%CO2.For scale-up experiments, in which the incubated cells were infused in vivo, donor (8 X 10‘ cellslml) and irradiated host splenocytes (8 X IO6 cellslml) were added together and placed into 225-cm2 flasks (Costar, Cambridge, MA), and incubated in 10% CO2 for 3 days. In some experiments, an aliquot of cells was removed after the completion of the culture period for assessment of proliferation and/or phenotyping by flow cytometry as described in detail below. For experiments with naive splenocytes, hCTLA4-Ig (SO pLg/mL) was added on the day of culture initiation (day 0) before a 2.5- to 4-day bulk MLR culture period. Control cultures were incubated in the absence of hCTLA4-Ig, but were otherwise treated identically to the hCTLA4-Ig-treated cultures. At the end of the culture period, the bulk population was washed three times with medium to remove any nonviable cells, resuspended, and added at a 1:l ratio to antiThyl.2 plus C’-treated BM. For experiments using primed splenocytes, hCTLA4-Ig (SO pg/mL final concentration), and/or anti-LFAl MoAb ( 1 50 pglmL final concentration), or an equal volume of PBS were added to the BMS preparation for 3 hours at 37°C. The BMS preparation was infused without additional manipulation. All recipients received 25 X 10‘ anti-Thy1.2 plus C’-treated BM cells along with 25 X 10‘ manipulated naive or primed splenocytes or nonmanipulated fresh splenocytes via caudal vein injection in 0.5 mL v01ume on the day after lethal irradiation. Pathologic examination of tissues. Mice were killed, autopsied, From www.bloodjournal.org by guest on February 6, 2015. For personal use only. 2609 CTLA4 Ig and LFAl COBLOCKADE IN MURINE GVHD and tissues were taken for histopathologic analysis. All samples were placed in 10% neutral-buffered formalin, imbedded in paraffin, sectioned, and stained with hematoxylin and eosin for histopathologic assessment. Organs were scored positive for GVHD if there was single cell necrosis (skin, colon), crypt dropout (colon), periportal infiltrate with acute necrosis (liver), or endothelialitis with a lymphocytic infiltrate ( l ~ n g ) . ~ ’In* ~previous * studies, these features were present only in mice with active GVHD and not in irradiated recipients of either syngeneic or anti-Thyl.2 plus C’-treated allogeneic BM. Immunohistochemistry. Tissues (spleen, liver, and colon) were embedded in OCT compound (Miles, Inc, Elkhart, IN), snap frozen in liquid nitrogen and stored at -80°C. Serial 4-mm sections were cut (IEC Minotome), thaw mounted onto glass slides, and fixedfor 5 minutes in acetone. After blocking with normal horse serum, sections were incubated with biotinylated MoAbs specific for the following molecules: B7-1 (clone 1G10, rat IgGZa, PharMingen, San Diego, CA), B7-2 (clone GLl, rat IgG2a, PharMingen), ICAM-l (clone Be2961, rat IgG2a, provided by Dr Adrienne Brian, University of California San Diego)33and ICAM-2 (clone 3C4, rat IgG2a, PharMingen). Detection with peroxidase-conjugated avidin-biotin complex and 3,3’-diaminobenzadine as chromogen was performed essentially as describedM with reagents purchased from Vector Laboratories, Inc (Burlingame, CA). The frequency of immunoperoxidase positive cells in tissue sections was quantitated using standard point counting technique as previously described35 and expressed as the number of positive staining cells per square millimeter averaged in at least seven representative fields. Analysis of splenocyte proliferationin response to host and thirdparty alloantigens. After in vitro culture, bulk MLR cell populations were washed, resuspended, and l@ cells plated into 96-well round-bottom (Costar) plates. Tritiated thymidine (1 pCi) was added 16 hours before harvesting and counting, in the presence of scintillation fluid, on a beta counter. Freshly obtained or in vitro cultured splenocyte preparations were subjected to limiting dilution analysis (LDA) to estimate the precursor frequency of proliferating T lymphocytes (pPTL) of post-BMT splenocytes to host and third-party alloantigens. Splenocytes were placed in 96-well flat round plates (Costar) in threefold dilutions ranging from 1.5 X lo5 to 5 X 10’ cells per well in media plus sera as described above for bulk MLR cultures. A fixed number (5 X l@) of irradiated (30 Cy), anti-Thyl.2 plus C’-treated host or third party (bm12) was added along with 10 U/mL recombinant IL-2 (Hoffman LaRoche, Nutley, NJ) in a final volume of 200 pL/well. The plates were pulsed with tritiated thymidine after 7 days of incubation and harvested the next day. Using Poisson distribution statistics according to the method of Taswell,” the likelihood of a single hit was confirmed and a frequency estimate calculated. Confidence intervals were obtained based upon the chisquare minimization and maximum likelihood methods to compare the frequencies between the groups. Flow cytometry analysis. Monoclonal antibodies (MoAbs) were directly labeled with biotin, fluorescein isothiocyanate (FITC), or phycoerythrin (PE) as previously described.” Single-cell splenocyte preparations were suspended in buffer (PBS plus 5% colostrum-free bovine serum + 0.015% sodium azide). Pelleted cells were incubated for 15 minutes at 4°C with 0.4 pg of an anti-Fc receptor MoAb (clone 2.462, provided by Dr Jay Unkless, Rockefeller University, New York, NY) to prevent Fc binding.38Optimal concentrations of directly conjugated (biotin-, PE-,and FITC-labeled) MoAbs were added to a total volume of 100 to 130 pL and incubated for 1 hour at4°C. For the biotin-labeled MoAb, fluorescence was indirectly measured by adding streptavidin-labeled Red 613 (GIBCO) for an additional hour at 4°C. After final washing, cells were fixed in 1% paraformaldehyde. For analysis ofin vitro cultured bulk population cells, one- or two-color flow cytometry was performed using the following MoAbs (obtained, unless otherwise indicated, from PharMingen): anti-CD3e (clone 145-2C11, hamster IgG), anti-CD8 (clone 53-6.72, rat IgG2a. provided by Dr Jeffrey Ledbetter, Bristol-Myers-Squibb, Seattle, WA),39 anti-CD4 (clone GK1.5, rat IgG2b), anti-CD45R (B220) (clone Ra3-6B2, rat IgG2a), anti-B7-l (clone 16.10A1, hamster IgG),” antiCD28 (clone 37.5 1, hamster IgG), and an irrelevant rat IgG2 antihuman antibody (3AlE):’ For analysis of donor or host origin of repopulating cells postBMT, two- or three-color flow cytometry was performed. Splenocytes were costained with anti-H-2b-SA-PE (clone EH144, mouse IgG, provided by Dr T.V. Rajan, Albert Einstein University, New York, NY) and anti-H-2’-FITC (clone 11-4.1, mouse IgG2a, American Type Culture Collection, Rockville, MD) after resuspending single cell suspensions in buffer. The following antibodies and reagents were also used in conjunction witha11ti-H-2~ or anti-H-2k MoAbs: anti-CD3e, anti-CD8, anti-CD4, B220, Mac1 alpha subunit (CD18, clone MU70.15.11, rat IgG2b; PharMingen), and an irrelevant rat IgG2 antihuman antibody (3AlE). All samples were analyzed on a FACScan (Becton Dickinson, Mountain View, CA) using consort-30 software. A minimum of 20,000 events were examined. Background subtraction using a directly conjugated irrelevant antibody control was performed for each sample. Statistical analyses. Group comparisons of continuous data were made by Student’s t-test. Survival data were analyzed by lifetable methods using the Mantel-Peto-Cox summary of ~hi-square.~’ Actuarial survival rates (the proportion of mice surviving on each day post-BMT) are shown. P values < .05 were considered significant and P values < .l, but > .05 were considered tobe a statistical trend. RESULTS In vitro hCTLACIg addition to CS7BU6anti-B1O.BRbulk cultured splenocytes inhibits the in vitro primary MLR, but does not lead to a signijcantly reduced anti-GVHD effect when the treated cells are infused in vivo. The proportion of CD4+ or CD8’ T cells in the splenocyte population used to initiate the bulk MLR culture was maintained at the end of the culture period ( C M + T cells, 20% v 21%; CD8+ T cells, 14% v 16%, respectively, for initiation v end of bulk MLR). However, the proportion of T cells coexpressing the IL-2R or B7-1 ligand substantially increased (IL-2R, 3% v 12%; B7-1, 3% v 9%, respectively, for initiation v end of bulk MLR), consistent with donor splenocyte priming. We have consistently observed an increase in the proportion of cells at the endofthe primary MLR IL-2RtandB7-1+ reaction in this strain combination. Because B7-2is constitutively expressed at low levels on splenocytes and B7-1 expression is inducible on professionalAPCS;~wetested whether the addition of hCTLA4-Ig (to block B7 binding to CD28lCTLA4) atthe initiation of bulk culture would reduce T-cell activation as measured by proliferationandIL-2R expression on T cells. Initial studies were performed to determine the optimal concentration (5, 20, 25, or 50 pg/mL) of hCTLA4-Ig necessary to inhibit the primary MLR response of C57BLJ6 anti-B10.BR bulk cultured splenocytes. In these experiments, as well as others involving different MHC disparate donor antihost strain combinations, maximum inhibition assessed o n the day of peak proliferation of the control cultures was consistently observed at concentrations of 2 20 pg/mL(datanotshown).Therefore,inallsubsequent From www.bloodjournal.org by guest on February 6, 2015. For personal use only. BLAZAR ET AL 2610 Table 1. Addition of hCTLA4-Ig Protein t o Allogeneic Stimulator Cells Decreases In Vitro Primary MLR Alloresponses Vitro In 18 Treatment Experiment 1 Sham incubated 33.1 hCTLA-Ig6.1 50 pg/mL Experiment 2 100 Sham incubated 40.4 hCTLA-Ig 13.1 50 pg/mL % Recovery A cpm 63 56 % Control cpm 100 L > a 3 v) 63 54 31 C57BU6 splenocytes (8 x 106/mL) were incubated with irradiated anti-Thyl.2 + C'-treated B1O.BR stimulators (1:l ratio)in the presence or absence of hCTLA4-Ig (50 pg/mL). The percentage of recovery is number of responders remaining divided by input cell number x 100%. Splenocytes were maintained for 3 days, plated at lo5cell per well. and pulsed overnight. A cpm ~cPms.p.,im..t.l - cpm,..wnd.r.alons) Standard are listed as mean values for 12 wells per group x deviations were 515% of mean values. _.", Table 2. Coblockade of the LFA1:ICAM Pathway in Context of hCTLAClg Protein Further Reduces Alloresponsiveness In Vitro In Vitro Treatment Experiment 1 Sham incubated hCTLA4-Ig Anti-LFAl hCTLA4-Ig + anti-LFAl Experiment 2 Sham incubated hCTLA4-Ig Anti-LFAl hCTLA4-Ig + anti-LFAl Experiment 3 Sham incubated hCTLA4-Ig hCTLA4-la anti-LFAl + A cpm % Control cpm 43' 38' 37 ' 25' 36.8 17.1 6.2 0.0 100 46 17 52 38 38 27 39.0 12.3 4.8 0.9 100 32 14 2 116 55 37 14.2 3.8 1.6 100 27 11 % Recovery 0 C57BU6 splenocytes (8 x 106/mL) were incubated with irradiated B1O.BR stimulators (1:l ratio)in the presence or absence of hCTLA4lg (50 pg/mL) alone, anti-LFAl antibody (150 pglmL), or hCTLA4-Ig plus anti-LFAl antibody. The percentage recovery is number of responders remaining divided by input cell number x 100%. Splenocytes were maintained for 4 days, plated at lo5 cell per well, and pulsed overnight. A cpm (cpmexw,imen,al- cpm,..wnondenalons) are listed Standard deviations as mean values for 12 wells per group x were 515% of mean values. 'Recoveries in experiment 1 were determined after Ficoll-Hypaque density gradient centrifugation and washing. , . 10 I 20 . I - . . I 30 40 - I 50 DAYS POST-BMT B ' O experiments, the highest concentration tested (50 pg/mL) of hCTLA4-lg was used. The results offive representative experiments in which hCTLA4-Ig (50 pg/mL) was added to a C57BL/6 antiBIO.BR bulk MLR culture are summarized in Tables 1 and 2. As compared with the sham-treated control cultures, responder cells were induced into a state of hyporesponsiveness by hCTLA4-Ig blockade of irradiated fully allogeneic stimulators (54% to 82% reduction) and led to cell recoveries ranging from 38% to 55% of input cell number. To determine if the reductions in donor antihost proliferative responses were sufficient for GVHD protection in vivo, . 0 f .* Y m . .. 0 BM (No Sploon) Fig 1. The effect on in vivo GVHD-inducing capacity by inhibiting of CD28/CTLA4B7 alone or in conjunction with thesignaling LFA1:ICAM in donor splenocytes during exposure t o host alloantigens in vitro: C57BL/6 splenocytes (8 x 106/mL) were cultured with irradiated B1O.BR splenocytes (8 x 106/mL)for 3 days in the presence or absence of hCTLA44g protein (50 pg/mL) alone or hCTlA4-Ig plus anti-LFAl (150 pg/mL) antibody, as indicated. Cells were then washed extensively, and 25 x 10' splenocytes were infused along with 25 x lo6 C57BL/6 anti-Thyl.2 plus C-treated BM into lethally irradiated B1O.BR recipients. Eight miceper group weretransplanted. Actuarial survival is plotted. incubated cells from the two groups in experiment 2 (Table 1 ) were extensively washed to remove all detectable nonviable cells, resuspended, and 25 X 10' cells reinfused along with 25 X 10' anti-Thy1.2 plus C' treated, freshly harvested donor BM into BIO.BR recipients (n = 8 mice/group). Flow cytometry analysis of donor splenocytes that were sham cultured with irradiated stimulator cells or cultured in the presence of hCTLA4-Ig-treated showed highly similar proportions ofCD4' (12% v 14%, respectively) or CD8' (9% v 8%, respectively) just before in vivo infusion. The actuarial survival rate is shownin Fig I . In vitro (3-day) donor antihost stimulated cells were sufficiently alloreactive to result in uniform lethality within the first 5 weeks post-BMT. As compared with the recipients of sham-treated inocula, recipients of either hCTLA4-Ig-treated stimulators had a marginallyprolonged ( P = .057) actuarial survival rate in vivo, From www.bloodjournal.org by guest on February 6, 2015. For personal use only. CTLA4 lg and LFAl COBLOCKADE IN MURINEGVHD although none of the mice that received the hCTLA4-Igtreated inocula survived the observation period. Examination of the post-BMT weight curves did not show any obvious benefit of infusing the treated as compared with the control cell populations. At no time period post-BMT did the mean weight curves of recipients of treated inocula appear to be superior to the sham-incubated control group (data not shown). The marginal actuarial survival rate prolongation conferred by infusing donor splenocytes exposed to hCTLA4-Ig-treated stimulator cells, as compared with the control group (Fig 1) was found not to be reproducible when the treated cells from experiment 3 (Table 2) were infused in vivo (data not shown). When data from these two identically performed experiments were pooled for analysis, there was no significant difference noted in actuarial survival rates when comparing recipients of sham-incubated donor splenocytes to recipients of hCTLA4-Ig-treated splenocytes, with the latter group having an actuarial survival rate that was not significantly (P= .5 1, respectively) higher than the control group. We concluded from these experiments that the reduction in the proliferative responses observed in hCTLA4-Ig-treated cultures was insufficient to provide any reproducible evidence of protection from GVHD-induced mortality in vivo. Coblockade of CD28/CTLA4:B7 and LFA1:ICAM pathways in naive donor splenocytes leads to a variable and incomplete reduction of GVHD in lethally irradiated recipients of fully allogeneic donor grafts. From the in vitro experiments discussed above as well as our previous studies in which hCTLA4-Ig was infused in vivo as a means of preventing GVHD, we concluded that in vitro blockade of the CD28/CTLA4:B7 pathway alone would be inadequate in this GVHD model system. One possible explanation for these findings is that other T cel1:APC pathways capable of providing a T-cell signal could overcome the partially inhibitory effects of CD28/CTLA4:B7 blockade. One such candidate interaction could be the LFA1:ICAM pathway. We focused on this pathway in part because of studies by other investigators who have shown that the in vivo blockade of LFA1:ICAM interaction facilitates the acceptance of solid organ grafts and results in a diminished GVHD response in mice. To determine if B7 ligands and ICAM counterreceptors were potentially involved in GVHD generation in vivo, we embarked upon immunohistologic analysis of tissues obtained from lethally irradiated B1O.BR recipients of 25 X IO6 anti-Thy1.2 plus C'-treated BM with 25 X lo6nontreated (GVHD positive control) or anti-Thy 1.2 plus C'-treated splenocytes (GVHD-negative, BMT control). The frequency of B7-I, B7-2, ICAM-1, or ICAM-2 expressing cells in spleen, colon, and liver of recipients of nondepleted or antiThy 1.2 plus C'-treated supplemental splenocytes is shown in Table 3. These data indicate that the graft-versus-host reaction in this model system is associated with the upregulation of B7-1, B7-2, andlor ICAM-1 in all three tissues, suggesting that these ligands may be involved in GVHD generation in vivo. Figure 2 is a photomicrograph of the immunohistologic results of B7 and ICAM ligand upregulation in representative splenic sections taken from recipients of anti-Thyl.2 plus C'-treated (Fig 2A) or untreated (Fig 2B) donor splenocytes. 2611 Table 3. Quantitative Tissue In S i Immunohistologic Assessment of Cell Sudace Expressionof B7 and C A M Ligands DuringGVHD In Vitro Treatment Tissue (anti-Thyl.2+ C')" Spleen No Yes No Yes No Yes Colon Liver 87-1 87-2 ICAM-l 329 296 0 5 0 85 45 0 310 150 243 37 317 0 43 0 9 0 35 2 CAM-2 306 56 80 24 Frozen-tissue sections were obtained from B1O.BR recipients of C57BU6 donor BMand splenocytes on day 14 post-BMT. In situ immunoperoxidase staining of the indicated tissues were performed and the proportion of cells expressing the indicated ligands were quantitated as described in Materials and Methods. Thedata shown are expressed as the number of positive cells/mm*. In vitro treatment (anti-Thyl.2 + C') refers to whether or not the donor splenocytes were treated in vitro with anti-Thyl.2 + C' to remove GVHD-causing cells. Based upon our in vivo GVHD immunohistologic data in this model system, we proceeded to experiments designed to coblock both the CD28:B7-1 and LFA1:ICAM pathways in vitro. The results of three primary MLR experiments in which the effect of hCTLA4-Ig andlor anti-LFAl antibody blockade of bulk MLR cultured cells was tested are shown in Tables 2 and 4. Both hCTLA4-Ig and anti-LFAl antibody alone were inhibitory to the C57BL/6 anti-B 10.BR proliferative response. The combination of hCTLA4-Ig and antiLFAl provided the highest degree of inhibition, resulting in only a 0% to 11% of control response. Cells from experiment 1 (Table 2) were isolated by Ficoll-Hypaque density-gradient centrifugation and the bulk population subjected to flow cytometry analysis. These data show that hCTLA4-Ig and antiLFAl antibody can reduce the expansion of CD8+ and IL2R+ cells from 19% to 2% in this MLR culture (Table 4). LDA analysis of C57BU6 responders that are nonmanipulated or depleted of either CD4+ or CD8+ T cells before culturing withBlO.BR stimulators has shown that CD8+ cells account for a majority of the proliferative responses (data not shown). These results are both consistent with early (day 5 through 7 post-BMT) studies in this model system that have shown a predominance of CD8+ T cells as GVHD effector cells as detected by in situ immunoperoxidase staining of GVHD target tissues (lung, liver, and colon) and flow cytometry analysis of cells obtained by thoracic duct cannulation (data not shown). In each experiment, 225% of the number of input cells was recovered, which permitted testing of the anti-GVHD effect of hCTLA4-Ig plus anti-LFAl in vivo. C57BL/6 antiB 10.BR in vitro bulk-cultured cells incubated under control conditions or in the presence of hCTLA4-Ig alone or hCTLA4-Ig plus anti-LFAl antibody were washed and 25 X lo6 supplemental cultured splenocytes infused in vivo along with anti-Thyl.2 plus C'-treated BM.As compared with sham incubated control cells, neither hCTLA4-Ig alone nor hCTLA4-Ig plus anti-LFAl antibody treatment provided any evidence of GVHD protection by actuarial survival rates or mean weight data (data not shown). The absence of any From www.bloodjournal.org by guest on February 6, 2015. For personal use only. BLAZAR ET AL 2612 A Fig 2. Upregulation of B7 and C A M ligands in the spleen obtained from mice undergoing graft-versus-host a reaction. lmmunoperoxidase staining for B7 and ICAM ligands is shownin representative splenic sections obtained from BIO.BR recipients of anti-Thyl.2 plus C'-treated C57BL/6 BM and anti-Thyl.2 plus C'-treated (A) or nontreated (B) C57BL/6 splenocytes on day 14 post-BMT. As compared with the controls (Al. there is an increase in the proportion of splenocytes expressing 87-1, 87-2, ICAM-1, and ICAM-2 ligands in mice that have histologically documented GVHD in the colon andliver. The frequency of cells expressing these ligands in the spleen, colon, and liver are listed in Table 3. obvious decrement in GVHD capacity was observed despite the 89% inhibition of donor antihost proliferation observed when an aliquot of these same cells was analyzed in vitro (Table 2, experiment 3) and despite infusing a lower proportion of T cells (CD4' or CD8') that coexpress IL-2R (Table 4). These data indicate that combination hCTLA4-Ig + antiLFA I treatment in vitro was entirely ineffective in reducing GVHD. Coblockode of CD28/CTLA4:B7 and LFA1:ICAM pathways in donor antihost primed splenocytes lead.7 to a significant reduction of GVHD in lethallv irradiated recipients q f j d l v allogeneic donor grajis. At this juncture, we considered the possibility that naive murine splenocytes would bemore difficult than primed cells to induce into a state of sufficient antigen-specific hyporesponsiveness. A large proportion of the published murine literature has examined anergy induction of antigen-specific and hence primed Tcell clones. Moreover, Turka et al" have shown that whereas hCTLA4-Ig-treated rat recipients of cardiac allografts maintained their graft rejection capacity, rats first primed in vivo Table 4. Effect of Coblockade of the LFA1:ICAM Pathway in the Context of hCTLA4-1g Protein T-, B-, and NK Cell Expansion and Activation In Vitro %IL-ZR' CD4' or In VitroTreatment 10 Sham incubated 59 5 hCTLA4-Ig Anti-LFAl 4 hCTLA4-Ig + anti-LFAl %CD4' %CDB' 19 8 CD8' %B220' 16 71 9 SbNK1.1' 5 7 14 6 67 5 217 6 64 6 Cells from experiment 1, Table 2 were phenotyped after the 4-day MLR culture as described in Materials and Methods. Listed are the percentages of positive cells in these bulk cultures. Abbreviation: NK, natural killer. with irradiated donor leukocytes and then subsequently given hCTLA4-Ig achieved permanent and specific cardiac allograft acceptance. Therefore, rather than to make naive donor splenocytes tolerant, we examined the effects ofCD281 CTLA4:B7 and/or LFAIKAM in vitro blockade on inhibiting the GVHD-induced lethality of C57BL/6 anti-B10.BR bulk-cultured splenocytes. At theend of a 3-daypriming period, hCTLA4-lg protein and/or anti-LFAI antibody was added to the cultured cells for 3 hourstopermitligand saturation, the cells washed. and 25 X 10' control or treated splenocytes infused in vivo along with anti-Thy 1.2 plus C'treated freshly harvested BM. An aliquot of the bulk cultured cells was analyzed for tritiated thymidine incorporation after an additional 24- or 48-hour incubation period. which documented their proliferative capacity (counts per minute [cpm] sllopencir rc.ipont~en. aulo~oCoarh g c k g n d = 12.0 and 14.9 x 10' cpm, respectively). The proportion of IL-2R' cells in these cultures was 2796, consistent with the functional data and indicating that alloreactivity had taken place. Becausethe3-hourexposureperiodwouldlikely be too short to fully induce antigen-specific hyporesponsiveness~.".JJ and because B7 and ICAM ligands are upregulated during GVHD, we continued the same reagents in vivo for 28 days post-BMT to complete the hyporesponsiveness induction process. Our previous experiments indicated that the in vivo combination of hCTLA4-Ig plus anti-LFA 1 alone was ineffective in preventing lethal GVHD (data not shown). The actuarial survival results of the current studies are depicted in Fig 3A and the mean weight curves in Fig 3B. As compared with the control group, recipients receiving donor antihost primed splenocytes exposed to hCTLA4-lg in vitro and in vivo had a marginally ( P = . l ) higher actuarial survival ratewith12% (v 0% of controls) surviving the138-day observation period. In contrast, 88% of the recipients of hCTLA4-lg plus anti-LFAI survived, a significantly higher proportion than either the PBS (P < .0002) or hCTLA4-lg only (P = .001) groups. Mice receiving hCTLA4-lg plus From www.bloodjournal.org by guest on February 6, 2015. For personal use only. CTLA4 lg and LFAl COBLOCKADE IN MURINEGVHD pas , . 0.0 20 0 60 40 80 l00 120 140 l11 138 DAYS POST-BM1 2613 cpm . ,,,," h:lckpn d = 16.6 X IO', assessed after pulsing an aliquot of cells for 18 hours). Control recipients of in vitro plus in vivo hCTLA4-Ig or in vivo anti-LFA 1 alone had significantly ( P = .00025, .032, respectively) higher actuarial survival rates as compared with the PBS control group (Fig 4A). Both reagents were required for the optimal biologic effect in this experiment because the addition of hCTLA4-Ig to anti-LFA 1 in vitro and in vivo resulted in a significantly higher actuarial survival rate in these mice as compared with those receiving only anti-LFAl or hCTLA4-Ig given in vitro plus in vivo ( P = .033, .OOS I , respectively) (Fig 4A). Whereas the actuarial survival rates of recipients of hCTLA4-Ig plus anti-LFAI given in vitro and in vivo were 100%, most of the mice were not GVHD free. Mean weight in this group averaged 90% of pre-RMT weight values (Fig 4B). The mean weight and immunologic data (see below) are indicative of a mild, sublethal GVHD process. Taken together, these data indicate that there is an A ' O 12 !z . . 10 . a 30 58 78 91 I'.." DAYS POST-BUT Fig 3. hCTLAClg plus anti-LFAl in vitro treatment of donor antihost primedsplenocytes followed byin vivo administration of these same reagents markedly reduces the GVHD capacity of the donor inocula: C57BL/6 splenocytes (8 x 106/mL) were culturedwith irradiated B1O.BR splenocytes (8 x 10E/mL)for 3 days. After the priming period, cells were exposed t o saline, hCTLA4-Ig protein (50 pg/mLl alone or hCTLA4-Ig plus anti-LFAl (150 pg/mL) antibody, as indicated, for a period of 3 hours on ice t o permit receptor saturation. After the additional culture period, cells were washed extensively, and 25 x lo6 splenocytes were infused along with 25 x lo6 C57BL/6 anti-Thyl.2 plus C-treated BM into lethally irradiated B1O.BR recipients. Post-BMT, recipients were injected in vivo with the same reagents used for the particular in vitro treatmentused for thatgroup. Recipients were treated with PBS and hCTLA4-Ig (250 p g on days -1 and 0, and 100 p g thrice weekly through day 28 post-BMT) with or without anti-LFAl antibody (300 p g twice weekly on days "-1 through day 29 post-BMTI as indicated. Eight mice per group were transplanted. (A) Actuarial survival is plotted;(B) mean weight data for the same experiment are plotted. anti-LFAI treatment had mean weight values that exceeded the pre-BMT bodyweightat the end of the observation period (Fig 3B), consistent with their healthy appearance and lack of clinical evidence of GVHD. These mean weight values were significantly higher than the PBS control group beginning on day 20 post-BMT. To determine whether the anti-GVHD effect of hCTLA4Ig plus anti-LFAI was reproducible and to sort outwhat themore important components of our approach are, we performed a second experiment according to the protocol shown in Fig 3 and included a number of additional control groups. These additional groups consisted of mice receiving anti-LFAI antibody alone in vitro and in vivo, anti-LFAI antibody plus hCTLA4-lg together in vivo alone, or no cultured splenocytes. C57BW6 anti-B IO.BR-primed splenocytes had a comparable proliferative indexto those cells usedto infuse mice in Fig 3 (in the current experiment, i............. ."...."....".. I O 0.0 0 B . n 2 0 llqroup 2 40 DAYS POST-BUT EO L 10 9+ hCtU4-l a l00 anll-LCAl In VIVO mll-LFAl antl-LFAl In vllm md In VIVO 120S "1 Fig 4. hCTLAClg plus anti-LFAl in v-ho treatment of donor antihost primedsplenocytes followed byin vivo administration ofthese same reagents markedly reduces the GVHD capacity of the donor inocula: C57BL16 splenocytes (8 x 106/mL)were culturedwith irradiated B1O.BR splenocytes (8 x 1OE/rnL)for 3 days. After the priming period, cells were exposed t o saline, hCTLA4-Ig protein (50 pg/mL) and/or anti-LFAl (150 pg/mLl antibody, as indicated, for a period of 3 hours on ice t o permit receptor saturation. After the additional culture period, cells were washed extensively, and 25 x lo6 splenocytes were infused along with 25 x lo6 C57BL/6 anti-Thyl.2 plus C'-treated BM into lethally irradiated B1O.BR recipients. Post-BMT, recipients were injectedin vivo with thesame reagents used for the particular in vitro treatment used for that group. Recipients were treated with PBS, hCTLA4-Ig (250 p g on days -1 and 0, and 100 p g thrice weekly throughday 28 post-BMTI, and/or anti-LFAl antibody (300 p g twice weekly ondays - 1through day 29 post-BMT) as indicated. Eight miceper group weretransplanted. (A) Actuarial survival is plotted;(B) mean weight data for thesame experiment are plotted. From www.bloodjournal.org by guest on February 6, 2015. For personal use only. BLAZAR ET AL 2614 ...-." l34 .I ~~. 0 20 ab40 ab . .=.."".""" -a PBS , l9 39 59 l9 99 l00 DAYS POST-BMT DAYS POST-BUT Fig 5. hCTLA4-lg plus anti-LFAlin vitro treatment of donor splenocytes followed byin vivo administration of these same reagents markedly reduces the GVHD capacity of the donorinocula effect of priming beforein vitro treatment on in vivo GVHD generation: C57BL/6 splenocytes 18 x 1O6/mL1were cultured with irradiated B1O.BR splenocytes (8 x 106/mL). To generate primed cells, a 3-day incubation period was used prior t o a 3-hour in vitro treatment. For nonprimed cells, reagents were added on the day of culture initiation. In both instances, bulk MLR cultures were exposed to saline, hCTLA4-Ig protein (50 pg/mL) and/or anti-LFAl 1150 pg/mL) antibody, as indicated, for a period of 3days Inonprimed cells) or 3 hours (primedcells). After the additional cultureperiod, cells were washed extensively, and 25 x lo6 splenocytes were infused along with 25 x lo6 C57BL/6 anti-Thyl.2 plus C'-treated BM into lethally irradiated B1O.BR recipients. Post-BMT, recipients were injected in vivo with the same reagents used for the particular in vitro treatment used for that group. Recipients were treated with PBS, hCTLA4-Ig (250p g o n days -1 and 0, and 100 p g on thrice weekly through day 28 post-BMTI, and/or anti-LFAl antibody1300 p g twice weekly weretransplanted. (A) Actuarialsurvival is plotted; (B) mean weight on days -1 through day 29 post-BMTI as indicated. Eiqht - miceDer . aroup data for the same experiment are plotted. apparent requirement for hCTLA4-Ig + anti-LFAI to be added in vitro before in vivo infusion because recipients receiving splenocytes exposed to these reagents in vitro and in vivo had a statistical trend ( P = .07)toward a higher actuarial survival rate as compared with recipients of nontreated splenocytes and hCTLA4-Ig plus anti-LFA1 given in vivo (Fig 4A). Of greater importance, the combination of hCTLA4-lg plus anti-LFA1 in vitro and in vivo did provide a reproducible and significantly higher degree of protection from GVHD-induced mortality confirming the data shown in Fig 3A. The in vitro and in vivo coblockade of CD28/CTLA4:B7 and LFA1:ICAM pathways i s more effective in preventing GVHD when primed rather than when naive donor splenocytes are infused in vivo. To address whether recipients of primed cells might experience a less potent anti-GVHD effect conferred by hCTLA4-lg plus anti-LFAI treatment when compared with recipients of naive splenocytes, additional groups of mice in the previously described experiment were given naive splenocytes that were sham-incubated or incubated withhCTLA4-Ig plus anti-LFAI in vitro for 3 hours before in vivo infusion. Mice receiving primed or naive cells incubated with hCTLA4-Ig plus anti-LFA1 also received these reagents in vivo for 28 days post-BMT. Recipients of nonprimed cells and hCTLA4-Ig plus anti-LFAI had a significantly ( P = .045)higher actuarial survival rate (Fig 5A) as compared with their respective PBS controls. Recipients of primed cells and in vitro and in vivo hCTLA4-Ig plus anti-LFAI had a higher ( P = .074) survival rate than those receiving nonprimed cells (100% v 75%. respectively) (Fig 5A), suggesting that the use of primed cells offers an advantage over naive cells, even though control recipients of primed or nonprimed splenocytes had a comparable ( P = .39) actuarial survival rate. The in vitro and in vivo coblockade of CD28/CTLA4:B7 and LFA1:ICAM pathways does not lead to a prolonged state of irnmunocompromiseas compared with recipients of anti-Thyl.2 plus C'-treated BMS. We sought to ask whether recipients of primed splenocytes exposed to hCTLA4-lg plus anti-LFA1 in vitro and in vivo or in vivo only were more immunosuppressed than recipients of antiThyl.2 plus C'-treated BM without supplemental splenocytes. Representative mice (mean weights were 2 20 g in the three groups) were analyzed at 5 months post-BMT. Despite the histologic evidence of mononuclear cell infiltration in some of these mice (data not shown), flow cytometry analysis of the spleenat this time post-BMT showedno evidence of hypocellularity, T- or B-cell lymphopenia, or increases in the relative proportion of macrophages typically associated with acute, active GVHD in this model system (Table 5). Recipients that received hCTLA4-lg plus antiLFAl in vitro and/or in vivo hadno evidence of residual host cells, in contrast with animals that did not receive supplemental splenocytes and either of these proteins. Two additionalmiceper group thatreceived anti-Thy1.2 plus C'treated BM andno supplemental splenocytes or primed splenocytes incubatedwith hCTLA4-lg plus anti-LFAI in vitro followed by in vivo infusions of these reagents were studied at 18 1 days post-BMT. By this time, in the recipients of hCTLA4-Ig plus anti-LFA1 in vitro and in vivo, there was evidence of a modest decrease in overall splenic cellularity and a modest T-cell lymphopenia. The B-cell and macrophage compartments appeared to be relatively spared from the graft-versus-host reaction. Taken together, these data are consistent with the clinical appearance of a nonlethal form of GVHD (Table 5). We concurrently pursued the issue as to whether the recipients of hCTLA4-lg plus anti-LFAI were globally immunosuppressed. When analyzed at 5 months post-BMT, splenic pPTL analysis of donor antihost and anti-third party (bm 12) alloreactive proliferating cells (Table 6) showed comparable frequencies against both types of stimulator cells in each From www.bloodjournal.org by guest on February 6, 2015. For personal use only. LFAl CTLA4 Ig and2615 COBLOCKADE GVHD IN MURINE + Table 5. Effect of CD28/CTLA4 B7 LFA1:ICAM Blockade on Splenic Phenotypic Reconstitution in Long-term Post-BMT Survivors Receiving In V i 0 Donor Antihost Primed Cells Total no. Group Day 151 post-BMT BM without splenocytes hCTLA4-lg anti-LFAI in vitro in vivo hCTLA4-lg anti-LFAl in vivo only Day 181 post-BMT BM without splenocytes 65.2 hCTLA4-Ig anti-LFAl in vitro in vivo + + + + + 64.4 179.2 104.4 80.0 66.5 H-2‘ H-2b H-Zb CD4 H-Zb CD8 10.9 5.2 1.3 57.3 166.7 14.3 0.0 9.4 0.0 96.0 41.2 120.1 25.1 77.312.5 5.2 16.1 6.3 14.4 74.4 6.0 42.6 4.7 8.0 4.1 1.6 0.0 H-Zb B220 H-2b Mac1 52.8 8.8 Spleens were obtained at 151 or 181 days post-BMT from two representative mice per group from Table 4 and Fig 4 were pooled together at each time point before analysis by two-color flow cytometry. Values listed are absolute numbers of cells x lo-’. Details of the in vitro and in vivo treatments are described in the legend to Fig 4. of the three groups. Because C57BU6 anti-BlO.BR splenic pPTL precursor frequencies in nontransplanted mice were more than 13-fold higher on days 151 and 191 post-BMT, mice in each group appeared to be hyporesponsive to host alloantigens. In contrast, anti-third party (anti-bml2) responses in nontransplanted controls were less than 1.5-fold higher than transplanted mice at both time periods in all groups with the single exception of recipients of hCTLA4Ig plus anti-LFAl on day 181. The latter data could be indicative of GVHD-induced immunosuppression, consistent with the weight data and splenic flow cytometry data and indicative of a graft-versus-host reaction. DISCUSSION The major finding of this study is that B7 and ICAM ligands are upregulated during GVHD generation and the combination of hCTLA4-Ig plus anti-LFAl antibody can significantly and reproducibly decrease acute GVHD-induced in lethally irradiated recipients of fully allogeneic donor grafts. In addition, we have shown that the in vitro exposure of donor splenocytes to host alloantigens under conditions in which CD28/CTLA4:B7 signaling is limited does not eliminate GVHD induced across MHC class I, 11, and miH disparities. These are the first in vivo GVHD studies that directly test the theory that the in vitro blockade of the CD28/CTLA4:B7 pathway alone or in combination with coblockade of the LFA1:ICAM T-cell costimulatory pathway would be. sufficient for GVHD prevention in mice. One of the goals of studies such as ours is to develop approaches that are encouraging for clinical applications. Theobald et a14’ have shown that the frequency of IL-2producing T cells (precursor helper T lymphocytes [pHTL]) correlates with the incidence of grades 11-IV GVHD in recipients of histocompatible sibling donor grafts. Whereas we do not know in the strain combination used in this study whether an in vitro MLR response or pFTL frequency would be predictive of the incidence andor severity of GVHD, reducing the pHTL frequency of donor antihost alloreactive cells by CD28/CTLA4:B7 and LFA1:ICAM coblockade in vitro and/or in vivo could be exploited as a means of GVHD prevention in humans. Tan et a l l ’ showed that the CD4+ T cell MLR response to fully allogeneic stimulators was reduced by 50% to 85%, consistent with our murine bulk MLR cultures. Similar primary MLR results have been reported by Gribben et a1,46 who also showed that pHTL responses between histoincompatible donors and recipients could be reduced up to 1,000-fold. These data would imply that hCTLA4-Ig inhibition of donor antihost alloreactive cells in Table 6. Effect of CD28/CTLA4B7 + LFA1:ICAM Blockade on Splenic Antihost and Anti-Third Party pPTL Frequencies in Long-term Post-BMT Survivors Receiving In Vitro Donor Antihost Primed Cells Group Day 151 post-BMT BM without splenocytes hCTLAGlg/anti-LFAl hCTLA4-lg/anti-LFAl Nontransplanted B6 Day 181 post-BMT BM without splenocytes hCTLAClg/anti-LFAl Nontransplanted B6 In Vitro Tx None Yes None NA None Yes NA Vivo In Tx PBS 1/5,913 Yes 117,165 1/4,761 Yes NA 1/3,696 PBS Yes NA Anti-B1O.BR - Anti-bml2 1t17.964 1/18,654 1/20,638 1t1.234 V19.842 V22.653 111.364 1t8.198 V31.780 V6.310 Spleens were obtained at 151 days post-BMT from two representative mice per group shown in Fig 4 and Table 4 were pooled at each time point and analyzed in an LDA for determination of antihost (B1O.BR) and anti-third party (anti-bml2) pPTL precursor frequencies. In addition, nontransplanted C57BU6 (66) control micewere used on each assay day for pPTL precursor frequencies. Responder cells were incubated for 7 days in the presence of T-cell-depleted irradiated stimulators and IL-2 (10 U/mL) in mcirotiter wells. Cells were labeled overnight with tritiated thymidine, harvested, and counted. All frequencies conformed to the Poisson distribution. Abbreviations: Tx, treatment: NA, not applicable. From www.bloodjournal.org by guest on February 6, 2015. For personal use only. 2616 vitro might be effective in reducing GVHD. However, our results would suggest that the anti-GVHD effect of hCTLA4Ig in vitromight be incomplete. The incomplete effectis not likely caused by suboptimal CD28/CTLA4:B7 blockade because (1) hCTLA4-Ig was used at concentrations severalfold higher than necessary to achieve optimal MLR inhibition; (2)concurrentexperiments withparaformaldehydefixation (0.15% or 0.5%) ofhost splenocytes provideda similar MLR inhibition (n = 3 experiments)and in vivo GVHD effect (n = 2 experiments) as hCTLA4-lg treatment Paraformaldedyde-fixation precludes of hostsplenocytes. cellsurface upregulation of ligands generated during an MLR so that B7-1 is nondetectable on the host A P C S ~at~ the end of the culture period and only the low level of B72constitutivelyexpressed on hostAPCs”couldbind to CD28/CTLA4 on the donor splenic T cells. At this point,weconcludedthatthedonor antihost GVHD-causing T cells may have been partially resistant to CD28/CTLA4:B7blockadebecause of theparticular cell of the prestype involved in the GVHD response or because ence of T-cell-associated determinants capableof providing costimulation. It is possible that use of another strain combination for MLR and GVHD generation might be more susceptible to hCTLA4-Igblockade. For example, becauseThl (IL-2-producing) cells appear to be dependent upon CD28/ CTLA4:B7 signaling, perhapsin contrast with IL-4-producing Th2 cell^,"^'"^^ it will be important to test this approach in a GVHD system in which only T h l cells are required for in vitro alloresponse and lethality in V ~ V O . ~ ”We ~ ~ ’focused on the coblockade T cell LFAl and APC ICAM interaction because studies by other investigators have shown that signaling through the LFA1:ICAM pathway can costimulate T cells. Damle eta1” have shownthat LFAl :ICAM and CD28/ CTLA4:B7pathways synergistically activateeither nonprimed or primed antigen-specific CD4’ T cells, suggesting that the coblockade of these pathways may be highly desirable for efficiently inhibiting donor antihost alloresponses. Greenet als4 concludedfrom aseries of experiments in which the TCR is triggered by staphylococcal enterotoxins, immobilized anti-CD3 epsilon antibody, or phorbol esters, that LFAI, in contrast with CD28, does not function as a costimulatory molecule, but serves primarily to modulate the signal delivered through the TCR. As related to our studies, the use of hCTLA4-Igandanti-LFAI antibodytogether would modify signals transduced via the CD28/CTLA4:B7 and the TCR pathways that could be additive or synergistic in shutting down alloresponses. Both our in vitro and in vivo data support this hypothesis. Our mostpromisingresultswith hCTLA4-Ig plusantiLFAl usedprimedrather than naive cellsandcontinued infusion of these reagents in vivo. With this combined approach, acute GVHD-induced lethality has been eliminated and the majority, but not all, of long-term post-BMT chimeras are clinically healthy without evidence of GVHD. Why some T cells may have escaped this combined blockade is not known. Perhaps the T-cell costimulatory properties of one of the alternative 1igand:receptor interactions discussed earlier was sufficient to partially compensate for the potent inhibitory effects of hCTLA4-Ig plus anti-LFA I . In that re- BLAZAR ET AL gard, the mechanisms involvedin the beneficial anti-GVHD effects observed withthisapproach arealso unknown.A prolonged in vitro incubation period would be required to show specificity of inhibition of alloresponsiveness with a minimum of 3 daysfor priming, 2 to 4 daysfor anergy induction, and 3 to 5 days for assessment of antihost and anti-third party responses (total 2 8 to13days). lnitial experiments to assess whether antihost-specific hyporesponsiveness has been obtained have been uninterpretable, and additional studies will be needed to elucidate the exact mechanisms involved. A potential drawback is that the in vivo administration of hCTLA4-Ig andor anti-LFA1 antibody could result in a state of prolonged immunosuppression in the recipient. However, we elected to administerthese reagents in vivo after incubating the primed donor splenocytes in vitro for 2 reasons: (1) the kinetics of antigen-specifichyporesponsivenessinduction have been estimated to require at least 1 to 2 days for full biologic effects. We have not as yet been able to test how long primed splenocytes will retain their GVHD capacity beyond the 4-day culture period we have used for these studies. (2) We have observed the upregulation of B7-1 and B7-2 ligands andICAM-I in vivo during theearlypostBMT GVHD response in this strain combination. Thus, if donor antihost alloresponsiveness is not completely achieved in vitro, it is possible that costimulation of the small proportion of donor cells that have escaped anergyinduction in vitro could be sufficient for inducing GVHD lethality. As assessed by phenotype and function, we could find no indication that hCTLA4-Ig plus anti-LFA1 antibody infused in vivo was immunosuppressive. However, although ourresults have shown that recipients of hCTLA4-Ig plus antiLFAl are not globally immunosuppressed as measured by LDA analysis of anti-third party pPTL, these studies were performed in the presence of exogenous IL-2 added to support the typically reduced pPTL frequency observed postBMT. Therefore, based upon these data, we are unable to assess whether the recipients that received hCTLA4-Ig plus anti-LFAI in vitro and in vivo had antigen-specific hyporesponsiveness. The possibility that anti-LFA1 -directed approaches may lead to specific hyporesponses is suggested by the experiments of Isobeetal” who showedspecific cardiac allograft acceptance with preservation of third-party cardiac rejectioncapacityupon challenge in micetreated with antibodies to block the LFA1:ICAM pathways. Other studies havesuggestedthattheblockade of ICAM-I-induced signaling of LFAl on CD4’ T cells causes immunosuppressionratherthananergyinduction.I3 However, because these latter studies were performed with human CD4’ T cells andartificial APCs consistingof NIH-3T3 fibroblasts transfected with a single MHC class I1 antigen (HLA-DR7), it is unclear how extrapolatable these results are to professional APCs that may have multiple non-B7 ligands capable ofcostimulating T cells or to our murine GVHD studies. Finally, it isnoteworthythatthe in vivo blockade of the LFA1:ICAM signalingpathwaymight be advantageous in BMT because other studies in mice have shown that antiLFAl or anti-ICAM antibodies can facilitate alloengraftment” and reduce murine lethal GVHD.” In human From www.bloodjournal.org by guest on February 6, 2015. For personal use only. CTW4 Ig and LFAI COBLOCKADE IN MURINE GVHD BMT studies, anti-LFAl a chain antibody administered in vivo to recipients of T-cell-depleted nonsibling donor grafts has been shown to facilitate alloengraftment and survival in children with immunodeficiency disorders.56The infusion of anti-LFA1 a chain antibody to patients with steroid-resistant severe acute GVHD was also found to be temporarily efficacious in 30% of patients5’ In summary, these data complement our recent studies in which we have shown that the in vitro and/or in vivo modulation of the TCR on donor splenocytes by exposure to a highly mitogenic anti-CD3c antibody is partially effective in reducing lethal GVHD.58Both the previous5*and current studies are examples of how the functional manipulation of GVHD-causing donor T cells can reduce GVHD and potentially avoid the complications of pan-T-cell depletion. ACKNOWLEDGMENT We thank Dr Marc Jenkins for valuable discussions and critical reading of this manuscript. REFERENCES 1. Schwartz RH: Costimulation ofT lymphocytes: The role of CD28, CTLA-4, and B7/BB1 in interleukin-2 production and immunotherapy. Cell 71:1065, 1992 2. Jenkins MK, Schwartz RH: Antigen presentation by chemically modified splenocytes induces antigen-specific T cell unresponsiveness in vitro and in vivo. J Exp Med 165:302, 1987 3. Jenkins MK, Pardoll DM, Mizuguchi J, Chused TM, Schwartz RH: Molecular events in the induction of a nonresponsive state in interleukin 2-producing helper T-lymphocyte clones. Proc Natl Acad Sci USA 84:5409, 1987 4. Jenkins MK, Ashwell JD, Schwartz RH: Allogeneic non-T spleen cells restore the responsiveness of normal T cell clones stimulated with antigen and chemically modified presenting cells. J Immuno1 140:3324, 1988 5. DeSilva DR, Urdahl KB, Jenkins M K Clonal anergy is induced in vitro by T cell receptor occupancy in the absence of proliferation. J Immunol 147:3261, 1991 6. June CH, Ledbetter JA, Gillespie MM, Lindsten T, Thompson CB: T-cell proliferation involving the CD28 pathway is associated with cyclosporine-resistant interleukin 2 gene expression. Mol Cell Biol 74472, 1987 7. Thompson CB, Lindsten T, Ledbetter JA, Kunkel SL, Young HA, Emerson SG, Leiden JM, June CH: CD28 activation pathway regulates the production of multiple T-cell-derived lymphokines/ cytokines. Proc Natl Acad Sci USA 86:1333, 1989 8. June CH, Ledbetter JA, Linsley PS, Thompson CB: Role of the CD28 receptor in T-cell activation. Immunol Today 11:211,1990 9. Liu Y, Janeway CA: Cells that present both specific ligand and costimulatory activity are the most efficient inducers of clonal expansion of normal CD4 T cells. Proc Natl Acad Sci USA 89:3845, 1991 10. Linsley PS, Brady W, Urnes M, Grosmarie LS, Ledbetter JA, Damle N K CTLA-4 is a second receptor for the B cell activation antigen B7. J Exp Med 174:561, 1991 11. Tan P, Anasetti C, Hansen JA, Melrose J, Brunvand M, Bradshaw J, Ledbetter JA, Linsley PS: Induction of alloantigen-specific hyporesponsiveness in human T lymphocytes by blocking interaction of CD28 with its natural ligand B7/BBI. J Exp Med 177:165, 1993 12. Gimmi CD, Freeman GJ, Gribben JG, Gray CS, Nadler LM: Human T cell clonal anergy is induced by antigen presentation in the absence of B7 costimulation. Proc Natl Acad Sci USA 90:6586, 1993 2617 13. Boussiotis VA, Freeman GJ, Gray G, Gribben J, Nadler LM: B7 but not intracellular adhesion molecule-l costimulation prevents the induction of human alloantigen-specific tolerance. J Exp Med 178:1753, 1993 14. Blazar BR, Taylor PA, Linsley PS, Vallera D A In vivo blockade of CD28/CTLA4B7/BBl interaction with CTLA4-Ig reduces lethal murine graft-versus-host disease across the major histocompatibility complex (MHC) barrier in mice. Blood 83:3815, 1994 15. Blazar BR, Taylor PA, Gray CS, Vallera DA: The role of T cell subsets in regulating the in vivo efficacy of CTLA4-Ig in preventing graft-versus-host disease in recipients of fully MHC or minor histocompatibility (miH) only disparate donor inocula. Transplantation 58:1422, 1994 16. Turka LA, Linsley PS, Lin H, Brady W, Leiden JM, Wei R-Q, Gibson ML, Zheng X-G, Myrdal S , Gordon D, Bailey T, Boling SF, Thompson CB: T-cell activation by the CD28 ligand B7 is required for cardiac allograft rejection in vivo. Proc Natl Acad Sci USA 89:11102, 1992 17. Lin H, Boloing SF, Linsley PS, Wei R-Q, Gordon D, Thompson CB, Turka LA: Long-term acceptance of major histocompatibility complex mismatched cardiac allografts induced by CTLA4-Ig plus donor-specific transfusion. J Exp Med 178:1801, 1993 18. Jenkins MK: The role of cell division inthe induction of clonal anergy. Immunol Today 13:69, 1992 19. Lenschow DJ, ZRng Y,Thistlethwaite R, Montag A, Brady W, Gibson MC, Linsley PS, Bluestone JA: Long-term survival of xenogeneic pancreatic islet grafts by CTLA4-Ig. Science 257:789, 1992 20. Ferrara JLM, Deeg HJ: Graft-versus-host disease. N Engl J Med 324:667, 1991 21. Isobe M, Yagita H, Okumura K, Ihara A: Specific acceptance of cardiac allograft after treatment with antibodies to ICAM-1 and LFA-I. Science 255:1125, 1992 22. Qin L, Chavin KD, Lin J, Yagita H, Bromberg JS: Anti-CD2 receptor and anti-CD2 ligand (CD48) antibodies synergize to prolong allograft survival. J Exp Med 179:341, 1994 23. Dune FH, Fava RA, Foy TM, Aruffo A, Ledbetter JA, Noelle RJ: Prevention of collagen-induced arthritis with an antibody to gp39, the ligand for CD40. Science 261:1328, 1993 24. Foy TM, Shepherd DM, Dune FH, Aruffo A, Ledbetter JA, Noelle RJ: In vivo CD40-gp39 interactions are essential for thymusdependent humoral immunity. 11. Prolonged suppression of the humoral immune response by an antibody to the ligand for CD40, gp30. J Exp Med 178:1567, 1993 25. Schlegel PG, Vaysburd M, Chao NJ: Selective inhibition of T cell costimulation mediated by VCAM-1 prevents murine graftversus-host disease. Blood 82:456a, 1993 (suppl, abstr) 26. Haming R, Pelletier J, Lubbe K, Takei F, Merluzzi VJ: Reduction in the severity of graft-versus-host disease and increased survival in allogeneic mice by treatment with monoclonal antibodies to cell adhesion antigens LFA-I alpha and MALA-2. Transplantation 52:842, 1991 27. Blazar BR, Soderling CCB, Koo GC, Vallera DA: Absence of a facilitory role for NK 1.1 positive donor cells in engraftment across a major histocompatibility barrier in mice. Transplantation 45:876, 1988 28. Blazar BR, Thiele DL, Vallera DA: Pretreatment of murine donor grafts with L-leucyl-L-leucine methyl ester: Elimination of graft-versus-host disease without detrimental effects on engraftment. Blood 72798, 1990 29. Blazar BR, Carroll SF, Vallera DA: Prevention of murine graft versus host disease and bone marrow alloengraftment across the MHC harrier following donor graft preincubation with anti-LFAl immunotoxin. Blood 78:3093, 1991 30. Blazar BR, Quinones RR, Heinitz KJ, Sevenich E, Filipovich From www.bloodjournal.org by guest on February 6, 2015. For personal use only. 2618 AH: Comparison of three techniques of T-cell depletion in human bone marrow. Exp Hematol 13:123, 1985 3 1. Vallera DA, Carroll SF, Snover D, Carlson GJ, Blazar BR: Toxicity and efficacy of anti-T cell ricin toxin A chain (RTA) immunotoxins in a murine model of ongoing GVHD across themajor histocompatibility barrier. Blood 77: 182, 1990 32. Blazar BR, Taylor PA, Snover DC, Bluestone JA, Vallera DA: Non-mitogenic anti-CD3F(ab’)2 fragments prevent lethal murine graft-versus-host disease induced across the major histocompatibility barrier. J Immunol 150:265, 1993 33. Kuhlman P, Moy VT, Loll0 BA, Brian AA: The accessory cell function of murine intracellular adhesion molecule-l in T lymphocyte activation. Contributions of adhesion and coactivation. J Immunol 146:1773, 1991 34. Hsu S-M, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in imunoperoxidase techniques. J Histochem Cytochem 29577, 1981 35. Weibel ER: Point counting methods, in Weibel ER (ed): Sterological Methods: Practical Methods for Biological Morphometry, v01 I . New York, NY, Academic, 1979, p 101 36. Taswell C: Limiting dilution assays for the determination of immunocompetent cell frequencies. I. Data analysis. J Immunol 126:1614, 1991 37. Blazar BR, Hirsch R, Gress RE, Carroll SF, Vallera DA: In vivo administration of monoclonal antibodies or immunotoxins in murine recipients of allogeneic T-cell depleted marrow for the promotion of engraftment. J Immunol 147:1492, 1991 38. Unkless JC: Characterization of a monoclonal antibody directed against mouse macrophage and lymphocyte Fc receptors. J Exp Med 150:580, 1979 39. Ledbetter JA, Rouse R, Spedding-Micklem H, Herzenberg L: T cell subsets defined by expression of Lyt-l, 2,and 3 and Thy-l antigens. J Exp Med 152:280, 1980 40. Razi-Wolf 2, Freeman GJ, Galvin F, Bendcerraf B, Nadler L, Reiser H: Expression and function of the murine B7 antigen, the major costimulatory molecule expressed by peritoneal exudate cells. Proc Natl Acad Sci USA 89:4210, 1992 41. Haynes BF, Eisenbarth GS, Fauci AS: Production of a monoclonal antibody that defines functional thymus-derived subsets. Proc Natl Acad Sci USA 765829, 1979 42. Mantel N: Evaluation of survival data in two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163, 1966 43. Lenschow DJ, Su GH-T, Zuckerman LA, Nabavi N, Jellis CL, Gray GS, Miller J, Bluestone JA: Expression and functional significance of an additional ligand for CTLA-4. ProcNatlAcad Sci USA 90:11054, 1993 44. Quill H, Schwartz RH: Stimulation of normal inducer T cell clones with antigen presented by purified Ia molecules in planar lipid membranes: Specific induction of a long-lived state of proliferative nonresponsiveness. J Immunol 138:3704, 1987 4.5. Theobald M, Nierle T, Bunjes D, Arnold R, Heimpel H: Hostspecific interleukin-2-secreting donor T-cell precursors as predictors BLAZAR ET AL of acute graft-versus-host disease inbonemarrow transplantation between HLA-identical siblings. N Engl J Med 327: 1613, 1992 46. Gribben JG, Boussiotis V, Freeman G, Gray G, NadlerL: Blockade of the B7 family of costimulatory molecules significantly reduces alloreactive T-cells capable of inducing GVHD. Blood 82:456a, 1993 (suppl, abstr) 47. McArthur JG, Raulet DH: CD28-induced costimulation of T helper 2 cells mediated by induction of responsiveness to interleukin4. J Exp Med 178:1645, 1993 48. DeWitD,Van Mechelen M, Ryelandt M, Figueiredo C, Ambramowicz D, Goldman M, BazinH,Urbain J , Leo 0: The injection of deaggregated gamma globulins in adult mice induces antigen-specific unresponsiveness of T helper type I but not type 2 lymphocytes. J Exp Med 175:9, 1992 49. Seder RA, Germain RN, Linsley PS, Paul WE: CD28-mediated costimulation of interleukin 2 (IL-2) production plays a critical role in T cell priming for IL-4 and interferon y production. J Exp Med 179299, 1994 50. Rolink AG, Gleichmann E: Allosuppressor- and allohelperT cells in acute and chronic graft-vs.-host (GVH) disease. 111. Different Lyt subsets of donor T cells induce different pathological syndromes. J Exp Med 158546, 1983 51. Allen RD, Staley TA, Sidman CL: Differential cytokine expression in acute and chronic murine graft-versus-host-disease. Eur J Immunol 23:333, 1993 52. Fowler DH, Kurasawa K, Husebekk A, Cohen PA, Gress RE : Cells of Th2 cytokine phenotype prevent LPS-induced lethality during murine graft-versus-host disease reaction: Regulation of cytokines and CD8+ lymphoid engraftment. J Immunol 152:1004, 1994 53. Damle NK, Klussman K, Linsley PS, Aruffo A: Differential costimulatory effects of adhesion molecules B7, ICAM-I, LFA-3, and VCAM-I on resting and antigen-primed CD4’ T lymphocytes. J Immunol 148:1985, 1992 54. Green JM, Zheng XG, Shimizu Y, Thompson CB, Turka LA: T cell receptor stimulation, but not CD28 costimulation, is dependent on LFA-l-mediated events. Eur J Immunol 24: 265, 1994 SS. Van Dijken PJ, Ghayur T, Mauch P, Down J, Burakoff SJ, Ferrara JL: Evidence that anti-LFA 1 - 1 in vivo improves engraftment and survival after allogeneic bonemarrow transplantation. Transplantation 49:882 56. FischerA, Landais P, Friedrich W, Gerritsen B,Fasth A, Porta F, Vellodi A, Benkerrou M, Jais JP, Cavazzana-Calvo MM, Souillet G, Bordigoni P, Morgan G, Van Dijken P, Vossen J, Locatelli F, di Bartolomeo P: Bone marrow transplantation (BMT) in Europe for primary immunodeficiencies other than severe combined immunodeficiency: A report from the European Group for BMT and the European Group for Immunodeficiency. Blood 83: 1149, 1994 57. Stoppa AM, Maraninchi D, Blaise D, Viens P, Him M, Olive D, Reiffers J, Milpied N, Gaspard MH, Mawas C: Anti-LFAl monoclonal antibody (25.3) for treatment of steroid-resistant grade 111-IV acute graft-versus-host disease. Transplant Int 4:3, 1991 58. Blazar BR, Taylor PA, Vallera DA: In vivo or in vitro antiCD3e chain monoclonal antibody therapy for the prevention of lethal murine graft-versus-host disease across the major barrier in mice. J Immunol 152:366S,1994 From www.bloodjournal.org by guest on February 6, 2015. For personal use only. 1995 85: 2607-2618 Coblockade of the LFA1:ICAM and CD28/CTLA4:B7 pathways is a highly effective means of preventing acute lethal graft-versus-host disease induced by fully major histocompatibility complex-disparate donor grafts BR Blazar, PA Taylor, A Panoskaltsis-Mortari, GS Gray and DA Vallera Updated information and services can be found at: http://www.bloodjournal.org/content/85/9/2607.full.html Articles on similar topics can be found in the following Blood collections Information about reproducing this article in parts or in its entirety may be found online at: http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests Information about ordering reprints may be found online at: http://www.bloodjournal.org/site/misc/rights.xhtml#reprints Information about subscriptions and ASH membership may be found online at: http://www.bloodjournal.org/site/subscriptions/index.xhtml Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036. Copyright 2011 by The American Society of Hematology; all rights reserved.
© Copyright 2024