Prevention of Acute Graft-Versus-Host Disease in a Xenogeneic SCID

Biol Blood Marrow Transplant 19 (2013) 28e39

Prevention of Acute Graft-versus-Host Disease in a Xenogeneic SCID Mouse Model by the Humanized ASBMT American Society for Blood Anti-CD74 Antagonistic Antibody Milatuzumab and Marrow Transplantation

Xiaochuan Chen 1,*, Chien-Hsing Chang 2, Rhona Stein 1, Thomas M. Cardillo 2, David V. Gold 1, David M. Goldenberg 1,*

1 Center for Molecular Medicine and Immunology, Garden State Cancer Center, Morris Plains, New Jersey 2 Immunomedics, Inc., Morris Plains, New Jersey

Article history: abstract Received 15 August 2012 Prevention and treatment of graft-versus-host disease (GVHD) remains a major challenge, given that current Accepted 20 September 2012 T-cell depletion and mainstay immunosuppressive therapies compromise preexisting T-cell immunity, often leading to severe infections and disease relapse. Thus, there is a critical need for novel anti-GVHD agents that Key Words: can spare protective T-cell memory. Here we show that milatuzumab (hLL1), a humanized anti-CD74 Milatuzumab antagonist monoclonal antibody, can moderately reduce the numbers of CD74-expressing B cells and CD74 myeloid dendritic cells, but has no effect on the survival of T cells that are CD74 . Consequently, milatuzumab Alemtuzumab inhibits allogeneic T-cell proliferation in mixed leukocyte reactions. In a human/mouse xenogeneic SCID Graft-versus-host disease þ mouse model in which GVHD is induced and mediated by engrafted human CD4 T cells and dendritic cells, milatuzumab effectively prevents the onset and manifestations of acute GVHD, suppresses serum levels of human IFN-g and IL-5, eliminates the infiltration of human lymphocytes into GVHD target organs (ie, lung, liver, and spleen), and significantly promotes survival (90% versus 20% for controls; P ¼ .0012). Importantly, exposure to milatuzumab does not affect the number of cytomegalovirus-specific, IFN-geproducing human þ CD8 T cells in allogeneic mixed leukocyte reactions. These encouraging results warrant further exploration of milatuzumab as a possible new therapeutic agent for GVHD. Ó 2013 American Society for Blood and Marrow Transplantation.

INTRODUCTION antigen-presenting cells (APCs), which include B cells, Graft-versus-host disease (GVHD) is the major cause of monocytes, macrophages, Langerhans cells, dendritic cells mortality and morbidity in patients undergoing allogeneic (DCs), and endothelial and certain epithelial cells [7,14]. hematopoietic stem cell transplantation (allo-HSCT), Because both recipient and donor APCs, including non- a powerful and currently the sole curative therapy for many hematopoietic APCs, play critical roles in the initiation of malignant and nonmalignant hematologic diseases [1]. GVHD [15-24], we reasoned that milatuzumab might have Prevention and treatment of GVHD remains a major chal- therapeutic potential for GVHD by affecting recipient and/or lenge [2], given that contemporary T-cell depletion donor APCs. The anti-GVHD potential of anti-CD74 mAb is and immunosuppressive therapies, although effective in also supported by evidence that macrophage migration controlling GVHD in certain patients, compromise preexist- inhibitory factor (MIF), the ligand of CD74 [25], is involved in ing T-cell immunity, leading to severe infections and disease the development of acute GVHD in a murine model of allo- relapse [3-6]. Thus, there is a critical need for novel anti- geneic stem cell transplantation [26]. GVHD agents that spare protective T-cell memory. In this study, we demonstrate that milatuzumab selec- Milatuzumab (hLL1) is a humanized IgG1k mAb that tively reduces myeloid DCs (mDCs) and B cells, but not reacts with human CD74 [7-9], the HLA class IIeassociated plasmacytoid DCs (pDCs), monocytes, or T cells, in human invariant chain [10]. Previous studies found that milatuzu- peripheral blood mononuclear cells (PBMCs). As a result, mab has potent cytotoxicity against CD74-expressing malig- milatuzumab inhibits allogeneic mixed lymphocyte reac- nant B cells in vitro and in xenograft models [7,11], which has tions (allo-MLRs); and in a human PBMC-transplanted SCID led to the ongoing clinical evaluation of milatuzumab in mouse model (hu-PBL-SCID) [27], milatuzumab effectively relapsed or refractory B-cell malignancies [12]. More recent prevents acute GVHD, suppresses human cytokine storm, preclinical studies [13] have demonstrated that milatuzumab eliminates infiltration of human lymphocytes in GVHD target is capable of modulating human B cell proliferation, migra- organs, and significantly improves survival. Unlike alemtu- tion, and adhesion molecule expression, suggesting the zumab, an anti-CD52 mAb currently used clinically for GVHD therapeutic potential of this mAb in autoimmune diseases. prevention, milatuzumab does not affect cytomegalovirus As an HLA class II invariant chain molecule, CD74 is widely (CMV)-specific T-cell immunity in vitro, suggesting that it expressed in both hematopoietic and nonhematopoietic might be exploited as a novel agent for GVHD without compromising preexisting antiviral immunity.

Financial disclosure: See Acknowledgments on page 38. * Correspondence and reprint requests: Xiaochuan Chen or MATERIALS AND METHODS David M. Goldenberg, Garden State Cancer Center, Center for Molecular Antibodies and Reagents Medicine and Immunology, 300 The American Road, Morris Plains, Milatuzumab (hLL1), labetuzumab (hMN-14; a humanized anti- NJ 07950. CEACAM5 mAb) [28], epratuzumab (hLL2; a humanized anti-CD22 mAb) E-mail addresses: [email protected] (X. Chen), [email protected] [29], IMMU-T3 (a humanized anti-CD3 mAb), and hLL1-Fab-A3B3 (a fusion (D.M. Goldenberg). protein of milatuzumab Fab fragment with CEACAM5 [CD66e] A3B3 domain 1083-8791/$ e see front matter Ó 2013 American Society for Blood and Marrow Transplantation. http://dx.doi.org/10.1016/j.bbmt.2012.09.015 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 29

Figure 1. Milatuzumab selectively reduces myeloid DCs in human PBMCs. Human PBMCs were incubated with 5 mg/mL of milatuzumab, control mAbs, or medium only for 3 days. The effect of each treatment on APC subsets was evaluated by costaining the cells with PE-labeled anti-CD14 and anti-CD19 IgG, in combination with allophycocyanin-labeled antieBDCA-1 IgG, for analysis of mDC1, or a mixture of FITC-labeled antieBDCA-2 and allophycocyanin-labeled antieBDCA-3 IgG for simultaneous analysis of mDC2s and pDCs, respectively. PBMCs were gated to exclude the debris and dead cells on the basis of their forward scatter and side scatter þ þ characteristics. (A) Subpopulations of PBMCs were gated as follows: monocytes, CD14 SSCmedium; B cells, CD19 SSClow; T cells, CD19 CD14 SSClow; mDC1, þ þþ þ CD14 CD19 BDCA-1 ; mDC2, CD14 CD19 BDCA-3 ; pDCs, CD14 CD19 BDCA-2 . (B-E) Representative ﬂow cytometry data for B cells (B), monocytes (B), non-B lymphocytes (B), mDC1s (C), mDC2s (D), and pDCs (E) in PBMCs after mAb treatment. (F-H) Mean percentages of live mDC1s, mDC2s, and pDCs in PBMCs after mAb treatment (n ¼ 7 donors). Error bars indicate SD. P values were determined by the paired t test. **P<.05; ***P<.01 versus hMN-14 control.

[30]), were provided by Immunomedics (Morris Plains, NJ). Rituximab (anti- labeled anti-CD14 and anti-CD19 in combination with allophycocyanin- CD20) was purchased from IDEC Pharmaceuticals (San Diego, CA). Alemtu- labeled anti-BDCA-1. After washing, the cells were analyzed by flow zumab (anti-CD52) was obtained from Genzyme (Ridgefield, NJ). The mAbs cytometry using the gating strategy described below. Live PBMCs were used for flow cytometry were obtained from BD Pharmingen (San Jose, CA): gated based on forward scatter and side scatter signals. Within live PBMCs, þ FITC-labeled anti-CD74 (M-B741), FITC-labeled mouse IgG1, PE-labeled anti- type 1 mDCs (mDC1s) were identified as CD14 19 BDCA-1 cell pop- CD25, PerCp-labeled anti-CD8, and allophycocyanin-labeled anti-CD4 and ulations [31], and the lymphocyte population was analyzed for B cells þ CD3; or from Miltenyi Biotec (Auburn, CA): PE-labeled mAbs to CD19 (LT19) (CD19 SSClow), non-B lymphocytes (primarily T cells) (CD19 14 SSClow), þ and CD14 (TÜK4) and allophycocyanin-labeled mAbs to BDCA-1 (AD5-8E7), and monocytes (CD14 SSCmedium). To measure the frequencies of pDCs and BDCA-2 (AC144), and BDCA-3 (AD5-14H12). Human IgG was obtained from type 2 mDCs (mDC2s), PBMCs were stained with PE-labeled anti-CD14 and Jackson ImmunoResearch (West Grove, PA). HLA-A*0201 restricted CMV anti-CD19 IgG, in combination with FITC-labeled antieBDCA-2 and pp65 peptide (NLVPMVATV) and HIV gag peptide (SL9, SLYNTVATL) were allophycocyanin-labeled antieBDCA-3 IgG. Within the live PBMCs, mDC2s þ obtained from Anaspec (Fremont, CA). were identified as the CD14 19 BDCA-3 cell population, whereas pDCs þ were identified as the CD14 19 BDCA-2 cell population [31]. Flow cytometry was performed using a FACSCalibur (BD Biosciences, San Jose, Assessment of APC Subsets in PBMCs CA), and data were analyzed with FlowJo software (Tree Star, Ashland, OR). Buffy coats from expired whole blood were obtained from healthy donors at the Blood Center of New Jersey (East Orange, NJ), after approval by the New England Institutional Review Board. PBMCs were isolated from Assessment of T-Cell Apoptosis and Survival buffy coats by standard density-gradient centrifugation over Ficoll-Paque. Human PBMCs were treated with hMN-14, milatuzumab, or alemtuzu- The isolated PBMCs were treated with milatuzumab or other mAbs at mab at different concentrations for 18 hours in the presence of phytohe- 37 Cin5%CO2 for 3 days. After incubation, the cells were stained with PE- magglutinin (final concentration, 2.5 mg/mL) and IL-2 (final concentration, 30 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39

Figure 1. (continued).

50 U/mL). After treatment, the cells were stained with Alexa Fluor 28) were euthanized, and livers, spleens, and lungs were obtained for 488elabeled annexin V, PE-labeled anti-CD25, and allophycocyanin-labeled analysis. Before euthanasia, peripheral blood was collected for serum anti-CD3. After washing, the cells were stained with 7-aminoactinomycin-D cytokine measurements using a chromatic bead assay (CBA) kit (BD (7-AAD; Life Technologies, Grand Island, NY), followed by flow cytometry Biosciences). Livers, spleens, and lungs were fixed in 10% formalin, þ þ analysis of early (annexin V 7-AAD ) and late apoptotic cells (annexin V 7- embedded in paraffin, and sectioned for hematoxylin and eosin histological þ AAD ) and live cells (annexin V 7-AAD ) [31] in activated T cells examination of periportal lymphocytic infiltration, as well as immunohis- þ þ þ þ þ (CD3 CD25 ) and nonactivated T cells (CD3 CD25 ) within PBMCs. tochemical staining for human CD3 and CD74 cells, using the humanized mAbs IMMU-T3 and milatuzumab, respectively, as described previously [33]. In addition, blood samples collected from cardiac puncture of anes- Two-Way Allo-MLR thetized animals were used for measuring circulating Th1/Th2 cytokines by Proliferation of T cells in the presence of APCs was evaluated by 2-way CBA (BD Biosciences). allo-MLR without irradiation of PBMCs from either donor [31,32]. PBMCs from different donors were labeled with 1 mM carboxyfluorescein succini- Quantitation of CMV-Specific T Cells in Allo-MLR by Assessing CMV- midyl ester (CFSE; Life Technologies) following the manufacturer’s instruc- Specific IFN-g Response tions. After extensive washing, cells from different donors were mixed with PBMCs from donors with HLA-A*0201erestricted, CMV-specific IFN-g one another at a 1:1 ratio and then cultured with milatuzumab or control responses, as precharacterized by Cellular Technology (Shaker Heights, OH), mAbs at 37 Cin5%CO2 for 7-10 days. Starting on day 4, human IL-2 (final were mixed with PBMCs from other donors, irrespective of HLA type and concentration 50 U/mL; Roche Applied Science, Penzburg, Germany) was CMV status, in the presence of milatuzumab or control mAbs at 5 mg/mL. The added to the culture, and the cells were split every other day by adding fresh mixed cells were cultured for 4 days in RPMI-1640 medium with 10% fetal medium without mAbs. To analyze cell proliferation, cells were harvested, bovine serum (FBS), followed by the addition of 50 U/mL of IL-2, and then and the CFSElow cell frequencies were quantitated by flow cytometry. In some cultured for another 2 days. Cells were then harvested and incubated with experiments, cells from MLRs were stained with anti-CD4 and anti-CD8 mAbs CMV pp65 peptide (NLVPMVATV) [27] or HIV gag peptide (SL9; SLYNTVATL) for flow cytometry analysis of the percentages of T-cell subsets in allo-MLRs. [27] at 10 mg/mL for 6 hours in the presence of brefeldin A at a final concentration of 1 mg/mL. The cells were then stained with PerCp-CD3 and Hu-PBL-SCID Mouse Model of GVHD FITC-CD8, followed by permeabilization and fixation using BD Cytofix/ All animal procedures were approved by the Institutional Animal Care Cytoperm solution (BD Pharmingen) and intracellular staining with allo- þ and Use Committee of the University of Medicine and Dentistry of New phycocyanin-IFN-g mAb. The percentages of IFN-g cells stimulated by CMV þ þ Jersey. The hu-PBL-SCID mouse model of GVHD was established by pp65 peptides in total MLR cells and CD3 CD8 cells were determined by following the method of Wilson et al. [27]. SCID mice (C.B-Igh-1b/IcrTac- flow cytometry. Prkdcscid; 5-7 weeks old) were purchased from Taconic (Rensselaer, NY), housed in sterile microisolation cages, and given autoclaved food and water. Statistical Analyses After a 10-day acclimation period, on study day 1, each mouse was injected Statistical significance between mAb treatment and controls was i.p. with 20 mL of asialo-GM-1 (Wako Chemicals, Richmond, VA) and irra- assessed using the paired t test. Animal survival data were analyzed with 7 diated (325 cGy). On the next day (day 0), 5 10 fresh human PBMCs in Prism (GraphPad Software, La Jolla, CA). Body weight changes were analyzed 200 mL of serum-free therapeutic-grade AIM-V medium (Life Technologies) by analysis of variance using a generalized linear model approach, and were transplanted into each mouse by i.p. injection. Milatuzumab or control GVHD clinical scores were analyzed using a generalized estimation equa- Abs were administered by i.p. injection at 3 h before human PBMC trans- tions approach [34] with SAS software (SAS Institute, Cary, NC). plantation, with doses of 350 and 700 mg/mouse, equivalent to 1.46 and 2.92 mg/kg in humans. Mice were then assessed daily using a GVHD scoring RESULTS system that measures 6 items related to clinical signs of GVHD: weight loss, Milatuzumab Selectively Reduces mDCs posture, activity, fur texture, skin integrity, and diarrhea [27]. The overall score for each mouse was the sum of the 6 individual scores (0-2 points for To test the potential cytotoxicity of milatuzumab on DCs each). Mice with severe GVHD (overall score 5) or at the end of study (day and other hematopoietic APC subsets, human PBMCs were X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 31

Figure 2. Milatuzumab suppresses T-cell proliferation in allo-MLR. CFSE-labeled PBMCs from different donors were mixed with one another and incubated with 5 mg/ mL of milatuzumab or other mAbs (hMN-14, rituximab, or hLL1-Fab-A3B3), or medium only, for 8 days. IL-2 was added to the culture on day 4. On day 8, the cells were harvested and analyzed by flow cytometry. Proliferating cells were quantitated by measuring CFSElow cell frequencies. Representative data from one combination of stimulator/responder PBMCs are shown in (A), and statistical analysis of all combinations is shown in (B). (C) In another experiment, allo-MLRs and mAb treatments were the same as in (A) and (B), but stained with PerCp-labeled CD4 and allophycocyanin-labeled CD8, followed by flow cytometry analysis of the þ þ þ þ CFSElow cell frequencies in gated CD4 and CD8 T cells. (D) Statistical analysis of CD4 and CD8 T cells with diluted CFSE. CD4high T cells in total live cells of allo- MLRs were gated and analyzed (n ¼ 10 stimulator/responder combinations from 5 donors). Error bars indicate SD. Statistical significance was determined by the paired t test.

treated in vitro with milatuzumab or other mAbs for 3 days, (Figure 1B and E). Similar results were obtained in human followed by the enumeration of various APC subsets in whole blood, in which milatuzumab at 5 mg/mL decreased PBMCs. As shown in Figure 1A-C, mDC1 cells (defined as the mDC1 and B cells, but not T cells, after 2 days of incubation þ CD14 CD19 BDCA-1 cell population [31]) in milatuzumab- in vitro (Supplemental Figure S1A). The differential effects of treated PBMCs were reduced by 59.8% compared with milatuzumab on different PBMC subsets may be related to control anti-carcinoembryonic antigen (CEACAM5; CD66e) the different expression levels of CD74 on these cells. As antibody hMN-14etreated cells (P ¼ .0158, n ¼ 4 donors) shown in Supplemental Figure S2A and B, the CD74 expres- (Figure 1C), and B cells were reduced by 21.1% (P ¼ .0035, sion was highest on mDC2s and mDC1s and lowest on noneB þ n ¼ 4 donors), whereas in monocytes (CD14 SSChigh cell lymphocytes, which is correlated with the efficacy of population) and noneB lymphocytes (CD14 CD19 cell reduction on APC subsets by milatuzumab (Supplemental population, >80% T cells [33]) were not affected (Figure 1B). Figure S2C and D). þ In separate experiments, mDC2 cells (CD14 CD19 BDCA-3 Compared with milatuzumab, the Fab fragment of mila- cell population [31]) were reduced by 53.8% (P < .05, n ¼ 7 tuzumab (hLL1-Fab-A3B3), which is linked to an irrelevant þ donors) (Figure 1D), and pDCs (CD14 CD19 BDCA-2 cell A3B3 protein domain of CEACAM5 [30], did not exhibit any population) were not affected (Figure 1E). As expected, rit- effect on all APC subsets (Figure 1F-H), indicating that either uximab, the chimeric anti-CD20 mAb, significantly reduced B divalent binding or the Fc Ab portion is required for the cells (53.2% reduction; P ¼ .0129, n ¼ 4 donors) (Figure 1B) action of milatuzumab. However, milatuzumab failed to kill but had no effect on monocytes, DCs, or non-B lymphocytes purified mDC1s or mDC2s (Supplemental Figure S3A and B), 32 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39

Figure 3. Milatuzumab has no effect on activated or nonactivated T cells. Human PBMCs were treated with milatuzumab, control mAb hMN-14, or alemtuzumab at the indicated concentrations for 18 h in the presence of 2.5 mg/mL of phytohemagglutinin and 50 U/mL of IL-2. The cells were then stained with Alexa Fluor 488elabeled annexin V, PE-labeled anti-CD25, allophycocyanin-labeled anti-CD3, and 7-AAD for ﬂow cytometry analysis. (A) Gating proﬁle of early apoptotic cells, late apoptotic cells, and live cells in activated T cells (CD3þCD25þ) and nonactivated T cells (CD3þCD25). (B and C) Apoptotic and live cells in activated T cells (B) and nonactivated T cells (C) (n ¼ 6 PBMC donors). *P < .05; **P < .01; ***P < .001 versus hMN-14 control. X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 33

Figure 4. Milatuzumab prevents acute GVHD in SCID mice undergoing transplantation with human PBMCs. After sublethal total body irradiation (3.25 Gy), 40 SCID mice were grouped into 10 blocks following a randomized block design. The 4 mice in each block received 5 107 human PBMCs i.p. from the same donor. At 3 hours before transplantation, each of the 4 mice in each block was either untreated or given an i.p. injection of milatuzumab (350 or 700 mg) or alemtuzumab (5 mg). Throughout the 28-day study period, GVHD scores and body weight were assessed daily. Mice were euthanized at a GVHD score of 5. (A) Euthanized or spontaneously expired mice were recorded for survival analysis by each treatment. (B) Daily recordings of total GVHD scores and survival over the full 28-day study period. (C) Summary of total GVHD scores. (D) Body weight loss. Signiﬁcant prevention of mortality (A) and body weight loss (D) was seen in the milatuzumab- and alemtuzumab-treated mice. Milatuzumab also reduced GVHD scores in a dose-dependent manner (B and C). The survival data (A) were analyzed using GraphPad Prism. GVHD clinical scores (B and C) were analyzed using the generalized estimation equation approach, and body weight changes (D) were analyzed by analysis of variance. suggesting that the reduction of mDC1s and mDC2s in Milatuzumab Inhibits Allo-MLR human PBMCs may occuring through the ligation of other Because DCs are the most potent APCs for stimulating cell types by the Fc of milatuzumab. T cells, we next investigated whether the reduction of mDC1s 34 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39

Figure 4. (continued). and mDC2s in PBMCs by milatuzumab could be translated to this possibility in a xenogeneic SCID mouse model in which þ reduced T-cell proliferation in 2-way allo-MLRs [31,32]. After acute GVHD is mediated by engrafted human donor CD4 T 8 days of culture, allo-MLRs treated with the control anti- cells and DCs [27]. As shown in Figure 4A, 90% (9 of 10) of mice CEACAM5 mAb hMN-14 exhibited T-cell proliferation, char- injected i.p. with either 350 or 700 mg of milatuzumab at 3 h acterized by 23.7% of cells with CFSE dilution. In contrast, before transplantation survived for the full 28-day experi- T-cell proliferation was detected in only 8.4% of cells in mental period, compared with only 20% (2 of 10) of untreated allo-MLRs treated with milatuzumab (Figure 2A). Statistical mice (P < .0012). All of the 8 untreated mice that did not analysis of 10 different stimulatoreresponder combinations survive died spontaneously (ie, were found dead at the daily showed a significant reduction (P < .0001) in T-cell prolif- checkup) on day 10-15 (Figure 4B). In the mice treated with eration in milatuzumab-treated allo-MLRs (Figure 2B). milatuzumab 350 mg and 700 mg, only 1 of the 10 mice in each þ Milatuzumab also inhibited proliferation of both CD4 and group did not survive the full study period (28 days). Both of þ CD8 T cell subsets in allo-MLRs (Figure 2C and D). The these mice died spontaneously, one on day 20 (the 350 mg proliferation of CD4high T cells, a subpopulation of activated milatuzumab-treated mouse) and the other on day 13 (the þ CD4 T cells [35], was inhibited as well (Figure 2D). No 700 mg milatuzumab-treated mouse) (Figure 4B). Before they inhibition was found in hLL1-Fab-A3B3etreated allo-MLRs, were found dead, the animals exhibited hunching, fur ruffling, suggesting that the inhibition of T-cell proliferation of allo- reduced activity, and greater weight loss, but no diarrhea or MLRs by milatuzumab also requires the Fc portion or diva- skin lesions. In the same experiment, 5 mg of the clinically used lent binding. anti-GVHD anti-CD52 mAb alemtuzumab protected 100% (10 To exclude the possibility that milatuzumab might have of 10) of the mice at day 28. direct cytotoxicity on activated T cells, we tested the effect of We also performed direct temporal comparisons of GVHD þ milatuzumab on apoptosis and survival of activated (CD25 ) clinical scores and weight loss between treatments over the and nonactivated (CD25 ) T cells in phytohemagglutinin- full 28-day experiment. Throughout the study, GVHD scores, stimulated PBMCs. The results showed that milatuzumab body weight changes, and survival were assessed on a daily has no effect on early and late apoptosis or survival in either basis. Mice were euthanized at a GVHD score of 5. Figure 4B activated or nonactivated T cells (Figure 3A-C), which is shows daily-recorded GVHD scores in each hu-PBL- consistent with the lack of CD74 expression in total T cells SCID mouse over the full 28-day study period, which (Supplemental Figure S2). In comparison, the anti-CD52 mAb demonstrate that both milatuzumab and alemtuzumab alemtuzumab induced early apoptosis and reduced survival reduced GVHD scores in this xenogeneic animal model, and of T cells (Figure 3A-C). Thus, the inhibitory effect of mila- in 9 of 10 blocks of mice (Figure 4B, a-d and f-j), milatuzumab tuzumab on allo-MLRs is likely due not to a direct cytotoxic reduced the scores in a dose-dependent manner. Whereas effect on T cells, but rather to the reduction of mDCs and B untreated control hu-SCID mice experienced severe GVHD, cells, or to the antagonistic blocking of MIF by milatuzumab. mice treated with either milatuzumab (350 or 700 mg) or As the ligand of CD74, MIF plays an essential regulatory role alemtuzumab (5 mg) had significantly lower GVHD scores in T-cell activation [36]. (Figure 4B and C; P < .001) and less body weight loss (Figure 4D; P < .001). Similar results also were obtained from Milatuzumab Prevents Acute GVHD in hu-PBL-SCID Mice an earlier pilot animal study, showing that 2 of 2 The reduction of mDCs and inhibition of allo-MLRs suggest milatuzumab-treated hu-SCID mice had much lower GVHD that milatuzumab may be therapeutic for GVHD. We tested scores, less or no body weight loss, and no deaths throughout X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 35

and alemtuzumab effectively controlled acute GVHD in this xenogeneic model. þ þ Like human CD3 cells, human CD74 cells were detected in the lungs, liver, and spleen of untreated mice (Figure 5A), but not in milatuzumab-treated mice (Figure 5B) or alemtuzumab-treated mice (data not shown). Because CD74 is expressed in both APCs and a subset of activated T cells [7], þ the human CD74 cells represent donor APCs, including mDCs, and a subset of activated T cells in this xenogenic model. The disappearance of these cells in GVHD target organs in milatuzumab-treated mice provides evidence that milatuzumab controls GVHD by a CD74-dependent mechanism. It remains to be determined whether milatuzumab, in an in vivo environment, can directly kill donor-derived APCs and activated T cells in the GVHD target organs of this model, or can simply inhibit the recruitment of these cells to these organs through the antagonistic blocking of MIF on CD74.

Milatuzumab Suppresses Human Cytokine Storm in hu- PBL-SCID Mice In addition to clinical signs and pathological changes, acute GVHD is also characterized by a human “cytokine storm,” characterized by elevated serum levels of IFN-g, TNF- a, IL-1, IL-5, IL-6, and IL-8 [27,37]. These inflammatory cytokines are responsible for or contribute to the tissue damage and clinical manifestations of acute GVHD. The anti-CD52 mAb alemtuzumab and anti-CD83 polyclonal antibody that targets mature DCs reportedly can suppress this cytokine storm in parallel to their efficacy against GVHD in hu-PBL- SCID mice [27]; thus, we examined whether milatuzumab affected serum levels of proinflammatory cytokines at the Figure 5. Milatuzumab eliminates human T-cell and APC infiltration in GVHD target organs. At the end of the experiment shown in Figure 3, all surviving end of the animal study (ie, day 28 posttransplantation). The mice were euthanized. GVHD target organs (lungs, livers, and spleens) were results demonstrate that milatuzumab at 350 or 700 mg collected for pathological analysis. Human T cells and APCs within the GVHD substantially reduces the serum levels of IFN-g (by 4w5-fold; fi target organs were identi ed by immunohistochemical labeling of human CD3 P < .001) and IL-5 (by 5w7-fold; P < .001), but has no and CD74, respectively. Representative specimens are shown, with indications þ þ fi of human CD3 T cells and CD74 APCs in the specimens from untreated mice signi cant effect on TNF-a, IL-2, IL-4, and IL-10 (Figure 6A (A), but not in those from milatuzumab-treated mice (B). Perivascular and B). Similar results were obtained by treatment with lymphocytic engraftment is also evident in the hematoxylin and eosine- alemtuzumab. Thus, milatuzumab is capable of suppressing stained sections (arrows in A). the cytokine storm in acute GVHD.

Milatuzumab Does Not Affect CMV-Specific T-Cell the study period, compared with severe GVHD clinical signs, Immunity body weight loss, and spontaneous death or mandatory We showed that milatuzumab reduces mDC1s and mDC2s euthanasia in both untreated mice (Supplemental Figure S4A but spares CD74 non-B lymphocytes (Figure 1B), which are and B). In another animal study, all (5 of 5) hu-PBL-SCID mice composed mainly of Tcells [31], and has no effect on activated treated with milatuzuamb 350 mgor700mg or alemtuzumab or nonactivated T cells (Figure 1C). These results led us to 5 mg survived to day 11 when the study was terminated, postulate that milatuzumab may preserve preexisting whereas only 2 of 5 untreated or control antibody pathogen-specific memory T cells, which are essential for hMN-14etreated (700 mg) hu-PBL-SCID mice survived at this immunologic control of posttransplantation opportunistic time point. infections. To verify this, we examined milatuzumab’s effects on the survival and function of CMV-specific T cells in 2-way þ allo-MLRs in which hPBMCs from CMV , HLA-A*0201 þ þ Milatuzumab Eliminates Infiltration of CD3 and CD74 donors (n ¼ 5) were mixed with allogeneic PBMCs from Donor Cells in GVHD Target Organs other donors (n ¼ 5) with uncharacterized HLA type and CMV At the end (day 28) of the study shown in Figure 3, all status. The results, shown in Figure 7A and Supplemental surviving mice were euthanized to collect GVHD target Figure S5A and B, demonstrate that on stimulation with organs (ie, lungs, liver, and spleen) and peripheral blood for CMV pp65 peptide (NLVPMVATV), the frequencies of IFN- histopathological and cytokine analyses, respectively. geproducing cells in total MLR cells were no different in þ Immunohistochemistry showed human CD3 T cells infil- milatuzumab-treated (0.474 0.327%), control hMN- trating the lungs, liver, and spleen in untreated mice 14etreated (0.486 0.336%), and rituximab-treated (Figure 5A), but not in milatuzumab-treated mice (Figure 5B) (0.452 0.355%) MLRs. In contrast, alemtuzumab-treated or alemtuzumab-treated mice (data not shown). These MLRs had much lower frequencies of IFN-geproducing cells results provide histopathological evidence that acute GVHD than the control antibody hMN-14etreated MLRs was present in the untreated mice, and that milatuzumab (0.076 0.032% versus 0.486 0.336%; P ¼ .044, n ¼ 5). 36 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39

Figure 6. Milatuzumab suppresses the human cytokine storm in hu-PBL-SCID mice. At the end of the same experiment shown in Figures 3 and 4, mouse serum samples obtained at the end of the study (day 28) were analyzed for human Th1/Th2 cytokines by CBA. Data in (A) were obtained from 1 of the 2 blocks of mice given human PBMCs from 2 different donors. Only 2 of the 10 untreated mice survived to the end of the study (day 28), compared with 9 of the 10 milatuzumab-treated mice and all 10 of the alemtuzumab-treated mice. Therefore, only 2 blocks of mice, including the 2 surviving untreated mice together with their corresponding treated mice, were available for paired analyses. The statistical analysis of serum cytokine levels in all surviving mice is shown in (B) (n ¼ 2 untreated surviving mice versus n ¼ 9-10 milatuzumab- or alemtuzumab-treated surviving mice). Error bars indicate SD. Statistical signiﬁcance was determined with the t test. ***P < .001 versus untreated.

Similarly, there was no difference in IFN-geproducing cells in This is the first report showing that GVHD can be þ þ gated CD3 CD8 cells in milatuzumab-treated (1.83 1.09%), controlled by an antagonist mAb against CD74. Importantly, hMN-14etreated (1.91 1.12%), and rituximab-treated this antibody does not affect the survival of T cells, including (1.54 0.95%) MLRs (Figure 7B and Supplemental activated T cells, or antiviral T-cell immunity, as evidenced by þ Figure S5C and D). Alemtuzumab-treated MLRs had lower the preserved CMV-specific CD8 T cells in treated allo- þ þ frequencies of IFN-geproducing cells in gated CD3 CD8 cells MLRs, suggesting that milatuzumab may not compromise compared with the control antibody hMN-14etreated MLRs protective T-cell immunity against opportunistic infections. (0.81 0.83% versus 1.91 1.12%; n ¼ 5), but the difference did Because the graft-versus-leukemia (GVL) effect, the principal not reach statistical significance (P ¼ .1169), because alemtu- benefit and major therapeutic goal of allo-HSCT, is also zumab depleted both IFN-geproducing cells and the mediated by donor T cells, the lack of effect of milatuzumab þ þ CD3 CD8 cells, lessening its effect on the frequencies of IFN- on the survival of T cells implies that milatuzumab also might þ þ geproducing cells in CD3 CD8 cells. These results suggest not affect GVL. These promising results support further that milatuzumab maintains protective antiviral memory T- preclinical studies to examine this novel antibody for cell immunity, thus differing from the clinically used T- prophylactic and/or therapeutic control of GVHD. celledepleting mAb alemtuzumab, which causes high rates of Acute GVHD developed in hu-PBL-SCID mice is mediated CMV infection and reactivation [3-6]. by the engrafted human T cells and DCs [27]. In this animal model, the host (murine) APCs play a minor role in mediating DISCUSSION GVHD [27]. In GVHD patients, donor APCs contribute to the In this study, we have shown that milatuzumab, development of GVHD through the pathway of “indirect a humanized anti-CD74 antagonistic mAb, can reduce presentation” by processing and presenting host antigens to myeloid DCs and B cells, suppress the proliferation of allo- donor T cells [21-23], among which mDCs are a critical subset MLRs, and effectively prevent acute GVHD in a humanized for presenting alloantigens after experimental bone marrow SCID mouse model (hu-PBL-SCID). In this human/mouse transplantation [24]. Thus, the effective control of GVHD in xenogeneic GVHD animal model, a single i.p. administration hu-PBL-SCID mice by milatuzumab may be related, at least in of milatuzumab before human PBMC transplantation lowered part, to its ability to selectively reduce donor mDCs, thereby GVHD clinical scores, prevented weight loss, eliminated affecting the indirect presentation for GVHD development human lymphocyte infiltration in GVHD target organs, sup- [17]. This is similar to RA83, a polyclonal anti-CD83 antibody pressed the cytokine storm, and significantly improved the that effectively controls acute GVHD in hu-PBL-SCID mice survival of hu-PBL-SCID mice (90% versus 20% untreated through a proposed mechanism in which RA83 selectively controls). depletes activated donor human DCs that express CD83 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 37

þ þ Figure 7. CMV-specific CD8 T cells are preserved in allo-MLRs after exposure to milatuzumab. PBMCs from CMV HLA-A*0201 donors (n ¼ 5) were mixed with allogeneic PBMCs from other donors (n ¼ 5) with uncharacterized HLA type and CMV status. The mixed cells were cultured in the presence of milatuzumab or other mAbs for 8 days, followed by stimulation with CMV pp65 peptide (NLVPMVATV) or HIV gag peptide at 10 mg/mL for 6 hours. The cells were first stained with PerCp- labeled anti-CD3 and FITC-labeled anti-CD8, and then, after permeabilization and fixation, were intracellularly stained with allophycocyanin-labeled anti-IFN-g. (A) þ þ þ þ þ Frequencies of CD3 IFN-g cells in total MLR cells. (B) IFN-g cells in gated CD3 CD8 cells. Data shown are representative of 5 donorerecipient pairs of MLRs from 1 of 3 independent experiments.

[27,38]. CD83 is also expressed in activated T cells [38], the MIF gene in donor cells significantly prevents the onset of however, and thus RA83 could deplete activated donor T cells acute GVHD [28], suggesting that MIF/CD74 is actively as well, which might be an additional mechanism contrib- involved in GVHD. MIF induces inflammation through the uting to its therapeutic efficacy against GVHD. binding of CD74 [43]. It also binds CXCR2, CXCR4, and CD74, In typical allo-HSCT, recipient hematopoietic and non- forming a complex to trigger chemotaxis of monocytes and T hematopoietic APCs play critical roles in the development of cells [44], which may promote the recruitment of leukocytes GVHD [15-21,23] via “direct presentation” [17,21]. Because of into GVHD target organs to cause tissue damage. Moreover, the lack of animal models, we have not evaluated the effect MIF increases leukocyteeendothelial interactions through of milatuzumab on recipient hematopoietic and non- up-regulation of adhesion molecules on epithelial cells [45], hematopoietic APCs. The hu-PBL-SCID mouse model only possibly via CD74. In addition, after engagement of allows us to evaluate the effect of milatuzumab on donor CD74, MIF stimulates the production of IL-8 [46], a proin- APCs. However, because milatuzumab is cross-reactive with flammatory cytokine involved in various human inflamma- monkey CD74 [39], we would expect monkey GVHD models, tory diseases [47] and a biomarker for GVHD [48]. MIF is if available, to be useful for testing the effect of milatuzumab a chemoattractant that attracts neutrophils to local sites of on recipient APCs in vivo. Therapeutic evaluation of milatu- inflammation [47], possibly contributing to the infiltration zumab in monkey GVHD models may provide more relevant of donor leukocytes to GVHD target organs. MIF also stimu- information for the future clinical use of this mAb in GVHD lates TNF-a production, which precedes and involves prevention. the pathogenesis of acute GVHD [49]. As a CD74 antagonist Along with the reduction of mDCs, other possible mech- antibody, milatuzumab may target these pathways to anisms may be implicated in the therapeutic control of GVHD inhibit local inflammation and prevent the recruitment by milatuzumab. One of these mechanisms may be through of leukocytes, including T cells and APCs, into GVHD local MIF, the natural ligand of CD74 and a pleiotropic proin- sites. flammatory protein participating in many inflammatory Given MIF/CD74’s essential role in T-cell activation [36],it diseases [40]. Patients with acute GVHD reportedly have is also expected that milatuzumab, although having no effect higher serum MIF levels [41] and up-regulated local expres- on T-cell survival, may modulate the activation status of T sion [42], and in a murine model of allogeneic SCT, knockout of cells via antagonistic blocking of MIF/CD74. Because 38 X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39

activated alloreactive T cells are the direct mediator of GVHD, 2. Perez L, Anasetti C, Pidala J. Have we improved in preventing and modulation of the activation status of donor human T cells treating acute graft-versus-host disease? Curr Opin Hematol. 2011;18: 408-413. may be another mechanism contributing to the anti-GVHD 3. Chakrabarti S, Mackinnon S, Chopra R. High incidence of cytomegalo- efficacy of milatuzumab in hu-PBL-SCID mice. virus infection after nonmyeloablative stem cell transplantation: Furthermore, milatuzumab down-regulates the expres- potential role of Campath-1H in delaying immune reconstitution. Blood. 2002;99:4357-4363. sion of adhesion molecules in human B cells, including L- 4. Thompson Carter SL, Kernan NA. Effect of graft-versus-host disease selectin (CD62L) and b7-integrin [13]. It remains to be shown prophylaxis on 3-year disease-free survival in recipients of unrelated whether milatuzumab has a similar effect on these molecules donor bone marrow (T-cell Depletion Trial): a multi-centre, rando- mised phase II-III trial. Lancet. 2005;366:733-741. in T cells, given that a4b7-integrin plays a central role in the 5. Buyck HC, Prentice HG, Griffiths PD, et al. The risk of early and late CMV homing of T cells to the gut, a major GVHD target organ, and DNAemia associated with Campath use in stem cell transplant recipi- that the absence of b7-integrin decreases the homing of ents. Bone Marrow Transplant. 2010;45:1212-1219. 6. Kanda J, Lopez RD, Rizzieri DA. Alemtuzumab for the prevention activated T cells to the intestine, resulting in less GVHD [50]. and treatment of graft-versus-host disease. Int J Hematol. 2011;93: In addition, it has been reported that L-selectin and b7- 586-593. þ integrin on donor CD4 T cells are required for the early 7. Stein R, Mattes MJ, Cardillo TM, et al. CD74: a new candidate target for migration of these cells to gut-associated organized lymphoid the immunotherapy of B-cell neoplasms. Clin Cancer Res. 2007;13: 5556s-5563s. tissues, such as mesenteric lymph nodes and Peyer patches. 8. Stein R, Qu Z, Cardillo TM, et al. Antiproliferative activity of a human- Inactivation of the genes encoding these two molecules ized anti-CD74 monoclonal antibody, hLL1, on B-cell malignancies. markedly reduces the accumulation of donor T cells in gut- Blood. 2004;104:3705-3711. 9. Burton JD, Ely S, Reddy PK, et al. CD74 is expressed by multiple associated organized lymphoid tissues and the severity of myeloma and is a promising target for therapy. Clin Cancer Res. 2004; GVHD [51]. Thus, it would be of interest to investigate 10:6606-6611. whether milatuzumab down-regulates these molecules on T 10. Peterson M, Miller J. Invariant chain influences the immunological recognition of MHC class II molecules. Nature. 1990;345:172-174. cells as well. 11. Stein R, Smith MR, Chen S, et al. Combining milatuzumab with borte- In summary, this is the first report showing that GVHD zomib, doxorubicin, or dexamethasone improves responses in multiple can be controlled by an antagonist mAb targeting CD74 myeloma cell lines. Clin Cancer Res. 2009;15:2808-2817. 12. Berkova Z, Tao RH, Samaniego F. Milatuzumab: a promising new without impairing the protective T-cell immunity essential immunotherapeutic agent. Expert Opin Investig Drugs. 2010;19:141-149. for antiviral and GVL effects after allo-HSCT. Our results 13. Frölich D, Blabfeld D, Reiter K, et al. The anti-CD74 humanized support further investigation of CD74 as a novel therapeutic monoclonal antibody, milatuzumab, which targets the invariant chain of MHC II complexes, alters B-cell proliferation, migration, and adhe- target for GVHD and of the prophylactic and/or therapeutic sion molecule expression. Arthritis Res Ther. 2012;14:R54. use of the anti-CD74 mAb milatuzumab for the control of this 14. Freudenthal PS, Steinman RM. The distinct surface of human blood challenging disease. dendritic cells, as observed after an improved isolation method. Proc Natl Acad Sci USA. 1990;87:7698-7702. 15. Shlomchik WD, Couzens MS, Tang CB, et al. Prevention of graft-versus- ACKNOWLEDGMENTS host disease by inactivation of host antigen-presenting cells. Science. We thank Robert M. Sharkey, PhD, and Jack Burton, MD, 1999;285:412-415. for insightful discussions; Steven Devine, MD, for a critical 16. Duffner UA, Maeda Y, Cooke KR, et al. Host dendritic cells alone are sufficient to initiate acute graft-versus-host disease. J Immunol. 2004; reading of the manuscript; Hong Ding and Simon Mendoza- 172:7393-7398. Agostini for SAS analyses; Anju Nair and Maria Zalath for 17. Chakraverty R, Sykes M. The role of antigen-presenting cells in trig- daily animal care and observations; Marisol Hernandez for gering graft-versus-host disease and graft-versus-leukemia. Blood. 2007;110:9-17. hematoxylin and eosin and immunohistochemistry sections; 18. Toubai T, Tawara I, Sun Y, et al. Induction of acute GVHD by sex- and Susan Chen for experiments with human whole blood. mismatched H-Y antigens in the absence of functional radiosensitive Financial disclosure: This work was supported in part by host hematopoietic-derived antigen-presenting cells. Blood. 2012;119: 3844-3853. US Public Health Service grant 1R41AI093082-01 (to Chien- 19. Koyama M, Kuns RD, Olver SD, et al. Recipient nonhematopoietic Hsing Chang and Xiaochuan Chen on subcontract) from antigen-presenting cells are sufficient to induce lethal acute graft- NIAID, NIH. Chien-Hsing Chang, Thomas M. Cardillo, and versus-host disease. Nat Med. 2011;18:135-142. 20. Li H, Demetris AJ, McNiff J, et al. Profound depletion of host conven- David M. Goldenberg are employed by and have stock and/or tional dendritic cells, plasmacytoid dendritic cells, and B cells does not stock options with Immunomedics, which is developing prevent graft-versus-host disease induction. J Immunol. 2012;188: milatuzumab and dendritic celletargeting antibodies. The 3804-3811. fl 21. Stenger EO, Turnquist HR, Mapara MY, et al. Dendritic cells and regu- other authors have no con icts of interest to disclose. lation of graft-versus-host disease and graft-versus-leukemia activity. Authorship statement: Xiaochuan Chen designed and Blood. 2012;119:5088-5103. performed research, analyzed data, and wrote the manu- 22. Matte CC, Liu J, Cormier J, et al. Donor APCs are required for maximal script. Chien-Hsing Chang designed and performed research, GVHD but not for GVL. Nat Med. 2004;10:987-992. 23. Anderson BE, McNiff JM, Jain D, et al. Distinct roles for donor- and host- analyzed data, and wrote the manuscript. Rhona Stein derived antigen-presenting cells and costimulatory molecules in analyzed data and wrote the manuscript. Thomas M. Cardillo murine chronic graft-versus-host disease: requirements depend on designed and performed research and analyzed data. target organ. Blood. 2005;105:2227-2234. 24. Markey KA, Banovic T, Kuns RD, et al. Conventional dendritic cells are David V. Gold performed pathological analyses. David the critical donor APC presenting alloantigen after experimental bone M. Goldenberg stimulated the evaluation of milatuzumab, marrow transplantation. Blood. 2009;113:5644-5649. supervised the project, advised on all aspects of the study, 25. Leng L, Metz CN, Fang Y, et al. MIF signal transduction initiated by binding to CD74. J Exp Med. 2003;197:1467-1476. reviewed data, and wrote the manuscript. 26. Toubai T, Tanaka J, Nishihira J, et al. Effect of macrophage migration inhibitory factor (MIF) on acute graft-versus-host disease in a murine SUPPLEMENTARY DATA model of allogeneic stem cell transplantation. Transpl Immunol. 2006; 16:117-124. Supplementary data related to this article can be found 27. Wilson J, Cullup H, Lourie R, et al. Antibody to the dendritic cell surface online at http://dx.doi.org/10.1016/j.bbmt.2012.09.015. activation antigen CD83 prevents acute graft-versus-host disease. J Exp Med. 2009;206:387-398. 28. Sharkey RM, Juweid M, Shevitz J, et al. Evaluation of a complementarity- REFERENCES determining region-grafted (humanized) anti-carcinoembryonic 1. Blazar BR, Murphy WJ, Abedi M. Advances in graft-versus-host disease antigen monoclonal antibody in preclinical and clinical studies. Cancer biology and therapy. Nat Rev Immunol. 2012;12:443-458. Res. 1995;55:5935s-5945s. X. Chen et al. / Biol Blood Marrow Transplant 19 (2013) 28e39 39

29. Jacobi AM, Goldenberg DM, Hiepe F, et al. Differential effects of epra- 41. Toubai T, Shono Y, Nishihira J. Serum macrophage migration inhibitory tuzumab on peripheral blood B cells of patients with systemic lupus factor (MIF) levels after allogeneic hematopoietic stem cell trans- erythematosus versus normal controls. Ann Rheum Dis. 2008;67: plantation. Int J Lab Hematol. 2009;31:161-168. 450-457. 42. Lo JW, Leung AY, Huang XR, et al. Macrophage migratory inhibitory 30. Hefta LJ, Chen FS, Ronk M, et al. Expression of carcinoembryonic factor (MIF) expression in acute graft-versus-host disease (GVHD) in antigen and its predicted immunoglobulin-like domains in HeLa cells allogeneic hematopoietic stem cell transplant recipients. Bone Marrow for epitope analysis. Cancer Res. 1992;52:5647-5655. Transplant. 2002;30:375-380. 31. Chen X, Chang CH, Stein R, et al. The humanized anti-HLA-DR moAb, 43. Takahashi K, Koga K, Linge HM, et al. Macrophage CD74 contributes to IMMU-114, depletes APCs and reduces alloreactive T cells: implications MIF-induced pulmonary inflammation. Respir Res. 2009;10:33. for preventing GVHD. Bone Marrow Transplant. 2012;47:967-980. 44. Bernhagen J, Krohn R, Lue H, et al. MIF is a noncognate ligand of CXC 32. Sato T, Deiwick A, Raddatz G, et al. Interactions of allogeneic human chemokine receptors in inflammatory and atherogenic cell recruit- mononuclear cells in the two-way mixed leucocyte culture (MLC): ment. Nat Med. 2007;13:587-596. fl in uence of cell numbers, subpopulations and cyclosporin. Clin Exp 45. Cheng Q, McKeown SJ, Santos L, et al. Macrophage migration inhibitory Immunol. 1999;115:301-308. factor increases leukocyteeendothelial interactions in human endo- 33. Gold DV, Stein R, Burton J, et al. Enhanced expression of CD74 in thelial cells via promotion of expression of adhesion molecules. gastrointestinal cancers and benign tissues. Int J Clin Exp Pathol. 2010; J Immunol. 2010;185:1238-1247. 4:1-12. 46. Binsky I, Haran M, Starlets D, et al. IL-8 secreted in a macrophage 34. Zeger SL, Liang KY. Longitudinal data analysis for discrete and migration-inhibitory factor- and CD74-dependent manner regulates B continuous outcomes. Biometrics. 1986;42:121-130. cell chronic lymphocytic leukemia survival. Proc Natl Acad Sci USA. 35. Martins SL, St John LS, Champlin RE, et al. Functional assessment and 2007;104:13408-13413. specific depletion of alloreactive human T cells using flow cytometry. 47. Harada A, Mukaida N, Matsushima K. Interleukin 8 as a novel target for Blood. 2004;104:3429-3436. intervention therapy in acute inflammatory diseases. Mol Med Today. 36. Bacher M, Metz CN, Calandra T, et al. An essential regulatory role for macrophage migration inhibitory factor in T cell activation. Proc Natl 1996;2:482-489. Acad Sci USA. 1996;93:7849-7854. 48. Levine JE, Logan BR, Wu J, et al. Acute graft-versus-host disease 37. Hill GR, Crawford JM, Cooke KJ, et al. Total body irradiation and acute biomarkers measured during therapy can predict treatment outcomes: graft-versus-host disease: the role of gastrointestinal damage and a Blood and Marrow Transplant Clinical Trials Network study. Blood. inflammatory cytokines. Blood. 1997;90:3204-3213. 2012;119:3854-3860. 38. Munster DJ, MacDonald KP, Kato M, et al. Human T lymphoblasts and 49. Holler E, Kolb HJ, Moller A, et al. Increased serum levels of tumor activated dendritic cells in the allogeneic mixed leukocyte reaction are necrosis factor a precede major complications of bone marrow trans- susceptible to NK cell-mediated anti-CD83-dependent cytotoxicity. Int plantation. Blood. 1990;75:1011-1016. Immunol. 2004;16:33-42. 50. Waldman E, Lu SX, Hubbard VM, et al. Absence of b7 integrin results 39. Sapra P, Stein R, Pickett J, et al. Anti-CD74 antibody-doxorubicin in less graft-versus-host disease because of decreased homing of conjugate, IMMU-110, in a human multiple myeloma xenograft and alloreactive T cells to intestine. Blood. 2006;107:1703-1711. in monkeys. Clin Cancer Res. 2005;11:5257-5264. 51. Dutt S, Ermann J, Tseng D, et al. L-selectin and b7 integrin on donor 40. Morand EF, Leech M, Bernhagen J. MIF: a new cytokine link between CD4 T cells are required for the early migration to host mesenteric rheumatoid arthritis and atherosclerosis. Nat Rev Drug Discov. 2006;5: lymph nodes and acute colitis of graft-versus-host disease. Blood. 2005; 399-410. 106:4009-4015.