In Vivo Imaging Using Bioluminescence: a Tool for Probing Graft-Versus-Host Disease

PERSPECTIVES

known as GVHD2,3. Severe GVHD limits the INNOVATION overall effectiveness of HCT and precludes the application of this life-saving therapy to In vivo imaging using other clinical settings, such as for the treatment of severe autoimmune disorders4 or for the induction of tolerance to organ transplan- bioluminescence: a tool for tation. The risks of GVHD are substantial, such that 20–60% of patients will develop probing graft-versus-host disease this complication following allogeneic HCT, depending on disease-related factors, such as Robert S. Negrin and Christopher H. Contag the stage of disease, the age of the recipient and the degree of genetic disparity between Abstract | Immunological reactions have a key role in health and disease and are the donor and recipient. The reasons why complex events characterized by coordinated cell trafficking to specific locations some patients develop severe GVHD after throughout the body. Clarification of these cell-trafficking events is crucial for HCT, whereas others do not, remain unclear. improving our understanding of how immune reactions are initiated, controlled Given the substantial risk of GVHD, applica- and recalled. As we discuss here, an emerging modality for revealing cell tion of current HCT procedures requires that the donor is a fully matched histocompatible trafficking is bioluminescence imaging, which harnesses the light-emitting sibling or unrelated donor. Unfortunately, properties of enzymes such as luciferase for quantification of cells and uses low- many patients that need HCT are unable to light imaging systems. This strategy could be useful for the study of a wide range secure a well-matched donor, resulting in of biological processes, such as the pathophysiology of graft-versus-host and lethal consequences for the patient due to graft-versus-leukaemia reactions. disease progression. Therefore, the study of GVHD and GVL reactions provides insight into both normal and pathological immune Normal immune function is crucial for as for genetic and acquired immune deficien- reactions and has significant implications for maintaining health in a hostile environ- cies1. HCT involves the transfer of the entire the development of new and more effective ment containing many potential pathogens. haematopoietic and immune systems from a strategies for clinical management of disease. To protect against pathogens, immune donor to a recipient. The procedure involves responses must be both rapid and sustained. pretreatment of the patient with high doses Experimental models of HCT In addition, effector cells of the immune of chemotherapy, with or without irradiation, HCT has been widely studied in both rodent system provide immune surveillance against to eliminate malignant or defective and canine models, and these studies have malignancy and promote tissue remodelling haematopoietic cells. This is followed by the been crucially important in developing the and repair. The importance of a normal transfer to the patient (recipient) of donor- theoretical basis of HCT. Initial studies in immune response is further exemplified derived haematopoietic cells, which home mice established the concept of allogeneic by the observation that patients with many to the bone marrow and re-establish immune responses and GVHD. Mouse diseases, including cancer and infectious and haematopoiesis. After HCT, all of the haema- studies have also been crucial for studying autoimmune diseases, have dysfunctional topoietic cells in the recipient, including allorecognition, and for exploring effector-cell immune responses. Therefore, a greater professional antigen-presenting cells (APCs) populations and mechanisms, owing to the understanding of normal and pathological and immune effector cells, are of donor defined genetics and availability of strains immune responses will not only provide origin. The functional consequences for the that lack key effector molecules. Early stud- insights into basic biological mechanisms but recipient of the new donor-derived immune ies in dogs showed that in some litter-mates also will aid in the development of effective system are dramatic and include the ability to long-term engraftment occurred, whereas treatments for a range of diseases. reject the underlying malignancy or replace in others GVHD developed. The outbred In this Innovation article, we discuss damaged haematopoietic-cell populations canine model has been useful in the develop- how bioluminescence imaging (BLI) can be with normal cells. For example, the damaged ment of preparative regimens for transplan- used to analyse aspects of complex immune red blood cells in patients with thalassaemia tation, which were translated to the clinic5,6, reactions in living animals. We use graft-versus- or sickle-cell disease can be replaced with and to study the use of drug prophylaxis leukaemia (GVL) reactions and graft-versus-host healthy cells. The rejection of malignancy is for both acute and chronic GVHD in large disease (GVHD) as examples of what BLI can known as the GVL effect. The antigens on animals. More recent studies have led to teach us about clinically relevant immune the tumour cells that are recognized by the the development of non-myeloablative HCT, responses. donor leukocytes are largely unknown but in which the intensity of the preparative they include major and minor histocompatibility regimen of the recipient is reduced and Haematopoietic-cell transplantation antigens (depending on the genetic disparity replaced with immuno suppressive Nowhere in clinical medicine is the impor- between the donor and the recipient) as well medications to prevent graft rejection. This tance of effective immune responses clearer as potential tumour-specific antigens. strategy of HCT is associated with reduced than after allogeneic haematopoietic-cell Despite its success in promoting the rejec- transplant-related morbidity and mortality7. transplantation (HCT). Over the past several tion of malignancy, allogeneic HCT carries Non-myeloablative HCT has been widely decades, HCT has emerged as an effective the significant risk that the donor-derived applied in the clinic; however, GVHD and often life-saving treatment for a broad immune cells will recognize and respond to remains a serious complication of the array of haematological malignancies, as well recipient tissues and result in a syndrome therapy with an associated mortality rate

PERSPECTIVES

Box 1 | Molecular imaging in immunology: watching and waiting for an immune response small animals also has several advantages over other approaches: it can usually accommodate Imaging modalities. Two-photon intravital microscopy offers resolution of cells in vivo11, but it is constrained by small fields of view and motion artefacts. Non-invasive measures of immunological multiple animals in a single image at each processes in vivo have been accomplished using positron emission tomography (PET)49–51, single time point; it is user-friendly so dedicated photon emission computed tomography (SPECT)52, magnetic resonance imaging (MRI)53, imaging technicians are not required; and bioluminescence imaging (BLI) and fluorescence imaging (FLI)49,54–56. Ultrasound and X-ray it is less expensive than the instrumenta- computed tomography (CT) provide anatomical information, and when used in combination with tion required for SPECT, MRI and PET. other modalities, this information improves localization of the signals obtained by PET, SPECT The use of two-photon intravital microscopy or optical imaging57. PET, SPECT, ultrasound MRI and CT have potential clinical uses, and therefore offers high-resolution images of cell–cell are useful in translational studies. interactions in tissues11, but is constrained Optical methods. BLI and FLI can be used to refine and accelerate studies of animal models, but by a limited field of view and can be severely they have limited clinical application. Imaging times for optical imaging methods are generally hampered by motion artefacts. As a result short, which facilitates the analysis of greater numbers of animals. Optical methods also allow a of these limitations and constraints, only range of image resolutions from microscopic to macroscopic, produce images without the use of certain tissues can be observed using this ionizing radiation, offer the choice of many reporters and dyes, and benefit from user-friendly and approach. 58 59,60 inexpensive instrumentation . Signal-to-noise ratios (SNRs) for BLI are excellent , enabling In the field of optical imaging, there are detection of subtle changes non-invasively, thereby obviating the need to remove overlying tissue. several approaches that have been used to Reporter genes. The use of dyes and contrast agents allows visualization of the early events, but generate whole-body images of biological they are diluted by cell division. To prevent loss of labels during cell division, genes that encode processes in rodents. These include the use reporter proteins can be integrated into the genome. Reporter genes are available for PET, SPECT, of light scatter, absorbance, fluorescence MRI and optical imaging49,57, each with strengths and weaknesses. The radiotracers used for PET and bioluminescence. In this Perspective and SPECT often produce signals from kidney, liver and bladder that can obscure the target tissue, and MRI is generally less sensitive than imaging of reporter-gene expression with PET or SPECT. article, we focus on the use of biolumin- In summary, optical imaging of reporter-gene expression in vivo offers the greatest versatility, escence to generate images of cell migra- sensitivity and SNR of all of the modalities used for small animals. tion and proliferation in vivo and on the

of 10–20% (REF. 8). Recent studies in which Visualizing immune responses Select labelled the immune environment of the recipient is Until recently, much of our understanding cells altered before transplantation using total of immune-cell trafficking and the factors lymphoid irradiation (TLI) and depleting that control this process has been obtained Transgenic antibodies that target T cells (anti-thymocyte by using culture systems, in which the influ- donor mouse globulin) — an approach pioneered in ence of intact organ structure, circulation, mouse model systems — have been trans- endothelial barriers and tissue effects have lated to the clinic with an apparent reduced been removed. Insights into the specific risk of acute GVHD9. locations and timing of immune-cell migra- Inject luciferase reagent Transfer cells to Mouse studies have also been important in tion and proliferation that can be gained unlabelled wild-type the development of a conceptual framework using imaging methods in living animals recipient for GVHD reactions. As in humans, GVHD hold promise for providing new informa- in animal models shows an unusual tissue tion on physiology and pathophysiology. tropism; it mainly affects the skin, gastrointes- Various imaging modalities are emerg- tinal tract and liver. Tissue injury in the recip- ing for the study of small animal models of Low-light ient owing to the preparative regimen results human biology and disease, and several imaging system Localize cells in the release of pro-inflammatory cytokines of these have been applied to the study of that fuel the GVHD reaction2,3. Alloreactive immune-cell migration. There are preclinical Image whole animal Further analysis donor-derived T cells can become activated in versions of clinical imaging systems, such as Figure 1 | Schematic representation of a bio- the recipient, and can then infiltrate and sub- magnetic resonance imaging (MRI), positron luminescence imaging strategy using cells sequently cause damage to tissues of the skin, emission tomography (PET) and single from a transgenic donor mouse. Cells express- gut and liver, resulting in the pathophysiology photon emission computed tomography ing the transgene encoding a luciferase–GFP– of GVHD. In mice, this alloreactivity trans- (SPECT). For example, MRI has recently been green fluorescent protein (GFP) fusion protein lates as end-organ damage such as hair loss, used to track cardiac-graft rejection in a rat are isolated from a transgenic donor animal and ruffled fur, weight loss, diarrhoea and eventu- model10. The use of these tools in studies of selected by cell-sorting technologies, using the ally mortality. However, these experimental small animals allows ready translation of new GFP signal or fluorescent antibodies specific for end-points are reflective of end-stage disease reagents and imaging approaches to the clinic. selected cell-surface markers. Luciferase–GFP- and provide little information about the However, the use of optical markers, such as positive cells are then transferred to recipient spatial and temporal events in the induction those that are bioluminescent or fluorescent, syngeneic or allogeneic animals. Recipient animals are then injected with the luciferase sub- of GVHD at time points when intervention to assess cell fate and function in whole strate to allow serial imaging of the biolumines- could affect the ultimate outcome. To explore animals offers several advantages that can cent signal in vivo. The tracking of effector cells GVHD and GVL reactions in greater detail, be used to refine and accelerate the study of involved in graft-versus-host disease or graft- imaging strategies have been used to visualize mouse models of disease (BOX 1). The instru- versus-leukaemia reactions can be carried out in the spatial and temporal events in GVHD mentation for acquiring whole-body images recipient animals that have been prepared using induction. using optical reporters that are expressed in myeloablative or non-myeloablative regimens.

PERSPECTIVES

a b cd ) Mouse 1 5 Number of cells per well 10 1,000 Mouse 1 02510 20,000 Liver Spleen × 1,000 1,000 100 100 100

CD19 10 10 10 Mouse 2 1 1 1 0.1 0.1 4 5 6 7 8 9 5,000 0.11 10 100 1,000 0.1 1 10 100 1,000 Relative intensity of bioluminescent signal ( 10 10 10 10 10 10 Number of GFP-positive cells GFP GFP

Mouse 2 Liver Spleen 5,000 1,000 1,000 e Number of cells per animal 0 100 1,000 10,000 100 100

10 10 CD19 11

0.1 0.1 0.11 10 100 1,000 0.1 1 10 100 1,000 500 GFP GFP Figure 2 | Sensitivity of detection in bioluminescence imaging studies. disease course. One time point is shown for two mice. b | After recovery from The detection of weak optical signals, such as those generated during bio- the animals, tumour cells from the liver and spleen of these animals can then luminescence imaging (BLI), from inside small animals is influenced by sev- be quantified by flow cytometry using the GFP signal and fluorescent anti- eral factors that include: the level of cell brightness (photon flux from bodies specific for the B-cell marker CD19; the results for the two animals source), the depth at which the bioluminescent source is located in the tis- in part a are shown. c | The quantity of GFP-positive cells detected by flow sue, the wavelength of the emitted light, the quantum efficiency and noise cytometry correlated well with the bioluminescent signals detected in vivo, of the detector, the nature of the collection optics and the background showing that BLI is a sensitive and reliable measure of cell number in vivo. emission levels from the live animal. a | In this example, studies of the detec- d | Using the same detector, the detection sensitivity of BLI can also be ana- tion sensitivity of BLI were carried out in a mouse B-cell lymphoma model, lysed by measuring the bioluminescent signal emitted from known numbers in which the tumour cells were labelled with luciferase and green fluor- of cells in culture or following transfer in vivo (e). Images are adapted, with escent protein (GFP), by imaging whole animals at various times during the permission, from REF. 24 © (2003) the American Society of Hematology. use of these data to guide sampling of the the body, and this has a marked influence into animals (FIG. 2d,e). These measures appropriate tissues at times when biological on the transmission of light through tissues. have indicated that, in mice, a minimum of changes are occurring. The influence of tissue on the detection of 100–1,000 cells can be detected at specific bioluminescent reporters in vivo has been anatomical sites. Among the parameters BLI strategies studied using four luciferases in three animal that affect the sensitivity of detection are The use of in vivo bioluminescence was first models19. The data generated were consist- the wavelength of light emission, expression shown in tracking bacterial pathogens12. ent with those reported using external light levels of the enzyme in the target cells, the In vivo BLI has since been applied to the sources, and they showed that the longer location of the source of bioluminescence in study of gene-expression patterns13, as a wavelengths of bioluminescence (>600 nm), the animal, the efficiency of the collection measure of successful gene transfer14, to in the red and near-infrared regions of the optics and the sensitivity of the detector. indicate the extent of tumour growth and spectrum, are transmitted through mam- Detection of internal biological sources response to therapy15, to assess the extent of malian tissues more efficiently than the of light requires sensitive detection systems protein–protein interactions in vivo16 and to shorter wavelengths of light, in the blue and with spectral sensitivity in the red region of determine the location and proliferation of green regions of the visible spectrum. This the spectrum. Charge-coupled device (CCD) stem cells17. These examples show the versa- indicates that using luciferases that have a cameras (similar to a home camcorder) in tility of the approach and some of the basic significant portion of their emission greater general, have a spectral range that is appropri- principles of the method (FIG. 1). than 600 nm, such as luciferase derived from ate for detecting biological light sources. But BLI is based on the expression of a fireflies and click beetles (approximately most CCD detectors are not sensitive enough light-emitting enzyme (such as luciferase) 60% of the light emitted from these two to detect the light from inside the body that in target cells and tissues. In the presence of enzymes has wavelengths greater than 600 is needed for the study of cell migration or its substrate (such as luciferin), an energy- nm), will lead to more-sensitive detection other biological processes. To increase their dependent reaction releases photons that of the labelled cells in vivo. In each animal sensitivity, these detectors can be cooled. can pass through tissues and be detected model, the sensitivity of detection should Alternatively, approaches can be used to using sensitive detection systems. As with be measured by determining cell numbers amplify or intensify the signal, but many all optical imaging approaches, BLI is sub- (FIG. 2), assessing luciferase activity in excised intensifiers used for amplifying optical signals ject to the optical properties of tissues18. For tissues20 or using other imaging modalities21. are not sensitive in the red region of the spec- example, in the visible region of the spec- Cells can be counted after recovery from trum. So, CCD cameras, in which the CCD trum, haemoglobin is the main absorber in animals (FIG. 2a–c) or before introduction chip is thinned, back illuminated, placed in

PERSPECTIVES

Syngeneic recipient suffer from variable efficiency, especially 8,000 7,000 40,000 when attempting to introduce genes into 6,000 30,000 certain primary cells. These techniques also 20,000 5,000 10,000 often require cell activation and culturing 4,000 for variable periods of time, and this might 3,000 alter the biological activity of the cells. Transduction of haematopoietic stem cells (HSCs) has been carried out at high frequen- cies using viral vectors; these transduced cells 123456 14 Days after transplantation of luciferase-positive splenocytes can then be used directly in vivo to visualize 26 Allogeneic recipient engraftment . In addition, transduced HSCs 8,000 can be transplanted into immunodeficient 7,000 mice, such as recombination-activating 6,000 gene 2 (Rag2)–/– animals (which lack B and 5,000 4,000 T cells), allowed to engraft and then effector 3,000 cells can be isolated for secondary transfer27. This approach can be particularly useful for studying the migration of cells derived from particular strains of mice that lack key effec- 123456 tor molecules, as backcrossing these animals Days after transplantation of luciferase-positive splenocytes to reporter-gene-expressing mice is expensive Figure 3 | Imaging of graft-versus-host disease. Bioluminescence imaging of luciferase-positive and time consuming. However, if single HSCs splenocytes transplanted to either irradiated syngeneic (top panels) or allogeneic (bottom panels) or small numbers of HSCs are to be studied, animals are shown. Serial images show markedly different patterns of lymphocyte trafficking, pro- then uniform integration of the reporter gene liferation and tissue infiltration. At defined time points, tissue sites of interest, as determined by into a given genomic site17 is preferred to bioluminescence imaging, can then be further analysed. Images are reprinted, with permission, from retroviral transduction, which results in each REF. 25 © (2005) the American Society of Hematology. cell having the reporter gene integrated into different sites. a vacuum and cooled to temperatures as low all enzymes characterized so far that use An alternative approach to gene transfer as –105°C, are at present the most common coelenterazine emit blue light, which is highly has been the generation of transgenic mice cameras for imaging weak biological sources absorbed by mammalian tissues. As more that express a luciferase–green fluorescent of light in the body. enzyme–substrate pairs are characterized and protein (GFP) fusion protein under the For BLI, cells must be engineered to existing reactions are optimized for in vivo control of the chicken β-actin promoter and express the reporter luciferase and the sub- applications, more reagents will become avail- the cytomegalovirus enhancer in all haemato- strate for the reaction must be injected into able for accelerating and refining animal stud- poietic cells. Cells from these transgenic mice the animal for light to be emitted. Luciferases ies using BLI. At present, the luciferin-using (known as L2G85 mice) can then provide have been cloned from both marine (such enzymes from fireflies and click beetles that a source of luciferase-positive donor cells as Renilla luciferase) and terrestrial emit light greater than 600 nm offer the great- for transplantation studies17. The presence (such as firefly and click-beetle luciferases) est sensitivity, and the coelenterazine-using of GFP facilitates the isolation of these cells eukaryotic organisms. The substrates that enzymes can be used as secondary markers, from the donor animals and can also be used these luciferases use seem to group with their despite their severe limitations as convenient as a marker in fluorescence microscopy or origins — marine bioluminescent organ- and sensitive markers. flow-cytometry studies of transplanted cells isms use coelenterazine as a substrate and in tissues from recipient animals. Although terrestrial organisms use d-luciferin. The Using BLI to explore GVHD transgenic animals expressing GFP alone biodistribution of these substrates has been BLI is an effective means of evaluating com- have been used to provide donor cells in studied in animals and differs significantly. plex biological processes such as stem-cell studies of GVHD that show widespread infil- d-luciferin has a longer circulation time than engraftment, GVHD and GVL reactions17,24,25. tration of tissues by donor-derived cells28, it is coelenterazine and there is little catalysis BLI is remarkably sensitive; it can detect possible that the cells of interest lose expres- of d-luciferin by mammalian proteins22. as few as 10 cells in vitro and 100–1,000 sion of the transgene or are recognized by These two differences determine the use cells in vivo (FIG. 2). The ability to track cell the immune system and deleted29. Therefore, of the enzyme–substrate pair for in vivo populations serially and non-invasively, so control experiments with wild-type cells are BLI in which luciferase–luciferin reactions as to define key time points and locations for important to verify these results. provide a longer duration of signal at a longer further analysis, has provided important new In studies with cells from L2G85 mice, wavelength that is less influenced by tissue insights. Whole-body imaging of cell migra- syngeneic (genetically identical; FVB; H2q) absorption than enzyme–substrate reactions tion guides investigators to specific times and or allogeneic (genetically different; BALB/c; that emit blue light (for example, that are pro- organs for more labour-intensive assays. H2d) recipient animals first received lethal duced by luciferases from marine organisms A central limitation of all reporter-gene irradiation (to delete the existing haematopo- such as Renilla spp.). Coelenterazine-using strategies is the need to introduce the reporter ietic cells and simulate the clinical setting of enzymes can provide a short-lived signal construct into the cell populations of interest. HCT) followed by injection of T-cell-depleted that is useful when combining the assays Gene transfer using viral vectors or non-viral bone-marrow cells from wild-type donor with luciferin-using enzymes22,23. However, strategies have been useful; however, they animals (to re-establish haematopoiesis) and

PERSPECTIVES

Cells Bone T cells and Visualizing early migration of donor T cells. that there are several possible approaches to transferredmarrow only T cells T cells Reg Before donor-cell infiltration of GVHD target control GVHD, which include blocking or organs, such as the skin, gut and liver, infiltra- limiting access of T cells to priming sites, tion of secondary lymphoid structures was controlling alloreactive T-cell proliferation Day 5 seen. Histological analysis showed that, in the and blocking entry to GVHD target organs. gut, these early sites of donor-cell prolifera- Because access to secondary lymphoid tion were the Peyer’s patches and mesenteric structures was key to activation and prolif- lymph nodes. These structures are extremely eration of alloreactive T cells, several studies difficult to detect in irradiated animals but have indicated that different populations became easily visible by BLI following HCT25. of T cells might have variable access to Death from + Day 15 GVHD In all of the lymphoid tissues analysed, CD4 such sites and, therefore, differential abil- T cells infiltrated first (as early as 24 hours ity to induce GVHD25,33,34. For example, after transfer), followed (several days later) effector memory T cells (defined by a + + hi low Figure 4 | Effect of transfer of conventional by CD8 T cells. CD4 CD44 CD62L phenotype) can CD4+ and CD8+ T cells with and without CD4+ Previous studies have indicated a key role provide increased immune reconstitution CD25+ regulatory T cells on tumour progres- for Peyer’s patches in the induction of GVHD and GVL effects but have limited capacity to sion. Leukaemia cells expressing a transgene as, under certain experimental conditions, induce GVHD33,34. By contrast, naive T cells encoding the fusion protein luciferase–yellow animals that lack such structures do not (defined by a CD4+CD44lowCD62Lhi pheno- fluorescent protein were injected into recipient develop disease30. Using a myeloablative- type) readily induce GVHD. Using BLI to mice and, using bioluminescence imaging, can be conditioning regimen, we have found that follow the trafficking and survival of these observed infiltrating the bone marrow. Recipient alloreactive T-cell activation and proliferation T-cell populations, it was shown that, unlike mice that received irradiation and T-cell-depleted occurs in several sites, including Peyer’s naive T cells, effector memory T cells did not bone marrow only have progressive tumour patches, mesenteric lymph nodes and other infiltrate and proliferate in secondary lymph growth at day 5 and 15 (left panels). Animals that 25 received T-cell-depleted bone marrow and con- nodal sites as well as the spleen (A. Beilhack, nodes . Other studies have highlighted the ventional T cells die rapidly due to acute lethal S. Schulz and R.S.N., unpublished observa- importance of expression of the leukocyte- graft-versus-host disease (GVHD). By contrast, tions). This is consistent with results reported adhesion molecule CD62L by T cells in recipient mice that received both conventional by others indicating that animals that lack GVHD induction, also indicating that there T cells and CD4+CD25+ regulatory T cells Peyer’s patches still develop GVHD31. These is a need to access secondary lymphoid 35 (TReg cells) in equal proportions retain the ability studies indicate that priming (activation) of structures to initiate GVHD . to reject the tumour without significant GVHD. alloreactive T cells seems to occur in second- Other populations of T cells and natural Images are reproduced, with permission, from ary lymph nodes and the spleen. Following killer (NK) cells with cytolytic activity have © Nature Medicine REF. 27 (2003) Macmillan activation, alloreactive cells upregulate also been studied in models of GVHD Publishers Ltd. the expression of key molecules, such as and GVL. The generation of T cells with α β 4 7-integrin, that are required for entry to defined reactivity — for example, against splenocytes from L2G85 animals (to induce GVHD target organs. A central question minor histocompatibility antigens that are GVHD)25. To visualize the donor cells, luci- is whether cells are imprinted for entry to expressed exclusively by malignant cells or ferin, the substrate for luciferase, was injected specific sites at their site of priming — for against viral antigens in Epstein–Barr-virus- into the recipient just before imaging. Serial example, are cells that are activated in Peyer’s associated diseases — is currently being imaging showed striking differences between patches and mesenteric lymph nodes destined explored and holds significant promise syngeneic and allogeneic recipients (FIG. 3). to enter the gut? And similarly, are other sites for inducing GVL reactions36. Another In the syngeneic animals, a waxing and of activation required for cell infiltration of T-cell population that can be expanded waning BLI signal from the transplanted the skin? In vitro studies support this hypo- ex vivo, known as cytokine-induced killer luciferase-positive cells was observed, which thesis and have indicated that APCs from (CIK) cells, has a limited capacity to induce ultimately resulted in bone-marrow engraft- particular sites, such as the Peyer’s patches, GVHD in mouse models due, at least in ment, probably from residual stem cells in activate cells that can infiltrate the gut but not part, to the production by CIK cells of the splenocyte preparations25. By marked the skin32. interferon-γ (IFNγ)37. BLI studies of CIK contrast, the transplanted cells in allogeneic BLI is therefore extremely useful to cells have indicated that they might have recipients showed early (in the first 24–48 define the time points and sites of donor-cell reduced proliferative capacity compared hours) infiltration of cervical lymph nodes infiltration, and to direct further analyses with naive T cells, which is consistent with and structures in the gut. At 2–4 days after using phenotypic and functional assays. their attenuated capacity to induce GVHD. transplantation, marked proliferation of the Validation of these concepts in vivo with NK cells have also been observed to lack donor cells was observed at these lymph node re-transplantation of luciferase-positive cells the capacity to induce GVHD but can have and gut sites, indicated by the increase in activated at specific sites and analysis by BLI GVL effects38,39. Accordingly, infusion of BLI signal. By day 6, infiltration of the skin will be invaluable for directly answering the NK cells from donors to HLA-mismatched (most obviously in the ears and tail) was remaining questions. recipients with relapsed malignancies results readily apparent25. This study identified the in very limited GVHD, but in some recipi- key target structures and organs involved in Using BLI to evaluate strategies to reduce ents, especially those with acute myeloid the induction of GVHD, so further analyses, GVHD. A major goal of these studies is to leukaemia, beneficial GVL responses were using ex vivo BLI-, immunofluorescence- and develop strategies that can reduce the risk of observed40. The precise mechanisms under- flow-cytometry-based approaches, can focus GVHD but do not interfere with GVL reac- lying the inability of NK cells to induce on these tissues. tions. The studies described earlier indicate GVHD are unclear. BLI studies of NK cells

PERSPECTIVES

proliferation assay and express FOXP3; Glossary + however, only CD62L TReg cells can suppress Allorecognition mature cells of all lineages: erythrocytes, myeloid cells GVHD in vivo46,48, indicating that CD62L- Allorecognition occurs when the host immune system (such as macrophages, mast cells, neutrophils and mediated homing of TReg cells is required detects same-species, non-self antigens and triggers eosinophils), B and T cells, and natural killer cells. to control GVHD. The clinical application allograft rejection. It can occur by direct or indirect pathways: the direct pathway involves recognition of Minor histocompatibility antigens of TReg cells is under active development by foreign MHC molecules on donor cells, and the indirect Polymorphic peptides derived from normal cellular several groups. pathway involves processing and presentation of donor- proteins that can be recognized in the context of derived MHC molecules by host antigen-presenting cells. MHC molecules. Immune responses to these Concluding remarks polymorphic antigens can result in graft-versus-host Graft-versus-host disease reactions, graft rejection or beneficial antitumour BLI has provided new insights into complex (GVHD). Tissue damage in a recipient of allogeneic tissue responses. biological processes that, until now, could (usually a bone-marrow transplant) that results from the not be evaluated. Future goals include activity of donor cytotoxic T cells recognizing the tissues Non-myeloablative haematopoietic-cell improvement of the techniques that allow transplantation of the recipient as foreign. GVHD varies markedly in extent, visualization of more than one population but it can be life threatening in severe cases. Damage to An allogeneic haematopoietic-cell transplantation in a the liver, skin and gut mucosae are common clinical recipient who has received a conditioning regimen to of cells simultaneously and improvement manifestations. achieve immunosuppression and prevent graft rejection in quantification of cell numbers. Improved without the complete ablation of host haematopoiesis. quantification might be possible through Graft versus leukaemia The recipient might develop (transient) mixed chimerism, spectral imaging, in which the differential Hosts with leukaemia who receive an allogeneic bone- owing to haematopoietic recovery of the host and marrow transplant have far fewer disease relapses than engraftment of donor haematopoietic cells. transmission of blue and green components individuals who obtain autologous bone-marrow of the luciferase emission spectra relative transplants. This results from the transplanted T cells Two-photon intravital microscopy to the red components can be used to recognizing alloantigens expressed by the leukaemia. Laser-scanning microscopy that uses pulsed infrared laser determine the depth of the signal in the light for the excitation of conventional fluorophores or body. Reconstruction of three-dimensional Haematopoiesis fluorescent proteins. The main advantage is deep tissue The commitment and differentiation processes that lead penetration of the infrared light, owing to the low level of images from multiple views will also from a haematopoietic stem cell to the production of light scattering in the tissue. improve the quantification of bioluminescence signals. Although the ability to carry out imaging studies using luciferase in in mouse models are ongoing, and it is and have resulted in reduced levels of acute humans will be limited to very superficial hoped that they will provide further clues GVHD with retention of GVL activity9. sites in which expression of a foreign gene to why this cell population has such limited Another well-characterized population is not problematic, lessons learned from capacity for GVHD induction. of regulatory T cells is the subset of naturally using BLI in animal models will greatly occurring CD4+ T cells that express the IL-2 affect the design of clinical trials for cellular Monitoring regulatory T-cell function. An receptor α-chain (also known as CD25)43 transplantation. Perhaps the greatest benefit alternative approach to controlling GVHD and the transcription factor forkhead box of BLI to clinical medicine will be through has been to harness regulatory mechanisms P3 (FOXP3): that is, CD4+CD25+ regula- accelerating and refining preclinical mod- in an effort to control alloimmune responses. tory T cells (TReg cells). Several groups have els. Evaluating the fate of transferred cell Several populations of regulatory T cells shown that the infusion of grafts containing populations is likely to prove crucial to the that can control immune reactions, such as an equal mix of TReg cells and conventional evaluation of HCT and other cell-based a mixed lymphocyte reactions in vitro and T cells results in the control of GVHD44–46; therapeutics. autoimmune diseases in vivo, have been crucially, GVL reactions are retained in ani- Robert S. Negrin and Christopher H. Contag are at the 27,47 described. One such population of regulatory mals treated in this way . Using BLI, active Departments of Medicine, Center for Clinical Research T cells, known as natural killer T (NKT) cells, rejection of the leukaemia could be observed Building, 269 West Campus Drive and Pediatrics, Clark co-expresses both NK-cell and T-cell markers, following the infusion of equal numbers Center, 318 Campus Drive, Stanford University, recognizes CD1 through an invariant T-cell of luciferase-positive conventional T cells Stanford, California 94305, USA. Correspondence to R.S.N. 27 (FIG. 4) receptor and produces large amounts of and TReg cells . The mechanism of e-mail: [email protected] 41 interleukin-4 (IL-4) after activation . Mouse control of GVHD and retention of GVL doi:10.1038/nri1879 recipients treated with TLI and anti- seems to be due to the ability of TReg cells to 1. Negrin, R. S. & Blume, K. in Williams Hematology thymocyte serum (ATS) to delete the existing suppress conventional T-cell proliferation, as 7th edn Ch. 22 (eds Lichtman, M. A. et al.) 209–247 T cells have reduced numbers of conven- shown using BLI to evaluate donor-derived (McGraw-Hill Professional, 2005). + + 2. Ferrara, J. L. & Deeg, H. J. Graft-versus-host disease. tional CD4 and CD8 T cells but increased conventional T-cell trafficking and numbers N. Engl. J. Med. 324, 667–674 (1991). 42 27 3. Kaitin, K. I. Graft-versus-host disease. N. Engl. J. Med. numbers of NKT cells . Interestingly, these in the presence and absence of TReg cells . 325, 357–358 (1991). animals are resistant to GVHD induction By contrast, GVL reactions mainly required 4. Sykes, M. & Nikolic, B. Treatment of severe such that 1,000-fold more allogeneic donor activation of CD8+ T cells, and this occurred autoimmune disease by stem-cell transplantation. Nature 435, 620–627 (2005). cells can be transferred without GVHD even in the presence of TReg cells. However, 5. Thomas, E. D., Storb, R. & Clift, R. A. Bone-marrow developing42. The use of BLI to follow the fate important questions remain: where, how transplantation. N. Engl. J. Med. 292, 832–843 (1975). of donor cells transplanted to animals treated and for how long do TReg cells exert their 6. Thomas, E. D., Storb, R. & Clift, R. A. Bone-marrow with TLI and ATS compared with animals immunological control? Clues have come transplantation. N. Engl. J. Med. 292, 895–902 (1975). receiving total body irradiation is under from studies of TReg cells that are divided into 7. Storb, R. et al. Stable mixed hematopoietic chimerism active investigation. Recipient preparation subsets on the basis of CD62L expression. in dogs given donor antigen, CTLA4Ig, and 100 cGy + – total body irradiation before and pharmacologic strategies involving TLI and T-cell depletion Both CD62L and CD62L TReg cells sup- immunosuppression after marrow transplant. Blood before HCT have been translated to the clinic press cell proliferation in an in vitro T-cell 94, 2523–2529 (1999).

PERSPECTIVES

8. McSweeney, P. A. et al. Hematopoietic cell graft-versus-host disease after bone marrow 45. Taylor, P. A., Lees, C. J. & Blazar, B. R. The infusion of transplantation in older patients with hematologic transplantation. Nature Med. 9, 1144–1150 ex vivo activated and expanded CD4+CD25+ immune malignancies: replacing high-dose cytotoxic (2003). regulatory cells inhibits graft-versus-host disease therapy with graft-versus-tumor effects. Blood 28. Panoskaltsis-Mortari, A. et al. In vivo imaging of graft- lethality. Blood 99, 3493–3499 (2002). 97, 3390–3400 (2001). versus-host-disease in mice. Blood 103, 3590–3598 46. Cohen, J. L., Trenado, A., Vasey, D., Klatzmann, D. & 9. Lowsky, R. et al. Protective conditioning for (2004). Salomon, B. L. CD4+CD25+ immunoregulatory acute graft-versus-host disease. N. Engl. J. Med. 29. Murakami, T. et al. Immune evasion by murine T cells: new therapeutics for graft-versus-host disease. 353, 1321–1331 (2005). melanoma mediated through CC chemokine J. Exp. Med. 196, 401–406 (2002). 10. Wu, Y. L. et al. In situ labeling of immune cells with receptor-10. J. Exp. Med. 198, 1337–1347 47. Trenado, A. et al. Recipient-type specific CD4+CD25+ iron oxide particles: an approach to detect organ (2003). regulatory T cells favor immune reconstitution and rejection by cellular MRI. Proc. Natl Acad. Sci. USA 30. Murai, M. et al. Peyer’s patch is the essential site in control graft-versus-host disease while maintaining 103,1852–1857 (2006). initiating murine acute and lethal graft-versus-host graft-versus-leukemia. J. Clin. Invest. 112, 11. Sumen, C., Mempel, T. R., Mazo, I. B. & reaction. Nature Immunol. 4, 154–160 (2003). 1688–1696 (2003). von Andrian, U. H. Intravital microscopy: visualizing 31. Welniak, L. A. et al. Peyer patches are not required for 48. Ermann, J. et al. Only the CD62L+ subpopulation immunity in context. Immunity 21, 315–329 acute graft-versus-host disease after myeloablative of CD4+CD25+ regulatory T cells protects from lethal (2004). conditioning and murine allogeneic bone marrow acute GVHD. Blood 105, 2220–2226 (2005). 12. Contag, C. H. et al. Photonic detection of transplantation. Blood 107, 410–412 (2005). 49. Herschman, H. R. Molecular imaging: looking at bacterial pathogens in living hosts. Mol. Microbiol. 32. Mora, J. R. et al. Selective imprinting of gut-homing problems, seeing solutions. Science 302, 605–608 18, 593–603 (1995). T cells by Peyer’s patch dendritic cells. Nature (2003). 13. Contag, C. et al. Visualizing gene expression in 424, 88–93 (2003). 50. Ponomarev, V. et al. Imaging TCR-dependent NFAT- living mammals using a bioluminescent reporter. 33. Anderson, B. E. et al. Memory CD4+ T cells do mediated T-cell activation with positron emission Photochem. Photobiol. 66, 523–531 (1997). not induce graft-versus-host disease. J. Clin. Invest. tomography in vivo. Neoplasia 3, 480–488 (2001). 14. Lipshutz, G. S., Flebbe-Rehwaldt, L. & Gaensler, K. M. 112, 101–108 (2003). 51. Dubey, P. et al. Quantitative imaging of the T cell Reexpression following readministration of an 34. Foster, A. E. et al. Human CD62L– memory T cells antitumor response by positron-emission tomography. adenoviral vector in adult mice after initial in utero are less responsive to alloantigen stimulation Proc. Natl Acad. Sci. USA 100, 1232–1237 (2003). adenoviral administration. Mol. Ther. 2, 374–380 than CD62L+ naive T cells: potential for adoptive 52. van Montfrans, C. et al. In vivo evaluation of 111 In- (2000). immunotherapy and allodepletion. Blood 104, labeled T-lymphocyte homing in experimental colitis. 15. Sweeney, T. J. et al. Visualizing the kinetics of tumor cell 2403–2409 (2004). J. Nucl. Med. 45, 1759–1765 (2004). β clearance in living animals. Proc. Natl Acad. Sci. USA 35. Dutt, S. et al. L-selectin and 7 integrin on donor 53. Bulte, J. W. & Kraitchman, D. L. Iron oxide MR 96, 12044–12049 (1999). CD4 T cells are required for the early migration to contrast agents for molecular and cellular imaging. 16. Massoud, T. F., Paulmurugan, R. & Gambhir, S. S. host mesenteric lymph nodes and acute colitis of NMR Biomed. 17, 484–499 (2004). Molecular imaging of homodimeric protein–protein graft versus host disease. Blood 106, 4009–4015 54. McCaffrey, A., Kay, M. A. & Contag, C. H. Advancing interactions in living subjects. FASEB J. 18, (2005). molecular therapies through in vivo bioluminescent 1105–1107 (2005). 36. Rooney, C. M. et al. Infusion of cytotoxic T cells imaging. Mol. Imaging 2, 75–86 (2003). 17. Cao, Y. A. et al. Shifting foci of hematopoiesis during for the prevention and treatment of Epstein-Barr virus- 55. Contag, C. H. & Ross, B. D. It’s not just about reconstitution from single stem cells. Proc. Natl Acad. induced lymphoma in allogeneic transplant recipients. anatomy: in vivo bioluminescence imaging as an Sci. USA 101, 221–226 (2004). Blood 92, 1549–1555 (1998). eyepiece into biology. J. Magn. Reson. Imaging 18. Jobsis, F. F. Noninvasive, infrared monitoring of cerebral 37. Baker, J., Verneris, M. R., Ito, M., Shizuru, J. A. & 16, 378–387 (2002). and myocardial oxygen sufficiency and circulatory Negrin, R. S. Expansion of cytolytic CD8+ natural 56. Ntziachristos, V., Ripoll, J., Wang, L. V. & Weissleder, R. parameters. Science 198, 1264–1267 (1977). killer T cells with limited capacity for graft-versus-host Looking and listening to light: the evolution of 19. Zhao, H. et al. Emission spectra of bioluminescent disease induction due to interferon γ production. whole-body photonic imaging. Nature Biotechnol. reporters and interaction with mammalian tissue Blood 97, 2923–2931 (2001). 23, 313–320 (2005). determine the sensitivity of detection in vivo. 38. Asai, O. et al. Suppression of graft-versus-host 57. Hildebrandt, I. J. & Gambhir, S. S. Molecular J. Biomed. Opt. 10, 41210 (2005). disease and amplification of graft-versus-tumor effects imaging applications for immunology. Clin. Immunol. 20. Lipshutz, G. S. et al. In utero delivery of adeno- by activated natural killer cells after allogeneic bone 111, 210–224 (2004). associated viral vectors: intraperitoneal gene marrow transplantation. J. Clin. Invest. 101, 58. Contag, C. H. & Bachmann, M. H. Advances in in vivo transfer produces long-term expression. Mol. Ther. 1835–1842 (1998). bioluminescence imaging of gene expression. Annu. 3, 284–292 (2001). 39. Ruggeri, L. et al. Effectiveness of donor natural Rev. Biomed. Eng. 4, 235–260 (2002). 21. Rehemtulla, A. et al. Rapid and quantitative killer cell alloreactivity in mismatched hematopoietic 59. Rice, B. W., Cable, M. D. & Nelson, M. B. In vivo assessment of cancer treatment response using in vivo transplants. Science 295, 2097–2100 (2002). imaging of light-emitting probes. J. Biomed. Opt. bioluminescence imaging. Neoplasia 2, 491–495 40. Miller, J. S. et al. Successful adoptive transfer and 6, 432–440 (2001). (2000). in vivo expansion of human haploidentical NK cells 60. Troy, T., Jekic-McMullen, D., Sambucetti, L. & Rice, B. 22. Zhao, H. et al. Characterization of coelenterazine in patients with cancer. Blood 105, 3051–3057 Quantitative comparison of the sensitivity of detection analogs for measurements of Renilla luciferase activity (2005). of fluorescent and bioluminescent reporters in animal in live cells and living animals. Mol. Imaging 3, 43–54 41. Bendelac, A. et al. CD1 recognition by mouse NK1+ models. Mol. Imaging 3, 9–23 (2004). (2004). T lymphocytes. Science 268, 863–865 (1995). 23. Bhaumik, S. & Gambhir, S. S. Optical imaging of 42. Lan, F., Zeng, D., Higuchi, M., Higgins, J. P. & Competing interests statement Renilla luciferase reporter gene expression in living Strober, S. Host conditioning with total lymphoid The authors declare competing financial interests: see web mice. Proc. Natl Acad. Sci. USA 99, 377–382 irradiation and antithymocyte globulin prevents version for details. (2002). graft-versus-host disease: the role of CD1-reactive 24. Edinger, M. et al. Revealing lymphoma growth and the natural killer T cells. Biol. Blood Marrow Transplant. efficacy of immune cell therapies using in vivo 9, 355–363 (2003). DATABASES bioluminescence. Blood 101, 640–648 (2003). 43. Sakaguchi, S., Sakaguchi, N., Asano, M., Itoh, M. & The following terms in this article are linked online to: 25. Beilhack, A. et al. In vivo analyses of early events in Toda, M. Immunologic self-tolerance maintained by Entrez Gene: acute graft-versus-host disease reveal sequential activated T cells expressing IL-2 receptor α-chains http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene infiltration of T-cell subsets. Blood 106, 1113–1122 (CD25). Breakdown of a single mechanism of self- CD4 | CD8 | CD25 | CD44 | CD62L | FOXP3 | interferon-γ | IL-4 (2005). tolerance causes various autoimmune diseases. 26. Wang, X. et al. Dynamic tracking of human J. Immunol. 155, 1151–1164 (1995). FURTHER INFORMATION hematopoietic stem cell engraftment using in vivo 44. Hoffmann, R., Ermann, J., Edinger, M., Fathman, C. G. Christopher H. Contag’s laboratory: bioluminescence imaging. Blood 102, 3478–3482 & Strober, S. Donor type CD4+CD25+ regulatory http://sci3.stanford.edu/lab/ (2003). T cells suppress lethal acute graft-versus-host Robert S. Negrin’s homepage: 27. Edinger, M. et al. CD4+CD25+ regulatory T cells disease after allogeneic bone marrow transplantation. http://med.stanford.edu/profiles/robert_negrin preserve graft-versus-tumor activity while inhibiting J. Exp. Med. 196, 389–399 (2002). Access to this links box is available online.