Towards the development of human immune-system-on-a-chip platforms
Alessandro Polini, Loretta L. del Mercato, Adriano Barra, Yu Shrike Zhang, Franco Calabi, Giuseppe Gigli
DDrug
Discovery
Today ! Volume
Number ! February
REVIEWS
Towards the development of humanimmune-system-on-a-chip platforms Alessandro
Polini , Loretta L. del Mercato , Adriano
Barra ,Yu
Shrike
Zhang , Franco
Calabi and Giuseppe
Gigli Dipartimento di Matematica e Fisica E. De Giorgi,
University of Salento,
Campus
Ecotekne, via
Monteroni,
Lecce,
Italy CNR
NANOTEC – Institute of Nanotechnology c/o
Campus
Ecotekne, via
Monteroni,
Lecce,
Italy INFN,
Sezione di Lecce,
Campus
Ecotekne, via
Monteroni,
Lecce,
Italy INdAM (GNFM),
Sezione di Lecce,
Campus
Ecotekne, via
Monteroni,
Lecce,
Italy Division of Engineering in Medicine,
Department of Medicine,
Brigham and
Women ’ s Hospital,
Harvard
Medical
School,
Cambridge, MA USA
Organ-on-a-chip (OoCs) platforms could revolutionize drug discovery and might ultimately becomeessential tools for precision therapy.
Although many single-organ and interconnected systems have beendescribed, the immune system has been comparatively neglected, despite its pervasive role in the bodyand the trend towards newer therapeutic products (i.e., complex biologics, nanoparticles, immunecheckpoint inhibitors, and engineered T cells) that often cause, or are based on, immune reactions. Inthis review, we recapitulate some distinctive features of the immune system before reviewingmicrofluidic devices that mimic lymphoid organs or other organs and/or tissues with an integratedimmune system component. Introduction
Tissue chips (TCs),
OoCs, or microphysiological systems (MPSs) [1–3] are microfluidic in vitro systems that aim to replicate keystructural and functional features of organs or tissue units in aconvenient format for extended analysis and manipulation. Theirobjective is to provide standardized, economical, yet highly rele-vant ex vivo models to both enable the investigation of basicbiological processes and improve the efficiency of the drug dis-covery process. By using patient-derived induced pluripotent stemcells (iPSCs) and their progeny as cell sources, they could ultimate-ly lead to on-chip replication of an individual’s disease for diag-nostic and therapeutic testing and, hence, could become essentialtools for personalized and/or precision therapies [4,5].Numerous single-organ systems have been proposed, and effortsare under way to produce interconnected multiorgan platforms [6–8]. However, so far, the development of OoC systems to emulate theimmune system has lagged behind, which is surprising for severalreasons. First, immune mechanisms have a significant role in major diseases, including cancer, atherosclerosis, and the neurodegenera-tive diseases, in addition to being the primary cause of other com-mon conditions, such as chronic infections and autoimmunediseases [9]. Second, modern approaches to molecularly targetedtherapies often comprise complex entities such as monoclonalantibodies, stem cells, and nanoparticles, which are well abovethe size radar of the immune system. Such molecules can evokeunwanted immune responses, which can be catastrophic [10,11].Third, selective activation or blockade of the immune system hasemerged as a powerful therapeutic approach in its own right: signifi-cantresults inat leastsome typesof cancer havebeen reported for theuse of immune checkpoint inhibitors [12], bispecific antibodies [13],and engineered T cells [14]. However, their use comes at a price, asexemplified by adverse effects such as the cytokine-releasesyndrome(CRS), and central nervous system (CNS)-related and immune-relat-ed adverse events (IRAE) [15]. The causes and remedies of theseunintended reactions remain incompletely understood and repre-sent an active area of investigation [16,17]. Thus, adequate models ofthe human immune system are needed to investigate its role in bothpathogenesis and therapy. R e v i e w s ! G E N E T O S C R EE N Corresponding author:
Polini, A. ([email protected]) ã The
Authors.
Published by Elsevier
Ltd.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).https://doi.org/10.1016/j.drudis.2018.10.003 mall rodents, such as mice and rats, have for decades providedconvenient models for immunologists, and have been widely usedfor preclinical testing. However, they show significant differenceswith the human immune system [18]. In an effort to provide morehuman-relevant models, humanized mouse strains have beenengineered that are engrafted with human hemopoietic precursorsand populated mostly by mature human blood cells. These micehave proven useful in dissecting human-specific infections as wellas IRAE [19].
However, they require special handling, are costlyand pose ethical problems, quite apart from the fact that they donot yet fully reproduce the human system because of persistentmismatches [e.g., in major histocompatibility complex (MHC)antigens] between grafted human cells and host mouse cells. Thus,the development of a wholly human ex vivo immune system wouldprovide significant added value.In this review, we first briefly recapitulate some of the distinctiveanatomo-functional features of the immune system and thenreport on the latest developments and applications of on-chipmicrofluidic devices specifically designed for the immune system. The human immune system: mechanisms and districts
The immune system has evolved to maintain the body homeosta-sis by recognizing and eliminating abnormal components whetherof exogenous (microorganisms and macromolecules) or endoge-nous origin (e.g., cancer) [20]. Two interconnected subsystemsmay be distinguished: a phylogenetically older innate and a neweradaptive immune system. The former provides a fast response, butlacks fine specificity and memory, which are characteristics of thelatter. Here, we describe some features and mechanisms of thehuman immune system: for a more comprehensive overview,please see Refs [9,20].The innate immune system acts via preformed soluble proteinsand phagocytic cells (macrophages and granulocytes) expressinggermline-encoded receptors with broad specificity for shared mo-lecular features of pathogens (pathogen-associated molecular pat-terns, PAMPs).
Activated phagocytes not only kill microorganisms,but also release several factors (cytokines and chemokines) thattrigger an acute inflammatory response and eventually the adap-tive immune response. The latter proceeds via the generation ofantigen-presenting cells (APCs, primarily macrophages and den-dritic cells,
DCs), their migration to local lymph nodes (LNs) andthe priming of antigen-specific T and B cells. This leads to cellproliferation (clonal expansion) and differentiation and/or matu-ration into highly specific cellular effectors: from B cells intoantibody-secreting plasma cells via steps that include isotypeswitching and affinity maturation; and from T cells into thevarious classes of helper (T H ), cytotoxic (CTLs),and regulatory(T reg ) T cells. T and B cell priming also gives rise to memory cells,which ensure a faster and more effective response upon re-en-countering the same antigen.In mammals, most cells of the immune system are produced inthe bone marrow (BM) from multipotent hemopoietic stem cells(HSCs): most cells of the innate system derive from the myeloidlineage and most cells of the adaptive system from the lymphoidlineage. Unlike the former, most T and B cells undergo develop-ment and triggering in specialized structures. This is because theirprecursors clonally express large repertoires of random receptorsderiving from somatic recombination and must undergo sequen- tial steps of positive and negative selection to gate out cells witheither excessive or insufficient affinity for self-molecules to ensureboth self-tolerance and proper immune responsiveness. This most-ly occurs in primary lymphoid organs (BM and thymus). Bycontrast, the secondary lymphoid organs (LN, spleen, and muco-sa-associated lymphoid tissues,
MALTs) are where mature T and Beffectors of the adaptive immune response are generated.Hemopoiesis depends on specialized BM niches that are orga-nized into distinct functional microdomains [21,22]. Whereasearly T cell progenitors migrate to the thymus, B cell developmentoccurs entirely in the BM, and requires interactions with differenttypes of resident stromal cell. Distinct niches in the BM promotehoming of antigen-triggered B cells and the survival of antibody-producing plasma cells and of memory immune cells, while alsoproviding a reservoir of myeloid-lineage immune cells. In thethymus,T cell development proceeds from the outer cortex tothe medulla. Contact of T cell precursors with cortical epithelialcells has a key role in positive selection, and with APCs in themedulla in negative selection of self-reactive T cells.The architecture of peripheral lymphoid organs is generallyorganized in distinct B cell and T cell zones, with specific classesof DCs.
The former comprises follicles and germinal centers (GCs),which represent the areas where clonal expansion of antigen-triggered B cells occurs in parallel with affinity maturation (so-matic hypermutation). LNs intercept the lymph draining fromperipheral tissues and carrying tissue-derived antigen and
APCs,while also receiving a constant supply of lymphocytes enteringthrough the walls of specialized blood vessels, the high endothelialvenules (HEVs), and leaving via the efferent lymphatics. Special-ized cells (M cells) occur in the epithelia overlying MALTs andchannel antigens from the lumen.
Human immune-system-on-a-chip platforms: a longway to go In recent years, the development of increasingly sophisticated andflexible microfabrication techniques has led to major progress inboth basic and applied immunological research. This is particular-ly evident in biochip-based approaches to the manipulation ofsingle immune cell populations for high-throughput analysis:currently, microfluidic devices are largely used to study the in-flammatory responses in vitro down to the single cell level, provid-ing interesting insights into the activation, adhesion,transmigration, phagocytosis, and secretion of immune cells(reviewed in Refs [23,24]).
Here, we report the latest developmentsin those OoC platforms where an immune system component hasbeen introduced and on progress relating to lymphoid OoCdevices.
Organ-on-a-chip platforms with an immune system component Most of the work in this area has focused on three systems: (i) theinteraction of immune cells with tumors; (ii) the interaction ofleukocytes with endothelial cells; and (iii) inflammation.Key advantages of tumor-on-chips with respect to other in vitro models, such as classical cultures and spheroids, are a controlled3D architecture and dynamic microflow conditions, with ex-change of oxygen and nutrients enabling operation for up toseveral days. In particular, reduced thickness and greater homo-geneity with respect to spheroids facilitates both real-time imaging REVIEWS
Drug
Discovery
Today ! Volume
Number ! February R e v i e w s ! G E N E T O S C R EE N nd drug and/or cell access to tumor cells [25]. The addition ofimmune cells, either as an adjacent extracellular matrix (ECM)-embedded compartment or perfused through a ‘surrogate bloodvessel’, has allowed the detailed study by time-lapse microscopy ofthe different migratory phenotypes of macrophages and breastcancer cells [26] and of activated natural killer (NK) cells penetrat-ing a glioblastoma tumor [27] in a manner that would have beenimpossible in vivo . Thus, the role of direct cell interactions versusparacrine signaling can be addressed. Moore and colleagues de-scribed a microfluidic model ( Ex Vivo
Immuno-oncology
DynamicENvironment for
Tumor biopsies;
EVIDENT) for imaging the interactions between tumor fragments and flowing autologoustumor-infiltrating lymphocytes (TILs) over multiple days [28](Fig. By automatic quantitative image analysis (Fig. fraction of cell death attributable to TILs was estimated andfound to respond to anti-PD-1 (an immune checkpoint inhibitor).Thus, the device has potential for the initial assessment of immu-notherapeutics. Unlike animal models, on-chip systems are readilyamenable to controlled manipulations of individual variables. Thepower and flexibility of the approach enabled Pavesi et al . to dissectthe effects of low O and inflammatory cytokines (IFN- g and TNF- a ) on the antitumor activity of engineered T cells, and to compare Drug
Discovery
Today ! Volume
Number ! February
REVIEWS
TIL reservoir Bubble traps
Sample trapdetail Microfluidic connectorFluidicresistorTray
Sample insertion wellsMagnetic stirrerAir pumpControl PCB sample traps WastechamberMicrobial microchamberOptodeOxygen andbiomoleculegradients Electricalresistancemeasurement CD - F I T C APC-Cy7APC-Cy7
MucinNanoporous membraneEpithelial cell microchamberCollagenMicroporous membranePerfusion microchamber
61% alive cells 35% deadcells –10 –10 CD - F I T C
59% alive cells 35% deadcells –10 –10 (a)(d) (e)(f) (b)(c) Drug Discovery Today
FIGURE Examples of microfluidic devices for studying cell – cell interactions. (a) Illustration of the microfluidic model Ex Vivo
Immuno-oncology
Dynamic
ENvironment forTumor biopsies (EVIDENT) and its control system. (b)
Map of tumor death through one plane of a z-stack confocal image. (c) Perimeter of tumor-infiltratinglymphocyte (TIL) infiltration versus time, showing the advancing front of TIL penetration into the tumor. (d)
Conceptual diagram of the human-microbialcrosstalk (HuMiX) device comprising three co-laminar microchannels: a medium perfusion microchamber; a human epithelial cell culture microchamber; and amicrobial culture microchamber. Flow cytometry analysis of the viability of CD4 + T cells cultured alone (e), or co-cultured for h with LGG (Lactobacillusrhamnosus
GG) (f).
Reproduced, with permission, from
Ref. [28] (a – c). Adapted, with permission, from
Ref. [38] (d – f). Abbreviation:
PCB,
XXXX. R e v i e w s ! G E N E T O S C R EE N ifferent technologies for T cell engineering [29]. Thus, screens canbe devised to optimize the latter process. By contrast, by neglectingT cell migration, classical static cultures provide misleadingpotency data.Interactions between leukocytes and the endothelium have animportant role in both the immune response and in triggeringinflammation and vascular pathology (atherosclerosis). An inflam-mation-on-a chip device was realized comprising a central, endo-thelial-lined chamber flanked by ECM-filled compartments [30].
Itwas used to study the effects on neutrophil extravasation ofprecisely controlled and stable chemokine gradients and of vari-able ECM stiffness.
Unlike classical
Boyd chambers, the geometryavoids the confounding effect of gravity and provides high-reso-lution in situ imaging. A microvessel-on-chip has been describedthat reproduces lumen structure, barrier function, and secretion ofangiogenic and inflammatory mediators, and can be maintainedin culture for longer than weeks [31]. Thus, it provides a highlyrelevant in vitro model, and was used to investigate transendothe-lial extravasation of both purified neutrophils and whole-bloodcells in response to individual chemotactic factors. It could findapplication for the identification of both intrinsic and exogenousregulators of inflammation.Rolling of activated T cells along the endothelial lining and theirsubsequent adhesion are affected by the local shear stress resultingfrom the blood flow rate and viscosity and from the vessel geome-try. A dynamic microfluidic system to investigate the role ofchanges in the flow rate was reported by Park et al . [32]. A parallelmultichannel architecture was designed to induce different celladhesion molecules (CAMs) on human umbilical vein endothelialcells (HUVECs) and test the adhesion of primary T cells fromblood. Adhesion of lymphocytes from patients with systemiclupus erythematosus (SLE) was higher than in controls, and wasdramatically reduced by immunosuppressants. The device wasproposed as a tool for high-content screening of novel drugs thatrequires less time, cost, and labor than conventional methods,while using human primary cells. Wu et al . realized a capillary-endothelium-mimetic microfluidic chip comprising a silicon mi-cropore array lined on one side with an endothelial cell line andsandwiched between two microchannels [33]. It enabled theauthors to investigate the effects of flow rate and chemotacticgradients on leukocyte extravasation over an extended period oftime. Chemotactic factors were found to predominate at low flowrates, whereas extravasation was limited at higher flow ratesbecause of cell aggregation near the side wall resulting fromhydrodynamic forces. The device is also applicable to the studyof cancer metastasis, atherosclerosis, and other angiopathies.To simulate atherosclerosis, Menon et al . devised an originalapproach to pattern microchannels of complex geometry inmicropillar-free ECM [34] and studied the effect of luminal steno-sis on leukocyte and platelet interactions with normal and in-flamed endothelium under different shear stress.A humanized dynamic in vitro model based on polypropylenehollow microfibers featuring transmural microholes (2–4 m m)under pulsatile flow was designed to investigate the transendothe-lial trafficking of immune cells across the blood–brain barrier (BBB)[35]. In the presence of circulating monocytes, flow cessationfollowed by reperfusion caused BBB opening with abluminalextravasation of monocytes paralleled by a significant increase in proinflammatory cytokines and activated matrix metallopro-teinases.A microfluidic device (kit-on-a-lid-assay, KOALA) was describedfor the rapid purification of neutrophils from nanoliter volumes ofblood and the reproducible generation of chemotactic gradients[36]. Its simple operation and versatility are useful for both clinicaland research applications.Other categories of microfluidic devices have been specificallydeveloped for the study of inflammation. For example, interac-tions between the gut microbiome, intestinal mucosa, and im-mune components, and reduced peristalsis are believed tocontribute to inflammatory bowel disease (IBD). Independentcontrol of these factors in either animal studies or standard invitro models would be difficult, if not impossible. Therefore, aphysiologically relevant human gut-on-a-chip microfluidic devicewas developed that replicated the architectural and histotypicalfeatures of the human intestine in addition to peristalsis-likemotions [37]. It enabled the co-culture of gut microbes in directcontact with the intestinal epithelium for more than weeks, incontrast to static Transwell cultures or organoid cultures. Usingthis device, immune cells were found to synergize with lipopoly-saccharide (LPS) or bacteria in inducing damage to both villi andthe epithelial barrier through secretion of a set of proinflammatorycytokines. By manipulating individual factors (purified cytokinesalone or in different combinations), a requirement for high levelsof IL-8 for tissue damage was established.
Moreover, bacterialovergrowth was found in the absence of cyclic mechanical defor-mations, even with constant luminal flow. This human gut on-a-chip could be further developed by incorporating primary or iPSC-derived epithelial cells, various subsets of immune cells, andmicrobial populations, and applied to identify and test potentialtherapeutic targets and drug candidates for IBD.A different gut-on-chip was described by Shah et al . [38], namedhuman-microbial crosstalk (HuMiX) and designed to allow aerobicconditions for human cells and anaerobic conditions for bacteria.As shown in Fig. it comprises three separate microchannels,one for perfusion, one for the human epithelial cells, and one forthe microbes. Primary
CD4 + T cells were cultured for over h inthe presence of a facultative anaerobe commensal without anysignificant decrease in viability (Fig. is also at the root of common respiratory diseases,such as asthma, chronic bronchitis, and emphysema. The clinicalrelevance of animal models has been questioned because of struc-tural and functional peculiarities, whereas standard in vitro generally lack properly arranged endothelial and immunesystem components and an active, shear stress-inducing fluid flow.To overcome these limitations, a human ‘small airway-on-a-chip’was fabricated comprising an upper (airway) channel and a lower(microvessel) channel separated by a thin, porous polyester mem-brane colonized with lung airway epithelial cells and lung micro-vascular endothelial cells, respectively [39]. The device reproducedseveral structural and functional features of bronchioles for per-iods of weeks while enabling independent control of systemparameters and analysis of human organ-level responses in realtime with molecular-scale resolution. For example, both indepen-dent and collective responses of the lung epithelium and endo-thelium to a viral mimic [poly(I:C)] were identified, which wouldhave been impossible in humans. The model was also used to REVIEWS
Drug
Discovery
Today ! Volume
Number ! February R e v i e w s ! G E N E T O S C R EE N nvestigate new therapies. Notably, the action of a novel inhibitorof nuclear factor (NF)- k B signaling in suppressing neutrophiladhesion was found to be largely microflow dependent, suggestingpreferential inhibition of early events of the neutrophil-recruit-ment cascade, and consistent with a reduction in the expression ofadhesion molecules on endothelial cells. Lymphoid organ-on-a-chip platforms
Although not ideal to study the complexity behind the lymphoidorgan physiology and pathology, protocols for static cellcultures have been known for a long time and have had a majorrole both in research and clinical settings: for example, long-termbone marrow cultures, LTBMCs, have been essential to study long-term in vitro myelopoiesis and lymphopoiesis [40,41]. More recent-ly, scaffold-based perfusion systems have been introduced forthe study of BM [42,43]. Among the applications of microfluidictechnology, an engineered hemopoietic bone marrow-on-a-chip(eBM) was produced by implanting into mice a cylindrical poly-dimethylsiloxane (PDMS) device containing a collagen scaffoldenriched with bone morphogenetic proteins and demineralizedbone powder, and with a single open end placed in the proximityof muscle (Fig. This resulted in the formation of a shell ofcortical bone surrounding hemopoietic marrow. Constructs trans-ferred into a microfluidic device were maintained for up to dayson-chip with no decrease in hemopoietic stem/precursor cells(HSPCs) compared with fresh BM, as demonstrated by phenotyp-ing and functional reconstitution of both myeloid and lymphoidlineages in g -irradiated mice [44]. Moreover, the effect of g -irradi-ation on the eBM was closely related to that on in vivo BM, unlikethat on cultures, proving the efficacy of this approach as a BM in vitro model.Aiming at improving the lifetime of such artificial in vitro models for long-term testing, Sieber et al . achieved the successful long-term culture of HSPCs (up to days, far longer than theconventional days reached by other approaches) with full multi-lineage differentiation by means of co-cultivation with humanmesenchymal stromal cells in a hydroxyapatite-coated zirconiumoxide scaffold inside a multiorgan microfluidic platform [45].Although no studies have focused on the adaptive immunityniches at the BM level, efforts have been devoted to modeling BMmalignancies, such as leukemia and multiple myeloma, in micro-fluidic devices. For example, a microfluidic acute lymphoblas-tic leukemia model, where leukemic cells were introduced alongwith stromal cells and osteoblasts in a collagen gel, showed howdifferent drug responses are elicited by changes in the microenvi-ronment [e.g., mechanical forces (flow) and biochemical cues][46]. BM models can be also used as innovative tools for testingcellular immunotherapy approaches. Mesenchymal stromal cellsand their osteogenic progeny were co-cultured with endothelialprecursors in a BM niche model facilitating the stable out-growth of primary CD138 + myeloma cells for up to days [47].The model was analyzed to assess the effects on primary myelomacells of ab T cells engineered to express a tumor-specific gd T cellreceptor (TCR).In contrast to the BM, there are no published studies on engi-neering thymus-like architectures. The latest achievements inreproducing organ-like structures in vitro following cell biologyapproaches have developed, among many other organs, an orga-nized and functional thymus-like organoid, able to support thedevelopment of CD4 + and CD8 + T cells: this was generated byreprogramming mouse embryonic fibroblasts into thymic epithe-lial cells in vitro with the upregulation of transcription factorforkhead box N1 (FOXN1) [48]. Further studies have to be designedto interrogate these thymus-like organoids to obtain fundamentalknowledge on the function of the whole organ and, hopefully, touse these artificial structures in drug-testing platforms. Drug
Discovery
Today ! Volume
Number ! February
REVIEWS
ThrombopoieticareaErythroblasticareaWhitebonemarrow
Fat cellsEndosteal lining:Bone lining cellsReticular cellsBarrier cellsArterioleVenule T cellPlasma cellMonoblastmonocyteMyeloblastgranulocyteStromal cell ErythrocyteErythroblastPlateletsMegakaryoblastB lymphoblastT lymphoblastB cell
Lymphoblastoidarea Immune cellhoming area (niche)MyeloidareaMonocytopoieticarea
Bone-inducingmaterials Subcutaneousimplantation 8 weeks eBM
Insert eBM
In vivo engineeringof bone marrow (a) (b)(c)
Drug Discovery Today
FIGURE Bone marrow (BM) structure and an example of a BM-on-a-chip platform. (a)
Schematic cross-section of the BM highlighting the different hematopoietic areas(where platelets, erythrocytes, lymphocytes, monocytes, and granulocytes are continuously created and released into the blood), the immunological niches(where memory T and B cells locate), and the large vasculature, necessary for the massive cell movement in the BM. (b) An engineered BM (eBM) was producedby implanting a polydimethylsiloxane (PDMS) device in vivo and later transferring it into a microfluidic platform. (c) A bone-inducing material is placed in a PDMSstructure (left), implanted for weeks to form a visible pink marrow (center) and then integrated in a microfluidic system (right). Scale bars: mm. Adapted, withpermission, from
Ref. [75] (a) and
Ref. [44] (b,c). R e v i e w s ! G E N E T O S C R EE N ost research on secondary lymphoid organ equivalents hasbeen targeted towards LN-like architectures (Fig.
The spatialcomplexity of the LN has been investigated in a microfluidicdevice capable of addressing individual B and T cell zones in an ex vivo slice of an isolated mouse LN [49], allowing local drugadministration. Given the complexity of the LN architecture andfunctions, tissue-engineered LNs are currently designed to mimiconly a few aspects rather than a complete artificial organ. Forexample, the possibility to replicate GCs in vitro provides anefficacious tool for understanding the maturation of B cells anddesigning innovative antibody-based therapeutics. In this context,Purwada et al . encapsulated murine naı¨ve B cells and CD40L- andBAFF-expressing stromal cells into an arginylglycylaspartic acid(RGD)-functionalized gelatin hydrogel, reinforced with silicatenanoparticles for modulating the scaffold stiffness and porosity(Fig. [50,51]. The presence of the structure potentiatedCD40L/BAFF signaling and, upon addition of exogenous interleu-kin-4, led to a " increase in the proliferation of CD19 + GL7 + GC B cells compared with conventional co-culture sys-tems, and a corresponding or higher increase in isotype-switchedimmunoglobulin (Ig)G1 cells by day By replacing gelatin with asynthetic polymer displaying integrin ligands at a controlleddensity, the system was adapted to investigate the differential roleof a b - and a v b integrins in the differentiation of GC B cells[52]. In a lymph node-on-a-chip flow device, a simple perfusionsystem applying a controlled tangential shear was introduced forthe in vitro study of the mechanical forces between antigen-pre-senting DCs and different classes of T cells (CD4 + versus CD8 + ) orantigen-specific and unspecific T cells, undetectable with tradi-tional in vitro systems [53]. Perhaps the most elaborate prototype ofa human artificial lymph node (HuALN) was realized by a minia-turized, membrane-based perfusion bioreactor, hosting a hydrogelmatrix preloaded with DCs through which T and B lymphocytesare made to recirculate continuously. A planar set of microporoushollow fibers provides continuous nutrient and gas exchange.Upon in vitro immunization, cell proliferation, antigen-dependentlocal cytokine release, assembly of lymphoid follicle-like structures (albeit not typical GCs), and an increase and decrease in IgMproduction mimicking a primary response were observed [54].This model was further improved by the addition of mesenchymalstromal cells [55] and used to evaluate the efficacy of single vaccinecandidates [56].Efforts have also been made to realize cancer-on-a-chip modelsof solid lymphoid malignancies for drug discovery or mechanisticstudies. A lymphoma-on-a-chip model was obtained by seedingtumor cells in a vascularized hyaluronic acid hydrogel and used toinvestigate the crosstalk between tumor cells, immune cells, andendothelial cells as well as the response to drug treatment [57]. Amicroreactor specifically engineered to provide uniform flow pat-terns and shear stress with a pressure gradient close to physiologi-cal values was used to show that fluid flow upregulates surface Igand integrin receptors in subsets of diffuse large B cell lymphomas(DLBCL) via a mechanism mediated by CD79B and signalingtargets [58].The spleen is connected directly to the blood and has a centralrole in preventing sepsis conditions by cleansing the blood ofpathogens, in maintaining erythrocyte homeostasis by removingolder erythrocytes, as well as in the maturation of T and Blymphocytes.
Whereas the first and second functions have beenmimicked successfully in microfluidic devices, the third has not.A blood-cleansing microfluidic device has been designed forsepsis therapy [59]: it exploits the selectivity of magnetic nano-beads coated with an engineered human opsonin [mannose-binding lectin (MBL)] for removing a range of pathogens andtoxins from the blood of an infected individual without activat-ing complement factors. Its efficacy was demonstrated in a ratmodel of endotoxemic shock, where it increased animal survivalrates after a treatment. Technological efforts have been alsodevoted to reproducing in a miniaturized system the peculiarhydrodynamic forces and physical constraints present in thespleen and fundamental for its filtering activity. Different devices[60,61] have been designed and successfully tested as in vitro platforms for assessing the deformability of red blood cells (RBCs)in different conditions, such as old and fresh RBCs,
Plasmodium - REVIEWS
Drug
Discovery
Today ! Volume
Number ! February
Lymph node
Afferent lymph flow(cells, pathogens, and biomolecules)
FRC FDCHEV
T cellzoneBcellsGerminalcenter Efferentlymphatics
3D B cell follicularorganoid
CD40 /CD40L40LB cells NaiveB cell
Fluidics /Shear stress / solid stress/stiffness/ intranodal pressure T FH T FH C ellBAFF Crosslinkedgelatin-SiNPSilicatenanoparticles(SiNP)Gelatin (37˚C) G e l a t i n S i N P N a i v e CD + B c e ll s Liquidgelatin Hydrogel μ m Calcein37˚CBAFF FDCNaiveB cell CD40L
Mechanicalgradient(unknown ?)
Lymphoidtissue 37˚C (a) (b) (c) (d)(e)
Drug Discovery Today
FIGURE Lymph node structure and an example of an ex vivo engineered B cell follicle organoid. (a) Schematic of a lymph node showing the vessels that allow the in-and-out movement of lymphocytes, antigen-presenting cells, pathogens, and biomolecules. Specialized antigen-sampling, T cell, and B cell zones are indicated. (b)Schematic of the in vivo follicular interaction between mature naïve B cells, follicular T helper (T FH ) cells and follicular dendritic cells (FDCs), supporting thematuration of naïve B cells through the B cell activation factor (BAFF). (c) The cells were encapsulated in a silica nanoparticle-gelatin composite, which crosslinksat C and allows a proper B cell viability as shown in (d) and (e) (green calcein marking active B cells. Scale bar: mm. Adapted, with permission, from
Ref. [78](a) and
Ref. [50] (b – e). Abbreviations:
FRC, fibroblastic reticular cells;
HEV, high endothelial venules. R e v i e w s ! G E N E T O S C R EE N arasitized RBCs, and
RBCs from patients with hereditary spher-ocytosis.Aiming at the creation of a versatile, multifunctional in vitro assayplatform for studying the humane innate and adaptive immuneresponses within one system, the modular immune in vitro construct(MIMIC ) technology proposes a modular approach comprisingthree different modules [62]: (i) the peripheral tissue equivalent(PTE) module, resembling the innate immune response that occursin peripheral tissues (e.g., skin and muscles), gives insights into thetoxicity and immunostimulatory potential of a variety of biologicaland chemical compounds; (ii) the lymphoid tissue equivalent (LTE)module, which can include activated DCs from the
PTE module,
Tcells, B cells, and follicular DCs, generates activated T cells, anti-bodies,and cytokines;and (iii)the functional assays ordiseasemodelmodule, which includes specific functional in vitro tests for studyingvaccine candidates and pathogens. Theoretical perspectives and concluding remarks An interesting application of immune-system-on-a-chip devices isthe modeling of cell network properties. Thus, parameters extractedfrom extended imaging data sets of operating devices could beanalyzed with mathematical tools from stochastic processes andstatistical mechanics to provide a quantitative and predictive de-scription of immune cell behavior. In one approach, time-lapsemicroscopy was used to examine the motility of either immuno-competent (wt) or immunodeficient [interferon-regulatory factor-8(IRF-8)-knockout] splenocytes in the presence of melanoma cells in atwo-chamber microfluidic device [63]. From images recorded at asampling frequency of every min over h and with a spatialresolution of m m, several parameters were extracted: step length,time correlations, mean displacement, tortuosity, and ergodicity.The behavior of IRF-8-knockout splenocytes was characterized as asimple randomwalk, lacking any evidence of collective organizationeven in the presence of the target. By contrast, wild-type splenocytescould be modeled by biased random walks with a mean displace-ment of the ensemble growing linearly with time, which supportsthe hypothesis of a highly coordinated motion for the system as awhole. The analysis was extended to model the interaction betweenhuman peripheral blood mononuclear cells (PBMCs) and drug-treated breast cancer cells [64]: wild-type cells performed biasedrandom walks towards the target, whereas heterozygous mutantsheterozygous for formyl peptide receptor (FPR1, a receptor forannexin A1 on dying cells) showed a weaker bias, and homozygousmutants performed uncorrelated random walks. Moreover, by per-forming stability analysis, novel reliable descriptors (Lyapunovcoefficients) of both stable and metastable interactions betweenPBMCs and cancer cells were demonstrated. The motility analysisprovided a quantitative correlate of the response to cytotoxic che-motherapy, which highlights the role of the immune system in thistreatment. Thus, such an approach could be applied to predict theefficacy of chemotherapy in individual patients.Microfluidic systems could also provide experimental tools tovalidate theoretical models whereby the balance between an ef-fective immune response to foreign antigens and the suppressionof autoimmunity depends on the degree of connectivity of the Bcell network [65] or on collective decisions by the T cell populationvia a requirement for locally controlled quorum thresholds [66–69]. Furthermore, although necessarily replicating a limited frac- tion of the whole immune system, microfluidic devices couldprovide a sufficient set of averaged parameters to enable buildingof maximum entropy models predictive of the collective proper-ties of the whole ensemble [70]. An intriguing possibility is to usesuch systems to create real, immune cell-based, pattern-dilutednetworks with experimentally tunable variables to enable differentdegrees of parallel processing [71,72]. The operational output ofsuch ‘immune processors’ (i.e., either the relative expansion ofindividual clones or more sensitive indicators) might not onlyvalidate theoretical predictions, but also suggest novel approachesto manipulate the immune system.Although the need to add immune system components to OoCplatforms has been repeatedly underlined [7], achievements so farhave been limited. This might be due in part to competition withhumanized mice, which are likely seen by immunologists as amore relevant model, and partly to the complexity of the immunesystem itself, with its wide structural and functional diversity.Further progress calls for a closer collaboration between profes-sional technologists and immunologists.As reviewed here, a first generation of devices with limitedfunctionality has been reported. Not surprisingly, they targetpreferentially the innate, fast response system (macrophage andneutrophils) rather than the slower and more complex adaptivesystem [73,74].
There is some progress in replicating myelo- andlymphopoiesis (BM) and antigen-priming sites in LN (with evi-dence of a primary immune response). However, no OoC platformexists that reproduces the key steps of repertoire selection as wellthe generation of functionally distinct sublineages (e.g., T H H H T FH and T reg ).Designing a functional immune system on microchips is clearlya daunting task of extreme complexity. The human immunesystem has been estimated to comprise " $ lymphocytes,of a similar order of magnitude to the number of neurons in thebrain. Ig and TCR repertoires in mature lymphocytes are of theorder of , with the average size of a virgin (unprimed) clonebeing " –10 cells. Microfluidic systems generally handle sub-ml volumes, with fewer than cells. It has been argued that thelimitation of cell numbers and, hence, of receptor repertoires,could strongly bias or skew the results [75]. This is clearly a keycriticism that could be addressed by theoretical and experimentalinvestigations of the (likely considerable) extent and functionalconsequences of redundancy in the immune system, as in otherbiological networks [76,77].As briefly sketched above, the human adaptive immune systemcomprises a vast number of integrated components in a logisticallyhighly distributed, yet precise spatial organization. The anatomi-cal connections between individual parts, be they molecules orcells, are highly specific and must enable appropriate, efficientcell–cell interactions at both short range (e.g., during developmentand antigen presentation) and long range (the homing and recir-culation of T cells). Although a microfluidic device could emulate afunctional unit of the BM, thymus, or LN, dozens of devices arelikely required to simulate even a rudimentary immune system.Moreover, they must be peripherally interfaced with all otherorgans, including the microbiota, given that fundamental inputsto the collective system are provided by the latter. However,although in-depth investigations of basic immunology are likelyto require complex systems, basic systems, such as those illustrated Drug
Discovery
Today ! Volume
Number ! February
REVIEWS R e v i e w s ! G E N E T O S C R EE N ere, offer potential over the shorter term, particularly for theinitial assessment of novel personalized immunotherapies. Acknowledgments
This work was supported by the Progetto
FISR – C.N.R. ‘Tecnopolodi
Nanotecnologia e Fotonica per la Medicina di Precisione’ – CUPB83B17000010001.
L.L.d.M. acknowledges the
European
Union (ERC-StG project
INTERCELLMED,
Contract
No. forpartial funding.
A.B. acknowledges supports from the
MatchNetwork through
Progetto
Pythagoras (CUP:J48C17000250006)and partial (basal) support from
INFN and
MIUR.
Y.S.Z.acknowledges supports by the National
Institutes of Health(K99CA201603,
R21EB025270) and the
New
England
Anti-Vivisection
Society (NEAVS).
References Bhatia,
S.N. and
Ingber,
D.E. (2014)
Microfluidic organs-on-chips.
Nat.
Biotechnol.
Polini, A. et al. (2014) Organs-on-a-chip: a new tool for drug discovery. Expert
Opin.Drug
Discov. Esch,
E.W. et al. (2015) Organs-on-chips at the frontiers of drug discovery. Nat.
Rev.Drug
Discov.
Low,
L.A. and
Tagle,
D.A. (2018) ‘You-on-a-chip’ for precision medicine.
Expert
Rev.Precis.
Med.
Drug
Dev. Yesil-Celiktas, O. et al. (2018) Mimicking human pathophysiology in organ-on-chipdevices. Adv.
Biosyst. online August
Balijepalli, A. and Sivaramakrishan, V. (2017) Organs-on-chips: research andcommercial perspectives.
Drug
Discov.
Today
Watson,
D.E. et al. (2017) Fitting tissue chips and microphysiological systems intothe grand scheme of medicine, biology, pharmacology, and toxicology. Exp.
Biol.Med.
Zhang,
Y.S. et al. (2017) Multisensor-integrated organs-on-chips platform forautomated and continual in situ monitoring of organoid behaviors. Proc.
Natl.
Acad.Sci. U. S. A. E2293–E23029
Rich,
R.R. et al. (2019) Clinical
Immunology (5th edn.),
Elsevier10
Guan, M. et al. (2015) Adverse events of monoclonal antibodies used for cancertherapy. Biomed.
Res.
Int.
Baldo,
B.A. (2013)
Adverse events to monoclonal antibodies used for cancertherapy: Focus on hypersensitivity responses. Oncoimmunology e2633312 Ribas, A. and Wolchok,
J.D. (2018)
Cancer immunotherapy using checkpointblockade.
Science
Batlevi,
C.L. et al. (2016) Novel immunotherapies in lymphoid malignancies. Nat.Rev.
Clin.
Oncol.
June,
C.H. et al. (2018) CAR T cell immunotherapy for human cancer. Science
Kroschinsky, F. et al. (2017) New drugs, new toxicities: severe side effects of moderntargeted and immunotherapy of cancer and their management. Crit.
Care
Norelli, M. et al. (2018) Monocyte-derived
IL-1 and
IL-6 are differentially required forcytokine-releasesyndromeandneurotoxicityduetoCAR
Tcells.
Nat.Med.
Giavridis, T. et al. (2018) CAR T cell-induced cytokine release syndrome is mediatedby macrophages and abated by IL-1 blockade.
Nat.
Med.
Brehm,
M.A. et al. (2016) Humanized mice in translational immunology. In Translational
Immunology (Tan,
S.-L., ed.), pp.
Elsevier19
Walsh,
N.C. et al. (2017) Humanized mouse models of clinical disease. Annu.
Rev.Pathol.
Murphy, K. et al. (2012) Janeway’s
Immunobiology (8th ed.),
Garland
Science21
Mercier,
F.E. et al. (2011) The bone marrow at the crossroads of blood and immunity. Nat.
Rev.
Immunol.
Nombela-Arrieta, C. and Manz,
M.G. (2017)
Quantification and three-dimensionalmicroanatomical organization of the bone marrow. Blood
Adv. Irimia, D. and Ellett, F. (2016) Big insights from small volumes: decipheringcomplex leukocyte behaviors using microfluidics. J. Leukocyte
Biol.
Shao, N. and Qin, L. (2018) Biochips-new platforms for cell-based immunologicalassays.
Small
Methods Boussommier-Calleja, A. et al. (2016) Microfluidics: a new tool for modeling cancer-immune interactions. Trends
Cancer Huang,
C.P. et al. (2009) Engineering microscale cellular niches for three-dimensional multicellular co-cultures.
Lab
Chip Ayuso,
J.M. et al. (2016) Development and characterization of a microfluidic modelof the tumour microenvironment. Sci.
Rep. Moore, N. et al. (2018) A multiplexed microfluidic system for evaluation ofdynamics of immune-tumor interactions. Lab
Chip
Pavesi, A. et al. (2017) A microfluidic model for preclinical evaluation of TCR-engineered T cells against solid tumors. JCI
Insight Han, S. et al. (2012) A versatile assay for monitoring in vivo -like transendothelialmigration of neutrophils. Lab
Chip
Ingram,
P.N. et al. (2018) An accessible organotypic microvessel model using iPSC-derived endothelium. Adv.
Healthc.
Mater. Park,
J.Y. et al. (2011) Monitoring the status of T-cell activation in a microfluidicsystem. Analyst
Wu,
W.-H. et al. (2015) A capillary-endothelium-mimetic microfluidic chip for thestudy of immune responses. Sens.
Actuators B Chem.
Menon,
N.V. et al. (2017) Micro-engineered perfusable vasculatures forcardiovascular diseases. Lab
Chip
Cucullo, L. et al. (2011) A dynamic in vitro BBB model for the study of immune celltrafficking into the central nervous system. J. Cereb.
Blood
Flow
Metab.
Sackmann,
E.K. et al. (2012) Microfluidic kit-on-a-lid: a versatile platform forneutrophil chemotaxis assays. Blood e45–e5337
Kim,
H.J. et al. (2016) Contributions of microbiome and mechanical deformation tointestinal bacterial overgrowth and inflammation in a human gut-on-a-chip. Proc.Natl.
Acad.
Sci. U. S. A. E7–E1538
Shah, P. et al. (2016) A microfluidics-based in vitro model of the gastrointestinalhuman-microbe interface. Nat.
Commun. Benam,
K.H. et al. (2016) Matched-comparative modeling of normal and diseasedhuman airway responses using a microengineered breathing lung chip. Cell
Syst.
Dexter,
T.M. (1979)
Hemopoiesis in long-term bone marrow cultures. A review. ActaHaematol.
Whitlock,
C.A. and
Witte,
O.N. (1982)
Long-term culture of B lymphocytes andtheir precursors from murine bone marrow. Proc.
Natl.
Acad.
Sci. U. S. A. Di Maggio, N. et al. (2011) Toward modeling the bone marrow niche using scaffold-based culture systems. Biomaterials
Kim, J. et al. (2015) Organ-on-a-chip: development and clinical prospects towardtoxicity assessment with an emphasis on bone marrow. Drug
Saf.
Torisawa,
Y.S. et al. (2014) Bone marrow-on-a-chip replicates hematopoietic nichephysiology in vitro . Nat.
Meth.
Sieber, S. et al. (2018) Bone marrow-on-a-chip: long-term culture of humanhaematopoietic stem cells in a three-dimensional microfluidic environment. J.Tissue
Eng.
Regener.
Med.
Bruce, A. et al. (2015) Three-dimensional microfluidic tri-culture model of the bonemarrow microenvironment for study of acute lymphoblastic leukemia. PLoS
One
Braham,
M.V.J. et al. (2018) Cellular immunotherapy on primary multiple myelomaexpanded in a bone marrow niche model. Oncoimmunology e143446548 Bredenkamp, N. et al. (2014) An organized and functional thymus generated fromFOXN1-reprogrammed fibroblasts. Nat.
Cell
Biol.
Ross,
A.E. et al. (2017) Spatially resolved microfluidic stimulation of lymphoid tissueex vivo. Analyst
Purwada, A. et al. (2015) Ex vivo engineered immune organoids for controlledgerminal center reactions. Biomaterials
Purwada, A. and Singh, A. (2017) Immuno-engineered organoids for regulating thekinetics of B-cell development and antibody production.
Nat.
Protoc.
Purwada, A. et al. (2017) Modular immune organoids with integrin ligand specificitydifferentially regulate ex vivo B cell activation. ACS
Biomater.
Sci.
Eng. Moura
Rosa, P. et al. (2016) The intercell dynamics of T cells and dendritic cells in alymph node-on-a-chip flow device. Lab
Chip
Giese, C. et al. (2010) Immunological substance testing on human lymphatic micro-organoids in vitro . J. Biotechnol.
Seifert, M. et al. (2012) Crosstalk between immune cells and mesenchymal stromalcells in a bioreactor system. Int. J. Artif.
Organs
Radke, L. et al. (2017) In vitro evaluation of glycoengineered RSV-F in the humanartificial lymph node reactor. Bioengineering E70
REVIEWS
Drug
Discovery
Today ! Volume
Number ! February R e v i e w s ! G E N E T O S C R EE N Mannino,
R.G. et al. (2017) microvascular model recapitulates the diffuse largeB-cell lymphoma tumor microenvironment in vitro . Lab
Chip
Apoorva, F. et al. (2018) How biophysical forces regulate human B cell lymphomas. Cell
Rep.
Kang,
J.H. et al. (2014) An extracorporeal blood-cleansing device for sepsis therapy. Nat.
Med.
Rigat-Brugarolas,
L.G. et al. (2014) A functional microengineered model of thehuman splenon-on-a-chip. Lab
Chip
Picot, J. et al. (2015) A biomimetic microfluidic chip to study the circulation andmechanical retention of red blood cells in the spleen. Am. J. Hematol.
WaxDesign
Corporation (2018)
MIMIC Technology.
WaxDesign
Corporation63
Agliari, E. et al. (2014) Cancer-driven dynamics of immune cells in a microfluidicenvironment. Sci.
Rep. Biselli, E. et al. (2017) Organs on chip approach: a tool to evaluate cancer–immunecells interactions. Sci
Rep. Agliari, E. et al. (2015) Anergy in self-directed B lymphocytes: a statistical mechanicsperspective. J. Theor.
Biol.
Burroughs,
N.J. et al. (2006) Regulatory T cell adjustment of quorum growththresholds and the control of local immune responses. J. Theor.
Biol.
Busse, D. et al. (2010) Competing feedback loops shape
IL-2 signaling betweenhelper and regulatory T lymphocytes in cellular microenvironments. Proc.
Natl.Acad.
Sci. U. S. A. Butler,
T.C. et al. (2013) Quorum sensing allows T cells to discriminate between selfand nonself. Proc.
Natl.
Acad.
Sci. U. S. A. Zikherman, J. and Au-Yeung, B. (2015) The role of T cell receptor signalingthresholds in guiding T cell fate decisions. Curr.
Opin.
Immunol.
Mora, T. et al. (2010) Maximum entropy models for antibody diversity.
Proc.
Natl.Acad.
Sci. U. S. A. Agliari, E. et al. (2012) Multitasking associative networks.
Phys.
Rev.
Lett.
Agliari, E. et al. (2013) Immune networks: multitasking capabilities near saturation. J. Phys. A Boneschansker, L. et al. (2018) Convergent and divergent migratory patterns ofhuman neutrophils inside microfluidic mazes.
Sci.
Rep. Reategui, E. et al. (2017) Microscale arrays for the profiling of start and stop signalscoordinating human-neutrophil swarming. Nat.
Biomed.
Eng. Giese, C. and Marx, U. (2014) Human immunity in vitro – solving immunogenicityand more. Adv.
Drug
Deliv.
Rev.
Tononi, G. et al. (1999) Measures of degeneracy and redundancy in biologicalnetworks. Proc.
Natl.
Acad.
Sci. U. S. A. Crumiller, M. et al. (2011) Estimating the amount of information conveyed by apopulation of neurons. Front.
Neurosci. Singh, A. (2017) Biomaterials innovation for next generation ex vivo immune tissueengineering. Biomaterials
Drug
Discovery
Today ! Volume
Number ! February