Abstract

   Protozoan parasites that infect humans are widespread and lead to
   varied clinical manifestations, including life-threatening illnesses in
   immunocompromised individuals. Animal models have provided insight into
   innate immunity against parasitic infections; however, species-specific
   differences and complexity of innate immune responses make translation
   to humans challenging. Thus, there is a need for in vitro systems that
   can elucidate mechanisms of immune control and parasite dissemination.
   We have developed a human microphysiological system of intestinal
   tissue to evaluate parasite-immune–specific interactions during
   infection, which integrates primary intestinal epithelial cells and
   immune cells to investigate the role of innate immune cells during
   epithelial infection by the protozoan parasite, Toxoplasma gondii,
   which affects billions of people worldwide. Our data indicate that
   epithelial infection by parasites stimulates a broad range of effector
   functions in neutrophils and natural killer cell–mediated cytokine
   production that play immunomodulatory roles, demonstrating the
   potential of our system for advancing the study of human-parasite
   interactions.
     __________________________________________________________________

   Engineered human intestinal/vascular models can be used to study
   dissemination and immune control of parasitic infections.

INTRODUCTION

   Infections caused by parasitic pathogens are a global health problem
   that affects more than a quarter of the world’s population ([38]1), yet
   effective antiparasitic therapeutics are limited and often come with
   severe adverse reactions. Similarly, vaccines are limited for any food
   or water-borne parasitic infection ([39]1). Immune responses initiated
   by the innate immune system in the intestinal tissue are a part of the
   frontline defense against parasitic infections. A more complex picture
   of their role in shaping protective immunity development and
   pathogenesis is beginning to emerge. Infection by the protozoan and
   intracellular parasite Toxoplasma gondii affects about a third of
   humans worldwide, making it one of the most widespread human pathogens
   in the world ([40]2). Human exposure to T. gondii infection occurs
   primarily through the oral route due to ingestion of contaminated food
   or water ([41]2). However, there are substantial knowledge gaps in how
   the innate immune system interacts with human intestinal parasites like
   T. gondii in the local microenvironment of the intestinal tissue that
   either reduce parasite dissemination or contribute to the development
   of invasive systemic disease. As responses to parasitic infection by
   the innate immune system play key roles in shaping protective immunity,
   improved knowledge of the mechanisms that initiate innate immunity and
   contribute to pathogenesis is central to developing effective
   therapeutics and vaccines.

   The gut epithelium and vascular barrier regulate what enters the host
   tissue beyond the intestinal epithelial barrier and what enters the
   circulation. Functionally, the intestinal epithelial barrier separates
   the luminal contents from immune cells found in the gut parenchyma and
   prevents the systemic dissemination of the microbiota and enteric
   pathogens to the liver, spleen, and other peripheral tissues ([42]3).
   During parasitic infection, immune cells in the gut parenchyma
   coordinate host-protective responses necessary for resolving acute
   infection and preventing tissue dissemination ([43]4). Currently,
   murine hosts are predominantly used as translational models for T.
   gondii infections in humans. Despite the vast knowledge of basic and
   translational human immunology obtained from mouse studies, several
   components of the mouse immune system are incongruent with the human
   immune system ([44]5, [45]6). Humanized mice allow for improved
   modeling of the human immune systems and have emerged as alternatives
   to traditional rodent models. However, challenges still remain because
   of the lack of critical adhesion molecules and other signaling proteins
   in humanized mouse models that are required for mounting human-specific
   immune responses to infections ([46]7, [47]8). While in vivo animal
   models replicate microenvironmental complexity and physiological
   conditions, several aspects including species-specific differences,
   limited ability to image pathogen trafficking, and the time/expense of
   mechanistic study make investigation of pathogen trafficking
   challenging in these models. Thus, there is a need for relevant human
   in vitro models to interrogate the innate immune responses to
   pathogens.

   Organ-on-chip devices and microphysiological systems (MPSs) have
   overcome some of the challenges of traditional in vitro approaches by
   enabling the integration of three-dimensional (3D) complexity, spatial
   organization, and relevant cell-cell and cell–extracellular matrix
   (ECM) interactions for modeling in vivo–like responses ([48]9). Several
   technologies for modeling pathophysiological processes in human tissue
   have emerged that represent important steps forward in the evolution of
   organ-on-a-chip platforms. MPS of the lung ([49]10), brain ([50]11),
   liver ([51]12), and intestine ([52]13) has been developed for studies
   of host interactions with viruses, fungi, bacteria, and parasites
   ([53]14). In the context of modeling host-microorganism interactions,
   viral infections and bacterial colonization in the epithelia have been
   the primary applications of MPSs, with emerging applications in testing
   antiviral therapeutics and modeling innate immune responses. Immune
   cell trafficking across endothelial and epithelial barriers has been
   studied using these models ([54]15), but difficulties persist with
   integrating membrane-free interfaces that mimic tissue composition of
   the gut epithelial and vascular barrier, including cellular and ECM
   components, and tissue architecture (i.e., endothelial vessel and
   epithelial lumen geometry) that can critically influence host and
   pathogen responses. Recent efforts to model the 3D tissue anatomy of
   intestinal epithelia have culminated in the establishment of
   organoid-on-a-chip systems that retain the cellular diversity and
   regenerative potential offered by organoid technology while addressing
   the problem of lumen accessibility using microengineering or tissue
   engineering approaches ([55]16–[56]18). Using advanced 3D bioprinting
   techniques, culture of intestinal tubes can be maintained long term in
   ECM gels within these systems ([57]16). These systems can support in
   vitro tissue homeostasis of the intestinal epithelium and its
   interactions with underlying tissue components. However, functional
   responses of innate immune cells and concomitant molecular analysis of
   responses to microorganisms in these models have not been
   characterized.

   In this work, we developed an MPS of the human intestinal tissue to
   study host-parasite interactions and innate immune cell response to
   parasite infection in vitro. By using a highly tractable micromolding
   technique for creating hollow structures, we generated tubes of
   intestinal epithelium and endothelium supported by ECM gel,
   recapitulating the lumen geometries of the gastrointestinal tract and
   blood vessels. We microengineered a human-relevant model of the gut
   epithelium and vascular barrier, within the intestinal tissue, that
   provides means for studying gastrointestinal parasitic infections and
   the associated interactions with innate immune cells. We demonstrate
   adaptation of this model to incorporate primary human intestinal stem
   cells from organoids derived from intestinal tissue resections. We used
   the system to model parasite invasion and replication within the
   intestinal epithelium, followed by leukocyte trafficking to the site of
   infection. Neutrophils and natural killer (NK) cells are critical for
   early protection and reducing parasite burden during T. gondii
   infection ([58]19, [59]20). By measuring gene transcription, metabolic
   changes, and cytokine secretions, we gained interesting insights into
   the function of human innate immune cells, including neutrophils and NK
   cells, during the initial stages of T. gondii parasite infection of the
   intestinal epithelium. Hence, the engineered MPS model of the
   intestinal tissue emulates critical features of host-parasite
   interactions, enabling us to address questions related to
   human-relevant tissue response, activation of innate immunity, and the
   role of immune cells in parasite dissemination.

RESULTS

Development of an intestinal tissue MPS for studying host-pathogen
interactions

   Infection by T. gondii is naturally acquired through oral ingestion of
   food or water that are infected with parasite cysts or oocysts ([60]2,
   [61]21). The intestinal epithelium, characterized by a villus-crypt
   axis consisting of a single layer of constantly renewing and
   differentiating epithelial cells, separates microbes in the lumen from
   the intestinal vascular systems ([62]Fig. 1A) ([63]22, [64]23).
   Experiments in mice and cell line/explant studies have shown that
   translocations of T. gondii from the apical surface of the intestinal
   epithelium to the basal side, during acute infection, may occur via
   epithelial transmigration or following epithelial invasion and
   intracellular replication ([65]24). Tissue damage caused by the T.
   gondii epithelial invasion and intracellular replication leads to the
   secretion of a host of inflammatory factors by intestinal
   tissue-resident cells ([66]Fig. 1A) ([67]25). The presence of these
   factors increases the expression of integrins and chemokine receptors
   on endothelial cells of local vasculature that promote immune cell
   extravasation into the lamina propria. Leukocytes, primarily
   neutrophils, are then recruited to the site of infection from
   neighboring vasculature where they encounter effector-enhancing
   cytokines and pathogen-derived products to neutralize the parasite by
   killing or controlling their replication ([68]Fig. 1A) ([69]21,
   [70]26). Other innate immune cells such as NK cells are also recruited
   from the vascular system that further amplifies the inflammatory
   response by producing effector-enhancing cytokines ([71]Fig. 1A)
   ([72]27, [73]28). A robust immune response during these early stages of
   the T. gondii infection is critical for shaping innate immunity, which,
   in turn, induces acquired immunity.

Fig. 1. Human intestinal tissue MPS for studying innate immune responses to
parasitic infection.

   [74]Fig. 1.
   [75]Open in a new tab

   (A) Schematic representation of the design rationale for modeling
   parasite infection, in the intestinal epithelium, and innate immune
   cell responses. (B) Three-dimensional rendered illustration of the
   intestinal tissue MPS, within a polydimethylsiloxane (PDMS) device that
   includes tubular intestinal epithelium and endothelium within an ECM
   gel. Immune cells are introduced into the lumen of the endothelium for
   modeling and elucidating innate immune responses to parasite infection
   of the epithelium. (C) Schematic representation describing the
   experimental approach to set up each component (i to iv) of the MPS
   used in this study. (D) Fluorescence image showing the formation of a
   confluent Caco-2 intestinal epithelium (magenta) and HUVEC endothelium
   (yellow). (E) The model retains phenotypic characteristics of tight
   junction markers (i, ZO-1), microvilli markers (ii, villin), and
   markers of mucus producing goblet cells (iii, MUC2A) for the
   epithelium. (F) HUVEC endothelium in coculture with the epithelium
   retains expression of endothelial marker (CD31).

   To model these initial stages of T. gondii infection, we bioengineered
   an intestinal tissue MPS that incorporates 3D tubular structures of the
   intestinal epithelium and the endothelium, and immune cell components
   to allow the analysis of host immune responses against intestinal
   pathogens in a more physiologically relevant culture microenvironment
   ([76]Fig. 1B). Inspired by previous work by ourselves and others, we
   generated lumen structures within a gas- and nutrient-permeable
   collagen-based ECM hydrogel using a micromolding technique for modeling
   tubular geometries of the intestinal epithelium and the gut vascular
   barrier within intestinal tissue ([77]16, [78]29, [79]30). The
   hydrogels are integrated into a polydimethylsiloxane (PDMS) elastomeric
   device with a central chamber, which contains the ECM gel, connected to
   two pairs of inlet and outlet ports that support PDMS rods used for
   molding the tubular structures ([80]Fig. 1Ci). Following ECM gel
   polymerization, the rods are pulled out of the chamber, leaving behind
   hollow lumen structures. The inlet and outlet ports connected to the
   chamber provide direct access to the lumens, which can be used for cell
   seeding and for the supply of medium and growth factors during culture.
   We injected and cultured intestinal epithelial cells inside one of the
   tubular structures to form an epithelial lumen with direct access to
   the apical surface. After the epithelial lumen is formed, in the same
   manner, we seeded and cultured vascular endothelial cells in the
   adjacent lumen to generate a biomimetic endothelial vessel ([81]Fig.
   1Cii). To model epithelial invasion of T. gondii, we seed parasites
   into the apical surface of the epithelium through the inlet port and
   culture them ([82]Fig. 1Ciii). Next, immune cells are introduced
   directly into the endothelial vessel and cultured for evaluating the
   immune response ([83]Fig. 1Civ). Together, our model mimics components
   of the intestinal tissue including the epithelium, adjacent blood
   vessels, and immune cell components that could provide insight into
   human innate immune response during infection by protozoan parasites in
   a more physiologically relevant context.

   For initial characterization of the model, we generated tubular
   intestinal epithelium using the human colon epithelial cell line,
   Caco-2, and formed a tube-shaped endothelial vessel using human
   umbilical vein endothelial cells (HUVECs) ([84]Fig. 1D). The ECM
   hydrogel provides structural support for the formation of continuous
   layers of epithelial and endothelial cells that functions as a barrier
   between the apical side of the lumen and the underlying matrix. The
   intestinal epithelium generated within our system expresses tight
   junction marker ZO-1, which is involved in barrier function, as well as
   villin 1 (VIL1) which is involved in microvillus formation and
   epithelial restitution after damage ([85]Fig. 1E, i and ii). The
   intestinal epithelium is increasingly recognized as a critical
   component of mucosal innate immunity against invading microorganisms
   through the secretion of mucin and antimicrobial proteins. We
   demonstrate the expression of mucin 2 (MUC2), which is secreted by
   goblet cells in the intestinal epithelium and forms a major component
   of the inner mucus layer ([86]Fig. 1Eiii). Expression of these cell
   lineage markers highlights the characteristic properties of the
   intestinal epithelium cultured within our system. To assess the
   characteristic features of the endothelial vasculature, we stained for
   and observed the expression of endothelial junction protein CD31
   ([87]Fig. 1F). Together, these results show that cocultures of tubular
   intestinal epithelium and endothelium supported by collagen-based ECM
   gel retain the phenotypic characteristics of a mature epithelium and
   endothelium.

Generation and characterization of primary small intestinal tubes in the
intestinal tissue model MPS

   In vitro cultures of patient or induced pluripotent stem cell–derived
   intestinal epithelial stem cells in 3D organoids can generate a fully
   differentiated and polarized epithelium ([88]31–[89]33). However, the
   relative inaccessibility of the apical surfaces makes the inoculation
   of larger pathogens, such as T. gondii compared to bacteria or viruses,
   into the cavity difficult to perform. Moreover, the variable 3D
   geometry of organoids makes real-time imaging of host-pathogen
   interactions technically challenging. Therefore, we adapted our
   intestinal tissue MPS to integrate the formation of tubular epithelium
   using primary human intestinal epithelial cells, from intestinal
   organoid fragments, to provide a model that retains the major
   epithelial cell types and provides easy access to the apical surface
   ([90]Fig. 2A). To incorporate primary human intestinal epithelial
   cells, we obtained surgical samples from macroscopically normal regions
   of the human small intestinal tissue. Human intestinal crypts
   containing functional stem cells, derived from the jejunum region of
   the small intestine, were used to generate organoids and cultured for
   >5 passages before use within our model. Jejunal organoids cultured in
   an expansion medium for 7 to 9 days were dissociated into a mix of
   fragments and single-cell suspensions before being introduced into the
   lumen tube of the microdevice ([91]Fig. 2B). To maximize coverage of
   the luminal surface, primary small intestinal cells were seeded into
   the device twice and cultured in expansion media for 24 hours between
   each cell seeding. The device was flipped upside down before the second
   round of cell seeding to facilitate cell adhesion to the top half of
   the lumen surface. Initially, the large majority of cells appear spread
   across the luminal surface within 1 to 2 days of cell seeding and
   progressively grow to form a continuous epithelium. Following cell
   adhesion to the matrix (approximately 72 hours from initial cell
   seeding), the culture medium is changed to a differentiation medium
   (DM), and the intestinal epithelium is cultured for 5 to 7 days to
   promote differentiation of epithelial subtypes. We used a previously
   published differentiation culture medium formulation and made further
   modifications (see Materials and Methods) to accommodate the culture of
   HUVECs in the adjacent lumen following differentiation of the
   epithelium. HUVECs were seeded into the adjacent lumen, and the device
   was inverted every 30 min over a 2-hour period to coat the lumen with
   endothelial cells. The tubes of intestinal epithelium and endothelium
   were cocultured for an additional 48 hours before infection with T.
   gondii. For modeling immune response to parasite infection, the
   intestinal epithelial tube was infected with T. gondii for 48 hours and
   immune cells including neutrophils and NK cells were added to the lumen
   of the endothelium followed by coculture within the model for 12 to 24
   hours ([92]Fig. 2B).

Fig. 2. Integrating primary intestinal epithelial cells in the human
intestinal tissue MPS.

   [93]Fig. 2.
   [94]Open in a new tab

   (A) Schematic showing spatial distribution of the intestinal epithelium
   and endothelium in the intestinal tissue MPS. (B) Optimized culture
   protocol and timeline to set up the intestinal tissue MPS. (C) Tubular
   HUVEC endothelium and primary intestinal epithelium cells as shown by
   immunostaining of F-actin. (D) Differentiated tubular epithelium shows
   retention of phenotypic characteristics including expression of
   epithelial cell adhesion protein (i, E-cadherin), enterocyte-specific
   marker [ii, fatty acid–binding protein 1 (FABP1)], marker for
   mucin-producing cells (iii, MUC2A), and marker for protein involved in
   formation of microvilli (villin 1, iv). (E) Schematic representing the
   culture and differentiation of human primary intestinal epithelial
   cells in Transwells and in the intestinal tissue MPS. (F and G) Bar
   graphs showing differential gene expression in cells isolated from
   intestinal crypts and in epithelium cultured in the intestinal tissue
   MPS and Transwells. Genes analyzed include markers associated with the
   crypt and villus compartment of the intestinal tissue. Values are
   presented as means ± SD from four independent experiments involving
   tubular or monolayer epithelium generated from human intestinal
   organoids (* = Transwell versus lumen, ****P ≤ 0.0001, ***P ≤ 0.001,
   and **P ≤ 0.01; # = isolated crypts versus lumen, ####P ≤ 0.0001 and #P
   ≤ 0.05; & = isolated crypts versus Transwell, &&&&P ≤ 0.0001, &&&P ≤
   0.001, &&P ≤ 0.01, and & P ≤ 0.05). ns, not significant.

   Cocultures of tubular small intestinal epithelium derived from human
   primary intestinal epithelial cells with an adjacent vascular lumen can
   also be established, as shown by immunofluorescent staining of F-actin
   ([95]Fig. 2C). Immunofluorescence staining confirmed the expression of
   adherens junction component (E-cadherin) and enterocyte marker (fatty
   acid–binding protein, FABP1), which are characteristic of intestinal
   epithelial tissue ([96]Fig. 2D, i and ii). Similar to the Caco-2
   epithelial tubes, protein expressions of MUC2 and villin were also
   observed in epithelial tubes generated from primary intestinal
   epithelial cells ([97]Fig. 2D, iii and iv).

   In vivo, the intestinal epithelium rests on a supporting basement
   membrane composed of structural and adhesive proteins that epithelial
   cells use to anchor, migrate, and differentiate ([98]34). Matrix
   properties such as ECM ligands, stiffness, and porosity are key factors
   that influence a wide range of cell behaviors including viability,
   tissue organization/architecture, and stem cell renewal and
   differentiation ([99]35, [100]36). The intestinal epithelium is
   characterized by a villus-crypt axis that is under constant cell
   renewal, with cells migrating from the proliferative crypt compartment
   to the functional villus compartment while differentiating. Major cell
   types of the intestinal epithelium include stem cells, typically
   enriched in the crypt compartment, to absorptive enterocytes, typically
   found in the villus compartment, and cells of secretory lineage such as
   goblet and Paneth cells. To further characterize the change in cell
   type composition following differentiation in the intestinal tissue
   MPS, we analyzed gene expression of several established marker genes
   characteristic of different cell types of the intestinal epithelia
   including stem cells, goblet cells, Paneth cells, enterocytes, and
   other markers associated with antimicrobial activity. Details of
   up-regulated/down-regulated genes and their function in the human
   intestinal tissue are highlighted in figs. S1 and S2. We compared
   differentiation of intestinal tubes in our platform and epithelial
   monolayers in a standard culture platform (Transwell) ([101]Fig. 2E)
   against stem cell–rich intestinal crypts isolated from small intestinal
   tissue resections. Small intestinal epithelial cells in the intestinal
   tissue MPS and the Transwell system were cultured under the same
   conditions: in the presence of expansion media for 2 days, followed by
   7 days of culture in the DM.

   Reverse transcription quantitative polymerase chain reaction (RT-qPCR)
   analysis showed that differentiation in both the tubular
   (microphysiological model) and the monolayer (Transwell) epithelium
   generally led to decreased expression of markers for cellular
   proliferation, MKI67 ([102]Fig. 2F). In contrast, levels of enterocyte
   marker, sucrase-isomaltase, and levels of genes that encode for the
   production of antimicrobial peptides, β-defensin 1 (DEFB1), were
   elevated in the differentiated epithelium in both platforms ([103]Fig.
   2G). Moreover, our analysis revealed that, compared to Transwells,
   differentiation in our model leads to increased gene expression of
   markers associated with microvilli formation, VIL1, and production of
   antimicrobial peptides, defensin beta 4A (DEFB4A) ([104]Fig. 2F). A
   further decrease in markers associated with stem cell renewal,
   leucine-rich repeat-containing G protein–coupled receptor 5 (LGR5) and
   crypt regeneration, and axis inhibition protein 2 (AXIN2) relative to
   intestinal crypts was also observed in our model compared to the
   Transwell condition ([105]Fig. 2G). Levels of absorptive cell marker,
   bestrophin 4 (BEST4), and marker for mucin-producing goblet cells,
   MUC2, decreased significantly in monolayers differentiated in
   Transwells but remained at similar levels in our tissue model.
   Expression of marker for Paneth cells, lysozyme, remained the same in
   both platforms. Overall, these results suggest that, compared to
   standard culture techniques, primary intestinal epithelial
   differentiation in our platform exhibits improved differentiation with
   attenuated crypt-based stem cell signatures (reflected by decreased
   levels of MKI67, LGR5, and AXIN2 transcripts) and elevated villus-like
   absorptive/goblet cell signatures (reflected by higher abundance of
   VIL1, BEST4, and MUC2 transcripts).

Modeling protozoan parasite infection in human intestinal epithelium

   Having established that the intestinal tissue MPS recapitulates tubular
   geometries of the intestinal epithelium and vasculature, the system has
   strong potential for modeling parasite invasion, replication, and
   translocation beyond epithelium into the lamina propria ([106]Fig. 3A).
   T. gondii infects the small intestinal epithelium cells following
   ingestion of parasites found in contaminated meat, produce, or water.
   During acute infection, T. gondii parasites that invade the intestinal
   epithelium differentiate into a rapidly replicating life stage,
   referred to as the tachyzoite stage, which then disseminates throughout
   the body ([107]21, [108]37). To model T. gondii infection of the
   intestinal epithelium, we introduced tachyzoites into the tubular
   epithelium of our intestinal tissue MPS. Our model provides an exposed
   luminal surface accessible to pathogens, allowing us to directly expose
   the cells to the parasite. Caco-2 epithelial tubes were exposed to
   approximately 8 × 10^7 transgenic mCherry-tagged T. gondii (type II
   ME49 strain) tachyzoites per tube. The lumen of the epithelial tube was
   washed with media 12 hours post-infection (hpi) to remove parasites
   that did not adhere to or invade the epithelium. Discrete and dense
   areas of parasites could be observed within a small proportion of
   epithelial cells using fluorescence microscopy at 24 hpi and were
   largely contained within the boundaries of the epithelial tube
   ([109]Fig. 3B). The T. gondii replicated within the epithelial cells
   between 24 and 48 hpi. On the basis of the T. gondii seeding density
   used in our study, bright-field images of Caco-2 epithelium do not
   indicate considerable damage to the epithelium at 48 hpi (fig. S3A).
   However, as parasite-harboring cells continue to increase, epithelial
   cell lysis is observed at 72 hpi ([110]Fig. 3C), releasing tachyzoites
   into the lumen of the epithelium and the basal side of the epithelium.
   Cell lysis was the primary mode of T. gondii translocation across the
   epithelium observed in our system; however, T. gondii transmigration
   across the epithelium was also observed at times, before cell lysis,
   when lower T gondii seeding densities were used. To confirm and
   quantify parasite replication, T. gondii genomic DNA in the infected
   epithelial tubes was quantified at 24 and 48 hpi via qPCR using the
   SAG1 (T. gondii surface antigen) primer set. qPCR analysis in the
   infected epithelium showed a sixfold increase in T. gondii genomic DNA
   at 48 hpi relative to 24 hpi ([111]Fig. 3D), confirming our
   microscopy-based observations. To examine the state of the epithelium,
   we analyzed the effects of T. gondii infection on the barrier function
   of the epithelium. As a control, uninfected epithelial tubes and
   epithelial tubes infected with heat-killed T. gondii were used (fig.
   S3, B and C). We measured the permeability characteristic of the
   epithelial tubes by quantifying transport of fluorescently labeled
   polysaccharides [fluorescein isothiocyanate (FITC)–dextran, 20 kDa]
   from the apical side to the basal side of the epithelium (fig. S4).
   Epithelial barrier function was analyzed at 48 hours for control
   (uninfected) epithelium and epithelium infected with live T. gondii and
   heat-killed T. gondii (fig. S4A). Diffusion of 20-kDa FITC-dextran was
   tracked over a period of 30 min to analyze epithelial permeability
   (fig. S4, B and C). Our analysis shows increased permeability in T.
   gondii–infected epithelium compared to uninfected and heat-killed T.
   gondii control conditions ([112]Fig. 3E), suggesting that the infection
   impairs epithelial barrier function. This also demonstrates our ability
   to analyze epithelial permeability directly in our platform. Together,
   these results demonstrate that tubular intestinal epithelium generated
   in our human intestinal tissue MPS can support the invasion,
   replication, and translocation of T. gondii beyond the epithelium,
   which are key initial stages of the infection process before the
   encounter with innate immune cells.

Fig. 3. Modeling protozoan parasite invasion and replication in the
intestinal epithelium.

   [113]Fig. 3.
   [114]Open in a new tab

   (A) Schematic representation of epithelial infection by T. gondii.
   Infection involves epithelial invasion, intracellular replication, and
   transmigration or epithelial cell lysis of T. gondii, which releases
   the parasites into the lamina propria containing immune cells and
   intestinal vasculature. (B) Caco-2 epithelial tubes were infected with
   mCherry-tagged T. gondii of the ME49 strain for 24 hours and imaged by
   fluorescence microscopy. The images depict time course images at 24 (i)
   and 48 hours (ii), showing that viable T. gondii was present within the
   epithelial lumen. White dashed line represents the epithelial boundary
   separating apical (lumenal) surface from the basal surface. (C)
   Fluorescent image depicting epithelial cell lysis 72 hours following
   infection with mCherry-tagged T. gondii. (D) Relative fold expression
   of T. gondii genomic DNA in the infected epithelial tubes, quantified
   at 24 and 48 hpi via qPCR using the SAG1 (T. gondii surface antigen)
   primer set. Values are presented as means ± SD from two independent
   experiments and 12 different devices (***P ≤ 0.001). (E) Comparison of
   dextran (20 kDa) permeability through Caco-2 intestinal epithelium
   under control and infection by heat-killed T. gondii or live T. gondii
   conditions. Values are presented as mean permeability + SD (****P ≤
   0.0001).

Neutrophil response to T. gondii infection of the intestinal epithelium

   Phagocytes involved in innate immunity (i.e., dendritic cells,
   macrophages, and neutrophils) are the first cells to encounter
   intracellular pathogens after they cross the epithelial barrier of the
   intestine. In mice, neutrophils are recruited in abundance to the sites
   of T. gondii infection and account for a high proportion of T.
   gondii–invaded phagocytes in the small intestine following oral entry
   ([115]Fig. 4Ai) ([116]26). However, little is known about how
   neutrophils traffic and behave at the gut infection site, and data in
   mouse studies remain to be validated in human models. Therefore, to
   gain insight into the behavior and response of human neutrophils to T.
   gondii infection, we used our intestinal tissue MPS to assess
   neutrophil recruitment behavior to the infection site from a
   neighboring blood vessel ([117]Fig. 4Aii). We infected Caco-2
   intestinal epithelial tubes in our intestinal tissue MPS with ME49
   tachyzoites for 48 hours, as previously described ([118]Fig. 3B).
   Neutrophils isolated from healthy donor blood were introduced to the
   endothelial lumen of infected systems and observed for migration
   behavior over a course of 16 hours. Time-lapse imaging of the
   intestinal tissue MPS revealed neutrophils migrating within the
   endothelium and some events of transendothelial migration across the
   endothelium into the surrounding matrix toward the infected epithelium
   ([119]Fig. 4B and movies S1 and S2). We observed increased neutrophil
   speed and displacement from the endothelium in our infected systems
   compared to uninfected controls (fig. S5). Neutrophil extravasation and
   migration were also observed in models without epithelium, where a
   slight, but not significant, increase in migration distance of
   neutrophils was observed in systems with an epithelium (fig. S6).
   Migrating neutrophils were seen in the matrix near the infected
   epithelium, and some small fractions of neutrophils were also observed
   within the lumen of the infected epithelium ([120]Fig. 4C). Confocal
   imaging also revealed the presence of fluorescently labeled neutrophils
   within the infected epithelium colocalized with mCherry-tagged ME49
   tachyzoites ([121]Fig. 4C, inset).

Fig. 4. Neutrophil response to T. gondii–infected epithelium.

   [122]Fig. 4.
   [123]Open in a new tab

   (A) Schematic depicting neutrophil trafficking and interaction with a
   T. gondii–infected intestinal epithelium as seen in vivo (i) and as
   modeled in the intestinal tissue MPS (ii). (B) Confocal image showing
   extravasated neutrophils and their migration toward a T.
   gondii–infected Caco-2 epithelium. Neutrophils were introduced at 48
   hpi and imaged 6 hours after neutrophil seeding. (C) Confocal image
   showing neutrophils trafficking to the infection site with some
   neutrophils interacting with the T. gondii–infected epithelium. Inset
   shows colocalization of neutrophils with T. gondii. Image was taken 24
   hours after neutrophil seeding. (D) Confocal image showing neutrophil
   extravasation and trafficking toward a T. gondii–infected epithelium
   generated from primary intestinal epithelial cells. Neutrophils were
   introduced at 72 hpi and imaged 6 hours after neutrophil seeding. (E)
   Grouped scatter plot showing distance migrated by neutrophils from the
   endothelium toward T. gondii–infected primary intestinal epithelium.
   Each dot represents the migration distance of a single neutrophil
   (****P ≤ 0.0001). (F) Clustergram comparing gene expression in
   neutrophils from models with T. gondii–infected and control epithelium.
   Neutrophils from three nondiseased donors were used. Gene expression
   was analyzed for each donor in both control and infected conditions.
   The dendrograms at the top indicate relationship among experimental
   conditions (control and infected) and, on the left of the figures,
   indicate relationship among genes (high and low expression). (G) The
   top five most relevant GO terms associated with the analyzed gene set,
   and their corresponding −log [false discovery rate (FDR)–adjusted P
   value] (i). The percentage of genes was associated with each GO term
   according to their fold changes in expression (increase in yellow and
   decrease in blue) (ii).

   When we infected primary small intestinal epithelial cells within the
   intestinal tissue MPS, a similar increase in end-to-end displacement
   was observed compared to uninfected systems ([124]Fig. 4, D and E). To
   directly examine the effects of T. gondii infection on gene
   transcription, we performed RT-qPCR analysis on neutrophils in our
   model that are responding to infected and uninfected primary small
   intestinal epithelium. We selected genes that are involved in
   establishing a protective immunity and activating immune effector
   functions against invading pathogens. Differential gene expression
   analysis revealed that infection of the intestinal epithelium with T.
   gondii up-regulates multiple genes in neutrophils involved in innate
   immune responses, including IL-10 (interleukin-10), MYD88 (myeloid
   differentiation primary response 88), TNF (tumor necrosis factor),
   IL-1B, IL-6, IL-1A, NOS2 (nitric oxide synthase 2), IL-12A, NLRP3
   (nucleotide-binding oligomerization domain, leucine rich repeat and
   pyrin domain containing 3), and IFN-γ (interferon-γ) ([125]Fig. 4F).
   Hierarchical clustering analysis revealed distinct gene expression
   profiles between infected and control conditions. Clustering analysis
   was conducted using average linkage clustering with Pearson correlation
   as the default distance metric. We identified the 10 most significant
   Gene Ontology (GO) terms related to biological processes (table S1A)
   based on the gene set enrichment analysis molecular signatures database
   ([126]www.gsea-msigdb.org/). The five most relevant GO terms ([127]Fig.
   4G) represent biological processes involving response to cytokines,
   defense response, positive regulation of cytokine production,
   regulation of leukocyte-mediated immunity, and positive regulation of
   immune effector functions ([128]Fig. 4Gi). Most of the genes chosen in
   our study are involved in these biological processes (table S1B), and
   the overall gene expression of each of these gene sets was generally
   up-regulated in neutrophils from the infected models ([129]Fig. 4Gii).
   These expression patterns suggest that pathways involved in cytokine
   response, defense response, regulation of leukocyte mediated immunity,
   and immune effector functions may be active in neutrophils responding
   to T. gondii–infected epithelium. Together, these results show that
   infection of the intestinal epithelium with T. gondii in our intestinal
   tissue MPS elicits responses consistent with observed in vivo behavior
   where increased transendothelial migration, trafficking of neutrophils
   toward the infection site, and activation of pathways involved in
   defense response to invading pathogens are observed.

IFN-γ–mediated neutrophil response to T. gondii infection

   In the context of T. gondii infection, IFN-γ is a cytokine that plays a
   major role in host resistance and is critical for coordinating
   protective immunity ([130]38). IFN-γ mediates its protective effects by
   stimulating lysosomal activity ([131]39), inducing expression of nitric
   oxide synthase and effector genes ([132]40), and modulating metabolic
   activity ([133]41) in phagocytes. Neutrophils, upon infiltration to the
   site of infection, execute a broad range of immune effector functions,
   which include phagocytosis, production of reactive oxygen species or
   antimicrobial peptides, and activation of programmed cell death
   pathways to reduce the pathogen’s chances of survival ([134]Fig. 5Ai)
   ([135]42). Here, we investigate the influence of stimulation and
   inhibition of IFN-γ on the effector functions and metabolic activity of
   neutrophils following an encounter with a T. gondii–infected
   epithelium. To evaluate this, we generated tubular epithelium with
   Caco-2 cells, infected them with T. gondii tachyzoites, as previously
   described, and introduced primary human neutrophils into the lumen of
   the epithelium ([136]Fig. 5Aii). As T. gondii is an intracellular
   pathogen, we infected the apical side of the Caco-2 epithelium and
   allowed incubation for 72 hours, with consistent media replenishment,
   which resulted in some epithelial cells undergoing cell lysis and
   exposing T. gondii to the luminal contents ([137]Fig. 3C). Neutrophils,
   isolated from healthy donor blood, were then added into the epithelial
   lumen to maximize encounter probability with T. gondii and cocultured
   for 6 hours before collection and assessment ([138]Fig. 5Bi).
   Neutrophil extracellular trap (NET) formation, apoptosis, and
   parasitization of neutrophils were examined ([139]Fig. 5B, ii to iv,
   and fig. S7). We first examined and quantified the fraction of
   neutrophils undergoing NET formation, which are released from
   neutrophils in response to invading pathogens. A key step in NET
   formation is the release of antimicrobial DNA complexes into the
   cytosol of the cell. Hence, the measurement of decondensed nuclei has
   been a recognized method for quantifying neutrophils undergoing NET
   formation ([140]43). We used Hoechst nuclear stain to distinguish
   between intact and decondensed nuclei, where intact was characterized
   by normal trilobed nuclei and decondensed was characterized by diffuse
   staining with large nuclear area ([141]Fig. 5Ci). Neutrophils were
   stained with Hoechst before introduction into the lumen of T.
   gondii–infected epithelium and removed after 6 hours for analysis.
   Overall, increased NET formation was observed for conditions with
   infection. IFN-γ stimulation significantly increased the fraction of
   neutrophils undergoing NET formation compared to control infection.
   Conversely, blocking IFN-γ significantly decreased the fraction of
   NET-forming neutrophils relative to immunoglobulin G controls and
   IFN-γ–stimulated systems. To evaluate the influence of IFN-γ on
   apoptotic cell death, we examined the induction of apoptosis using a
   caspase 3/7 activity assay. Our results indicate that blocking IFN-γ
   significantly triggered apoptosis, while IFN-γ stimulation suppressed
   it ([142]Fig. 5Cii). We then examined the effects of IFN-γ on the
   fraction of T. gondii–parasitized neutrophils. To eliminate parasites
   not in contact with neutrophils from analysis, we resuspended
   neutrophils collected from the epithelium in culture medium and
   centrifuged them at 200g for 3 min before imaging. Consistent with the
   trends observed with NET formation, IFN-γ stimulation significantly
   increased the fraction of T. gondii–parasitized neutrophils relative to
   unstimulated systems ([143]Fig. 5Ciii). Blocking IFN-γ decreased the
   fraction of T. gondii–parasitized neutrophils compared to
   IFN-stimulated systems, demonstrating the sensitivity of neutrophils to
   IFN-γ stimulation.

Fig. 5. Neutrophil effector functions and contribution to T. gondii
dissemination beyond the intestinal epithelium.

   [144]Fig. 5.
   [145]Open in a new tab

   (A) Schematic depicting neutrophil responses at the infection site, as
   seen in vivo (i), and as modeled in the intestinal tissue MPS (ii). (B)
   Assessment of neutrophil effector functions (i). Immunofluorescence
   image showing neutrophil nuclei. White circular line highlights
   neutrophils with decondensed nuclei (ii). Combined bright-field and
   fluorescence image showing apoptotic neutrophils. White circle
   highlights an Apopxin Green–positive neutrophil (iii). Combined
   bright-field and fluorescence image showing parasitized neutrophils.
   White arrow highlights a neutrophil directly interacting with a T.
   gondii parasite (iv). (C) Bar graphs showing percentage of neutrophils
   undergoing NET formation (i), apoptosis (ii), and parasitization (iii),
   in response to IFN-γ stimulation and blockade. Values are presented as
   means ± SD of neutrophil response from three nondiseased donors (****P
   ≤ 0.0001, **P ≤ 0.01, and *P ≤ 0.05; # = versus no infection control,
   ####P ≤ 0.0001 and ##P ≤ 0.01). IgG, immunoglobulin G. (D) Optical
   metabolic imaging was used to visualize intracellular NAD(P)H and FAD
   fluorescence intensities of neutrophils in infected systems without (i)
   and with IFN-γ stimulation (ii) [redox ratio = NAD(P)H intensity
   divided by FAD intensity]. Violin plots showing the analysis of
   neutrophil redox ratio based on NAD(P)H and FAD intensity (iii) (***P ≤
   0.001). (E) Combined bright-field and fluorescence image showing
   neutrophils within a T. gondii–infected epithelium. (F) Fluorescence
   image showing some neutrophils with internalized T. gondii after 6-hour
   coculture with infected epithelium (i). Fluorescence images showing T.
   gondii trafficking by neutrophils across the epithelial barrier. White
   dashed line indicates the epithelial boundary separating apical and
   basal surface (ii and iii).

   We also investigated the influence of IFN-γ on neutrophil metabolic
   activity by means of optical metabolic imaging, which quantifies
   relative amounts of reduced form of nicotinamide adenine dinucleotide
   phosphate [NAD(P)H] and flavin adenine dinucleotide (FAD) ([146]44,
   [147]45), and, by extension, the redox ratio. The optical redox ratio
   is used to obtain information on the dynamic changes in
   oxidation-reduction rates in cells and is sensitive to alterations in
   cellular metabolic rates ([148]44). Optical NAD(P)H/FAD redox ratios
   increased in neutrophils after 6 hours in our IFN-γ–stimulated systems
   indicating increased metabolic activity ([149]Fig. 5D), and these
   observations were conserved when the analysis was limited to T.
   gondii–internalized neutrophils (fig. S8). Relative to unstimulated
   systems, lower FAD intensity was observed in response to IFN-γ
   stimulation. Together, these findings suggest that neutrophil effector
   functions are IFN-γ dependent where IFN-γ stimulation enhances
   neutrophil effector functions by increasing events of T. gondii
   internalization, NET formation, and cellular metabolic activity of
   neutrophils while prolonging their life span. These results demonstrate
   the utility of our intestinal tissue MPS in testing the sensitivity of
   immune cells to stimulants that may influence their response to
   pathogens.

   Following an encounter at the infection site, engulfed microbes are
   carried by migratory phagocytes such as macrophages and neutrophils
   beyond the epithelia of barrier organs into deeper tissue and draining
   lymph nodes, contributing to pathogen dissemination to other organs
   ([150]Fig. 5A) ([151]46). In T. gondii–infected mice,
   parasite-containing neutrophils in the small intestine have been shown
   to transport the parasite across epithelial barriers to facilitate
   parasite spread both within the intestine to other regions and beyond
   to the spleen and mesenteric lymph nodes ([152]26, [153]46, [154]47).
   To investigate whether dissemination of T. gondii by migratory
   neutrophils beyond the epithelium could be observed in our intestinal
   tissue MPS, we introduced neutrophils into an infected intestinal
   epithelium generated from Caco-2 cells ([155]Fig. 5E). Within 6 hours,
   T. gondii–containing neutrophils could be seen within the lumen of the
   intestinal epithelium ([156]Fig. 5Fi). Time course imaging revealed the
   translocation of T. gondii–containing neutrophils from the apical side
   to the basal side of the epithelium ([157]Fig. 5F, ii and iii). These
   results show that our intestinal tissue MPS can recapitulate the
   mechanisms of parasite dissemination by migratory phagocytes as seen in
   vivo. Furthermore, our ability to visualize immune cell–mediated
   transport of pathogens within our intestinal tissue MPS could provide
   valuable insight into how intracellular pathogens disseminate and
   present opportunities to investigate previously unidentified targets
   for therapeutic intervention.

Innate immune cell response to parasite invasion and replication in the
intestinal epithelium

   The innate immune system coordinates the first immunological defense
   against an invading pathogen. As cytokine production by intestinal
   epithelial and endothelial cells is a key feature of early host immune
   response, we wanted to evaluate the capacity of our intestinal tissue
   MPS to produce the inflammatory mediators required for immune effector
   functions in response to parasite invasion ([158]Fig. 6A). We analyzed
   inflammatory cytokines secreted in culture media before the addition of
   immune cells using a Luminex multiplex bead-based enzyme-linked
   immunosorbent assay. Our results revealed that infection of the primary
   small intestinal epithelium with T. gondii induced significantly higher
   levels of proinflammatory cytokines/chemokines MCP-1 (monocyte
   chemoattractant protein-1), MIP-1α (macrophage inflammatory protein-1
   alpha), IL-1α, IL-1β, GM-CSF (granulocyte-macrophage colony-stimulating
   factor), IL-6, and IL-8 secretion in culture media at 48 hpi compared
   to uninfected controls ([159]Fig. 6B). Secretion of anti-inflammatory
   cytokines such as IL-4, IL-10, and IL-13, on the other hand, showed no
   change in the infected systems compared to control uninfected systems.
   Together, these cytokines make up critical components of the
   inflammatory milieu that contribute to coordinated immune defenses like
   immune cell trafficking, activation, and effector functions against
   parasites during acute infection.

Fig. 6. Immune cell response and cytokine secretions within the intestinal
tissue MPS.

   [160]Fig. 6.
   [161]Open in a new tab

   (A) Schematic conceptualization of immune cell response to soluble
   factor signaling during the initial stages of T. gondii infection, as
   seen in vivo (i), and as modeled in the intestinal tissue MPS (ii). (B)
   Cytokine concentrations measured in media collected from the intestinal
   tissue MPS consisting of endothelial vessel and primary intestinal
   epithelial tubes infected with T. gondii normalized to control,
   uninfected systems (****P ≤ 0.0001, **P ≤ 0.01, and *P ≤ 0.05). (C)
   Fluorescence image showing differences in PBMC adhesion events to the
   endothelial vessel in control (i) and infected (ii) Caco-2 epithelium.
   (D) Bar graphs showing percentage of PBMCs and NK cells adhering to the
   endothelium of control and infected systems following coculture for 2
   hours. PBMCs and NK cells were added to the lumen of the endothelial
   vessel after Caco-2 epithelial infection by T. gondii for 48 hours.
   Five devices were prepared on two different days for each condition.
   ****P ≤ 0.0001. (E) Fluorescence image showing T. gondii–infected
   primary intestinal epithelial tube in coculture with an endothelial
   vessel containing NK cells. Epithelial tubes were infected for 72 hours
   before adding NK cells into the endothelial vessel. (F) Cytokine
   concentrations measured in media collected from infected intestinal
   tissue MPS consisting of an endothelial vessel, NK cells, and primary
   intestinal epithelial tubes infected with T. gondii normalized to
   infected systems without NK cells. In all cytokine measurement
   experiments, nine devices were prepared on two different days for each
   paired conditions (infected versus control, and infection in the
   absence versus presence of NK cells); media from three devices were
   pooled to make one replicate. Values are presented as means ± SD, ****P
   ≤ 0.0001.

   Adhesion molecules on endothelial cells are known to mediate leukocyte
   rolling, adhesion, and subsequent transendothelial migration as part of
   an inflammatory response to injury or infection. We investigated the
   effects of infection on the firm adhesion of immune cells to the
   endothelium. Fluorescently labeled primary human peripheral blood
   mononuclear cells (PBMCs) were added to the endothelium 48 hpi with
   ME49 T. gondii and washed away after 2 hours to evaluate endothelial
   adhesion of immune cells ([162]Fig. 6C). The percentage of PBMC
   adhering to the endothelium doubled relative to systems without
   infection ([163]Fig. 6D). A similar increase in endothelial adhesion
   was also observed when NK-92 cells were added to the endothelial vessel
   ([164]Fig. 6D and fig. S9A). Overall, these results provide strong
   evidence of epithelial-endothelial cross-talk during parasite infection
   that promotes immune cell adhesion to endothelium, a critical step in
   endothelial transmigration and recruitment of immune cells to the
   infection site. Concurrent with the high levels of proinflammatory
   cytokines observed in our system, increased levels of soluble adhesion
   molecules, sP-selectin, sICAM-1, and sE-selectin were also found in the
   culture media of infected systems (fig. S9B).

   Before adaptive immunity is established, NK cells restrain the spread
   of infection by secreting inflammatory cytokines that are critical for
   stimulating protective immunity ([165]28). To explore NK cell–mediated
   production of cytokines in our T. gondii–infected intestinal tissue
   MPS, we compared cytokine secretions in infected systems in the
   presence and absence of NK cells. NK-92 cells were introduced in the
   endothelial lumen of infected systems at 48 hpi and cultured for 24
   hours ([166]Fig. 6E). Secreted factor analysis was performed on media
   collected from systems 24 hours after the introduction of NK cells into
   the lumen of the endothelium. Our results show that the presence of NK
   cells in infected systems led to significantly higher levels of
   proinflammatory factors including IFN-α, IL-12p70, TNF-α, IP-10,
   IL-17A, MIP-1α, MIP-1β, and IFN-γ ([167]Fig. 6F). The presence of NK
   cells in infected systems also induced significantly higher secretion
   of IL-10, IL-4, and IL-13, which are known to suppress proinflammatory
   cytokine production. While the induction of a robust inflammatory
   response is critical for host resistance against T. gondii, unregulated
   inflammation can result in exacerbated immunopathological reactions
   causing tissue damage. Thus, anti-inflammatory factors may have
   critical roles in reducing inflammatory reactions and suppressing
   collateral damage because of up-regulated inflammation during parasite
   infection.

DISCUSSION

   Development of improved treatment strategies against parasitic
   infections requires increased knowledge of the pathogen–tissue–immune
   system interactions. Murine models have made substantial contributions
   to our understanding of innate immunity against parasitic infections,
   but sufficient differences in the organization of the immune system
   between humans and mice warrant the need for improved models.
   Therefore, the development of tools that allow researchers to elucidate
   mechanisms involved in immune responses and defense strategies against
   parasites and other activators of the immune system in humans could
   accelerate the development of antiparasitic drugs and therapeutic
   strategies. In this regard, we have bioengineered an integrated MPS of
   the intestinal tissue for exploring human innate immune cell responses
   to parasite infection in the gut. Tubular tissue structures like the
   intestines and the vasculature system serve essential roles in
   separation of internal contents and pathogenic microbes in the
   intestinal lumen from systemic dissemination. Our human intestinal
   tissue MPS incorporates in vivo–like tubular geometries of the
   intestinal epithelium and endothelium, allowing us to explore
   host-parasite interactions and tissue dissemination that occur during
   the initial stages of infection. Using our model, we examined
   epithelial infection by T. gondii, an intracellular parasite that
   initially infects the small intestine following ingestion. We
   introduced a suspension of T. gondii tachyzoites directly onto the
   apical surface of the intestinal tubes, and, using live cell
   microscopy, we demonstrated that our tubular intestinal epithelium
   supports invasion and intracellular replication of T. gondii. As an
   important component of the intestinal immune network, the innate immune
   system is considered the most important defense line against pathogens
   breaching the epithelial barrier and plays a pivotal role in
   maintaining immunity and preventing tissue dissemination through the
   vascular system. Although incorporating tubular intestinal epithelium
   within scaffolding matrices has previously been reported ([168]16,
   [169]30, [170]48), the originality of our model relies on the
   integration of vascular and immune cell components. This enables us to
   investigate the various facets of innate immune cell responses,
   including extravasation and trafficking, to T. gondii infection of the
   intestinal epithelium and its translocation across the epithelial
   barrier. One additional design aspect of our system is the
   compatibility with live cell microscopy enabling dynamic visualization
   of host-microbe interactions occurring at the interface between the
   intestinal epithelium and the neighboring vasculature at high
   spatiotemporal resolution. Using these techniques and molecular
   analysis, we examined early responses from human innate immune cells
   like neutrophils and NK cells and investigated parasite-immune cell
   interactions during epithelial infection that influence immune effector
   functions and tissue dissemination.

   Current in vitro studies of parasitic infections primarily rely on
   human cell lines that often do not recapitulate in vivo phenotypes. On
   the other hand, maintaining long-term proliferative cultures of the
   human intestinal epithelium in vitro has met with continuing difficulty
   due to the complex interactions between cell types and the presence of
   molecular signaling required for stem cell maintenance. Organoid
   technologies enable the expansion of primary intestinal stem cells and
   are regarded as a powerful in vitro tool for modeling intestinal
   epithelial tissue due to their structural and functional resemblance to
   in vivo tissue ([171]49, [172]50). Several studies have used these
   techniques to study T. gondii infection in intestinal organoids
   ([173]51) as well and in organoid models of the brain and central
   nervous systems ([174]52, [175]53). While apical access can sometimes
   be limited when using these techniques, methods to form epithelial
   monolayers from intestinal organoid fragments have been reported
   ([176]54, [177]55), where the major epithelial cell types found in 3D
   organoids are retained when cultured on hard surfaces coated with
   Matrigel or collagen ([178]56). Building on these methods, here, we
   adapted our intestinal tissue MPS to integrate the formation of tubular
   epithelium using primary human intestinal epithelial cells.

   We used a simple micromolding-based technique to generate tubular
   structures within a hybrid matrix, composed of type I collagen, type IV
   collagen (Col-IV), and Matrigel. Our data indicate this
   matrix-supported culture and differentiation of human intestinal
   epithelial cells, both cell lines (e.g., Caco-2) and primary
   organoid-derived intestinal stem cells obtained from jejunal tissue
   resections. Maintenance of primary human patient–derived small
   intestinal cells also requires stimulation from the basal compartment
   by niche factors that help retain stem cell components while supporting
   differentiated enterocyte-goblet cell function ([179]32, [180]49). To
   support the integration of the endothelial vessel and immune cells
   (i.e., neutrophils, NK cells, and PBMCs) into our systems, we optimized
   culture protocols and media compositions for differentiating a primary
   intestinal epithelium and to support all cell types within the
   intestinal tissue MPS. Although several molecular and functional
   properties of the intestinal epithelium within the system are expected
   to change because of constantly evolving differentiation states and
   influence from coculture with other cell types, we confirmed through
   visual inspection of the system that the epithelium and endothelium
   remain intact. Furthermore, we show that phenotypic characteristics and
   barrier properties of both the epithelium and endothelium are retained
   for the duration of the experiments (5 days after endothelial cell
   seeding), which correlates with the turnover of intestinal epithelia in
   vivo ([181]57). Intestinal tubes generated within the hybrid matrix,
   both from the Caco-2 cell line and primary intestinal stem cells,
   exhibited key properties of the intestinal epithelium including a
   polarized epithelium expressing markers of tight junctions, goblet
   cells, and microvillus formation. Gene expression analysis showed that
   expansion of primary intestinal stem cells in tubular structures
   resulted in improved differentiation of the epithelium, with elevated
   villus-like absorptive/goblet cell signatures and attenuated
   crypt-based stem cell signatures, compared to monolayers formed in
   Transwells. These results agree with previous reports showing that a
   suitable matrix with relevant ECM proteins and biophysical properties
   are key factors in providing physical support and facilitating the
   differentiation of intestinal stem cells ([182]36).

   By integrating blood vessels, circulating innate immune cells, and
   primary intestinal epithelial cells within a biomimetic 3D scaffolding
   matrix, we incorporated key cellular components and relevant cell-ECM
   interactions of the intestinal tissue for improved modeling of
   host-parasite interactions and innate immune cell responses. We also
   demonstrated impaired epithelial barrier function caused by T. gondii
   infection. Initial responses against T. gondii are managed by innate
   immune cells, with neutrophils, dendritic cells, and monocytes playing
   a central role ([183]6, [184]58). Early depletion studies in mice using
   monoclonal antibodies showed that neutrophils are important for host
   survival during acute infection ([185]59). Further highlighting their
   importance, depletion of neutrophils showed decreased levels of IFN-γ,
   IL-12, and TNF, indicative of a weaker type I immune response, and
   development of lesions in multiple organs including the lung, liver,
   and brain. Using our intestinal tissue MPS, we recapitulated events
   with human neutrophils including extravasation and migration toward the
   T. gondii–infected epithelium, reminiscent of the process of leukocyte
   trafficking to the site of infection previously observed in mouse
   models ([186]26). To reduce donor-to-donor variability from adult
   intestinal stem cells, we used organoid-derived stem cells from a
   single donor to generate tubular intestinal epithelium, but neutrophils
   from multiple blood donors were used to ensure consistency in
   donor-to-donor response to infected epithelium. Our results show that
   infection of the primary small intestinal epithelium by T. gondii
   up-regulates the expression of several genes in neutrophils that are
   involved in cellular processes including positive regulation of
   cytokine production, response to cytokines, regulation of leukocyte
   mediated immunity, defense response, and positive regulation of immune
   effector functions.

   Although our pathway enrichment analysis was conducted on the basis of
   a small gene set, future studies could further explore these processes
   using Kyoto Encyclopedia of Genes and Genomes pathway analysis. Recent
   ex vivo work exploring immune evasion strategies of T. gondii suggests
   that the parasite may prolong the life span of neutrophils by delaying
   apoptosis in infected cells ([187]60). The study examined
   transcriptional responses of T. gondii–infected human neutrophils and
   showed up-regulation of pathways involved in mediating immune responses
   and cellular response to stress that are consistent with some of the
   pathways implicated in our analysis. However, down-regulation of
   pathways associated with myeloid leukocyte activation, cytokine
   secretion, and activation of apoptotic caspases was also observed in
   neutrophils after they have been actively infected by T. gondii. As our
   study primarily examined the responses of neutrophils to T.
   gondii–infected tissue, future studies exploring responses of T.
   gondii–infected neutrophils in our platform would provide insight into
   the mechanisms of human innate immune evasion strategies used by T.
   gondii.

   Our study also highlighted changes in the functional behavior of
   neutrophils that correlate with the biological process involved as
   determined by gene expression analysis. In line with the
   transcriptional changes observed, dynamic visualization of neutrophil
   behavior at the epithelial and vascular interface showed increased
   migration speed and end-to-end displacement from the endothelium in
   response to the infection of the epithelium. Notably, some instances of
   direct interaction between neutrophils and excysted T. gondii within
   and near the epithelium could also be observed, demonstrating the
   capability of our model in capturing parasite invasion, replication,
   and initial response of neutrophils in the same experiment.

   Host-directed therapies that target signaling pathways for parasite
   clearance bypass many problems encountered by antiparasitic drugs
   including poor patient compliance and the emergence of drug-resistance
   parasites ([188]61). By targeting host pathways that are redundant in
   the host but are critical for parasite survival, there is a reduced
   chance of developing treatment resistance due to the slow rates of
   mutations in molecules and processes in the host relative to parasites.
   In this work, we examined the effects of a host-targeted pathway on the
   effector functions of neutrophils. Once at the site of infection, local
   production of inflammatory mediators regulates intercellular
   communication that mobilizes several defense mechanisms in
   tissue-resident cells and immune cells. IFN-γ has a multitude of
   immunomodulatory functions and is considered one of the most potent
   pleiotropic cytokines ([189]38, [190]62, [191]63). IFN-γ interactions
   with T cells, NK cells, and activated macrophages have been widely
   researched; however, investigations into the role of IFN-γ during the
   initial responses of neutrophils and other innate immune cells to
   infection by parasites have been limited. We demonstrated the influence
   of IFN-γ on traditional neutrophil functions including parasitization,
   NET formation, and apoptosis of neutrophils. IFN-γ treatment favorably
   affected neutrophil functions with increased parasitization, NET
   formation, and overall metabolic activity in neutrophils. In contrast,
   neutrophil apoptosis, the process of programmed cell death, decreased,
   which limits exposure of destructive neutrophil productions to
   surrounding tissue. IFN-γ blockage reversed these trends, further
   highlighting the importance of IFN-γ on traditional neutrophil
   functions of antimicrobial activities and showing the utility of our
   model for investigation of human immune responses. We also demonstrate
   the ability of our model to investigate parasite dissemination
   facilitated by neutrophils, both within the lumen of the intestinal
   epithelium and beyond by crossing the epithelial wall into the
   surrounding tissue. These observations are also consistent with data in
   mice that show the involvement of neutrophils in the spread of T.
   gondii infection with the intestinal tissue ([192]26). Together, we
   demonstrate the sensitivity of our system in capturing the changes in
   various effector functions and antimicrobial activity of neutrophils in
   response to inflammatory stimulus and blockage.

   Our secreted factor analysis revealed that our intestinal tissue MPS
   produces a wide array of proinflammatory cytokines/chemokines (MCP-1,
   MIP-1α, IL-1α, IL-1β, GM-CSF, IL-6, and IL-8) that enhance and activate
   defense mechanisms in innate immune cells during T. gondii infection.
   In mice, these factors are known to have immunoregulatory roles during
   acute infection. For instance, the production of MCP-1 and MIP-1α in
   our system is consistent with in vivo data in mice showing higher
   expression of these chemokines in T. gondii–infected intestinal
   epithelial cells ([193]64). Also, mice deficient in MCP-1 and its
   chemokine receptor CCR2 fail to generate adequate immune response
   during acute T. gondii infection ([194]65), highlighting its critical
   role in regulating innate immunity. Elevated levels of IL-1α and IL-1b
   have been previously observed in the serum of T. gondii–infected mice
   (1 week after infection) and are known inflammatory
   mediators/regulators of host tissue homeostasis during acute infection
   ([195]66, [196]67). Moreover, chemokines like GM-CSF, IL-6, and IL-8,
   aside from their involvement in inflammation, play important roles in
   the recruitment and enhanced survival of phagocytic cells like
   neutrophils during acute innate response to T. gondii infection
   ([197]68–[198]70). Together, these cytokines make up critical
   components of the inflammatory milieu that contribute to coordinated
   immune defenses like immune cell trafficking, activation, and effector
   functions against parasites during acute infection. In accordance with
   the increased levels of these cytokines in our infected systems, we
   observed increased neutrophil trafficking and transcriptional changes
   associated with responses to cytokines and positive regulation of
   immune effector functions in neutrophils. While we have not explored
   the direct influence of these cytokines on the functional behavior of
   neutrophils, future studies using our model could explore spatial and
   temporal relationships between neutrophils and infection-derived
   factors that enhance their overall effector functions against
   parasites.

   During innate immune responses, T. gondii infection–induced production
   of inflammatory cytokines drives NK cells to produce a host of factors
   such as IFN-γ and TNF-α that amplify effector functions and
   inflammatory responses driven by other innate immune cells. NK cell
   production of immunosuppressive cytokines that counterbalance
   inflammatory responses during disseminated pathogenic infection with T.
   gondii has also been found ([199]71), shedding light on the role of NK
   cells in facilitating immunoregulation. While the mechanisms have not
   been clearly defined, the immunoregulatory roles of NK cells are likely
   important for preventing inflammation-dependent pathology during
   parasite infection. Further exploration of secreted factor analysis
   shows that incorporating NK cells into our systems up-regulates several
   immunoregulatory signaling factors critical for amplifying the
   production of inflammatory and enhancing effector functions in other
   innate immune cells. In the presence of NK cells, several cytokines
   (IFN-α, TNF-α, IFN-γ, IL-17A, IP-10, IL-4, IL-13, IL-12p70, and IL-10)
   were uniquely elevated in our infected systems. IFN- α, a member of the
   type I IFN family, has recently been shown to enhanced cytokine
   secretion and cytotoxic potential in NK cells ([200]72), and TNF-α
   together with IFN-γ are major proinflammatory cytokines produced by NK
   cells as means of host protection against T. gondii ([201]28).
   Moreover, these two cytokines are recognized as essential immune
   effectors against intracellular pathogens. IL-17 has been implicated in
   resistance to T. gondii where IL-17R^−/− mice showed defects in
   neutrophil recruitment and increased parasite burden ([202]73). As NK
   cells are often considered to function at the interface of the innate
   and adaptive immune response, NK cell production of chemokine, IP-10,
   can be critical for generating an adaptive immune response and
   influencing the recruitment of effector T cells to the T. gondii
   infection site ([203]74). Cytokines typically involved in the
   resolution of cell-mediated inflammation [IL-4, IL-13, and IL-10
   ([204]75)] were also elevated in our infected systems with NK cells.
   Conventionally, IL-4 and IL-13 are thought to counter NK cell effector
   functions by limiting the production of IL-12. However, recent work
   shows that stimulation with IL-4 or IL-13 enhances TNF-α and IFN-γ
   production in NK cells in the presence of IL-12, underscoring the
   immunoregulatory roles of these cytokines ([205]76). Elevated IL-10
   levels observed are also in line with previous reports showing that NK
   cells produce IL-10 during acute T. gondii infection, which interferes
   with the activation of the adaptive immune responses ([206]77), further
   highlighting the immunoregulatory roles of NK cells. While the source
   of these cytokines is unclear in our analysis, secretion of these
   factors is markedly higher in our infected intestinal tissue MPS with
   NK cells, indicating a direct or indirect contribution from NK cells.
   The local effects of this inflammatory environment on the effector
   functions of neutrophils and other innate immune cells may be tested
   with future experiments within our platform and using tools/methods for
   examining complex interactions in vitro.

   Collectively, our secreted factor analysis demonstrated a wide array of
   immunoregulatory cytokines, chemokines, and adhesion molecules
   generated within our primary intestinal tissue MPS during T. gondii
   infection, which are essential for generating a robust immune response.
   These molecules are critically involved in innate immunity against
   intracellular pathogens in the following ways: (i) promoting leukocyte
   adhesion to the endothelium and transendothelial migration; (ii)
   trafficking, recruitment, and survival of phagocytic cells; and (iii)
   enhanced activation, cytotoxic activity, and effector functions of
   immune cells involved in the innate and adaptive immune responses.
   These data confirm the presence of an immunomodulatory cytokine milieu
   that can support mounting a strong immune response to parasitic
   infections.

   However, this study has several limitations. First, while we have
   examined some aspects of neutrophil and NK cell responses in our model,
   other biological factors such as the microbiome, tissue oxygen levels,
   and other immune cell components (i.e., monocytes, macrophages, and
   dendritic cells) could influence mechanisms of parasite interaction
   with the epithelium and innate immune cell responses observed in our
   models. In future studies, these components may be integrated in our
   model to better recapitulate the in vivo microenvironment and to
   examine their influence on parasite immunity. Another limitation of our
   study is that our transcriptomic and cytokine profiling analysis
   examined a limited number of genes and factors that only represents a
   fraction of the total number of alterations involved in establishing
   protective immunity and generating a robust immune response. RNA
   sequencing analysis and examination of additional immunomodulatory
   factors can help provide a more in-depth molecular analysis of the
   biological processes activated in neutrophils and NK cells during T.
   gondii infection in our model. Last, human in vivo data on innate
   immune responses to T. gondii infection, in particular during the early
   stages of the infection that we are modeling, remains limited, making
   it challenging to compare our observations against in vivo data. The
   neutrophil and NK cell responses observed in our model will need to be
   validated using humanized mouse models or clinical studies.

   The primary objectives of this work were to create an MPS that
   recapitulates complex interactions between key components of the human
   intestinal tissue and to demonstrate its usefulness in studying innate
   immune responses to parasitic infection. Recreating the initial stages
   of parasitic infection by T. gondii and the accompanying innate immune
   responses of the intestinal tissue highlights the potential of our
   microphysiological model to fill the gaps in studying human tissue
   responses through mechanistic correlations between various molecular
   and functional cell data to both animal and human clinical data. Our
   readily accessible approach to generating 3D geometries of tubular
   organs within biomimetic ECM scaffolds could answer questions that have
   so far been difficult to address. Combined with the ability to
   dynamically visualize and explore immune responses to host-microbe
   interactions at a high spatiotemporal resolution, our model may have
   substantial potential for discovery of cell-mediated immune responses
   and development of host-directed therapies.

MATERIALS AND METHODS

Intestinal tissue MPS fabrication and assembly

   The intestinal tissue MPS was fabricated using a micromolding technique
   to generate tubular structures with an ECM scaffold. This approach
   enables fabrication of one or more lumen structures with variable size,
   configuration, and lumen spacing controlled by micromold design in soft
   lithography. Here, we use a two-lumen setup, each lumen equipped with a
   separate inlet and outlet port, within a single chamber with
   perpendicularly oriented side ports. Two stacked PDMS layers, with
   microscale features patterned into them, formed the culture chambers in
   which scaffolding ECM gel can be loaded, while removable PDMS rods
   formed the hollow lumen structures surrounded by the ECM gel. The
   master molds for the PDMS layers were made using SU-8 100 (Microchem),
   which were spin-coated onto wafers, soft-baked (i.e., heat at 65°C for
   30 to 40 min and then at 95°C for 90 to 120 min depending on layer
   thickness), exposed to ultraviolet (UV) through a mask of desired
   patterns, and post-baked at 95°C for 20 to 30 min. This procedure was
   repeated for additional layers before development in propylene glycol
   monomethyl ether acetate (MilliporeSigma). After developing, PDMS
   (Sylgard 184 Silicon Elastomer Kit, Dow Corning Corporation) was
   applied to the masters at a ratio of 10:1 base to curing agent and
   allowed to polymerize for 4 hours at 80°C. The rods were drawn from
   needles with gauge size, 23G (340-μm inner diameter). Before device
   assembly, the PDMS layers and rods were soaked in 100% ethanol for
   several days to extract any uncured PDMS oligomers. Following PDMS
   extraction, the rods were placed in between two layers, across the body
   of the chamber (3 mm in length) in ledge features stemming from the
   smaller inlet and larger outlet ports to hold the rods in the middle of
   the chamber. The side ports (4 mm apart) of the chamber were used to
   fill the chamber with ECM gel, and the height of the chamber was about
   1.25 mm. Once assembled, the PDMS layers were oxygen-plasma–bonded onto
   a glass-bottom MatTek dish using a Diener Electronic Femto Plasma
   Surface System.

ECM gel preparation and loading

   The bonded devices were UV-sterilized for 20 min and moved to the
   biosafety cabinet before ECM gel loading. To promote matrix adhesion to
   PDMS, the device chambers were treated with 1% polyethylenimine
   (MilliporeSigma) in deionized (DI) water solution for 10 min, followed
   by a 30-min treatment of 0.1% glutaraldehyde (MilliporeSigma) in DI
   water solution. Following surface treatment, devices were flushed with
   DI water solution five times to remove excess glutaraldehyde. A high
   concentration rat tail collagen I (Col-I) (Corning) neutralized with
   0.5 N sodium hydroxide (Thermo Fisher Scientific) was mixed with 7.5 pH
   5× phosphate-buffered saline (PBS), complete growth medium or organoid
   expansion medium, human placental Col-IV, and Matrigel to achieve a
   final ECM solution containing Col-I (4 mg/ml), 15% Matrigel, and Col-IV
   (50 μg/ml). The pH of the ECM mix was adjusted to pH 7.2 before loading
   the mix into the gel chamber of the device. The devices were first kept
   at room temperature for 20 min and then moved to an incubator at 37°C
   for at least 1 hour before cell loading. To prevent dehydration during
   polymerization, PBS was added to the MatTek dish surrounding the
   devices. PDMS rods were then removed, leaving behind hollow tubular
   structures within the ECM gel that can be lined with cells.

Crypt isolation and gut organoid culture

   Small intestinal crypts were isolated using a previously established
   protocol ([207]78) from jejunal tissue resection samples removed from
   nondiseased tissue of deidentified individuals. Jejunal tissue
   resections were performed at University of Wisconsin-Madison upon the
   donors’ informed consent, and methods were carried out in accordance
   with Institutional Review Board (IRB) (protocol no. 2016-0934). Small
   intestinal organoids were established from isolated crypts by
   resuspending in Matrigel (Corning; growth factor–reduced, phenol
   red–free formulation) and culturing in 24-well plates (Polystyrene,
   Nunc, nontreated multidishes, Thermo Fisher Scientific) at 37°C in the
   organoid expansion media. Expansion medium consisted of a mixture of
   base medium [BM; 45% (v/v)], L-WRN conditioned medium [CM; 45% (v/v)],
   supplement mix, 1× Primocin (InvivoGen), and 10% heat-inactivated fetal
   bovine serum (FBS; MilliporeSigma). BM was prepared from Advanced
   Dulbecco’s modified Eagle’s medium (DMEM)/F12 (Thermo Fisher
   Scientific) supplemented with GlutaMAX (2 mM; Thermo Fisher
   Scientific), Hepes (10 mM; Thermo Fisher Scientific),
   penicillin/streptomycin (Pen/Strep) (Thermo Fisher Scientific), B-27
   Supplement (Thermo Fisher Scientific), N-2 supplement (Thermo Fisher
   Scientific), and 1% cell-culture grade bovine serum albumin (BSA).
   L-WRN CM was prepared using L-WRN cells from American Type Culture
   Collection (ATCC; catalog no. CRL-3276) and a previously published
   protocol ([208]79). The supplement mix consisted of epidermal growth
   factor (50 ng ml^−1; PeproTech), N-2 supplement (Thermo Fisher
   Scientific), human [Leu^15]-gastrin I (10 nM; MilliporeSigma), N-acetyl
   cysteine (500 μM; MilliporeSigma), nicotinamide (10 mM; Thermo Fisher
   Scientific), A 83-01 (500 nM; Tocris), SB202190 (10 μM; Selleckchem),
   prostaglandin E2 (10 nM; Tocris), Y-27632 (10 μM; Seleckchem), and
   CHIR99021 (5 μM; Tocris). Jejunal organoids were split every 7 to 9
   days as previously described ([209]31). On the basis of this
   established protocol, Matrigel plugs containing organoids were digested
   in 0.5 μM EDTA (Invitrogen) and collected. After centrifugation at room
   temperature for 3 min at 300g, supernatant was removed followed by
   incubation in trypsin (Sigma-Aldrich) for 2 min at 37°C. Trypsin was
   neutralized using Advanced DMEM/F12 supplemented with 20% FBS and 1%
   Pen/Strep followed by mechanical dissociation of organoids by vigorous
   pipetting (about 30 times) to make small organoid fragments. After
   centrifugation at room temperature for 3 min at 300g, pellets were
   reconstituted in 50:50 mix of fresh Matrigel and expansion medium at
   1:3 to 1:4 ratio and cast in 25-μl droplets in a 24-well plate.
   Matrigel droplets containing organoid fragments were then polymerized
   for 15 min at 37°C before adding 500 μl of expansion media to each
   well. Expansion media were changed every 2 to 3 days, and organoids
   were used between passages 5 and 25. Y-27632 was only used for the
   first 48 hours after single-cell dissociation to prevent
   detachment-induced cell apoptosis.

Caco-2 epithelial tube formation

   Caco-2 cell line was acquired from the ATCC and maintained in Eagle’s
   minimum essential medium (EMEM; MilliporeSigma) supplemented with 20%
   FBS and 1% Pen/Strep. To generate tubular epithelium, Caco-2 cells were
   detached using a trypsin/EDTA solution and resuspended at 17 million
   cells/ml of supplemented EMEM. After removal of the PDMS rod, 3 μl of
   cell suspension was introduced into the hollow tubes through the inlet
   ports, and cells were allowed to fill the lumen structures by passive
   pumping. The microdevice was rotated every 30 min over a 2-hour period
   to line the hollow tube with epithelial cells. Nonadherent cells were
   washed off with culture medium, and excess medium was added to the
   inlet/outlet ports. After completion of tube formation (~24 to 48
   hours), the epithelial tubes were cultured in supplemented EMEM with
   reduced FBS to 10% for 5 to 7 days before coculture with endothelial
   vessels.

Primary intestinal epithelial tube formation

   Jejunal organoids were collected at days 7 to 9 after passaging and
   dissociated into small organoid fragments as described above. Pellets
   of organoid fragments were resuspended in expansion medium at a density
   of 5 million cells/ml. As previously described, 3 μl of organoid cell
   suspension was introduced into the hollow tubes through the inlet
   ports. Cells were allowed to settle on the bottom half of the tubular
   structure for 24 hours at 37°C in 5% CO[2] after expansion medium was
   added to the gel ports and the inlet/outlet ports during culture. After
   24 hours, nonadherent cells were gently washed with expansion medium.
   Fresh organoid fragments were prepared and seeded again into the tube.
   Expansion medium was added to the ports and the microdevice cultured
   for an additional 24 hours while flipped upside down to facilitate
   adhesion of cells to the top half of the hollow tubular structure.
   After completion of tube formation (~72 hours), the primary intestinal
   epithelium was cultured in DM for 5 to 7 days. For differentiation, the
   culture medium was replaced with BM [80% (v/v)], L-WRN CM [10% (v/v)],
   supplement mix (without Y-27632 and reduced CHIR 99021 to 500 nM), 1×
   Primocin (InvivoGen), and 10% heat-inactivated FBS (MilliporeSigma). DM
   was changed twice daily. After 5 to 7 days of differentiation, HUVECs
   (Lonza) were added to the adjacent hollow tube to generate cocultures
   of epithelial and endothelial tubes.

Coculture of epithelial tubes with endothelial tubes

   HUVECs were maintained in EGM-2 MV BulletKit medium (Lonza) and used
   until passage 8. To generate coculture of primary intestinal tubes with
   endothelial vessels, HUVECs were detached using a trypsin/EDTA solution
   and resuspended at 15 million cells/ml of modified DM. Modified DM was
   made by replacing BM in the DM formulation with EGM-2 MV BulletKit
   medium. Three microliters of endothelial cell suspension was introduced
   into the cell-free hollow tube next to the epithelial tube, and device
   was rotated every 15 min over a 2-hour period as previously described.
   Modified DM was added to the gel ports and epithelial inlet/outlet
   ports during this process to nourish the epithelial cells. After lining
   the tube with endothelial cells, nonadherent cells were gently washed
   with modified DM. The epithelial and endothelial tube cocultures were
   maintained in modified DM for the remainder of the experiments unless
   otherwise stated. To support coculture of HUVEC vessels with Caco-2
   epithelial tubes, the supplemented EMEM with reduced FBS was mixed with
   EGM-2 MV at 50:50% (v/v) dilution and used as culture medium for the
   remainder of the experiments.

Coculture of epithelial and endothelial tubes with NK cells

   NK-92 cell line was acquired from the ATCC and maintained in X-VIVO 10
   (Lonza) supplemented with 20% FBS and 0.02 mM folic acid
   (MilliporeSigma) dissolved in 1 N NaOH, 0.2 mM myo-inositol
   (MilliporeSigma), and IL-2 (100 U/ml; PeproTech). NK cells were
   collected and after centrifugation at room temperature for 3 min at
   300g, pellets were resuspended in modified DM supplemented with 0.02 mM
   folic acid, 0.2 mM myo-inositol, and IL-2 (100 U/ml). Resuspended cells
   were introduced into the endothelial vessel and coculture for 24 hours
   in excess medium.

Neutrophil isolation

   All blood samples were drawn according to IRB-approved protocols per
   the Declaration of Helsinki at University of Wisconsin-Madison.
   Neutrophils were purified from whole blood using the MACSxpress
   Neutrophil Isolation Kit per the manufacturer’s instructions (Miltenyi
   Biotec), and residual red blood cells were lysed using the MACSxpress
   Erythrocyte Depletion Kit (Miltenyi Biotec). Blood was drawn from a
   total of three nondiseased donors with informed consent obtained at the
   time of the blood draw according to requirements of the IRB. Before
   loading, the purified neutrophils were stained with calcein
   acetoxymethyl ester at 10 nM (Thermo Fisher Scientific) according to
   the manufacturer’s instructions. Neutrophils were resuspended in 50:50%
   (v/v) EMEM and EGM-2MV culture medium for coculture with Caco-2
   epithelium or resuspended in modified DM for coculture with primary
   intestinal epithelium.

Parasite cell culture and infection

   T. gondii ME49 tachyzoites were propagated in human foreskin fibroblast
   monolayers grown in DMEM containing 10% FBS and 1% Pen/Strep.
   Tachyzoites were harvested and pelleted by centrifugation at 2200g for
   10 min and resuspended in prewarmed modified DM or 50:50% (v/v) EMEM
   and EGM-2MV culture medium before injection into the lumen of the
   epithelial tubes for infect studies.

Immunofluorescence staining

   For immunostaining, cells were fixed with 4% (v/v) paraformaldehyde
   (Alfa Aesar) for 20 min and permeabilized with 0.2% (v/v) Triton X-100
   (MP Biomedicals) for 10 min with three 1× PBS wash steps between each
   solution. To reduce nonspecific background fluorescence from collagen,
   cells were incubated in 0.1 M glycine (Thermo Fisher Scientific,
   Pittsburgh, PA) for 30 min and washed with 1× PBS three times.
   Subsequently, cells were blocked with buffer solution [3% (w/v) BSA and
   0.1% (v/v) Tween 20 (Thermo Fisher Scientific)] overnight at 4°C.
   Primary antibodies (table S2) diluted in buffer solution were added to
   the microdevices and incubated at 4°C for 2 days and washed with 1× PBS
   three times. Secondary antibodies (table S2) were added to the buffer
   solution and incubated for 1 day at 4°C. For cytoskeletal actin and
   nuclear staining, Alexa Fluor-conjugated phalloidin (Thermo Fisher
   Scientific) and Hoechst 33342 (Thermo Fisher Scientific) at 1:100 were
   also added to the secondary antibody buffer solution. Lumens were then
   rinsed with 1× PBS three times over a 2-day period.

Image acquisition

   Bright-field and fluorescent images were captured on a Nikon Ti Eclipse
   microscope with a top stage incubator equipped with temperature and
   CO[2] control (set at 37°C and 5%, respectively). Neutrophil kinetic
   parameters including end-to-end displacement and migration speed were
   analyzed with Fiji ([210]https://imagej.net/Fiji) using the track-mate
   module. Confocal images were acquired at University of
   Wisconsin-Madison Optical Imaging Core using a Leica SP8 microscope.

Optical metabolic imaging

   A custom-built inverted multiphoton microscope (Bruker Fluorescence
   Microscopy, Middleton, WI) was used to acquire fluorescence intensity
   and lifetime images. The equipment consists of an ultrafast laser
   (Coherent, Chameleon Ultra II), an inverted microscope (Nikon, Eclipse
   Ti), and a 40× water immersion (1.15 numerical aperture, Nikon)
   objective. Fluorescence intensity and lifetime data were collected
   sequentially with a GaAsP PMT (Hamamatsu, H7422PA-40). FAD fluorescence
   was isolated using an emission band-pass filter of 550/100 nm and an
   excitation wavelength of 890 nm. NAD(P)H fluorescence was isolated
   using an emission band-pass filter of 440/80 nm and an excitation
   wavelength of 750 nm. The optical redox ratio was determined from the
   NAD(P)H and FAD lifetime data by integrating the photons detected at
   each pixel in the image to calculate the total intensity. For each
   pixel, the intensity of NAD(P)H was then divided by the intensity of
   FAD. Using Cell Profiler, an automated cell segmentation pipeline was
   created. This system identified pixels belonging to nuclear regions by
   using a customized threshold code. Cells were recognized by propagating
   out from the nuclei within the image. To refine the propagation and to
   prevent it from continuing into background pixels, an Otsu Global
   threshold was used. The cell cytoplasm was defined as the cell borders
   minus the nucleus. Values for NAD(P)H intensity, FAD intensity, and the
   optical redox ratio [NAD(P)H/FAD intensity] were averaged for all
   pixels within each cell cytoplasm. At least 100 cells per sample were
   analyzed, and every experiment was repeated at least three times.

Cell retrieval from microdevice

   To quantify gene expression related to proliferation, differentiation
   and a function of the intestinal epithelium primary intestinal
   epithelial cells were selectively retrieved from the intestinal tissue
   MPS consisting of epithelial and endothelial tubes. The upper half of
   the microdevice was removed to expose the collagen hydrogel. The
   hydrogel was then transferred to an Eppendorf tube containing 300 μl of
   type I collagenase (6 mg/ml). The sample was incubated on ice for 2 min
   to degrade the hydrogel and release the cells. Two microliters of
   biotinylated anti–epithelial cell adhesion molecule (EpCAM) (Thermo
   Fisher Scientific) was added, and the sample was incubated at 4°C for
   15 min. Ten microliters of SeraMAGS beads coupled to streptavidin was
   added, and the sample was incubated for another 10 min at 4°C. The
   SeraMAGS beads, with the epithelial cells (EpCAM-positive), were
   isolated using a magnet and lysed for PCR analysis.

   To quantify gene expression related to protective immunity, neutrophils
   were selectively retrieved from the intestinal tissue MPS 6 hours after
   introduction into the endothelial vessel. Nonadherent neutrophils were
   first collected from the endothelial vessels through the
   pipet-accessible ports (fig. S10). To isolate neutrophils that have
   migrated into the ECM gel, a 2-mm-diameter biopsy punch (Thermo Fisher
   Scientific) was used to cut out the hydrogel next to the endothelial
   vessel. The hydrogel punches were transferred to an Eppendorf tube and
   digested using collagenase as described above. To separate
   CD31-positive endothelial cells, Dynabeads CD31 (Thermo Fisher
   Scientific) endothelial cell depletion was conducted using the
   manufacturer’s guidelines, leaving behind neutrophils in suspension.
   These neutrophils were mixed with neutrophils collected from the
   endothelial vessel and lysed for PCR analysis.

Reverse transcription quantitative polymerase chain reaction

   To study how primary intestinal epithelial cells and neutrophils
   adapted to the microenvironment within the intestinal tissue MPS, the
   expression of multiple genes related to different pathways was analyzed
   by RT-qPCR. Briefly, mRNA was isolated from cells using the Dynabeads
   mRNA DIRECT Purification Kit (Invitrogen). Isolated mRNA was quantified
   using a Qubit fluorometer (Thermo Fisher Scientific) and the Qubit RNA
   BR Assay Kit (Thermo Fisher Scientific). Complementary DNA (cDNA) was
   generated using the High Capacity RNA-to-cDNA Kit (Applied Biosystem)
   and preamplified with SsoAdvanced PreAmp Supermix (Bio-Rad) using
   primers from Integrated DNA Technologies or Thermo Fisher Scientific
   (tables S3 and S4). cDNA was analyzed by RT-qPCR using iTaq Universal
   SYBR Green Supermix (Bio-Rad) or Roche LightCycler master mix according
   to the manufacturer’s protocols in Roche’s LightCycler 480 II. Gene
   expression was normalized using the ΔΔ^Ct method. To quantify genomic
   DNA of T. gondii, infected epithelial tubes were digested as described
   above, and genomic DNA was extracted using TRIzol according to the
   manufacturer’s instructions. DNA was purified by phenol/chloroform
   extraction followed by ethanol precipitation. Genomic DNA was used as
   the template for preamplification and RT-PCR, as described above, using
   T. gondii primers (table S2).

Multiplex cytokine/chemokine assays and analysis

   To measure NK cell–mediated cytokines secretion, media were collected
   from intestinal tissue MPS with T. gondii–infected epithelium in the
   presence and absence of NK cells. Media were collected after 24 hours
   in culture and cytokine/chemokine concentrations analyzed using the
   Inflammation 20-Plex Human ProcartaPlex panel (EPX200-12185-901, Thermo
   Fisher Scientific) following the manufacturer’s guidelines. Data were
   collected on a MAGPIX Luminex xMAP system (Luminex Corporation) using
   Luminex xPONENT software. Concentration of each analyte was determined
   from a standard curve, generated by fitting a five-parameter logistic
   regression of median fluorescence on known concentrations of each
   analyte.

Statistical analysis

   Data were analyzed (Prism 9.0; GraphPad Software). The normal
   distribution assumption for statistical tests was confirmed by the
   Shapiro-Wilk test. Statistical significance was assessed using
   Student’s t tests when comparing two conditions/groups, and when
   comparing more than two groups, significance was assessed using one-way
   analysis of variance (ANOVA) corrected using the Tukey’s test. For
   nonparametric comparisons, a Mann-Whitney U test or a Kruskal-Wallis
   test was performed.

Acknowledgments