A Novel In Vitro Human Granuloma Model of Sarcoidosis and Latent Tuberculosis Infection

Abstract

Many aspects of pathogenic granuloma formation are poorly understood, requiring new relevant laboratory models that represent the complexity (genetics and diversity) of human disease. To address this need, we developed an in vitro model of granuloma formation using human peripheral blood mononuclear cells (PBMCs) derived from patients with active sarcoidosis, latent tuberculosis (TB) infection (LTBI), or normal healthy control subjects. PBMCs were incubated for 7 days with uncoated polystyrene beads or beads coated with purified protein derivative (PPD) or human serum albumin. In response to PPD-coated beads, PBMCs from donors with sarcoidosis and LTBI formed robust multicellular aggregates resembling granulomas, displaying a typical T-helper cell type 1 immune response, as assessed by cytokine analyses. In contrast, minimal PBMC aggregation occurred when control PBMCs were incubated with PPD-coated beads, whereas the response to uncoated beads was negligible in all groups. Sarcoidosis PBMCs responded to human serum albumin–coated beads with modest cellular aggregation and inflammatory cytokine release. Whereas the granuloma-like aggregates formed in response to PPD-coated beads were similar for sarcoidosis and LTBI, molecular profiles differed significantly. mRNA expression patterns revealed distinct pathways engaged in early granuloma formation in sarcoidosis and LTBI, and they resemble molecular patterns reported in diseased human tissues. This novel in vitro human granuloma model is proposed as a tool to investigate mechanisms of early granuloma formation and for preclinical drug discovery research of human granulomatous disorders.

Clinical trial registered with www.clinicaltrials.gov (NCT01857401).

Clinical Relevance

We have developed an “in vitro human model of granuloma formation,” which is shown to recapitulate molecular mechanisms that have been established in corresponding diseased human tissues. The model is unique in that it allows us to explore the mechanisms of early granuloma formation, which is critical for determining disease phenotype and is not represented in human tissue samples. The in vitro model is readily adopted in most laboratories, requires minimal risk to the study subjects, and represents the complex genetics of each disease, features which have not been accurately modeled in animals. The model can be used for biomarker discovery, exploration of disease mechanisms, and for preclinical testing of novel therapeutics.

Progress in the field of granulomatous disease research is currently hindered by the lack of models that accurately replicate the complex genetics and multicellular immune responses characterizing the corresponding human conditions. The resultant reliance on human tissues for research relating to granulomatous infections or granulomatous disorders of noninfectious etiology (e.g., sarcoidosis) greatly increases the research cost and limits dynamic interrogation, efficiency, and reproducibility of the research. In 2004, an NHLBI working group on future directions in sarcoidosis research concluded that relevant laboratory models of sarcoidosis are needed to “…elucidate the underlying mechanisms, including predisposing factors, and defective immune responses” characteristic of the disease (1). Likewise, new laboratory models of tuberculosis (TB) are needed to expedite preclinical testing of novel therapies and better understand host–pathogen interactions (2, 3). However, no widely accepted animal or in vitro models have been developed to date.

Historically, rodent models were favored for studying human diseases; however, by and large, these models have not been suitable for investigating human granulomatous disorders. Rodent models, including genetically modified mice, are limited by an inability to accurately replicate polygenic immune-mediated human granulomatous diseases. Significant challenges also arise with humanized mouse models, wherein immune cell progenitors are introduced from human donors to replicate the complex genetic traits of diseased humans, because the engraftment of the human immune system is often incomplete or is associated with abnormal inflammatory cell activation (4). As a result of these and other practical limitations, prior attempts to model human granulomatous disorders, such as sarcoidosis or latent TB infection (LTBI), in rodents have not gained wide acceptance (5, 6). For example, attempts to sensitize mice to immunogenic TB antigens lead to self-limited granuloma formation, which equates to a normal Th1 immune response (5). By contrast, patients with sarcoidosis develop sustained and, in some cases, progressive granulomatous inflammation in response to TB and presumably other environmental antigens (7). Moreover, granuloma formation and related disease manifestations in murine TB models differs significantly from the human condition (8), and LTBI, common in humans, is not observed in rodents (6). Thus, animal models are largely not reliable surrogates for human granulomatous diseases.

Recent studies show that peripheral blood mononuclear cells (PBMCs) derived from patients with LTBI or active TB form granulomas in vitro in response to TB antigen challenge (9, 10). This should come as no surprise, given the evidence that granulomas are assembled and maintained through the recruitment of circulating immune cells (11, 12). These results indicate that PBMCs derived from patients with sarcoidosis, LTBI, and other granulomatous disorders manifest the complex and, as yet, undefined genetic features that predispose to these disorders, and could serve as a tractable in vitro model of granulomatous diseases.

Sarcoidosis and LTBI are examples of human disorders representing the opposite ends of a spectrum of granulomatous responses to common antigens. Specifically, patients with sarcoidosis who are naive to TB infection often manifest with abnormally intense Th1 immune responses to TB antigens (13), whereas LTBI is characterized by an impaired Th1 immune response to TB antigens, leading to failure to fully resolve the infection (14). Thus, we sought to determine whether PBMCs from patients with sarcoidosis and LTBI form granulomas in response to tuberculin purified protein derivative (PPD), and, if so, whether the granulomas conform to established characteristics of tissue granulomas in corresponding human disease, the ultimate goal being to develop a novel in vitro research model for granulomatous disorders.

Materials and Methods

Please refer to the online supplement for additional methodological details.

Human Blood Samples

With Institutional Review Board (The Ohio State University, Columbus, Ohio; no. 2014H0380) approval and informed, written consent, blood samples were obtained from three study groups: (1) six patients with active pulmonary sarcoidosis being TB skin test and/or IFN-γ release assay (IGRA) negative using the Quantiferon test; (2) five subjects with LTBI having a positive PPD skin and/or Quantiferon test, no clinical symptoms, and absent chest radiographic evidence of active TB infection; and (3) five matching disease-free, TB skin test and/or Quantiferon-negative control subjects.

PPD- and Human Serum Albumin–Coated Bead Preparation

Fluoresbrite carboxylated microspheres (bright blue fluorescence, 1.0 μm) were either resuspended in PBS or covalently coupled to select proteins (i.e., PPD, human serum albumin [HSA]) via the carbodiimide method (see Figures E1 and E2 and further description in the online supplement). Prepared beads were diluted into RPMI media and added to isolated PBMCs at a multiplicity of infection of 50:1 relative to the number of monocytes/macrophages present.

PBMC Isolation from Human Blood and Culture Conditions

Human PBMCs were isolated from whole-blood samples employing Ficoll-Paque PLUS (GE Healthcare Bio-Sciences, Piscataway, NJ) as previously described (10) and detailed further in the online supplement. PBMCs (2 × 10⁶ cells/ml) were cultured for 7 days in RPMI 1,640 medium containing 10% human AB serum and each of the following separate treatments: (1) volume-matched vehicle control; (2) uncoated (PBS-washed) beads; (3) HSA-coated beads; or (4) PPD-coated beads.

Measuring In Vitro Granuloma-Like Cell Aggregates

The stage of granuloma formation (granuloma index) was determined semiquantitatively at Day 7 after treatment for each treatment in each group, as described in the online supplement and Figure E3. The scoring system is based on published literature and established models (9, 10, 15), where there is a progressive increase in the size and specific cell distribution of the granulomas.

Cytokine Analysis

Human PBMC–cleared supernatants, collected 7 days after treatment, were analyzed for their granulocyte-macrophage colony–stimulating factor (GM-CSF), IFN-γ, IL-2, IL-10, IL-12p40/IL-23, IL-12p70, IL-13, and TNF-α concentrations by ELISA, according to the manufacturer’s recommendations (IFN-γ, IL-10, IL-12p70, IL-13, TNF-α [eBioscience, Inc., San Diego, CA], GM-CSF, IL-2, and IL-12p40/IL-23 [BioLegend, San Diego, CA]).

Granuloma Gene Expression Analysis

Granuloma cells were harvested into TRIzol reagent (Ambion, Life Technologies, Carlsbad, CA). Total RNA was purified and DNase treated using the SpinSmart RNA purification kit (Denville Scientific, Inc., Holliston, MA), according to the manufacturer’s protocol. Ion Torrent AmpliSeq Transcriptome Human Gene Expression kit (Thermo Fisher Scientific, Inc., Pittsburgh, PA) was used to prepare libraries from 10 ng granuloma RNA. Quantified barcoded libraries were combined in equal amounts for emulsion PCR on the Ion OneTouch2 instrument (Thermo Fisher Scientific, Inc.). Templated libraries were loaded onto Ion PI Chips and sequenced on the Ion Proton Sequencer, using the Ion PI Sequencing 200 Kit v2 reagents (Thermo Fisher Scientific, Inc.). Transcript quantification was performed with the mm Hgent Suite (Thermo Fisher Scientific, Inc.) analysis plugin, “AmpliSeqRNA”.

AmpliSeq gene expression data were normalized and postalignment statistical analyses performed using DESeq2 (16), with custom analysis scripts written in R. Comparisons of gene expression and associated statistical analyses were made between different conditions of interest, as presented subsequently here, using the normalized read counts. Transcripts were considered to be significantly differentially expressed using a 10% false discovery rate (DESeq2-adjusted P ≤ 0.05) and a fold-change cut-off of 2 between the uncoated bead–treated and PPD-treated samples.

Results

Study Subjects

The study was approved by the local institutional review board and was listed with the National Institutes of Health (ClinicalTrials.gov identifier: NCT01857401). Inclusion criteria for subjects with sarcoidosis were based upon the accepted standards (biopsy-proven noncaseating granulomatous tissue inflammation in the absence of other likely causes) (17), and all patients had evidence of active (nonfibrotic) lung disease and negative screening for LTBI (negative PPD skin test and/or negative IGRA). LTBI was characterized by a positive PPD skin test and/or positive IGRA with no signs or symptoms of active disease (e.g., fevers, cough, night sweats). All study participants provided informed, written consent. The demographic features of the study subjects are summarized in Table 1. Of note, none of the study subjects were active smokers or were being actively treated with immune suppressants. Additional information pertaining to the disease characteristics of the patients with sarcoidosis is provided in Table 2.

Table 1.

Patient Demographics

Groups	Age	Race	Sex
	(Yr)	(W/B/O)	(M/F)
Control (n = 5)	28.0 ± 2.5	3/0/2	3/2
LTBI (n = 5)	28.6 ± 3.9	1/1/3	3/2
Sarcoidosis (n = 6)	49.8 ± 6.4	4/2/0	2/4

Definition of abbreviations: LTBI, latent tuberculosis infection; W/B/O, white/black/other.

Table 2.

Disease Characteristics Patients with Sarcoidosis

Patient*	Scadding Stages I–IV	Extrathoracic Organ Involvement	Lymphopenia <1,200/μl
SARC11	IV		Yes
SARC12	II		No
SARC13	I	Skin	No
SARC14	I	Stomach	Yes
SARC16	II		No
SARC17	III	Skin	Yes

^*As identified in Figures 3 and 4. Because of a very limited obtained blood sample, SARC15 failed to yield enough peripheral blood mononuclear cells to perform the experiments and was not included.

In Vitro Cell Aggregation Resembling Granulomas

Control subjects, subjects with sarcoidosis, and those with LTBI demonstrated minimal cellular aggregation in response to beads alone. By contrast, treatment with PPD-coated beads elicited the formation of multicellular (granuloma-like) complexes in PBMCs derived from patients with sarcoidosis and LTBI, whereas comparatively minimal cell aggregation was apparent under these conditions in normal healthy control subjects (Figures 1 and Figure E3). In response to HSA-coated beads, patients with sarcoidosis displayed an intermediate cellular aggregation response (Figure 1), with comparatively minimal responses in LTBI and control subjects (data not shown). Immunofluorescence (see the online supplement for methods and antibodies used) of granuloma-like structures 7 days after PPD-coated bead stimulation in both patients with LTBI and those with sarcoidosis demonstrated the predominant presence of both centrally clustered macrophages and peripherally situated lymphocytes (Figure 2; cellular components [with isotype control antibodies] and three-dimensional video views of the granuloma shown in Figures E4 and E5, respectively).

Figure 1.

In vitro granuloma-like cell aggregate formation induced by purified protein derivative (PPD)–stimulated peripheral blood mononuclear cells (PBMCs) of patients with sarcoidosis and latent tuberculosis (TB) infection (LTBI). (A) Representative light photomicrographs (×20) obtained 7 days after sarcoidosis PBMCs were incubated with PPD-coated (PPD) or uncoated (UNC) beads. (B) Representative light photomicrographs (×20 and ×40, respectively) obtained 7 days after the indicated treatment of PBMCs in the sarcoidosis, LTBI, and control groups. Uncoated beads elicited no significant cellular aggregation in any of the study groups, and human serum albumin (HSA)–coated beads caused no perceptible aggregation in any group, with the exception of the sarcoidosis group, in which an intermediate response was observed. PPD-coated beads induced prominent cellular aggregation in the sarcoidosis and LTBI groups but not in the control subjects.

Figure 2.

Composite immunofluorescence imaging of granuloma-like structures demonstrates the presence of both macrophages and lymphocytes. Representative photomicrographs of immunostained PBMCs in granuloma-like structures 7 days after bead treatment using confocal microscopy. Composite images of 4′,6-diamidino-2-phenylindole, CD11b, and CD3 are shown in blue, red, and green, respectively. (A) PPD-coated and (B) uncoated bead–stimulated PBMCs obtained from a PPD⁻, healthy control subject showing CD11b⁺ macrophages and CD3⁺ lymphocytes in the absence of granuloma formation. (C and D) CD11b⁺ macrophages and CD3⁺ lymphocytes in a granuloma-like structure of cells obtained from a subject with LTBI after PPD-coated and uncoated bead stimulation, respectively. (E and F) CD11b⁺ macrophages and CD3⁺ lymphocytes in a granuloma-like structure of cells obtained from a patient with sarcoidosis after PPD-coated and uncoated bead stimulation, respectively. Uncoated beads did not induce granuloma-like structures in both the patients with LTBI and those with sarcoidosis. In each panel, a matching differential interference contrast image is shown next to each composite image. All images were acquired at ×60 magnification.

Extracellular Cytokine and Chemokine Profiles

Release of Th1-related cytokines, chemokines, and growth factors (IFN-γ, IL-2, IL-12p40, IL-13, GM-CSF, and TNF-α) increased, as expected, in response to PPD-coated beads in all treatment groups (Figure 3 and Figure E6). However, release of IFN-γ was significantly lower in subjects with sarcoidosis compared with those with subjects with LTBI or control subjects, whereas TNF-α release was significantly higher in subjects with sarcoidosis compared with those with LTBI or control subjects. HSA-coated beads elicited an immune response corresponding to selective monocyte/macrophage activation (GM-CSF, TNF-α) in patients with sarcoidosis (Figure 3). This response to HSA-coated beads was not observed in either subjects with LTBI or control subjects.

Figure 3.

Extracellular cytokine release from PPD-stimulated PBMCs revealed distinct patterns in sarcoidosis and LTBI. Compared with treatment with uncoated or HSA-coated beads, incubation of PBMCs in the sarcoidosis (SARC), LTBI, and control (CTRL) groups with PPD-coated beads for 7 days was associated with a significant increase in the extracellular release of T helper type 1 cytokines (IFN-γ [B], IL-2 [D]) and other proinflammatory cytokines (TNF-α [A], granulocyte–macrophage colony-stimulating factor [GM-CSF] [C]). Treatment of sarcoidosis PBMCs with HSA-coated beads induced a significant increase in TNF-α and GM-CSF release compared with HSA-coated bead treatment of PBMCs from the LTBI or control groups (*P < 0.05, compared with the corresponding [within group] uncoated and no bead controls; ^†P < 0.05, relative to the matching [between groups] bead treatments).

Gene Expression Analysis

One sample each from the LTBI and control groups had inadequate RNA quality and were excluded from AmpliSeq analysis (control, n = 4; LTBI, n = 4; sarcoidosis, n = 6). In separate quality-control assays, AmpliSeq data were highly reproducible (technical replicates with r = ∼0.99 and biological replicates with r > ∼0.96). Significantly differentially expressed genes (>twofold change and adjusted P ≤ 0.05) were identified by comparing normalized gene expression levels in cells treated with PPD-coated beads (test treatment) versus cells treated with uncoated beads (PBS control treatment), taking into account the donor patient, to perform a paired analysis (Figure 4). A total of 1,206 genes were significantly differentially expressed in cells from healthy human control subjects, with 3,229 in the LTBI group. Strikingly, the molecular response to PPD-coated beads was attenuated in cells harvested from patients with sarcoidosis with only 352 differentially expressed genes.

Figure 4.

Clustering of global transcriptional responses of PBMCs after treatment with PPD-coated beads. (A) Principal component (PC) analysis of differential gene expression (AmpliSeq) performed on PBMCs derived from patients with SARC, LTBI, or normal healthy control subjects (CTRL) after incubation with PPD-coated beads for 7 days compared with paired treatment with uncoated (PBS-washed) beads. (B) Euclidean distance between samples was plotted to assess overall similarity between samples. This analysis revealed a hierarchical clustering of four subjects with SARC, whereas the remaining two SARC samples aligned more closely with CTRL (see text for further discussion).

Principal component analysis (PCA) using fold-change ratios in PBMCs (PPD versus PBS control treatment) revealed distinct patterns within the LTBI and sarcoidosis groups. In particular, PCA showed distinct clustering of four sarcoidosis samples from LTBI, whereas no distinct clustering was observed in normal, healthy control subjects (Figure 4A). Like PCA, Euclidean distance between samples was calculated to assess overall similarity between samples (Figure 4B). This analysis demonstrated a hierarchical clustering of four patients with sarcoidosis, whereas the remaining two sarcoidosis samples aligned more closely with controls. Of note, the two sarcoidosis samples that clustered with controls had lower cytokine production and developed less cell aggregation after PPD treatment compared with the other four subjects with sarcoidosis.

A more detailed comparison of gene expression patterns further differentiated sarcoidosis from LTBI. Table 3 provides a list of differentially expressed genes with the greatest fold change when comparing treatment with PPD-coated beads to correspondingly matched uncoated (PBS-washed) beads. The differentially expressed gene data used to generate the summary data in Table 3 are also represented graphically in the form of heat maps (Figure 5). This analysis indicated that PBMCs derived from the control and LTBI groups had a similar transcriptional response to PPD in some cases, whereas the sarcoidosis group had an attenuated response. For example, many transcripts exhibited similar patterns of expression in control subjects and those with LTBI (e.g., GZMB, IFNG, SEPP1, CCL18, CD163, KIF2C, RASD1, TICRR, HIST1H2BB, RARRES1, STAB1, SEPP1, and ITGB5). In contrast, none of these genes were significantly differentially expressed in the patients with sarcoidosis.

Table 3.

The Most Significantly Differentially Expressed Genes Identified within the Granulomatous Structures in Each Donor Group 7 Days after Treatment, when Comparing Treatment with Purified Protein Derivative–Coated Beads to Correspondingly Matched Uncoated (PBS-Washed) Beads

Gene	Accession No.	Fold Change
		Control	LTBI	Sarcoid
Up-regulated genes
IFNG	NM_000619	156.8*	239.7*	ns
KIF2C	NM_006845	50.5	189.3*	ns
DMC1	NM_007068	88.4*	140*	6.7
RASD1	NM_016084	138*	40.8	ns
TICRR	NM_152259	25.4	131.2*	ns
HIST1H2BB	NM_021062	129*	48.1	ns
ITGA2	NM_002203	31.2	120.8*	ns
DAPK2	NM_014326	33.6	118.1*	ns
ASPM	NM_018136	38.6	108.8*	ns
GZMB	NM_004131	106.5*	93.6*	3.1
CDT1	NM_030928	35.3	100.6*	ns
DIAPH3	NM_001042517	14.5	97.2*	ns
CSF2	NM_000758	83.1*	35.7	12.8*
CTLA4	NM_005214	78.1*	57.2	7.7
HIST1H3H	NM_003536	74.2*	65.7	ns
EGLN3	NM_022073	74*	ns	ns
INHBA	NM_002192	69.9*	ns	ns
CXCR6	NM_006564	37.4	18.4	14.6*
IL8	NM_000584	7.4	ns	23.6*
IL3RA	NM_002183	22.7	ns	20.5*
C21orf7	NM_020152	5.4	ns	20.8*
PTGES	NM_004878	ns	ns	20.1*
CRLF2	NM_022148	6	ns	14.2*
TNFAIP6	NM_007115	5.3	ns	13.6*
C1S	NM_201442	ns	ns	13.1*
CXCL5	NM_002994	ns	ns	12*
Down-regulated genes
RARRES1	NM_206963	−482.4*	−50.3	ns
STAB1	NM_015136	−333.2*	−174.9*	ns
SEPP1	NM_001085486	−297.9*	−328.8*	ns
LILRB5	NM_001081442	−66.2	−275.1*	ns
PMP22	NM_153322	−8.4	−227.9*	ns
ITGB5	NM_002213	−161.8*	−34.6	ns
CCL18	NM_002988	−159.2*	−99.5	ns
CD163	NM_004244	−93.7	−158.3*	−3.6
FOLR2	NM_001113536	−137.1*	−105.8*	ns
MMP12	NM_002426	−34.8	−134*	ns
HS3ST2	NM_006043	−111.4*	−124*	ns
SPARC	NM_003118	−118.9*	−44	ns
MS4A6A	NM_152851	−60.1	−117.4*	ns
LPL	NM_000237	−108.6*	−38.4	−7.7
MMP7	NM_002423	−38.1	−108.3*	−3.1
MPEG1	NM_001039396	−106.3*	−51.5	ns
GAL	NM_015973	−50.5	−30.2	−11.6*
TIMP3	NM_000362	−38.9	−25.6	−13.1*
GALNT14	NM_024572	−37.5	−30.3	−11.8*
CLMP	NM_024769	−13.4	−29.9	−18.3*
FBLN5	NM_006329	−20.3	−10	−11.5*
CACNA2D3	NM_018398	−19.9	−16.8	−14.5*
S100A16	NM_080388	ns	ns	−18.8*
GPER	NM_001098201	−16.7	−10.3	−14.9*
HTR2B	NM_000867	ns	−9	−12.9*
GPD1	NM_005276	ns	ns	−10.5*

Definition of abbreviations: LTBI, latent tuberculosis infection; ns, not significantly changed (i.e., P > 0.05).

Within each of the 3 treatment groups, the top 20 differentially expressed genes were identified (top 10 up- and 10 down-regulated); the resulting gene lists were merged.

^*Differentially expressed genes falling within the top 10 of the given treatment group.

Figure 5.

Heat map analysis of gene expression data from PPD-treated PBMCs. Differential gene expression was determined in PBMCs derived from patients with SARC, LTBI, or normal healthy control subjects (CTRL) after incubation with PPD-coated beads for 7 days compared with paired treatment with uncoated (PBS-washed) beads. The heat map was generated using the top 10 most up-regulated and top 10 most down-regulated genes within each treatment group. Genes with up-regulated expression are shaded red, whereas those with down-regulated expression are shaded green, based upon the magnitude of the change in expression. Hierarchical clustering demonstrated a grouping of four of six SARC samples, which were readily distinguished from three of four LTBI samples, based upon gene expression patterns (see text for further discussion).

The relative change in gene expression, as reflected in Table 3, did not account for the absolute expression level of the transcripts or the statistical power of the observed changes in expression. These variables are accounted for in the “MvA” and “Volcano” plots when comparing treatment with PPD-coated beads to uncoated beads for the control, LTBI, and sarcoidosis groups (Figure 6).

Figure 6.

Gene expression comparison after PPD treatment of PBMCs derived from subjects with sarcoidosis, those with LTBI, and normal, healthy control subjects. The output type “MvA” plots are of normalized gene expression data wherein each point represents a gene, with the y-axis representing a log (base 2) fold change in expression after 7-day treatment with PPD-coated beads compared with uncoated beads for PBMCs derived from normal, healthy control subjects (A), subjects with LTBI (C), and subjects with sarcoidosis (E). The x-axis is the log average of the gene expression level. All genes with an adjusted P value of 0.05 (representing a 5% false discovery rate) and at least a twofold change (highlighted by the two horizontal gray lines) in the magnitude of gene expression between PPD and uncoated beads are shaded red. The blue line represents smoothed local mean expression (fitted using a generalized additive model), with the surrounding 95% confidence level interval shaded light blue. The dotted gray contour lines represent a two-dimensional density estimation. Genes most highly expressed in response to PPD-coated beads are highlighted in the top of the MvA plot, whereas those genes expressed at a lower level are at the bottom (the top five most differentially expressed genes in each case are labeled). Volcano plots of the same data wherein the y-axis corresponds with transcripts with high statistical significance (−log 10 of P value), and the x-axis corresponds with fold-change of gene expression (log base 2) were also generated for normal, healthy control subjects (B), subjects with LTBI (D), and those with sarcoidosis (F). Labeled transcripts on the upper left side of the plot have strong statistical significance with relatively low expression after treatment with PPD-coated versus uncoated beads. Transcripts on the upper right are more highly expressed with strong statistical significance after PPD treatment. The dotted line corresponds with a P value cutoff of 0.05, and genes failing to meet this cutoff are shaded gray.

Discussion

The human in vitro granuloma model presented herein recapitulates several key features of tissue granulomas in sarcoidosis and LTBI, including three-dimensional cell aggregation, extracellular cytokine profiles, and induction of genes regulating molecular pathways characteristic of Th1 immune responses in the context of human granulomatous diseases. Moreover, and in contrast to models that had been previously developed and not widely adopted, the human in vitro model attributes are likely influenced by genetic variables that predispose to human granulomatous disease. The ability of the model to account for genetic variability is supported by the distinct molecular patterns that were promoted in human PBMCs in response to TB antigens in the sarcoidosis, LTBI, and normal, healthy control groups.

PBMC-based research is inherently convenient, but is also highly relevant for immunological research as it relates to granulomatous disorders. PBMCs constitute the reservoir from which most immune cells are recruited to the sites of new tissue granuloma formation. Furthermore, the maintenance of tissue granulomas is also dependent upon “new recruits” to maintain immune cell populations and to sustain coordinated interactions. Once recruited to the site of granuloma formation, the PBMCs become activated, as reflected by the production of various cytokines and chemokines, and in many cases, the cells further differentiate to play specific roles in the regulation of granulomatous inflammation (18–20). Investigations focused on in vitro immune responses to TB infection have yielded important discoveries relating to the pathogenesis of granulomatous diseases, including identification of Mycobacterium tuberculosis antigens essential for initial induction of granuloma formation (21), altered expression of host genes regulating type I IFNs, B and T cell functions in the context of active TB and LTBI (22), and the identification of microRNAs predicted to suppress WNT and transforming growth factor-β pathways in patients with sarcoidosis (23).

Despite similar granuloma-like appearance, the in vitro human granuloma model revealed significant distinctions in terms of the patterns of cytokine and gene expression in PBMCs from patients with sarcoidosis and LTBI, particularly in response to TB antigen stimulation. For instance, the release of IFN-γ was significantly higher after PPD treatment in LTBI, whereas relatively higher TNF-α expression was observed in sarcoidosis. In this regard, IFN-γ and TNF-α contribute independently to regulation of granulomatous immune responses. IFN-γ–mediated activation of signal transducer and activator of transcription (STAT) 1 transcription factor promotes immune responses that are critical for initial granuloma formation in response to infection (24, 25), and IFN-γ is essential to preventing dissemination of TB infection in mice (26), including prevention of reactivation of LTBI (27). In contrast, TNF-α is reportedly not essential for the initial granulomatous response to M. tuberculosis in mice, but is required for the maintenance of well formed granulomas and for eventual clearance of the pathogen (19).

It is interesting to consider the implications of higher production of TNF-α and the increased expression of other molecules engaged in antimicrobial functions (IL8, CXCL1, CXCL3, CXCL5, CXCL16, CCL13, CCL18, MMP7, MMP12, MMP14, TLR2, and S100A8/S100A9) in humans with sarcoidosis compared with patients with LTBI. For instance, significantly higher TNF-α production in patients with sarcoidosis is predicted to enhance bacterial and fungal pathogen clearance, which would reduce the risk of infections. On the other hand, excess TNF-α production in response to TB antigens may perpetuate granuloma formation in the absence of infection (by definition) in patients with sarcoidosis. In contrast to sarcoidosis, differential gene expression in LTBI included extremely high fold-change expressions of IFNG and GZMB (Table 3 and Figure 6), encoding molecules that are instrumental in the granulomatous immune response to TB in humans. This finding may reflect adaptation of the immune system to chronic low-grade TB infection, as previously reported (28, 29).

Gene expression analyses of the in vitro model may provide novel insights and new hypotheses relating to disease mechanisms. Volcano plots of the transcriptional response to TB antigen in healthy control subjects (Figure 6B) and LTBI (Figure 6D) identified the genes with the highest statistical probability of differential expression, which included multiple genes coding for components of histone 1 (e.g., HIST1H3H, HIST1H2BM, HIST1H2BB, HIST1H2BL, HIST3H2BB, HIST1H2AB, HIST1H3B, and HIST1H1A), which were 40- to 140-fold up-regulated. In contrast, up-regulation of this group of genes was dramatically attenuated in sarcoidosis (two- to fivefold up-regulated). Lower expression (activity) of histone 1 has implications for the regulation of inflammatory and fibrotic responses (30, 31).

S100A9 was observed to be significantly down-regulated at least 20-fold in both the control and LTBI groups, but expression remained unchanged in sarcoidosis. This increased relative expression of S100A9 in the sarcoidosis group has been previously reported in sarcoidosis granulomas (32, 33), and is postulated to correspond with freshly recruited macrophages during the early phase of granuloma formation (34). Similarly, CD163 expression was dramatically reduced (−58.8- and −52.5-fold in control subjects and LTBI, respectively), but was observed to be only moderately down-regulated in sarcoidosis (−5.8 fold), reflecting relative polarization to a M2 macrophage phenotype. M2 polarization has implications for promoting a fibrotic disease phenotype (i.e., poor prognosis) in patients with sarcoidosis (35).

Among the genes uniquely differentially expressed in LTBI in this model was ABCA1 (−28.3-fold down-regulated), which is incriminated in the survival of macrophages and intracellular pathogens in the context of acute infection, presumably relating to the intracellular transport of cholesterol to maintain bacterial survival (36). The induction of ABCA1 occurs within hours of infection in normal mice (36), whereas expression was greatly reduced by Day 7 in this model, suggesting that ABCA1 expression was dynamically changing over time, or is perhaps fundamentally altered in patients with LTBI compared with those with active TB.

The nonspecific activation of sarcoidosis macrophages in response to HSA-coated beads observed herein is in keeping with previous reports of preactivated macrophages in patients with active pulmonary sarcoidosis (37, 38), including enhanced antigen-induced T cell activation (39). Moreover, prior studies have shown that blood macrophages derived from patients with sarcoidosis exhibit accelerated phagocytosis (40). The mechanisms by which HSA-coated beads promoted selective activation of the macrophages from patients with sarcoidosis are unclear.

We believe that the in vitro human granuloma model has the potential to rapidly advance discoveries in the fields of sarcoidosis, TB, and many other granulomatous diseases. One novel feature of the model is the opportunity to investigate disease mechanisms during the earliest phases of granuloma formation. For example, many genes regulated by IFN-γ/STAT1 signaling pathways were induced by PPD treatment in both sarcoidosis and LTBI PBMCs compared with controls. However, a subgroup of genes the expression of which remained unchanged in sarcoidosis in response to PPD antigens compared with significant down-regulation in LTBI (CXCL16, MMP9, MMP12) are specifically regulated by STAT1 through downstream transcription factors, interferon regulatory factor 8 (IRF8) (IRF8 expression was ∼ twofold higher in sarcoidosis compared with LTBI) and IRF1 (41, 42). Likewise, IFN-γ/STAT1 signaling and downstream nuclear transcription factors, Ikaros (expressed ∼ twofold higher in LTBI), cAMP response element-binding protein, and activator protein 1, regulate the transcription of GZMB (43), which was highly expressed in LTBI (but not in sarcoidosis) (Table 3). Thus, PPD activation led to distinct expression of IFN-regulated genes in human PBMCs derived from patients with sarcoidosis and LTBI. Future investigations using this model could consider genetic variables (e.g., STAT1 [44] or IRF8 [45] polymorphisms) or epigenetic factors (46) contributing to altered IFN-regulated responses in these patients.

As with any in vitro model, there are some limitations. It could be argued that PBMCs do not accurately represent the immune cell populations found in diseased tissues; however, most of the cells that ultimately contribute to tissue granuloma formation are derived from PBMCs (18–20, 47). This point was accentuated by the experimental data showing that chemokines made up the majority of the most highly expressed genes in PPD-induced granulomas in patients with sarcoidosis (Table 3). Another limitation is the lack of stromal cell support in the model. Stromal cells, particularly surrounding epithelial cells, contribute to macrophage recruitment to the granuloma in patients with TB (47), and increased stromal expression of a specific matrix metalloproteinase (a disintegrin and metalloproteinase domain-like protein decysin 1), a marker of monocyte differentiation (48), suggests a potential role in promoting pulmonary sarcoidosis granulomas (49). Nonetheless, as hypothesized, granuloma-like structures exhibiting molecular features previously reported in the context of diseased human tissues was demonstrated. Although the model does not account for all variables in vivo, it more closely approximates the complex interaction among immune cells during the early formation of granulomas in the context of sarcoidosis compared with existing laboratory models. In particular, this model is clearly more advanced in this regard compared with typical in vitro analyses involving a single cell line, which fails to account for dynamic immune cell interactions, which are the hallmark of granulomatous inflammation. This point is further emphasized by a preliminary analysis of gene expression data in the in vitro model compared with human tissue (Table E1) showing common differentially expressed gene expression characteristics. However, we are currently modifying the model to more closely replicate in vivo conditions by coculturing appropriate stromal cells during in vitro granuloma formation.

Another inherent limitation of sarcoidosis research is the lack of specific knowledge relating to disease-causing antigens. That said, previous studies show that approximately 60% of patients with sarcoidosis who are naive to TB infection respond abnormally to TB antigen challenge to promote an adaptive immune response (13), which is in keeping with the variable response to PPD-coated beads in this model (Figure 1). The study size was small, such that all sarcoidosis phenotypes were not likely to be represented. Nonetheless, two patterns emerged in the model that were in keeping with prior studies: four of six subjects with sarcoidosis studied herein were particularly high responders to PPD-coated beads in terms of in vitro granuloma formation and related Th1 molecular patterns (Figures 4 and 5). Furthermore, gene expression patterns of these four patients clustered together and were distinct from control subjects or subjects with LTBI after PPD-coated bead treatment. The remaining two patients with sarcoidosis clustered more closely with the control group (Figures 4 and 5). We had previously reported nonspecific accentuated Toll-like receptor (TLR) responses in patients with active sarcoidosis, including TLRs that are highly responsive to TB antigens (TLR2, TLR9) (50), leading us to conclude that patients with sarcoidosis would respond to other environmental challenges, such as propionibacterium (a TLR2 agonist) or DNA viruses (TLR9 agonists), as previously reported (51, 52). Furthermore, to more closely model the normal immune response to environmental antigens, which involves initial phagocytosis by antigen-presenting cells (dendritic cells, macrophages) and subsequent presentation of antigen to T cells, we used beads that are known to be processed by macrophages. Antigens loaded on beads are shown to be more efficient in forming productive immune synapses between macrophages and T cells and subsequent T cell activation and proliferation (53). Thus, we employed the antigen-coated beads (as opposed to soluble antigen) to more closely replicate the in vivo immune response wherein immunogenic antigens are often admixed with other environmental particles and to optimize antigen processing via phagocytosis and subsequent antigen presentation. Thus, the model will be useful for screening and validating environmental factors implicated in the pathogenesis of sarcoidosis (54).

The in vitro human granuloma model provides a novel research tool for exploring disease mechanisms and treatments in the context of sarcoidosis, TB, and, presumably, other human granulomatous disorders. The model incorporates key components of granuloma biology, including exposure to environmental antigens, highly variable host genetics, complex intracellular and intercellular signaling, and the variable of time, resulting in a model that, in several respects, replicates the human condition. As opposed to research conducted on diseased human tissues, the model is highly versatile in that any number of variables can be manipulated, and it is relatively convenient and inexpensive compared with conventional studies based upon diseased tissues. Although not demonstrated here, the model is conducive to preclinical testing of existing or exploratory therapeutics designed to normalize the granulomatous response.