Spatial heterogeneity for APRIL production by eosinophils in the small intestine

Abstract

Eosinophils may reside in the lower intestine to play several homeostatic functions. Regulation of IgA⁺ plasma-cell (PC) homeostasis is one of these functions. Here, we assessed regulation of expression for a proliferation-inducing ligand (APRIL), a key factor from the TNF superfamily for PC homeostasis, in eosinophils from the lower intestine. We observed a strong heterogeneity, since duodenum eosinophils did not produce APRIL at all, whereas a large majority of eosinophils from the ileum and right colon produced it. This was evidenced both in the human and mouse adult systems. At these places, the human data showed that eosinophils were the only cellular sources of APRIL. The number of IgA⁺ PCs did not vary along the lower intestine, but ileum and right colon IgA⁺ PC steady-state numbers significantly diminished in APRIL-deficient mice. Use of blood cells from healthy donors demonstrated that APRIL expression in eosinophils is inducible by bacterial products. Use of germ-free and antibiotics-treated mice confirmed the dependency on bacteria for APRIL production by eosinophils from the lower intestine. Taken together, our study shows that APRIL expression by eosinophils is spatially regulated in the lower intestine with a consequence on the APRIL dependency for IgA⁺ PC homeostasis.

1. Introduction

Eosinophils are best known for their effector functions in immunity against some helminthic parasites.¹ However, most circulating mature eosinophils produced by the bone marrow extravasate in tissues to become resident.² There, they may play diverse functions. One tissue well characterized for the presence of resident eosinophils is the lamina propria of the lower intestine.³ This eosinophil colonization is an early process in life starting before birth and any colonization of the lower intestine by commensal bacteria.^4,5 There, resident eosinophils fulfill several homeostatic functions, including maintenance of the mucus layer and interaction with the enteric nervous system. In the lower intestine, eosinophils also have immune functions with a role in the homeostasis of IgA⁺ PC mostly achieved by provision of survival signals to IgA⁺ PC.⁶

A proliferation inducing ligand (APRIL), a member from the TNF superfamily (TNFSF13), is a key factor positively regulating PC homeostasis.⁷ The APRIL protein is first produced as a transmembrane protein and early cleaved in the Golgi apparatus during its secretory path by furin proteases.⁸ Myeloid cells constitute the main cellular sources of APRIL with a constitutive production, which may be further upregulated by inflammatory signals sensed by toll-like receptors.^9,10 We and others previously reported APRIL production by eosinophils,^6,11,12 but the regulation of APRIL production by eosinophils is still poorly characterized. Here, we studied APRIL expression by eosinophils in the gut of adult mice and humans.

2. Material and methods

2.1 Human experimentation

Approval for research on human samples was obtained from the ethics committee of the Geneva University hospital. Human intestinal samples were obtained from patients with Merkel cell carcinoma outside tumor lesions between 2003 and 2010 after patients’ informed consent according to the Declaration of Helsinki. The presence of vili with Brunner glands identified the duodenum. The presence of vili without Brunner glands but Peyer's patches identified the ileum. The right colon was identified by the absence of vili.

2.2 Bacterial cultures

Escherichia coli (E. coli) was grown in standard Luria Broth medium. Bifidobacterium bifidum was grown anaerobically at 37 °C in a Hungate tube with degassed Yeast Casitone Fatty Acids, medium 26 supplemented with antifoam 0.05% (Sigma). Cultures were harvested, washed in PBS, and resuspended in Krebs–Ringer glucose (KRG) buffer to the desired concentrations after density measurement at 600 nm.

2.3 Cell purification and activation

Peripheral blood eosinophils were purified from buffy coats of healthy adult donors. After centrifugation on a Ficoll gradient (Pharmacia) and hypotonic lysis, eosinophils were purified from the granulocyte fraction by FACS-sorting the CD16⁻SSC^high population with a FACS-Aria (Becton-Dickinson). The purity of the eosinophils was routinely >95%. Human blood neutrophils were purified as previously described.⁹ Viability assessed by trypan blue exclusion was above 80%. 10⁵ Eosinophils resuspended in KRG were incubated in microwell plates with bacteria at a ratio of 1/1. Human IL-5, GMCSF, and IL-3 used at 25 ng/ml were from Peprotech. Cultures were harvested after 18 h of incubation at 37 °C for mRNA analyses. Production of reactive oxygen species was determined as previously described.⁹ Mouse lamina propria and purified eosinophils were prepared as previously described.⁶ Purity of eosinophils was above 90%.

2.4 Mouse experimentation

All animal care and procedures were approved by the Geneva veterinary office and the French national AFiS veterinary office. C57Bl/6 WT and germ-free mice were obtained from Charles River. WT C57Bl/6 mice were maintained in our animal facility at least 6 wk before use. APRIL^−/− mice on a C57Bl/6 background were previously described.¹³ Antibiotics treatment consisted in the addition of 333 mg/l of tylosin tartrate (Sigma) in the drinking water for 2 consecutive days. Intestinal samples were collected 1 wk after treatment. Duodenum samples originated from the area right distal to the stomach. Ileum samples originated from the small intestine harboring Peyer's patches. Right colons were analyzed only.

2.5 RT-PCR

Total RNAs were prepared using Trizol (Thermo Fisher Scientific), treated with RQ1 DNAse (Promega), and reverse transcribed with Superscript II and oligo-dT_12-18 (Thermo Fisher). The intron spanning for human and mouse APRIL has been described elsewhere.¹⁴ Denaturation was performed at 94 °C, annealing at 55 °C, and extension at 72 °C, 1 min each. Forty cycles were applied. Amplified PCR products were visualized on agarose gels and ethidium bromide staining. Quantitative RT-PCR was done with a light cycler system (Roche Diagnostics) and the Quantitect SYBR Green PCR kit solution (Qiagen). Expression levels were normalized to human 18S rRNA and to mouse actin; amplification was done in duplicate, and experiments were performed twice.

2.6 Immunohistochemistry

Formalin-fixed paraffin-embedded intestine sections of 3 µm were stained with Stalk-1 (rabbit polyclonal) detecting APRIL-producing cells and Aprily-2, -6, -8 detecting secreted APRIL as previously described.¹⁵ Eosinophil staining was performed with a mouse anti-major basic protein antibody (clone BM13, Thermo Fisher) after a citrate-based antigen retrieval. In the human system, IgA⁺ CD138⁺ PC was stained with a goat antihuman IgA (Vector laboratories) and the antihuman CD138 (clone Mi15, Agilent) after citrate-based antigen retrieval. In the mouse system, a rabbit antimouse IgA (Thermo Fisher) and antimouse CD138 (Life Technologies) were used after citrate-based antigen retrieval. Fluorochrome-conjugated secondary antibodies were from Life Technologies. DAPI (Thermo Fisher) was used for nucleus staining. Stainings were visualized as previously described.⁹

2.7 Flow cytometry

Surface and total cell staining was performed as previously described.¹⁵ All fluorescent-conjugated primary antibodies were from Becton-Dickinson. Fluorescence was measured with a FACSCanto and analyzed with the CellQuest software (Becton-Dickinson).

2.8 Statistical analysis

Statistical analysis was performed using GraphPad Prism software. D’Agostino and Pearson test (normality), ANOVA, and t-tests were performed. Significant differences were defined as P < 0.05.

3. Results and discussion

3.1 Spatial heterogeneity for APRIL production in eosinophils from the lower intestine

We assessed in the human lower intestine APRIL production with the Stalk-1 antibody recognizing the stalk fragment left in producing cells after APRIL processing. Notably, we found no APRIL-producing cells in the duodenum region, whereas ileum and right colon contained numerous cells (Fig. 1A, upper panel). The highest was the right colon. High magnification of a stained cell revealed an eosinophil morphology with a characteristic bilobated nucleus (Fig. 1A, upper panel, inserts). We confirmed eosinophil identification for these APRIL-producing cells by costaining with the major-basic protein, a protein specifically contained in eosinophil granules. These cells were never present in Peyer's patches, the site of T-cell dependent production of IgA⁺ PC in the gut (Supplementary Fig. 1A).¹⁶ Occasionally, they infiltrated the epithelium (Supplementary Fig. 1B). Eosinophil infiltration of a mucosal epithelium is at the basis of several pathologies, but also occurs in healthy small intestines.¹⁷ Eosinophils are known to preferentially infiltrate the lower intestine with an increase in distal zones.¹⁸ Our eosinophil quantification based on MBP staining was similar with an infiltration of 20 cells +/− 6 per mm² in the duodenum, 22 +/− 8 in the ileum, and 43 +/− 4 in the right colon. APRIL and MBP costaining showed that all eosinophils infiltrating the duodenum did not produce APRIL. In the ileum, APRIL-negative eosinophils were reduced to 9% +/− 6, and all eosinophils produced APRIL in the right colon. We also tested secreted APRIL reactivity with the Aprily-2 antibody. Duodenum was negative, while a faint reactivity was detected in the extracellular space of the duodenum close to APRIL-producing cells (Fig. 1C). Some eosinophils were also positive intracellularly for secreted APRIL, and PC could occasionally be detected with surface bound secreted APRIL (inserts in Fig. 1C). Similar reactivities were obtained with Aprily-6 and -8 antibodies also detecting secreted APRIL. The latter constituted evidence of APRIL secretion in the lower intestine and an on-target activity. IgA⁺ cells were obviously numerous in the lower human intestine, adjacent to APRIL producing cells (Fig. 1C, upper panel). Among these cells, we quantified IgA⁺ CD138⁺ PCs with 2-color immunofluorescence (Fig. 1C, bottom left panel). We excluded from the quantification the few IgA⁺ CD138⁻ cells present (arrow in Fig. 1C, bottom left panel), likely being memory B cells present in the lamina propria as reported by others.¹⁹ Despite the heterogeneity in APRIL expression between duodenum and ileum, we did not observe a significant difference in the density of IgA⁺ CD138⁺ PCs at these locations (Fig. 1C, bottom right panel).

We next assessed APRIL expression in the mouse gut at the mRNA level, since the Stalk-1 antibody did not cross-react with mouse APRIL. RT-PCR confirmed the human data with an almost undetectable amplicon in the mouse duodenum, while the lamina propria from ileum and right colon were positive (Fig. 2A, left panel). Analysis of purified eosinophils from these lamina propria gave identical results with negativity and positivity for duodenal and ileal/colonic cells, respectively (Fig. 2A, right panel). IgA⁺ cells were also numerous in the mouse lower intestine (Fig. 2B, upper panel). We also quantified the number of IgA⁺ CD138⁺ PCs among these cells (Fig. 2B, bottom left panel). Again, we found no quantitative differences in duodenum, ileum, or right colon (Fig. 2B, bottom right panel). However, we found significantly lower numbers for these cells specifically in the ileum and right colon when we performed our analysis in APRIL^−/− mice. Taken together, our results showed that the human and mouse small intestine harbor a similar expression pattern for APRIL production with an absence of expression in the duodenum. The absence of APRIL expression in the duodenum was confirmed in the mouse study by the lack of IgA⁺ PC number variation in APRIL^−/− mice.

3.2 APRIL production in eosinophils is induced by bacterial stimulation

We next assessed the regulation of APRIL expression to understand the heterogeneity described above. We used human blood eosinophils. Intracellular staining with the Stalk-1 antibody did not detect the APRIL protein in these cells, while blood neutrophils were positive as previously described (Fig. 3A, left panel).¹⁴ We obtained a similar result at the mRNA level (Fig. 3A, right panel). Eosinophils may be activated by cytokines.²⁰ The common IL-5, GM-CSF, and IL-3 did not induce APRIL mRNA in human blood eosinophils, whereas they were a potent inducer of an oxidative burst in these cells (Fig. 3B). Bacterial products may also activate eosinophils.²¹ Coculture of eosinophils with E. coli potently induced APRIL mRNA in these cells. We also detected intracellular APRIL in E. coli-activated eosinophils (Fig. 3C). In this experiment, B. bifidum was also able to induce intracellular APRIL. In vivo, gut eosinophils are exposed to products from commensal bacteria. We observed that detection of APRIL mRNA by RT-PCR in the lamina propria of ileum and right colon vanished in germ-free mice (Fig. 4A). Furthermore, modulation of the commensal bacterial load after antibiotics treatment significantly reduced APRIL gene transcription in the ileum and right colon in the lamina propria (Fig. 4B). We observed a similar reduction in mRNA with purified eosinophils. Taken together, these show that bacterial products from the commensal bacteria induce production of APRIL by eosinophils from the lower intestine.

The phenotype observed in mice genetically deficient for APRIL was impairment of IgA switching.²² Several groups confirmed this role with recombinant APRIL and showed that it is a T-cell independent process.^23–26 A role for APRIL has also been observed in T-cell dependent humoral responses.^27,28 The APRIL receptor involved in this switch process is the transmembrane activator and CAML interactor.^29,30 Besides a role in IgA switching, APRIL also has a role later on during B-cell differentiation at the level of PC survival. This second role has been identified in the bone marrow and mucosa.^13,14,31 Identification of the first patient deficient in APRIL largely confirmed all these preclinical studies.³² Our present study shows that eosinophils and their production of APRIL are important for IgA⁺ PC homeostasis. This is highly consistent with a report from Chu et al. showing that mice deficient for eosinophils harbored reduction in gut IgA⁺ PCs.⁶ However, Beller et al. recently reported that microbiota but not eosinophils are required for IgA production.³³ In the latter study, IgA class switching occurred due to the production of transforming growth factor β upon bacteria colonization of the gut. One should note that quantification of IgA⁺ PC was performed quite shortly after microbiota transfer (21 d), so the study likely focused on PC generation rather than on long-term survival as Chu et al. did. Regarding the cellular source of APRIL in the human gut, we did not find evidence for APRIL production in epithelial cells from the lower intestine and right colon. This contradicts a report from He et al.,²³ even though the Stalk-1 antibody used here identified epithelial cells producing APRIL from several mucosa.^13,34 The discrepancy might be due to different zones studied from the gut. We studied the right colon, whereas He et al. did not specify the region of the human colon they analyzed. Regarding eosinophils, we further found that their APRIL production is heterogeneous along the lower intestine. Such heterogeneity is because APRIL production by eosinophils is not constitutive as shown by the negativity of blood eosinophils, but inducible depending on the tissue the cells extravasate in, at least for the gut. We observed that E. coli and B. bifidum, facultative and strictly anaerobic strains, respectively, well present in the lower intestine, induced APRIL production in blood eosinophils. Hence, the spatial heterogeneity observed may be explained by either the different bacteria composition along the lower intestine or the higher bacterial density within the ileum that contains over three log10 more cells than the duodenum.^35,36 This observation is consistent with the fact that gut eosinophils change their phenotype once extravasated in the gut compared to their circulating blood counterpart.³⁷ Most recent studies further indicated that they do so by responding to the gut microenvironment, including the microbiota.^38,39 The absence of APRIL in the duodenum is puzzling if one considers that duodenum IgA⁺ PC may persist for decades.⁴⁰ Nevertheless, one should note that duodenum constitutes a second organ known to sustain PC survival in which APRIL is absent in addition to inflamed salivary glands from Sjögren syndrome.⁴¹ Further studies are warranted to identify the PC survival factor(s) substituting to APRIL.

Acknowledgments

This work was supported by the Finovi Foundation (BH) and by the French Society for Hepatology (AFEF) (NS).

Author Contributions

NS, BH, PL, MG, MRG, and FP performed experiments and analyzed data. BH designed the study and wrote the article.

Supplementary material

Supplementary material is available at Journal of Leukocyte Biology online.

References

Author notes