Immunomodulatory tetracyclines shape the intestinal inflammatory response inducing mucosal healing and resolution

Abstract

Background and Purpose

Immunomodulatory tetracyclines are well-characterized drugs with a pharmacological potential beyond their antibiotic properties. Specifically, minocycline and doxycycline have shown beneficial effects in experimental colitis, although pro-inflammatory actions have also been described in macrophages. Therefore, we aimed to characterize the mechanism behind their effect in acute intestinal inflammation.

Experimental Approach

A comparative pharmacological study was initially used to elucidate the most relevant actions of immunomodulatory tetracyclines: doxycycline, minocycline and tigecycline; other antibiotic or immunomodulatory drugs were assessed in bone marrow-derived macrophages and in dextran sodium sulfate (DSS)-induced mouse colitis, where different barrier markers, inflammatory mediators, microRNAs, TLRs, and the gut microbiota composition were evaluated. The sequential immune events that mediate the intestinal anti-inflammatory effect of minocycline in DSS-colitis were then characterized.

Key Results

Novel immunomodulatory activity of tetracyclines was identifed; they potentiated the innate immune response and enhanced resolution of inflammation. This is also the first report describing the intestinal anti-inflammatory effect of tigecycline. A minor therapeutic benefit seems to derive from their antibiotic properties. Conversely, immunomodulatory tetracyclines potentiated macrophage cytokine release in vitro, and while improving mucosal recovery in colitic mice, they up-regulated Ccl2, miR-142, miR-375 and Tlr4. In particular, minocycline initially enhanced IL-1β, IL-6, IL-22, GM-CSF and IL-4 colonic production and monocyte recruitment to the intestine, subsequently increasing Ly6C⁻MHCII⁺ macrophages, Tregs and type 2 intestinal immune responses.

Conclusions and Implications

Immunomodulatory tetracyclines potentiate protective immune pathways leading to mucosal healing and resolution, representing a promising drug reposition strategy for the treatment of intestinal inflammation.

Abbreviations

BMDM

bone marrow-derived macrophages

cLP

colonic lamina propria

DAI

disease activity index

DC

dendritic cell

DEX

dexamethasone

DSS

dextran sodium sulfate

DXC

doxycycline

IBD

inflammatory bowel disease

ILC

innate lymphoid cell

MNC

minocycline

Mφ

macrophage

RFX

rifaximin

TGC

tigecycline

Th

T helper cell

Treg

regulatory T cell

TTC

t1etracycline

Introduction

Immunomodulatory antibiotics are an interesting therapeutic strategy for intestinal inflammation, targeting both the altered microbiota and the exacerbated inflammatory response. In particular, minocycline and doxycycline have shown promising results in experimental colitis (Huang et al., 2009b; Garrido-Mesa et al., 2011a,b, 2015). These are well known tetracyclines with proven benefits in many inflammatory conditions (Garrido-Mesa et al., 2013a). Their intestinal anti-inflammatory effect has been mainly associated with a reduction in inducible NOS (iNOS) and MMP activity (Huang et al., 2009b) and direct immunomodulatory and antibiotic properties (Garrido-Mesa et al., 2011a,b, 2015). However, the relevance of these activities to the overall anti-inflammatory effect has not been specifically assessed. Of note, although their actions within the immune system are generally anti-inflammatory, a certain degree of controversy has been observed as regards their effects in monocytes and macrophages (Mφ): while immunomodulatory tetracyclines inhibit the inflammatory activity of microglia and peritoneal Mφs, increased activation and cytokine production has been observed in monocytes (Kloppenburg et al., 1996), alveolar Mφs (Bonjoch et al., 2015) and RAW264.7 colonic Mφ cell line (Dunston et al., 2011).

In this regard, although the inflammatory reaction may cause harm and tissue damage, a powerful intestinal mucosal immune system is also needed to protect and restore intestinal homeostasis (Mowat and Agace, 2014), where Mφs play a key role (Gross et al., 2015). Indeed, unlike other locations, the intestinal Mφ pool is continuously replenished from CCR2⁺Ly6C^hi blood monocytes, which then differentiate into Ly6C⁻MHCII⁺ resident Mφs in the steady state. In inflammatory conditions, however, their differentiation is arrested and Ly6C⁺MHCII⁻ Mφs accumulate (Bain et al., 2013), which display an M1/pro-inflammatory phenotype and produce cytokines that drive the inflammatory reaction. By contrast, intestinal resident Ly6C⁻MHCII⁺ Mφs are tolerogenic and display an M2-like phenotype, contributing to mucosal healing, resolution of inflammation and maintenance of intestinal homeostasis (Sherman and Kalman, 2004; Pull et al., 2005). Hence, a differential activity of tetracyclines on intestinal Mφs might be of special relevance, and a full understanding of their mechanisms is required to expand their therapeutic application to intestinal inflammatory conditions.

The present study aims to characterize the mechanisms of action of immunomodulatory tetracyclines in acute intestinal inflammation, by comparing their effects with other antibiotics or immunomodulatory drugs and studying their impact on the course of the immune response developed in dextran sodium sulfate (DSS)-induced colitis in mice. Our results confirm the relevance of their differential immunomodulatory activity for their anti-inflammatory effect and allow us to establish a link between the initial up-regulation of innate immunity and an improved mucosal healing and resolution. Thus, we have demonstrated that the enhancement of mucosal-protective immune pathways is a key immunomodulatory mechanism of tetracyclines in acute colitis, which is of great interest as an approach to prevent the chronification of intestinal inflammation.

Methods

In vitro studies

RAW264 murine macrophage and L929 murine fibroblast cell lines were obtained from the Cell Culture Unit of the University of Granada (Granada, Spain) and cultured in standard conditions. Bone marrow-derived macrophages (BMDM) were obtained from the bone marrow of C57BL/6J mice, cultured for 6 days in DMEM supplemented with 20% FBS and 30% L929-supernatant containing M-CSF factor. Cells were plated at 1 × 10⁶ cells·mL⁻¹, and the drugs were added for 24 h before stimulation with LPS (100 ng·mL⁻¹ for RAW cells or 10 ng·mL⁻¹ for BMDM) for either 3 h for RNA isolation or 24 h for cytokine determination by elisa (PeproTech EC Ltd, London, UK) or nitrite determination by Griess Assay (Green et al., 1982). Briefly, for nitrite determination, 100 μL of Griess reagent (0.1% N-(1-naphthy) ethylenediamine solution and 1% sulphanilamide in 5% (v^.v^-1) phosphoric acid solution, mixed in a proportion 1:1) was added to 100 μL of cell supernatant and incubated for 10 min. The concentration of the product of the reaction, a coloured azolic compound, can be determined by photometric measurement of the absorbance at 550 nm. Cell viability of tested conditions was measured by the MTS-based assay (Promega, Madison, WI, USA).

In vivo studies

All animal studies were carried out in accordance with the ‘Guide for the Care and Use of Laboratory Animals’ as promulgated by the National Institute of Health. Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny et al., 2010). Male C57BL/6J mice (6–8 weeks, 25 g) were obtained from Janvier (St Berthevin Cedex, France) and kept in specific pathogen-free facilities at University of Granada Biological Services Unit at 23 ± 1°C, with a relative humidity of 50–70% andon a regular 12 h dark/light cycle. Mice were housed in Makrolom cages (Ehret, Emmerdingen, Germany), with a maximum of eight mice per cage, with dust-free laboratory bedding and enrichment. They were fed standard chow diet (Panlab A04, Panlab, Barcelona, Spain) and provided drinking water ad libitum.

To investigate the mechanism behind the beneficial activity previously reported for tetracyclines (Garrido-Mesa et al., 2011a,b, 2015) and, in particular, to characterize their impact on the pathways involved in the initiation and resolution of acute intestinal inflammation, we focused on the experimental model of colitis triggered by DSS-induced mucosal injury, the most widely used model of acute colitis (Wirtz et al., 2007). A curative treatment protocol was used considering the lack of preventive effect observed in previous studies (Garrido-Mesa et al., 2011a) and taking into consideration the limitations on antibiotic usage in clinical practise. Colitis was induced by adding DSS (3% w^.v^-1) (36–50 KDa, MP Biomedicals, Ontario, USA) in the drinking water for a period of 5 or 6 days as indicated. Mice were then randomized and treated with different drugs for either 2, 4 or 6 days depending on the study. Disease evolution was monitored by a daily determination of the disease activity index (DAI), calculated as described in Table 1. Mice were anaesthetized with ketamine/xylazine (100 and 7.5 mg·kg⁻¹ respectively) for blood collection by cardiac puncture when required and killed by cervical dislocation, and the whole colon length was resected. Stools were collected aseptically for pyrosequencing. The colonic tissue was washed in PBS, and samples were collected for subsequent histological, biochemical and immunological evaluations.

Table 1. Scoring of DAI

Score	Weight loss	Stool consistency	Rectal bleeding
0	None	Normal	Normal
1	1–5%	Mucous traces	Perianal blood traces
2	5–10%	Loose stools	Blood traces on stools
3	10–20%	Diarrhoea	Bleeding
4	> 20%	Gross diarrhoea	Gross bleeding

• DAI value is the combined scores of weight loss, stool consistency and rectal bleeding divided by 3.

Histology

Representative colonic specimens were taken at 1 cm from the distal region, fixed in 4% paraformaldehyde and embedded in paraffin. Histochemical staining of mucins was performed by incubation of 4 μm re-hydrated sections in alcian blue 1% in acetic acid 3% for 30 min prior to conventional haematoxylin and eosin staining. Colonic microscopic damage was evaluated by a pathologist blinded to the experimental groups according to the criteria described in Table 2.

Table 2. Scoring criteria of full-thickness distal colon sections

Mucosal epithelium and lamina propia

Ulceration: none (0); mild surface (0–25%) (1); moderate (25–50%) (2); severe (50–75%) (3); extensive-full thickness (more 75%) (4).

Polymorphonuclear cell infiltrate

Mononuclear cell infiltrate and fibrosis

Oedema and dilation of lacteals

Crypts

Mitotic activity: lower third (0); mild mid third (1); moderate mid third (2); upper third (3)

Dilations

Goblet cell depletion

Submucosa

Polymorphonuclear cell infiltrate

Mononuclear cell infiltrate

Oedema

Vascularity

Muscular layer

Polymorphonuclear cell infiltrate

Mononuclear cell infiltrate

Oedema

Infiltration in the serosa

• Scoring scale: 0, none; 1 slight; 2, mild; 3, moderate; 4, severe. Maximum score: 59.

Colonic explant culture and cytokine determination by elisa

Whole thick colonic punch biopsies (three per specimen) (Miltex, York, PA, USA) were obtained from distal, medial and proximal regions and incubated in 0.5 mL of medium supplemented with gentamycin 50 μg·mL⁻¹ for 24 h. Cytokine concentration in culture supernatant was measured by elisa (PeproTech EC Ltd, London, UK).

RNA extraction and gene expression analysis

Representative colonic samples were taken for RNA extraction and gene expression studies. In the comparative pharmacological study on DSS-colitis where both microRNAs and mRNAs were evaluated, total RNA was isolated with miRNeasy mini Kit (Qiagen, Hilden, Germany), and 500 ng of RNA was reverse transcribed using the miScript II RT Kit from Qiagen (Qiagen, Hilden, Germany). For other studies, RNA was isolated using RNeasy^® Mini Kit (Qiagen, Hilden, Germany), and 3 μg of RNA was reverse transcribed using oligo (dT) primers (Promega, Madison, WI, USA). RT-qPCR of microRNAs was performed using the QuantiTect SYBR Green PCR Master Mix with miScript Universal Primers and the specific miRNA primer sequences (Qiagen, Hilden, Germany). For mRNA expression, RT-qPCR was performed using KAPA SYBR^® FAST qPCR Master Mix (KapaBiosystems, Inc., Wilmington, MA, USA). Detection was performed on optical-grade 48 well plates in an EcoTM Real-Time PCR System (Illumina, CA, USA). The small nucleolar RNA C/D box 95 (SNORD95) and GAPDH were measured to normalize microRNA and mRNA expression (ΔCt) respectively. Fold increases in values for gene expression analysis were calculated using normalized expression levels (2^−ΔCt) referred to the mean of NC control group (Fold Increase = 2^−ΔCt/2^−ΔCt_NC). SNORD95, miRNA and reverse universal primer for miRNA (Qiagen) and IL-22 (PrimerDesign, Chandler's Ford, United Kingdom) were sourced commercially. The remaining specific primer sequences (Sigma-Aldrich Quimica, Madrid, Spain) are presented in Table 3.

Table 3. RT-qPCR primer sequences

Gene	Gene ID		Sequence 5′-3′	Annealing T (°C)
Gapdh	14433	FW	5′-CCATCACCATCTTCCAGGAG	60
		RV	5′-CCTGCTTCACCACCTTCTTG
Muc-1	17829	FW	5′-GCAGTCCTCAGTGGCACCTC	60
		RV	5′-CACCGTGGGGCTACTGGAGAG
Mic-2	17831	FW	5′-GATAGGTGGCAGACAGGAGA	60
		RV	5′-GCTGACGAGTGGTTGGTGAATG
Muc-3	666339	FW	5′-CGTGGTCAACTGCGAGAATGG	60
		RV	5′-CGGCTCTATCTCTACGCTCTC
Ttf-3	21786	FW	5′-CCTGGTTGCTGGGTCCTCTG	60
		RV	5′-GCCACGGTTGTTACACTGCTC
Zo-1	21872	FW	5′-GGGGCCTACACTGATCAAGA	56
		RV	5′-TGGAGATGAGGCTTCTGCTT
Occludin	18260	FW	5′-ACGGACCCTGACCACTATGA	56
		RV	5′-TCAGCAGCAGCCATGTACTC
Mmp-9	17395	FW	5′-TGGGGGGCAACTCGGC	60
		RV	5′-GGAATGATCTAAGCCCAG
Nos2	18126	FW	5′-GTTGAAGACTGAGACTCTGG	56
		RV	5′-GACTAGGCTACTCCGTGGA
Alox15	11687	FW	5′-TTTTTGACAAGGAGGTGATGAGC	57
		RV	5′-GAAGCAAGTGTCAATATCCAG
Tlr2	24088	FW	5′-CCAGACACTGGGGGTAACATG	60
		RV	5′CGGATCGACTTTAGACTTTGGG
Tlr4	21898	FW	5′-GCCTTTCAGGGAATTAAGCTCC	60
		RV	5′-AGATCAACCGATGGACGTGTAA
Cxcl2	20310	FW	5′-CAGTTAGCCTTGCCTTTGTTCAG	62
		RV	5′-CAGTGAGCTGCGCTGTCCAATG
Ccl2	20296	FW	5′-CAGCTGGGGACAGAATGGGG	62
		RV	5′-GAGCTCTCTGGTACTCTTTTG
Ccl11	20292	FW	5′-AGTAACTTCCATCTGTCTCC	51
		RV	5′-TGGTGATTCTTTTGTAGCTC
Tnfα	21926	FW	5′-AACTAGTGGTGCCAGCCGAT	56
		RV	5′-CTTCACAGAGCAATGACTCC
Il-1β	16176	FW	5′-TGATGAGAATGACCTCTTCT	55
		RV	5′-CTTCTTCAAAGATGAAGGAAA
Il-6	16193	FW	5′-TAGTCCTTCCTACCCCAATTTCC	60
		RV	5′-TTGGTCCTTAGCCACTCCTTC
Il-2	16183	FW	5′-TGATGGACCTACAGGAGCTCCTGA	60
		RV	5′-GAGTCAAATCCACAACATGCC
Il-10	16153	FW	5′-TCCTTAATGCAGGACTTTAAGGG	56
		RV	5′-GGTCTTGGAGCTTATTAAAAT
Il-4	16189	FW	5′-AGCTAGTTGTCATCCTGCTC	53
		RV	5′-AGTGATGTGGACTTGGACTC
Gm-csf	16981	FW	5′-CTACTACCAGACATACTGCC	51
		RV	5′-GCATTCAAAGGGATATCAG
miR-142-3p		FW	5′-UGUAGUGUUUCCUACUUUAUGGA	55
miR-150-5p		FW	5′-UCUCCCAACCCUUGUACCAGUG	55
miR-155-5p		FW	5′-UUAAUGCUAAUUGUGAUAGGGGU	55
miR-375-3p		FW	5′-UUUGUUCGUUCGGCUCGCGUGA	55

Bacterial DNA pyrosequencing and analysis

DNA from faecal content was isolated using phenol : chloroform extraction and ethanol purification (Sambrook and Russell, 2006). 16S rRNA gene sequence recovery and integrity was PCR amplified using primers targeting regions flanking the variable regions 1 through 3 of the bacterial 16S rRNA gene (V1–V3), gel purified and analysed using the 454/Roche GS Titanium technology (Roche Diagnostics, Branford, CT, USA). The amplification of a 600-bp sequence in the variable region V1–V3 of the 16S rRNA gene was performed using barcoded primers. PCR was performed in a total volume of 15 μL for each sample containing the universal 27F and Bif16S-F forward primers (10 μmol·L⁻¹) at a 9:1 ratio, respectively, and the barcoded universal reverse primer 534R (10 μmol·L⁻¹) in addition to dNTP mix (10 mmol·L⁻¹), FastStart 10× buffer with 18 mmol·L⁻¹ of MgCl2, FastStart HiFi polymerase (5 U in 1 mL), and 2 μL of genomic DNA. The dNTP mix, FastStart 10× buffer with MgCl₂ and FastStart HiFi polymerase were included in a FastStart High Fidelity PCR System, dNTP Pack (Roche Applied Science, Penzberg, Germany). The PCR conditions were as follows: 95°C for 2 min, 30 cycles of 95°C for 20 s, 56°C for 30 s, 72°C for 5 min and final step at 4°C. After PCR, amplicons were further purified using AMPure XP beads (Beckman Coulter, Inc., Indianapolis, IN, USA) to remove smaller fragments. DNA concentration and quality were measured using a Quant-iT™ PicoGreen^® dsDNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Finally, the PCR amplicons were combined in equimolar ratios to create a DNA pool (10⁹ DNA molecules) that was used for clonal amplification (emPCR) and pyrosequencing according to the manufacturer's instructions.

The reads obtained from 16S ribosomal DNA sequencing were scored for quality, and any poor quality and short reads were removed. Sequences were trimmed to remove barcodes, primers, chimeras, plasmids, mitochondrial DNA and any non-16S bacterial reads and sequences <150 bp. MG-RAST (metagenomics analysis server) was used to analyse the sequences and make taxonomic assignments with Ribosomal Database Project. Operational taxonomic units (OTUs) were obtained with minimum e-value of 1e-5, minimum alignment length of 15 bp and minimum identity threshold was set at 95%. The relative abundance of OTUs of each sample was calculated on the output file and used for subsequent analysis, including the determination of ecological parameters indicative of α- and β-diversity, determined using Statistical Analysis of Metagenomic Profiles software package version 2.1.3.

Cell isolation and flow cytometry analysis

Cells from the colonic lamina propria (cLP) were isolated as described previously (Scott et al., 2017) using a digestion media composed of HBSS without Mg²⁺ or Ca²⁺, 10% of FBS and 0.5 mg·mL⁻¹ collagenase V (Sigma-Aldrich Quimica, Madrid, Spain), 0.65 mg·mL⁻¹ collagenase D, 30 μg·mL⁻¹ DNase I and 1 mg·mL⁻¹ dispase II (all Roche Diagnostics GmbH, Mannheim, Germany). Blood (300 μL) was collected, and red blood lysis was performed as needed. Surface-staining antibodies were added to the cell suspension together with a viability stain (Invitrogen, Carlsbad, CA, United States) and FcR blocking reagent (Miltenyi, Pozuelo de Alarcón, Madrid, Spain) for 20 min at 4°C. For intracellular cytokine expression, cells were pre-stimulated with PMA (50 ng·mL⁻¹) and ionomycin (1 μg·mL⁻¹) (Sigma-Aldrich) in the presence of GolgiPlug™ (eBioscience, Thermofisher, Carlsbad, CA, USA) for 4.5 h, at 37°C. For intracellular staining of cytokines and transcription factors, cells were fixed in Fixation/Permeabilization buffer (FoxP3 Staining Kit, eBioscience), and antibodies were added following the manufacturer's instructions. Antibodies were from Miltenyi unless otherwise stated: α-mouse CD45 (30F11), α-human/α-mouse CD11b (M1/70.15.11.5), α-mouse Ly6G (REA526), α-mouse SiglecF (REA798), α-mouse MHCII (M5/114.15.2), α-mouse Ly6C (1G7.G10), α-mouse CD103 (2E7), α-mouse CD11c (N418), α-mouse F4/80 (REA126), α-mouse B220 (RA3-6B2, BD Bioscience, San Jose, CA, United States), α-mouse CD3 (17A2), α-mouse CD8 (53–6.7), α-mouse CD4 (RM4–5, BD Bioscience), α-mouse IL-4 (BVD4-1D11), α-mouse IFNγ (XMG1.2, BD Pharmigen), α-mouse IL-17A (eBio17B7, eBioscience) and α-mouse FoxP3 (FJK-16s, eBioscience). Samples were acquired using a FACSVerse™ or FACSCanto II™ cytometers (Becton Dickinson, USA), and data were analysed using FlowJo software (Tree Star, USA). Percentages of the different populations referring to live cells were multiplied by the total count to provide the total number of each population.

Data and statistical analysis

The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2015). Statistical significance was only evaluated in data sets with n ≥ 5 with one way ANOVA and post hoc Tukey's multiple comparison tests. For the microbial analysis, reduced sample size (n = 4) had to be reported in the groups with lower therapeutic effect [rifaximin (RFX), tetracycline (TTC) and dexamethasone (DEX)] due to loss of sample recovery, owing to the diarrheic effect and lack of amplification. Therefore, to comply with the guidance for publication in BJP, statistical analysis was not applied for the microbial data sets. Survival curves were analysed with the Gehan–Breslow–Wilcoxon test. Non-parametric data were analysed using the Mann–Whitney U-test. All statistical analyses were carried out with the Statgraphics 5.0 software package (STSC, Maryland), with statistical significance set at P < 0.05.

Materials

All chemicals were obtained from Sigma-Aldrich Quimica (Madrid, Spain), unless otherwise stated. Drug doses used in mice were equivalent to the therapeutic dose in humans.

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a, 2017b).

Results

Immunomodulatory tetracyclines have a dual effect on macrophages in vitro

The immunomodulatory activity of different tetracyclines was initially compared in LPS-activated RAW246 macrophages. Tigecycline (TGC) was the most potent inhibitor of NO production, followed by minocycline (MNC) and doxycycline (DXC), although the activity of DEX was stronger. In contrast, no significant inhibitory effect was observed for TTC or RFX at the viable concentrations assayed (Figure 1A). Then, the immunomodulatory tetracyclines were evaluated in BMDM, to characterize the dual anti-/pro-inflammatory activity described in this cell type (Kloppenburg et al., 1996; Dunston et al., 2011; Bonjoch et al., 2015). LPS activation of BMDM induced the expression of Nos2 and the release of IL-1β, IL-6 and TNFα. DEX reduced all these markers, whereas the immunomodulatory tetracyclines reduced Nos2 mRNA levels but potentiated cytokine release (Figure 1B–E).

Figure 1

Comparative study of the effects of RFX, TTC, DXC, MNC, TGC and DEX on MΦ activity in vitro. (A) Nitrite production by LPS-stimulated RAW 264 MΦs. Cells were incubated with the different treatments at the indicated concentrations for 24 h and then stimulated with LPS (100 ng·mL⁻¹) for 24 h. Nitrite concentration in the culture supernatant was measured by the Griess Assay. Data are expressed as mean ± SEM (n = 6). (B) Nos2 (Inos) mRNA expression quantified by real-time PCR and (C–E) cytokine concentration in the culture supernatant quantified by elisa, in LPS-stimulated (10 ng·mL⁻¹) BMDM after 24 h of pre-incubation with the different treatments (25 μM). Data expressed as mean ± SEM (n = 6). Fold increase is calculated versus unstimulated untreated cells. *P < 0.05 versus stimulated untreated cells.

Immunomodulatory tetracyclines ameliorate DSS-colitis, showing a better profile than rifaximin, tetracycline and dexamethasone

Initially, the compounds were assayed following a fatal colitis protocol in mice induced by administering 3% DSS for 6 days. Then, mice were treated for 6 days with (i) RFX (200 mg·kg⁻¹·day⁻¹), a non-absorbable antibiotic; (ii) TTC (250 mg·kg⁻¹·day⁻¹), included as reference for tetracyclines' antibiotic activity; (iii–v) immunomodulatory tetracyclines: DXC (25 mg·kg⁻¹·day⁻¹), MNC (50 mg·kg⁻¹·day⁻¹) and TGC (25 mg·kg⁻¹·day⁻¹); and (vi) DEX (2.4 mg·kg⁻¹·day⁻¹), included as reference of an anti-inflammatory drug without antibiotic activity. The administration of DXC, MNC and TGC significantly reduced the disease activity index (DAI; Figure 2A) from the first day of treatment and throughout the study; however, DEX-treated mice experienced an increase in DAI values after the third day of treatment, in contrast with the initial reduction observed. TTC-treated mice showed a slight improvement, although no significant differences were observed in comparison with DSS-control, and RFX did not show any beneficial effect. Moreover, a high mortality rate was experienced from day 8 in colitic mice, with only 30% of the animals surviving until the end of the study, and only those colitic groups treated with immunomodulatory tetracyclines showed a significantly increased survival rate (Figure 2B).

Figure 2

Comparative study of the intestinal anti-inflammatory effects of RFX, TTC, DXC, MNC, TGC and DEX in DSS-fatal colitis. NC: non-colitic group. (n = 10, DSS n = 20). (A) Schematic illustration of the experimental design and DAI values (means) assigned based on the criteria described in Table 1. (B) Survival curves (%) of the different groups during the 6 days treatment period and their P values versus DSS-control group.

Subsequently, the effect of these drugs was tested in a less aggressive colitis protocol, in which DSS intake (3%) was maintained for 5 days, followed by a 4-day period of treatment. The DAI progression followed a similar pattern: only the immunomodulatory tetracyclines significantly ameliorated DAI values (Figure 3A). Histological analysis showed that DSS-colitis mainly affected the mucosa, with epithelial ulceration in more than 70% of the colonic surface. Major histological alterations affected the crypt structure, with high mitotic activity, mucin depletion in goblet cells, the presence of oedema and intense inflammatory cell infiltration. Immunomodulatory tetracyclines significantly reduced the microscopic damage score, preserving the mucosal layer and restoring the presence of mucus-filled goblet cells. However, no histological improvement was observed in colitic mice treated with RFX, TTC or DEX (Figure 3B, C). The mucin depletion observed in DSS-control mice was associated with reduced expression of Muc-1, Muc-2 and Muc-3 and of the epithelial barrier integrity markers Zo-1 and Occludin. Importantly, their expression was improved in animals treated with immunomodulatory tetracyclines, which also showed an up-regulation of Tff-3 expression. In contrast, RFX and DEX treatments did not restore the expression of these protective markers, which appeared even further reduced in RFX-treated mice (Figure 3D).

Figure 3

Comparative study of the intestinal anti-inflammatory effects of RFX, TTC, DXC, MNC, TGC and DEX in the DSS model of mouse colitis. NC: non-colitic group. (A) Schematic illustration of the experimental design followed and DAI mean values assigned based on the criteria described in Table 1, during the 9 days experimental period (n = 8, DSS n = 14). (B) Microscopic damage score assigned according the criteria described in Tab2. (C) Representative histological sections of colonic mucosa of the different experimental groups stained with haematoxylin, eosin and alcian blue (40× magnification). (D) Colon mRNA expression quantified by real-time PCR. Fold increase calculated versus NC group. Box plot graphs represent ±SEM range, median (middle line) and extreme values (whiskers) (NC, DSS, DXC, MNC and TGC, n = 6; RFX, TTC and DEX, n = 5). *P < 0.05 versus DSS-control group.

When different inflammatory mediators were considered, DSS-induced colitis was linked to an increased expression of Tnfα, Il-1β, Il-6, Mmp-9, Ccl2 and Cxcl2 (Figure 4). The treatment with immunomodulatory tetracyclines significantly reduced Il-1β, Il-6, Mmp-9 and Cxcl2 expression, while the other drugs showed no effect. Strikingly, Ccl2 expression was strongly potentiated in mice treated with DXC, MNC and TGC and to a lesser extent with TTC. Recent studies have highlighted the role of microRNAs in the regulation of intestinal inflammation (Biton et al., 2011; Pekow and Kwon, 2012). In our study, DSS-colitis induced a significant up-regulation of miR-142, miR-150 and miR-155 (Figure 4). The tetracyclines and DEX reduced miR-150 and miR-155 expression, which have been associated with T helper cell and humoral responses (Monticelli et al., 2005) and the NF-κB pathway (Tili et al., 2007) respectively. The administration of tetracyclines to colitic mice resulted in an up-regulation of miR-375, which aligns with increased Tff-3 expression, both related to goblet cell function (Biton et al., 2011). MiR-142, preferentially expressed in immune cells (Kramer et al., 2015), was the most up-regulated miRNA upon colitis induction, and strikingly, its expression was further increased in mice treated with the immunomodulatory tetracyclines, an effect similar to the one observed on Ccl2 expression. In addition, the colonic inflammatory process has been associated with changes in microbial sensing through Toll-like receptors (TLRs; Franchimont et al., 2004). We observed a significant reduction in Tlr2 and Tlr4 expression in DSS-colitic mice; while Tlr2 levels were restored by all antibiotics, Tlr4, highly expressed by enterocytes and required to preserve barrier function and promote its repair (Franchimont et al., 2004; Fukata et al., 2005), was significantly up-regulated by immunomodulatory tetracyclines (Figure 4).

Figure 4

Comparative study of the intestinal anti-inflammatory effects of RFX, TTC, DXC, MNC, TGC and DEX in the DSS model of mouse colitis. NC: non-colitic group. Colon mRNA expression quantified by real-time PCR. Fold increase calculated versus NC group. Box plot graphs represent ±SEM range, median (middle line) and extreme values (whiskers) (NC, DSS, DXC, MNC and TGC, n = 6; RFX, TTC and DEX, n = 5). *P < 0.05 versus DSS-control group.

When considering the colonic microbiota, while no statistical analysis could be applied due to the reduced sample size in some data sets, a different impact on the microbial composition could be identified among different groups. Despite no differences being observed in α-diversity at this time point (Table 4), inner taxonomic groups showed a higher degree of variation, as has been demonstrated at early stages of intestinal inflammation (Schwab et al., 2014). Bacteroidetes abundance was reduced in DSS-control compared to healthy mice, while Firmicutes abundance increased. Antibiotic administration to colitic mice counteracted these changes, while DEX treatment showed a minor effect (Figure 5A). Order level heatmap and clustering analysis illustrates these results: based on their different composition, mainly of Bacteroidales, DSS- and DEX-treated animals cluster in a separated branch from antibiotic-treated and healthy mice (Figure 5B). However, analysis at lower taxonomic levels revealed that antibiotic treatment showed a divergent impact. PCA at genus level delimitated three different clusters with different microbiota compositions: NC control, DSS- and DEX-treated colitic mice, and antibiotic-treated groups. No major differences were observed among the antibiotics, which exerted a higher impact on the microbiota composition (distributed along PC1 axis, which explains 39.4% of the variance) than the colitis itself (separated in PC2, accounting for 16.2% of variability) (Figure 5C). As an example of this divergence, the colitis-associated reduction in Bacteroidetes included families such as Porphyromonaceae and Prevotellaceae, while the impact of antibiotics was greater within Bacteroidaceae, mainly due to an increase in the abundance of Bacteroides acidifaciens. Within the Phylum Firmicutes, antibiotics counteracted the increase in Bacilli class observed in colitis and, within it, only MNC and TGC significantly reduced the Lactobacillaceae family (Figure 5A).

Table 4. α-Diversity measures of intestinal microbiota

INDEX	Margalef	Chao1	1-Simpson	Shannon	Pielou
NC	10.1 ± 2.02	115. ± 20.2	0.85 ± 0.04	2.70 ± 0.25	0.61 ± 0.04
DSS	8.5 ± 2.50	104.8 ± 33.2	0.77 ± 0.08	2.27 ± 0.32	0.55 ± 0.06
RFX	6.1 ± 0.31	59.1 ± 3.2	0.89 ± 0.01	2.64 ± 0.11	0.67 ± 0.03
TTC	5.6 ± 1.48	67.2 ± 15.6	0.81 ± 0.07	2.16 ± 0.34	0.55 ± 0.04
DXC	7.1 ± 2.01	83.5 ± 21.9	0.83 ± 0.07	2.43 ± 0.35	0.61 ± 0.06
MNC	5.4 ± 1.21	73.9 ± 23.0	0.69 ± 0.06	1.77 ± 0.26	0.46 ± 0.04
TGC	6.0 ± 1.36	81.8 ± 18.2	0.76 ± 0.11	2.11 ± 0.40	0.53 ± 0.07
DEX	10.0 ± 2.79	117.4 ± 31.3	0.92 ± 0.02	2.97 ± 0.22	0.67 ± 0.04

• Comparison of α-diversity measures of intestinal microbiota between non-colitic (NC) group (n = 8), DSS-colitic group (n = 8), and RFX (n = 4), TTC (n = 4), DXC (n = 5), MNC (n = 5), TGC (n = 5) and DEX (n = 4) treated groups in the DSS model of mouse colitis. Data expressed as means ± SEM.

Figure 5

Comparison of microbiota composition between non-colitic group (NC) (n = 8), DSS-colitic group (n = 8), and RFX (n = 4), TTC (n = 4), DXC (n = 5), MNC (n = 5), TGC (n = 5) and DEX (n = 4) treated groups in the DSS model of mouse colitis. (A) Relative abundance of various taxonomic groups. Data expressed as means ± SEM. (B) Heatmap with relative abundance of the 10 most abundant orders, include hierarchical clustering of samples based on order level composition analysed with the method of minimum variance of Ward. (C) PCA plot representation based on the ordination of the distance matrix build with a dissimilarity analysis at genus level using the taxon-based Bray–Curtis complementary algorithm. Green ellipse includes NC samples, red ellipse includes DSS samples and purple ellipse includes sample from antibiotic-treated groups.

Minocycline potentiates the early inflammatory response and promotes mucosal healing and resolution in DSS-colitis

Considering their effects on Mφs in vitro and the different immunomodulatory activity observed for tetracyclines in vivo, particularly the up-regulation of Ccl2 and miR-142 associated with the generation of type 2 immunity (Gu et al., 2000; Belz, 2013), we analysed the effects of MNC on the initial immunological events of the intestinal inflammatory process. Once colitis was established after 5 days of 3% DSS treatment, mice were treated with MNC (50 mg·kg⁻¹) for 2 days; at this time point, the colonic inflammatory status was evaluated biochemically, and circulating and cLP immune populations were isolated and analysed by flow cytometry. No major differences were observed in blood leukocytes between NC and DSS-treated colitic mice; however, a strong increase in circulating neutrophils, eosinophils and monocytic myeloid cells was observed in MNC-treated animals (Figure 6A). The analysis of cLP immune cells showed clear differences between healthy and colitic mice, with B cells, CD4⁺T cells and neutrophils, being raised in the latter. In particular, inflammatory Mφs (Ly6C⁺MHCII⁻) and FoxP3⁺Tregs accumulated in the colon of colitic mice. We did not detect major changes in cLP immune populations at this time point. However, gene expression and cytokine production analysis in colonic tissue showed important changes related to MNC treatment. As opposed to what was found at later time points (Figure 4), the characteristic higher production of IL-1β and IL-6 in DSS-control mice was further increased in MNC-treated mice. Additionally, other inflammatory mediators were also up-regulated in the MNC-treated group in comparison with DSS-controls, such as Il-10, Il-2, Ccl2 and Ccl11 expressions, and IL-4, GM-CSF and IL-22 concentrations in the supernatant of colonic explants from MNC-treated mice were similarly increased (Figure 5C,D).

Figure 6

Evaluation of the effects of 2 days of minocycline (MNC) treatment on the immune response in DSS-colitic mice. NC: non-colitic group. (A) % of the indicated immune cell populations in the blood of the different experimental groups. (B) Analysis of immune cell populations in the cLP of the different experimental groups. Absolute cell numbers of B cells (B220⁺), CD4⁺T cells (CD3⁺), Tregs (CD3⁺CD4⁺FoxP3⁺), neutrophils (Ly6G⁺) and Ly6C⁺MHCII⁻ Mφs. Scatter plots represent individual values (dots) and mean ± SEM. (C) Colon mRNA expression quantified by real-time PCR. Fold increase calculated versus NC group. (D) Cytokine concentration in the supernatant of colonic explant cultures quantified by elisa. Box plot graphs represent ±SEM range, median (middle line) and extreme values (whiskers) (n = 5). *P < 0.05 versus DSS-control group, ^#P > 0.05 versus NC control group.

Then, the effects of MNC were characterized after 4 days of treatment, when MNC intestinal anti-inflammatory activity was fully displayed, with lower DAI values and marked histological improvement (Figure 7A,B). At this time point, the systemic immune response in colitic mice was characterized by an increase in circulating myeloid cells, particularly neutrophils. Interestingly, MNC treatment significantly reduced the neutrophilia, while myeloid monocytic cells and eosinophils were still elevated in this group (Figure 7C,D).

Figure 7

Evaluation of the effects of 4 days of minocycline (MNC) treatment on the immune response in DSS-colitic mice. (A) Schematic illustration of the experimental design followed and DAI values (means ± SEM) over the 9 day experimental period, calculated based on the criteria described in Table 1. NC: non-colitic group (50 mg·kg⁻¹·day⁻¹) (n = 7). (B) Histological sections of colonic mucosa stained with haematoxylin, eosin and alcian blue (40× magnification) and microscopic damage score assigned according the criteria described in Table 2. Box plot graph represents ±SEM range, median (middle line) and extreme values (whiskers). (C) Representative flow cytometry analysis of circulating CD45⁺ cells. (D) % of the indicated cell populations within the CD45⁺ cells present in the blood of NC, DSS and MNC mice. Scatter plots represent individual values (dots) and mean ± SEM. (n = 7, DSS n = 6) *P < 0.05 versus DSS-control group, ^#P > 0.05 versus NC control group.

Flow cytometry analysis of cLP leukocytes showed that the CD45⁺ cell number was slightly higher in MNC-treated group than in DSS-control, mainly associated with increased presence of CD11b⁺ myeloid cells (Figure 8A,B). Among them, MNC treatment significantly reduced neutrophils while it increased the numbers of eosinophils and monocytic myeloid cells (Figure 8C, D). These findings align respectively with increased Ccl11 and Ccl2 transcripts detected in the colon of mice after 2 days of MNC treatment (Figure 6C). The phenotype of cLP Mφs (CD11b⁺Ly6G⁻SSC^loF4/80⁺) and dendritic cells (DCs) (SSC^loF4/80⁻CD11c^hiMHC⁺) was further characterized, confirming that MNC-treated mice presented an increased number of Mφs and DCs in the cLP (Figure 8E,H). The monocyte-Mφ differentiation waterfall (Figure 8F) illustrates the accumulation of the initial Ly6C^hi population in the inflamed intestine. Despite the finding that MNC-treated mice had higher numbers of Mφs in the cLP than the DSS-controls, both groups had similar numbers of inflammatory Mφs, while the intermediate and tissue-resident Mφ populations were significantly increased in MNC-treated mice (Figure 8G). Among intestinal DCs, DSS-colitis induced their polarization towards the CD11b⁺CD103⁺ phenotype (Figure 8I), the main migratory population. MNC treatment resulted in an increase in the total number of DCs in the cLP without altering the polarization of DCs (Figure 8H). When considering the lymphoid compartment, a strong B cell infiltrate was observed in colitic mice, which was not modified by MNC treatment. Within the CD3⁺lymphocyte compartment, no differences were observed in CD8⁺T cells numbers among the different groups (Figure 9A). However, the number of cLP CD4⁺T cells, and particularly of IL-17A⁺ and Foxp3⁺ CD4⁺T helper cells, was higher in colitic mice than in healthy controls, and these appeared further increased in MNC-treated mice (Figure 9A). As observed before, the production of IL-22, a Th17-related cytokine, was increased in colonic explants from colitic animals, being even higher in MNC-treated colitic mice (Figure 9B). Higher numbers of IL-4-producing Th2 lymphocytes were found in the cLP of MNC-treated mice, while no differences were observed among NC and DSS groups (Figure 9A). Additionally, colonic explants from MNC-treated mice produced higher levels of IL-4 than NC and DSS groups (Figure 9B). Interestingly, increased numbers and percentages of IL-4⁺IL-17A⁺ and IL-17A⁺FoxP3⁺ double positive CD4⁺T cells were also found in the cLP of MNC-treated mice when compared to control groups (Figure 9A), which may suggest a higher degree of plasticity between these T cell subsets after MNC treatment (Gagliani et al., 2015). In addition, and in contrast to what was observed after 2 days of treatment, IL-1β and IL-6 cytokine release by colonic explant cultures from the MNC-treated group was now reduced in comparison to the DSS-control (Figure 9B). Since eosinophils, Th2 cells and alternatively activated Mφs are actively associated with the resolution phase of acute inflammation, and considering the higher numbers of these cells found in the cLP of MNC-treated mice, we evaluated the expression of Alox15, which encodes for the enzyme 12/15-lipoxygenase, involved in the synthesis of pro-resolving lipid mediators (Wang and Colgan, 2017). Interestingly, Alox15 expression was significantly up-regulated in the colonic tissue of the MNC-treated group compared to the DSS-control (Figure 9C).

Figure 8

Evaluation of the effects of 4 days of minocycline (MNC) treatment on the cLP immune response during DSS-colitis. NC: non-colitic group (50 mg·kg⁻¹·day⁻¹) (n = 7, DSS n = 6). (A) Representative flow cytometry analysis of live cells from the cLP showing the CD11b⁺ and CD45⁺ cell populations. (B) Absolute cell numbers of immune cells (CD45⁺) and CD45⁺CD11b⁺ myeloid cells. (C) Representative flow cytometry analysis of CD11b⁺ cells from the cLP showing the Ly6G⁺ and SiglecF⁺ populations. (D) Absolute cell numbers of: neutrophils (Ly6G⁺), eosinophils (SiglecF⁺) and myeloid monocytic cells (CD11b⁺ Ly6G⁻ SiglecF⁻ SSC^lo). (E) Total number of MΦs (CD11b⁺Ly6G⁻SSC^loF4/80⁺ cells). (F) Representative flow cytometry plots showing the expression of Ly6G and MHCII by MΦs from the cLP and illustrating the monocyte-MΦ waterfall. (G) Percentage (left) and absolute cell numbers (right) of inflammatory MΦs (Ly6C⁺MHCII⁻ cells), intermediate MΦ population (Ly6C⁺MHCII⁺ cells) and resident intestinal MΦs (Ly6C⁻MHCII⁺ cells). (H) Absolute cell numbers of DCs (Ly6G⁻SSC^loF4/80⁻CD11c^hiMHCII⁺) and % of CD11b⁺CD103⁺ DCs. (I) Representative flow cytometry analysis of DCs from the cLP showing the expression of CD103 and CD11b. Scatter plots represent individual values (dots) and mean ± SEM. *P < 0.05 versus DSS-control group, ^#P > 0.05 versus NC control group.

Figure 9

Evaluation of the effects of 4 days of minocycline (MNC) treatment on the immune response in the cLP during DSS-colitis. NC: non-colitic group (50 mg·kg⁻¹·day⁻¹) (n = 7, DSS n = 6). (A) Absolute cell numbers of lymphocyte populations. Scatter plots represent individual values (dots) and mean ± SEM. (B) Cytokine concentration in the culture supernatant of colonic explants quantified by elisa. (C) Colonic mRNA expression of Alox15 quantified by real-time PCR. Box plot graphs represent ±SEM range, median (middle line) and extreme values (whiskers). *P < 0.05 versus DSS-control group, ^#P > 0.05 versus NC control group.

Discussion

Following previous reports describing a beneficial anti-inflammatory activity obtained with minocycline and doxycycline in preclinical models of intestinal inflammation (Huang et al., 2009a,b; Garrido-Mesa et al., 2011a,b, 2015), we aimed to further investigate the potential of this interesting therapeutic family of immunomodulatory antibiotics. We identified an important link between the effect of immunomodulatory tetracyclines and the activation of specific inflammatory pathways leading to the resolution of inflammation, which supports the potential of these molecules as organ protective agents (Griffin et al., 2010, 2011). Moreover, this study also constitutes the first description of the intestinal anti-inflammatory activity of tigecycline. Our results suggest that the antibiotic activity per se does not exert a significant contribution to the anti-inflammatory effect of tetracyclines in this model of colitis, since all antibiotics had a similar impact on microbiota but no beneficial effect was observed with RFX or TTC. However, immunomodulatory tetracyclines have demonstrated a prompt effect, driving a strong improvement in the epithelial barrier integrity and reducing the colitis-associated mortality rate. These findings support the idea that the activation of innate immune protection, as opposed to the immune inhibition caused by dexamethasone, could in fact constitute an advantage in the treatment of intestinal inflammation. In particular, considering the effects displayed by tetracyclines on macrophages in vitro, and the up-regulation of Ccl2 in the colonic tissue of tetracycline-treated colitic mice, we propose that a potentiation of the MΦ response might underlie their anti-inflammatory effect. Although sustainably elevated cytokine levels may perpetuate the inflammatory process, an adequate initial inflammatory response is required for an effective recovery. In fact, GWAS have shown that an immune deficit underlies the pathogenesis of inflammatory bowel disease (IBD) (Lees et al., 2011), characterized by diminished cytokine production by monocytes and an impaired ability of the inflammatory response to restore intestinal homeostasis, as reported in Crohn's disease patients (Marks, 2011). Of note, immunomodulatory tetracyclines potentiated innate cytokine release by LPS-activated BMDM, a mechanism that contributes to clear bacterial infection and promote epithelial barrier protection (Wittkopf et al., 2015). This hypothesis was confirmed in vivo when we observed increased innate cytokine release after 2 days of MNC treatment.

MΦ production of IL-1β induces cytokine release by innate lymphoid cells (ILC)-3 (Mortha et al., 2014), which are the initial source of IL-22 upon mucosal damage (Sanos et al., 2009) and the major source of GM-CSF in the gut (Mortha et al., 2014). Considering this, and in view of the enhanced levels of the aforementioned cytokines in MNC-treated mice, the initial events mediating its anti-inflammatory effects may include the promotion of the MΦ response and their crosstalk with ILCs. Systemically, we observed higher numbers of circulating myeloid cells after 2 days of MNC treatment, associated with higher levels of GM-CSF, IL-6 and IL-1β, important players in emergency myelopoiesis (Root and Dale, 1999; Hsu et al., 2011). Locally, GM-CSF increases Ccl2 expression and contributes to the maintenance of the MΦ, DC and Treg populations (Tanimoto et al., 2008; Mortha et al., 2014). IL-22 also has a crucial role in host defence and tissue recovery inducing epithelial proliferation, repair and production of protective molecules, such as mucins, IL-10 and the ‘alarmins’ IL-25, IL-33 and TSLP (thymic stromal lymphopoietin; Nagalakshmi et al., 2004; Zheng et al., 2008; Lindemans et al., 2015). Additionally, we detected increased IL-2 expression in MNC-treated animals, which may be related to Treg expansion and IL-10 production (Barthlott et al., 2005), as well as to the potentiation of ILC function, inducing IL-22 expression in ILC3s (Crellin et al., 2010) and, together with ‘alarmins’, driving ILC2 activation and type 2 immune pathways (Roediger et al., 2015; Halim et al., 2016). These favour mucosal protection as opposed to the detrimental immune response observed in chronic inflammatory disorders.

The protective consequences of these immune changes are reflected in the effects observed for tetracyclines at later time points, such as the increase in miR-375, Tff-3 and Tlr4 expression, recovered goblet cell function and improved epithelial barrier integrity (Biton et al., 2011). Subsequent to Ccl11 and Ccl2 up-regulation, eosinophils, macrophages and DCs accumulated in the cLP of MNC-treated mice. Eosinophil activity attenuates experimental colitis (Masterson et al., 2015), and increased eosinophils have been found during the remission phase of ulcerative colitis (Lampinen et al., 2005). MNC treatment promoted monocyte recruitment but also their differentiation into Ly6C⁻MHCII⁺ Mφs despite the surrounding inflammatory conditions. GM-CSF, IL-10 and IL-4 have been demonstrated to promote the polarization of inflammatory MΦs towards the homeostatic and alternatively activated Mφ phenotypes, implicated in bacterial and apoptotic cell clearance and supporting local regulatory responses and mucosal healing (Hunter et al., 2010; Bain et al., 2013). Similarly, MNC promoted DCs recruitment and the increase in migratory DCs (CD11b⁺CD103⁺) correlated with an increase in CD4⁺T cell priming in this group. Specificlly, higher numbers of Tregs, Th17 and Th2 subsets were present in cLP of MNC-treated mice. Although Th17 cells have initial protective functions in intestinal inflammation, exacerbated Th17 responses can lead to perpetuated inflammation and tissue damage. Of note, Th17 cells can differentiate into Treg cells during the resolution of inflammation (Gagliani et al., 2015), as well as into the Th2 subset in response to IL-4 (Lee et al., 2009), and MNC treatment promoted a higher degree of plasticity among these T cell subsets, particularly between Th17 and Th2. Additionally, CCL2 enhances IL-4 secretion by T cells and elicits Th2 polarising effects, and miR-142 has an important role in DC priming of Th2 responses (Gu et al., 2000; Belz, 2013), both up-regulated in MNC-treated mice.

Alternative therapeutic strategies that exploit counter-regulatory pathways, such as parasites used to skew mucosal immune responses and favour barrier protection, involve alternatively activated macrophages, eosinophils, Th2 cells and Tregs (Smith et al., 2007; Hunter et al., 2010; Gause et al., 2013; Driss et al., 2016). These cells play a key role in the resolution of intestinal inflammation, for example, by producing anti-inflammatory lipid mediators that activate this process (Sherman and Kalman, 2004; Wang and Colgan, 2017). Correlating with these immune changes, Alox15 was up-regulated in the MNC- treated group, suggesting the initiation of the resolution phase. In fact, at this time point, colonic IL-6 and IL-1β levels dropped, the number of cLP neutrophils was reduced and the efficacy of the treatment was evident both macroscopically and histologically.

All together, these results indicate that the pro-inflammatory actions of immunomodulatory tetracyclines in MΦs, rather than being detrimental, strongly contribute to mucosal protection. The benefits of DXC and MNC, previously reported in experimental colitis (Huang et al., 2009b; Garrido-Mesa et al., 2011a,b, 2015) and in a model of 5-FU induced intestinal mucositis (Huang et al., 2009a), were attributed to their antibiotic activity and other mechanisms such as inhibition of MMPs and antioxidant effects (Garrido-Mesa et al., 2013b). We have now demonstrated that immunomodulatory tetracyclines, by promoting the innate immune response, actively induce mucosal healing and lead to an accelerated resolution of the process. This mechanism represents the favourable effects of the inflammatory response, aimed at restoring tissue homeostasis (Sherman and Kalman, 2004; Rutgeerts et al., 2007). Even though therapies aimed at a specific target are of special interest in pharmacology due to the rationale to avoid side effects, there is increasing awareness of their lack of efficacy in complex pathologies such as IBD, due to counter-regulatory pathways that sustain inflammation (Biancheri et al., 2013). By contrast, the high therapeutic benefit observed with tetracycines in preclinical models of IBD and other multifactorial diseases might be precisely related to their pleiotropic properties, influencing different factors involved in the inflammatory response (Griffin et al., 2010; Garrido-Mesa et al., 2013a). The benefit of their non-antibiotic properties has already proved to be of clinical relevance, so it is reasonable to believe that tetracyclines can be repurposed for other non-infectious pathologies in the future, and we hope this report will contribute to the highlighting of these interesting drugs. This, together with their well-known and safe profile, makes them very promising candidates for future translational studies into human disease. Similarly, further research into multi-target drugs and ways of exploiting pro-resolving pathways warrants interesting results in complex chronic pathologies (Medina-Franco et al., 2013).

Considering the epidemiological association of antibiotic intake and IBD development (Ungaro et al., 2014), the use of antibiotics seems to be discouraged. However, the disruption caused by antibiotics is particularly relevant earlier in life (Miyoshi et al., 2017; Örtqvist et al., 2018), and when used to treat IBD patients, it has not been reported that antibiotics worsen or perpetuate the disease. We previously reported that, when taken in a preventative manner, no significant differences were observed, suggesting that the modification of the microbiota in this case had neither deleterious nor beneficial effects (Garrido-Mesa et al., 2011a). On this basis, here, we proposed a curative treatment (once inflammation is established). But considering our new findings, indicating that the microbiota composition was not restored in antibiotic-treated groups, and the risks of dysbiosis associated with antibiotic intake, for example, leading to C. difficile infection (Owens et al., 2008), the long-term administration of tetracyclines in this context might seem to be discouraged as well. Of note, their use has not been associated with increased C. difficile infection when compared to other antibiotics (Deshpande et al., 2013). With this in mind, the recently developed chemically modified tetracyclines, devoid of antibiotic activity, offer a promising tool to explore the potential of the immunomodulatory properties of tetracyclines, allowing for long-term therapy if needed (Lokeshwar, 2011). However, the mechanism that we describe here suggests that tetracyclines would be best indicated as short-term treatment to induce remission of acute episodes. Acute intestinal inflammation is the second leading cause of death worldwide, because of the high mortality rates found in developing countries (Liu et al., 2012), and it also entails significant costs to developed societies (Glass et al., 2014). Thus, by potentiating host-protective pathways, immunomodulatory tetracyclines become a promising strategy for the treatment of acute intestinal inflammation. Whether tetracyclines can induce remission on a chronified pathology, where the protective effect of innate immunity is overridden, or whether a long-term therapy could be applied are questions that require further investigation in a chronic setting. Nonetheless, considering the defects in innate protective mechanisms that underlie IBD (Marks, 2011) and that the majority of IBD patients experience relapsing acute inflammatory flares (Baumgart and Carding, 2007), therapeutic approaches that are aimed at restoring these protective mechanisms are the next logical approach in IBD and our findings along with the current understanding of the disease. The effects described for tetracyclines in this study could benefit the course of the pathology, used as short-term treatment to promote the resolution of the inflammatory flares. Once induced, remission could be maintained by additional therapies with fewer side effects, such as probiotics. Indeed, we have already evaluated the potential of this strategy in previous studies, where minocycline or doxycycline administration was followed by long-term maintenance treatment with probiotics, and showed very promising results (Garrido-Mesa et al., 2011b, 2015).

In conclusion, the pharmacological treatments available for IBD are mainly aimed at reducing the symptoms of inflammation under an ‘acceptable’ threshold, but they have not succeeded at modifying the course of the disease, and IBD patients without a robust mucosal healing have worse outcomes (Baert et al., 2010). Thus, by targeting the immune system, current pharmacological treatments may interfere with the natural protective pathways activated by the inflammatory response. We have now generated solid evidence for the benefit of strengthening these defensive mechanisms with the administration of immunomodulatory tetracyclines. This adds to the broad range of promising properties exerted by this safe and well-known family of compounds, offering an appealing drug-reposition strategy to manage intestinal inflammatory conditions.

Acknowledgements

The authors thank Dr Gustavo Ortíz Ferrón and staff of the flow cytometry core facility of University of Granada for technical assistance. We are grateful to Dr Desiré Camuesco Perez for suggestions and critically discussing the results. We acknowledge Louise Mary Topping for English language editing of this manuscript. This work was supported by the Junta de Andalucía (CTS 164) and by the Spanish Ministry of Economy and Competitiveness (SAF2011-29648 and AGL2015-67995-C3-3-R) with funds from the European Union. The CIBER-EHD and the Red de Investigación en SIDA are funded by the Instituto de Salud Carlos III.

Author contributions

G.-M.J., R.-N.A., A.F., V.T., H.-G.L., G.-B.M., R.-C.M.E. and U.M.P. performed the experiments and contributed to the acquisition and analysis of data; G.-M.J., R.-N.A., G.F. and C.N. contributed to the taxonomic analysis and data interpretation; G.-M.J., G.-M.N. and G.J. designed the experiments, performed the analysis of data and wrote the manuscript. The funders had no role in study design, data collection and analysis.

Conflict of interest

The authors declare no conflicts of interest.

Declaration of transparency and scientific rigour

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research recommended by funding agencies, publishers and other organisations engaged with supporting research.