Spectrum 00831-23
Spectrum 00831-23
Spectrum 00831-23
a Microbiome and Host Health Program, South Australian Health and Medical Research Institute, Adelaide, South Australia, Australia
b Infection and Immunity, Flinders Health and Medical Research Institute, College of Medicine and Public Health, Flinders University, Bedford Park, South Australia,
Australia
c Flinders Health and Medical Research Institute, College of Medicine and Public Health, Flinders University, Adelaide, South Australia, Australia
d Institute of Pharmaceutical Science, School of Cancer & Pharmaceutical Sciences, King’s College London, London, United Kingdom
e Department of Respiratory and Sleep Medicine, Mater Adult Hospital, Brisbane, Queensland, Australia
Respiratory and Infectious Disease Research Group, Mater Research Institute, Brisbane, Queensland, Australia
f
g Nutrition & Metabolism, South Australian Health and Medical Research Institute, Adelaide, South Australia, Australia
ABSTRACT Long-term low-dose macrolide therapy is now widely used in the treat-
ment of chronic respiratory diseases for its immune-modulating effects, although the
antimicrobial properties of macrolides can also have collateral impacts on the gut micro-
biome. We investigated whether such treatment altered intestinal commensal microbiol-
ogy and whether any such changes affected systemic immune and metabolic regulation.
In healthy adults exposed to 4 weeks of low-dose erythromycin or azithromycin, as used
clinically, we observed consistent shifts in gut microbiome composition, with a reduction
in microbial capacity related to carbohydrate metabolism and short-chain fatty acid bio-
synthesis. These changes were accompanied by alterations in systemic biomarkers relat-
ing to immune (interleukin 5 [IL-5], IL-10, monocyte chemoattractant protein 1 [MCP-1])
and metabolic (serotonin [5-HT], C-peptide) homeostasis. Transplantation of erythromy-
well-tolerated, and is considered to carry relatively little risk, and is now widely used to
treat asthma, bronchiectasis, chronic obstructive pulmonary disease, bronchiolitis oblit-
erans, chronic rhinosinusitis, cystic fibrosis, organizing pneumonia, and diffuse pan-
bronchiolitis (1, 7). However, macrolides are known to have pleiotropic effects, and the
potential consequences of long-term continuous exposure to these agents, which can
span decades in some cases, is poorly understood.
The agents used in long-term macrolide therapy, such as azithromycin and erythro-
mycin, have the capacity to directly influence many aspects of host physiology, includ-
ing local and systemic immune regulation (8). Their ability to reduce cellular expression
of pro-inflammatory mediators has been demonstrated using in vitro and animal mod-
els (9). Moreover, through the modulation of intracellular mitogen-activated protein ki-
nase and NF-k B pathways, these effects occur across multiple immune cell types (8,
10). Studies in macrolide recipients have similarly shown that macrolides dampen pro-
inflammatory airway responses through reduced neutrophilic inflammation in chronic
respiratory disease (11, 12), and influence circulating cytokine levels (13–15).
In addition to their influence on immune regulation, macrolide antibiotics have antimi-
crobial activity against many Gram-positive and Gram-negative bacteria (8). Disruption of
the gut microbiome as a result of antibiotic exposure can have a profound impact on host
physiology and has been linked to an increased risk of cardiometabolic conditions, includ-
ing cardiovascular disease, nonalcoholic fatty liver disease, type 2 diabetes, and obesity
(16). The gut microbiota and its metabolites can modulate host metabolic regulation by
activating enteric cells and promote the release of gut-derived peptides such as serotonin
(5-HT) and glucagon-like peptide 1 (GLP-1) (17). These molecules are involved in signaling
the enteric nervous system (ENS) to influence gastrointestinal physiological functions such
as motility and nutrient absorption, which play an important role in maintaining energy
and glucose homeostasis (17). It is therefore possible that long-term macrolide treatment
used in chronic lung disease could influence diverse health outcomes, either directly or
indirectly via alteration of the gut microbiome.
Changes in intestinal microbiology and microbiome-host interaction are well-recognized
consequences of short-term exposure to antibiotics at antimicrobial dosages (18–20).
However, the extent to which long-term, low-dose macrolide therapy alters the gut micro-
biome or influences aspects of host physiology is not known. While addressing these knowl-
RESULTS
Low-dose, long-term macrolide exposure is associated with alterations in sys-
temic immune and metabolic markers. Healthy adults received low-dose erythromy-
cin (ERY) or azithromycin (AZM) for 4 weeks (n = 10 per antibiotic group) and effects
on systemic immune and metabolic markers were assessed. Because the immuno-
modulatory properties of macrolides are recognized, our initial analysis focused on
FIG 1 Changes in serum levels of (A) systemic immune markers and (B) metabolic biomarkers associated with glucose regulation, lipid metabolism Downloaded from https://journals.asm.org/journal/spectrum on 25 June 2023 by 103.28.157.98.
and bile acid regulation in erythromycin (ERY) and azithromycin (AZM) groups in humans. Baseline levels of each serum biomarker were based on
the mean and standard deviation at baseline across all participants (n = 20). All values are in pg/mL except for 5-HT (ng/mL), glucose (mmol/L),
adiponectin (m g/mL), leptin (ng/mL), and C-reactive protein (CRP, ng/mL). Serum biomarkers were measured using commercially available single and
multiplex immunoassay panels as described in the Methods section. Bar graph and error bars represent the mean fold change (log2) and standard
deviation of biomarker levels at the end of the antibiotic treatment compared to their respective baseline values. Pairwise comparisons were
performed using a linear mixed model (lme4 v1.1-23 and lmerTest v3.1-1 packages in R). Statistical significance (P , 0.05) following correction for
multiple testing using the false discovery rate (FDR) method is indicated by an asterisk (*).
CSF), interferon gamma (IFN-g ), IL-4, IL-6, IL-7, IL-8, IL-12, and IL-17A showed decreas-
ing trends following exposure to either macrolide, but these did not achieve statistical
significance. Levels of lipopolysaccharide (LPS), fibroblast growth factor 2 (FGF-2), and
the cytokines hepatocyte growth factor (HGF), IL-1 b , and IL-9 were detected in less
than 60% of samples in either group and were not analyzed further.
The potential for macrolide exposure to influence human metabolic homeostasis
was also explored through an assessment of macrolide-associated changes to serum
markers of glucose regulation (gastric inhibitory polypeptide [GIP], 5-HT, C-peptide, in-
sulin, GLP-1, glucagon, glucose, and peptide YY [PYY]), lipid metabolism (adiponectin
and leptin), and bile acid regulation (FGF-19) (Fig. 1B, Fig. S2). Baseline levels were
within normal ranges for all participants, while PYY was detected in less than 60% of
samples in either group. Erythromycin and azithromycin were both associated with sig-
nificant increases in C-peptide levels, while significant increases in serum levels of 5-HT
were observed with erythromycin alone (FDR P , 0.05) (Fig. 1B, Fig. S2). The changes
in insulin levels following macrolide treatment were positively correlated with those of
C-peptide (r = 0.523, P = 0.018), consistent with the function of C-peptide as a stable
biomarker for insulin secretion. Changes in the levels of GIP, insulin, GLP-1, glucagon,
glucose, adiponectin, and leptin trended upwards, while a nonsignificant downward
trend was observed for FGF-19.
Macrolide exposure is associated with alteration of gut microbiota composi-
tion. Given the potential for changes in intestinal microbiology to influence host physiol-
ogy, we explored the impact of low-dose macrolide exposure on the gut microbiome.
Macrolide treatment was not associated with significant reductions in fecal bacterial load
(P . 0.05 for both ERY and AZM) (Table S1). However, changes in gut microbiota alpha di-
versity were observed with both antibiotics. Specifically, compared to the baseline, treat-
ment was associated with a substantial decline in the number of bacterial taxa detected
(ERY, P = 0.006; AZM, P = 0.006; combined, P , 0.0001) and their diversity (ERY, P = 0.010;
AZM, P = 0.004, combined, P , 0.0001), but not bacterial evenness (Pielou’s evenness;
P . 0.05) (Fig. S3A to C, respectively). Significant alterations in fecal microbiota composition
were also observed following macrolide treatment (ERY: P = 0.002, t = 2.16; AZM: P = 0.004,
t = 1.99; combined: P = 0.0002, pseudo-F = 7.82) (Fig. 2A, Table S2). Notably, the composi-
tion of the altered microbiota following macrolide exposure did not differ significantly
between the erythromycin and azithromycin groups (P = 0.826, t = 0.88). The impact of
FIG 2 (A) Nonmetric multidimensional plot based on fecal microbiota before (baseline, BL) and after 4 weeks of erythromycin or azithromycin treatment
(ABX). Ellipses for each group represent the standard deviation with 80% confidence limit (dotted and solid lines represent baseline and tretment groups,
respectively). (B to D) Microbial species that were significantly altered by both erythromycin and azithromycin treatment in humans, and those that were
altered either by (C) erythromycin or (D) azithromycin. Heatmap represents the square root relative abundances of samples within each group; bar graphs
represent the (log2) fold change in the relative abundance of specific taxa between baseline and the end of 4 weeks of macrolide treatment. Bars and error
bars depict the median and interquartile ranges. Significance was determined based on the Wilcoxon test at FDR P , 0.05 for comparisons involving both
erythromycin and azithromycin groups, and P , 0.05 for agent-specific comparisons.
(SCFA) precursors acetate and/or lactate as well as pathways involved in the citric acid
(TCA) cycle, which utilizes intermediates of fermentation pathways (such as acetyl coen-
zyme A, oxaloacetate, and citrate), were also observed for both groups (P , 0.05) (Fig. 3B).
Of the species which displayed changes in relative abundance following macrolide
treatment, a reduction in B. longum was a potential principal contributor to many of
FIG 3 (A) Nonmetric multidimensional scaling ordination plot of bacterial functional pathway abundance in humans following treatment with azithromycin
or erythromycin for 4 weeks. Ellipses for each group represent the standard deviation with 80% confidence limit (dotted and solid lines represent baseline
and treatment groups, respectively). (B) Pairwise comparison between baseline and the end of erythromycin (ERY) and azithromycin (AZM) treatment was
performed using the Wilcoxon test, and P values were corrected for multiple comparisons using the FDR method. Only pathways that were significantly
altered by a 1.5-fold change (log2 average fold change . j0.58j and FDR P , 0.05) compared to the respective baseline abundance, either in the
erythromycin or azithromycin group, are shown. Bars and error bars represent the median and interquartile ranges. Stratification of pathways based on
bacterial species that were significantly altered by either erythromycin or azithromycin is shown using a heatmap. None of the taxonomically stratified
pathways were significantly altered with treatment. Pathway abundance (counts per million, CPM) were determined using the Human Microbiome Project
Unified Metabolic Analysis Network (HUMAnN3) tool and annotated based on the Metacyc Metabolic Pathway database.
were not associated with the microbiome, except for a significant association between
IL-10 and the inosine-59-phosphate biosynthesis pathway (associated with nucleotide
biosynthesis). While exploratory, these findings suggest a potential mechanism by
which changes in the gut microbiome arising through macrolide exposure might alter
host metabolic homeostasis and, to a lesser extent, aspects of immune regulation.
Erythromycin-associated changes are evident in the murine fecal microbiome.
A murine model of long-term, low-dose erythromycin treatment was used to further
investigate relationships between macrolide-associated changes in intestinal microbi-
ology and host metabolism. We first established that changes in fecal microbial com-
munity structure in conventional mice following erythromycin treatment (ERY) were
largely consistent with those observed in humans. Specifically, there was no significant
change in total fecal bacterial load (P = 0.451) (Table S4), but erythromycin was associ-
ated with a reduction in observed species (P # 0.0001) and the Shannon diversity
index (P # 0.001) (Fig. S5). These changes were significant compared to the control
mice at day 90 (P # 0.05).
Fecal microbiota composition in erythromycin-treated mice at day 90 differed signifi-
cantly from baseline (PERMANOVA [permutational multivariate analysis of variance]:
P = 0.0001, t = 9.28) and from the control mice (P = 0.0001, t = 8.72). This compositional
difference was associated with a significant decrease in the relative abundance of anaero-
bic taxa, particularly those involved in SCFA biosynthesis (Lactobacillus, Faecalibaculum,
Clostridium sensu stricto 1, and members of the Ruminococcaceae and Lachnospiraceae
families) and carbohydrate utilization (Bacteroidales 24-7 group) in erythromycin-treated
mice compared to controls (FDR P , 0.05, fold change . j1.5j) (Fig. 4A). Notably, a signifi-
cant increase in the relative abundance of Akkermansia with treatment contrasted the
depletion observed in humans following macrolide exposure. Additionally, one taxon in
the order Bacteroidales decreased significantly in erythromycin-treated mice (FDR
P , 0.05), although no corresponding change was observed in human participants (FDR
P = 0.915) (Table S2).
The impact of macrolide-associated changes in intestinal microbiology on the produc-
tion of potential mediators of host-microbiome interaction was then assessed using fecal
metabolomics. More than half of the differentially abundant metabolites identified between
the groups at day 90 (26 out of 46 detected metabolites) were found to be reduced follow-
ing erythromycin treatment (FDR P , 0.05) (Fig. 4B), consistent with the reduced microbial
FIG 4 (A) Bacterial taxa and (B) metabolites that were significantly altered in fecal samples of erythromycin-treated mice compared to control mice at day 90.
Each bar represents the log2 average fold change of bacterial taxa (or metabolites) that were significantly higher (in red) or significantly lower (in gray) in
(following a similar direction as erythromycin-treated conventional mice), but this did not
achieve statistical significance (P = 0.209) (Fig. 5C).
The effects of erythromycin on metabolic homeostasis were also assessed based on
average respiratory quotient (an indicator of basal metabolic activity) and total body
weight (used here as an overarching metabolic indicator). Erythromycin was associated
with significantly increased respiratory quotient compared to controls during both day
and night cycles (P , 0.0001), suggesting effects on nutrient utilization (Fig. S6C). This
change was driven primarily by significant increases in the volume of exhaled CO2 dur-
ing the active night cycle (P = 0.038) and was independent of food or water intake and
of total energy expenditure (P . 0.05) (Fig. S6C). However, body weight did not
change significantly with erythromycin treatment in conventional mice (P = 0.654) (Fig.
S6A). The body weights of germ-free mice treated with erythromycin and germ-free
mice transplanted with erythromycin-modified gut microbiota remained unaltered
compared to their respective controls (P . 0.05) (Fig. S7A and B, respectively).
Erythromycin influences host metabolism through its alteration of the gut
microbiome. Despite the alterations in glucose homeostasis, serum levels of the meta-
bolic biomarkers 5-HT and C-peptide did not change significantly in response to eryth-
romycin treatment in conventional mice (Fig. 6A). However, germ-free model studies
indicated that 5-HT levels were significantly modulated, both directly through expo-
sure (P , 0.001) and indirectly through microbiota-dependent pathways (P = 0.002)
(Fig. 6B and C, respectively). These findings suggest that the modulation of 5-HT is not
the major driver mediating the changes in glucose homeostasis. No significant direct
or microbiota-mediated effects were observed with C-peptide (P = 0.316 and 0.107,
respectively).
Gut motility is an important regulator of glucose and energy metabolism through its
effect on nutrient absorption and is influenced by both intestinal microbiology and proki-
netic agents, such as erythromycin. Although direct erythromycin exposure in germ-free
mice did not alter gut motility significantly (based on gut transit time) (P = 0.209) (Fig. 7A),
FIG 7 Assessment of direct effects of erythromycin exposure and erythromycin-associated gut microbiota effects on host physiology. Body weight
over the 90-day study was assessed in (A) germ-free mice receiving erythromycin treatment or water and (B) germ-free mice colonized with
erythromycin-associated or control microbiota. Cecum weight (C and E) and gut transit times (D and F) were also measured for the respective
groups at the end of the 90-day study. Associations between the total gut transit time and cecal weight in (G) germ-free mice receiving
erythromycin (Fig. 6C), but no difference in IL-5 levels was identified between germ-
free recipients of erythromycin-treated or control microbiota (Fig. 6B). Finally, the
immune markers IL-10 and MCP-1 remained unaltered in both conventional and germ-
free models of erythromycin (Fig. 6).
DISCUSSION
The efficacy of long-term macrolide therapy at preventing exacerbations in chronic
lung disease has led to its increasing clinical use (22, 23). Macrolides carry a risk of QTc
interval prolongation and torsade de pointes (24), and can result in increased toxicity
when co-prescribed with drugs such as statins through inhibition of liver cytochrome
P450 enzymes (25). However, with appropriate screening, macrolide therapy is consid-
ered to carry minimal risk. Despite this, little is known about the extrapulmonary
effects of these agents, which can have pleiotropic effects, either directly or resulting
from their impact on the gut microbiome.
Utilizing studies involving healthy adults and preclinical models, we investigated
whether macrolide antibiotics at low doses, as used in the management of chronic re-
spiratory conditions, were associated with altered markers of systemic homeostasis. In
particular, we focused on an aspect of physiology known to be influenced by altered
host-microbiome interactions: metabolic control. In humans, we observed an associa-
tion between macrolide exposure and alteration of host metabolic markers, changes
that were correlated with changes in gut microbiome composition and function. Our
findings suggest that long-term, low-dose macrolide therapy is likely to influence sys-
temic metabolic control in recipients and potentially affect the risk of cardiometabolic
disease.
Broadly, we identified associations between macrolide exposure and changes in host
markers of immune and metabolic homeostasis in human participants. Modest but signifi-
cant decreases in serum immune markers IL-5, IL-10, and MCP-1 were observed following
macrolide exposure, an effect that was more pronounced with azithromycin. The immuno-
modulatory properties of macrolides are well-recognized; for example, asthma patients
receiving azithromycin exhibit lower IL-5 expression in CD41 cells isolated from peripheral
blood mononuclear cells (26). The absence of a pronounced anti-inflammatory effect in our
study likely reflects the absence of systemic inflammation in healthy participants.
In contrast, exposure to macrolides was associated with increases in host markers of
metabolic regulation. In particular, erythromycin significantly increased serum levels of 5-
HT and C-peptide, which reflect aspects of glucose regulation. Spore-forming gut micro-
biota are known to promote 5-HT biosynthesis by colonic enterochrommaffin cells (27). 5-
HT, in turn, can affect glucose homeostasis by influencing pancreatic b -cell production of
hormones and regulating lipid metabolism in hepatocytes (28). Erythromycin is a prokinetic
and therefore might influence glucose homeostasis by increasing gastrointestinal motility
and nutrient absorption. In our investigations in germ-free mice, no increase in motility
was observed at the erythromycin dose used, which suggests that erythromycin may influ-
ence the host physiology via modulation of the gut microbiota.
Alongside the association between erythromycin and azithromycin and aspects of host
physiology, both agents were associated with changes in the composition and functional
capacity of the intestinal microbiology. While total bacterial loads were unchanged, both
macrolides were associated with significant decreases in bacterial richness and diversity.
These findings are similar to the general effects of antibiotic exposure (20), although the
macrolide dose used in this study was lower than that prescribed for antibiotic purposes.
Treatment was associated with depletion of a number of important commensal anaerobes,
including Bifidobacterium species (B. longum and B. adolescentis), while species associated
with inflammation, such as Ruminococcus gnavus (29), increased following macrolide expo-
sure. These observations align with those reported following acute azithromycin (30–32) or
kit (2 300 bp) using the Illumina MiSeq platform at the South Australian Genomics Centre (SAGC) in
Adelaide (Australia). Sequence data were subsampled to 4,459 reads prior to downstream analysis.
Computational analysis of paired-end reads to measure alpha diversity (observed species, Pielou even-
ness, Shannon diversity H9), beta diversity (weighted UniFrac distances), and bacterial relative abun-
dance were performed using the QIIME2 platform (54), as described previously (46).
Serum protein quantification. Serum concentrations of immune-related biomarkers, including
cytokines and chemokines (TNF-a, IFN-g , GM-CSF, MCP-1, IL-1 b , IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12,
IL-13, IL-17A), growth factors (HGF, FGF-2), a marker of inflammation (C-reactive protein [CRP]), and lipo-
polysaccharide (LPS); as well as metabolic markers, including hormones and molecules associated with
host glucose metabolism (GIP, 5-HT, C-peptide, insulin, GLP-1, glucose, glucagon, and PYY), lipid metab-
olism (adiponectin and leptin), and bile acid homeostasis (FGF-19), were quantified using a combination
of commercially available multiplex immunoassay panels (Milliplex) and enzyme-linked immunosorbent
assays (ELISA). Assays were performed according to the manufacturer’s instructions with modifications
(supplemental material).
Fecal metabolome analysis. Murine fecal metabolome characteristics were determined by proton
nuclear magnetic resonance (1H NMR), as described previously (55). Fecal metabolome analysis was per-
formed with NMR spectral intensities subjected to probabilistic quotient normalization and Pareto scal-
ing (56).
Glucose tolerance test. Mice were fasted for 12 to 14 h by removing the feeding tray and housing in
fresh bedding with fasting trays. All mice had ad libitum access to plain drinking water or drinking water con-
taining erythromycin ethylsuccinate (equivalent to 20 mg/kg) during the fasting period. Glucose (2 g/kg of
body mass, dissolved in sterile 0.9% sodium chloride) was administered by intraperitoneal injection and blood
glucose was measured from the tail tip using a glucometer (Abbott Freestyle Freedom Lite, Australia). Blood
glucose measurements were performed at 0 (basal level), 15, 30, 60, and 120 min after glucose administration.
Metabolic and behavioral monitoring in mice. Mice were housed individually in a Promethion
Metabolic cage system (Sable Systems International, NV, USA) (n = 3 per group). Following a 24-h acclimatiza-
tion period, metabolic and behavioral information was recorded over two consecutive day-and-night cycles.
Statistical analysis. Normality of the data distribution was determined using the Shapiro-Wilk test.
Comparisons between paired samples were performed using a t test (parametric data) or a Wilcoxon test (non-
parametric data). Unpaired samples were analyzed using an unpaired t test (parametric data) or a Mann-
Whitney test (nonparametric data). Paired comparisons of pre- and post-antibiotic human fecal samples were
performed to minimize diet-associated effects on study outcomes. Between-group differences in microbiota
composition were assessed by PERMANOVA (57), on distance matrices using PRIMER7 (PRIMER-E, Plymouth),
based on Bray Curtis distances (shotgun metagenomics data) and weighted UniFrac distances for genus-level
relative abundances (16S rRNA amplicon sequence data). Murine microbial diversity analyses were performed
using a mixed model with cage as a random factor. Correlations between the distance matrices of murine gut
microbiota and metabolome data were assessed by testing for distance covariances using the R package dcov
(v0.1.1). For serum biomarker analysis, only those whose levels reached above the detection threshold for
$60% of samples across the erythromycin or azithromycin groups were analyzed. Pairwise comparisons of se-
rum biomarker levels were performed using a linear mixed model (lme4 v1.1-23 and lmerTest v3.1-1) in R.
SUPPLEMENTAL MATERIAL
Supplemental material is available online only.
SUPPLEMENTAL FILE 1, PDF file, 1.7 MB
ACKNOWLEDGMENTS
We thank Amanda Page for assistance on the Promethion Metabolic cage
studies, and Samay Trec and Mariah Turelli for their assistance in germ-free murine
studies. NMR experiments described in this paper were carried out at the Centre for
Biomolecular Spectroscopy, King’s College London, on instruments acquired with a
Multi-User Equipment Grant from the Wellcome Trust and an Infrastructure Grant
from the British Heart Foundation. We thank Andrew Atkinson for helping perform
NMR experiments.
J.M.C., L.D.B., and G.B.R. conceptualized the study; S.L. and M.M. contributed to
acquisition of samples and data for the human study; J.M.C. and A.R. contributed to
acquisition of samples and data for murine studies; J.M.C., A.M., E.S., F.M.M., and T.K.
conducted formal analysis and interpretation of data; J.M.C., S.L.T., L.D.B., D.J.K., A.J.M.,
and G.B.R. contributed to the preparation and revision of the manuscript. All authors
reviewed and approved the final manuscript.
We declare that we have no competing interests.
REFERENCES
1. Smith D, Du Rand IA, Addy C, Collyns T, Hart S, Mitchelmore P, Rahman N, 17. Waclawikova B, Codutti A, Alim K, El Aidy S. 2022. Gut microbiota-motility
Saggu R. 2020. British Thoracic Society guideline for the use of long-term interregulation: insights from in vivo, ex vivo and in silico studies. Gut
macrolides in adults with respiratory disease. Thorax 75:370–404. https:// Microbes 14:1997296. https://doi.org/10.1080/19490976.2021.1997296.
doi.org/10.1136/thoraxjnl-2019-213929. 18. Ruiz VE, Battaglia T, Kurtz ZD, Bijnens L, Ou A, Engstrand I, Zheng X, Iizumi
2. Saiman L, Marshall BC, Mayer-Hamblett N, Burns JL, Quittner AL, Cibene T, Mullins BJ, Müller CL, Cadwell K, Bonneau R, Perez-Perez GI, Blaser MJ.
DA, Coquillette S, Fieberg AY, Accurso FJ, Campbell PW 3rd, Macrolide 2017. A single early-in-life macrolide course has lasting effects on murine
Study Group. 2003. Azithromycin in patients with cystic fibrosis chroni- microbial network topology and immunity. Nat Commun 8:518. https://
cally infected with Pseudomonas aeruginosa: a randomized controlled doi.org/10.1038/s41467-017-00531-6.
trial. JAMA 290:1749–1756. https://doi.org/10.1001/jama.290.13.1749. 19. Lynn MA, Tumes DJ, Choo JM, Sribnaia A, Blake SJ, Leong LEX, Young GP,
3. Altenburg J, de Graaff CS, Stienstra Y, Sloos JH, van Haren EH, Koppers RJ, Marshall HS, Wesselingh SL, Rogers GB, Lynn DJ. 2018. Early-life antibi-
van der Werf TS, Boersma WG. 2013. Effect of azithromycin maintenance otic-driven dysbiosis leads to dysregulated vaccine immune responses in
treatment on infectious exacerbations among patients with non-cystic fi- mice. Cell Host Microbe 23:653–660.e655. https://doi.org/10.1016/j.chom
brosis bronchiectasis: the BAT randomized controlled trial. JAMA 309: .2018.04.009.
1251–1259. https://doi.org/10.1001/jama.2013.1937. 20. Schwartz DJ, Langdon AE, Dantas G. 2020. Understanding the impact of
4. Serisier DJ, Martin ML, McGuckin MA, Lourie R, Chen AC, Brain B, Biga S,
antibiotic perturbation on the human microbiome. Genome Med 12:82.
Schlebusch S, Dash P, Bowler SD. 2013. Effect of long-term, low-dose
https://doi.org/10.1186/s13073-020-00782-x.
erythromycin on pulmonary exacerbations among patients with non-
21. Gelfond D, Borowitz D. 2013. Gastrointestinal complications of cystic fi-
cystic fibrosis bronchiectasis: the BLESS randomized controlled trial.
brosis. Clin Gastroenterol Hepatol 11:333–342. quiz e330-331. https://doi
JAMA 309:1260–1267. https://doi.org/10.1001/jama.2013.2290.
5. Gibson PG, Yang IA, Upham JW, Reynolds PN, Hodge S, James AL, Jenkins C, .org/10.1016/j.cgh.2012.11.006.
22. Adriaenssens N, Bruyndonckx R, Versporten A, Hens N, Monnet DL,
Peters MJ, Marks GB, Baraket M, Powell H, Taylor SL, Leong LEX, Rogers GB,
Simpson JL. 2017. Effect of azithromycin on asthma exacerbations and qual- Molenberghs G, Goossens H, Weist K, Coenen S, ESAC-Net study group.
ity of life in adults with persistent uncontrolled asthma (AMAZES): a rando- 2021. Consumption of macrolides, lincosamides and streptogramins in
mised, double-blind, placebo-controlled trial. Lancet 390:659–668. https://doi the community, European Union/European Economic Area, 1997–2017. J
.org/10.1016/S0140-6736(17)31281-3. Antimicrob Chemother 76:ii30–ii36. https://doi.org/10.1093/jac/dkab175.
6. Wang D, Fu W, Dai J. 2019. Meta-analysis of macrolide maintenance ther- 23. Sun J, Li Y. 2022. Long-term, low-dose macrolide antibiotic treatment in
apy for prevention of disease exacerbations in patients with noncystic fi- pediatric chronic airway diseases. Pediatr Res 91:1036–1042. https://doi
brosis bronchiectasis. Medicine (Baltimore, MD) 98:e15285. https://doi .org/10.1038/s41390-021-01613-4.
.org/10.1097/MD.0000000000015285. 24. Teng C, Walter EA, Gaspar DKS, Obodozie-Ofoegbu OO, Frei CR. 2019. Tor-
7. Spagnolo P, Fabbri LM, Bush A. 2013. Long-term macrolide treatment for sades de pointes and QT prolongation associations with antibiotics: a
chronic respiratory disease. Eur Respir J 42:239–251. https://doi.org/10 pharmacovigilance study of the FDA Adverse Event Reporting System. Int
.1183/09031936.00136712. J Med Sci 16:1018–1022. https://doi.org/10.7150/ijms.34141.
8. Kanoh S, Rubin BK. 2010. Mechanisms of action and clinical application of 25. Westphal JF. 2000. Macrolide-induced clinically relevant drug interactions
macrolides as immunomodulatory medications. Clin Microbiol Rev 23: with cytochrome P-450A (CYP) 3A4: an update focused on clarithromycin,
590–615. https://doi.org/10.1128/CMR.00078-09. azithromycin and dirithromycin. Br J Clin Pharmacol 50:285–295. https://
33. Brismar B, Edlund C, Nord CE. 1991. Comparative effects of clarithromycin 47. Choo JM, Rogers GB. 2021. Gut microbiota transplantation for coloniza-
and erythromycin on the normal intestinal microflora. Scand J Infect Dis tion of germ-free mice. STAR Protoc 2:100610. https://doi.org/10.1016/j
23:635–642. https://doi.org/10.3109/00365549109105189. .xpro.2021.100610.
34. Parada Venegas D, De la Fuente MK, Landskron G, Gonzalez MJ, Quera R, 48. Choo JM, Tran CD, Luscombe-Marsh ND, Stonehouse W, Bowen J,
Dijkstra G, Harmsen HJM, Faber KN, Hermoso MA. 2019. Short chain fatty Johnson N, Thompson CH, Watson EJ, Brinkworth GD, Rogers GB. 2021.
acids (SCFAs)-mediated gut epithelial and immune regulation and its rele- Almond consumption affects fecal microbiota composition, stool pH, and
vance for inflammatory bowel diseases. Front Immunol 10:277. https://doi stool moisture in overweight and obese adults with elevated fasting
.org/10.3389/fimmu.2019.01486. blood glucose: a randomized controlled trial. Nutr Res 85:47–59. https://
35. Longo KA, Charoenthongtrakul S, Giuliana DJ, Govek EK, McDonagh T, doi.org/10.1016/j.nutres.2020.11.005.
Distefano PS, Geddes BJ. 2010. The 24-hour respiratory quotient predicts 49. Dey N, Wagner VE, Blanton LV, Cheng J, Fontana L, Haque R, Ahmed T,
energy intake and changes in body mass. Am J Physiol Regul Integr Comp Gordon JI. 2015. Regulators of gut motility revealed by a gnotobiotic
model of diet-microbiome interactions related to travel. Cell 163:95–107.
Physiol 298:R747–754. https://doi.org/10.1152/ajpregu.00476.2009.
https://doi.org/10.1016/j.cell.2015.08.059.
36. Venema K. 2010. Role of gut microbiota in the control of energy and carbo-
50. Choo JM, Leong LE, Rogers GB. 2015. Sample storage conditions signifi-
hydrate metabolism. Curr Opin Clin Nutr Metab Care 13:432–438. https://doi
cantly influence faecal microbiome profiles. Sci Rep 5:16350. https://doi
.org/10.1097/MCO.0b013e32833a8b60.
.org/10.1038/srep16350.
37. Wolfe AJ. 2015. Glycolysis for microbiome generation. Microbiol Spectr 3.
51. Choo JM, Abell GCJ, Thomson R, Morgan L, Waterer G, Gordon DL, Taylor
https://doi.org/10.1128/microbiolspec.MBP-0014-2014. SL, Leong LEX, Wesselingh SL, Burr LD, Rogers GB. 2018. Impact of long-
38. Sato E, Nelson DK, Koyama S, Hoyt JC, Robbins RA. 2001. Erythromycin term erythromycin therapy on the oropharyngeal microbiome and resist-
modulates eosinophil chemotactic cytokine production by human lung ance gene reservoir in non-cystic fibrosis bronchiectasis. mSphere 3:
fibroblasts in vitro. Antimicrob Agents Chemother 45:401–406. https://doi e00103-18. https://doi.org/10.1128/mSphere.00103-18.
.org/10.1128/AAC.8.2.401-406.2001. 52. Mobegi FM, Leong LE, Thompson F, Taylor SM, Harriss LR, Choo JM,
39. Portune KJ, Benitez-Paez A, Del Pulgar EM, Cerrudo V, Sanz Y. 2017. Gut Taylor SL, Wesselingh SL, McDermott R, Ivey KL, Rogers GB. 2020. Intesti-
microbiota, diet, and obesity-related disorders: the good, the bad, and nal microbiology shapes population health impacts of diet and lifestyle
the future challenges. Mol Nutr Food Res 61:1600252. https://doi.org/10 risk exposures in Torres Strait Islander communities. Elife 9:58407. https://
.1002/mnfr.201600252. doi.org/10.7554/eLife.58407.
40. Campaniello D, Corbo MR, Sinigaglia M, Speranza B, Racioppo A, Altieri C, 53. Beghini F, McIver LJ, Blanco-Míguez A, Dubois L, Asnicar F, Maharjan S,
Bevilacqua A. 2022. How diet and physical activity modulate gut micro- Mailyan A, Manghi P, Scholz M, Thomas AM, Valles-Colomer M, Weingart
biota: evidence, and perspectives. Nutrients 14:2456. https://doi.org/10 G, Zhang Y, Zolfo M, Huttenhower C, Franzosa EA, Segata N. 2021. Inte-
.3390/nu14122456. grating taxonomic, functional, and strain-level profiling of diverse micro-
41. Renna M, Schaffner C, Brown K, Shang S, Tamayo MH, Hegyi K, Grimsey bial communities with bioBakery 3. Elife 10:65088. https://doi.org/10
NJ, Cusens D, Coulter S, Cooper J, Bowden AR, Newton SM, Kampmann B, .7554/eLife.65088.
Helm J, Jones A, Haworth CS, Basaraba RJ, DeGroote MA, Ordway DJ, 54. Bolyen E, Rideout JR, Dillon MR, Bokulich NA, Abnet CC, Al-Ghalith GA,
Rubinsztein DC, Floto RA. 2011. Azithromycin blocks autophagy and may Alexander H, Alm EJ, Arumugam M, Asnicar F, Bai Y, Bisanz JE, Bittinger K,
predispose cystic fibrosis patients to mycobacterial infection. J Clin Invest Brejnrod A, Brislawn CJ, Brown CT, Callahan BJ, Caraballo-Rodríguez AM,
121:3554–3563. https://doi.org/10.1172/JCI46095. Chase J, Cope EK, Da Silva R, Diener C, Dorrestein PC, Douglas GM, Durall
42. Nahata M. 1996. Drug interactions with azithromycin and the macrolides: DM, Duvallet C, Edwardson CF, Ernst M, Estaki M, Fouquier J, Gauglitz JM,
an overview. J Antimicrob Chemother 37 Suppl C:133–142. https://doi Gibbons SM, Gibson DL, Gonzalez A, Gorlick K, Guo J, Hillmann B, Holmes
S, Holste H, Huttenhower C, Huttley GA, Janssen S, Jarmusch AK, Jiang L,
.org/10.1093/jac/37.suppl_c.133.
Kaehler BD, Kang KB, Keefe CR, Keim P, Kelley ST, Knights D, et al. 2019.
43. Elderman M, van Beek A, Brandsma E, de Haan B, Savelkoul H, de Vos P,
Reproducible, interactive, scalable and extensible microbiome data sci-
Faas M. 2016. Sex impacts Th1 cells, Tregs, and DCs in both intestinal and
ence using QIIME 2. Nat Biotechnol 37:852–857. https://doi.org/10.1038/
systemic immunity in a mouse strain and location-dependent manner.
s41587-019-0209-9.
Biol Sex Differ 7:21. https://doi.org/10.1186/s13293-016-0075-9.