Azithromycin blocks autophagy and may predispose cystic fibrosis patients to mycobacterial infection.

Abstract

Azithromycin is a potent macrolide antibiotic with poorly understood antiinflammatory properties. Long-term use of azithromycin in patients with chronic inflammatory lung diseases, such as cystic fibrosis (CF), results in improved outcomes. Paradoxically, a recent study reported that azithromycin use in patients with CF is associated with increased infection with nontuberculous mycobacteria (NTM). Here, we confirm that long-term azithromycin use by adults with CF is associated with the development of infection with NTM, particularly the multi-drug-resistant species Mycobacterium abscessus, and identify an underlying mechanism. We found that in primary human macrophages, concentrations of azithromycin achieved during therapeutic dosing blocked autophagosome clearance by preventing lysosomal acidification, thereby impairing autophagic and phagosomal degradation. As a consequence, azithromycin treatment inhibited intracellular killing of mycobacteria within macrophages and resulted in chronic infection with NTM in mice. Our findings emphasize the essential role for autophagy in the host response to infection with NTM, reveal why chronic use of azithromycin may predispose to mycobacterial disease, and highlight the dangers of inadvertent pharmacological blockade of autophagy in patients at risk of infection with drug-resistant pathogens.

Full text

Page 1

Research article
3554?The?Journal?of?Clinical?Investigation? ? ? http://www.jci.org? ? ? Volume 121? ? ? Number 9? ? ? September 2011
Azithromycin blocks autophagy
and may predispose cystic fibrosis
patients to mycobacterial infection
Maurizio Renna,1 Catherine Schaffner,2 Karen Brown,2,3 Shaobin Shang,4
Marcela Henao Tamayo,4 Krisztina Hegyi,2 Neil J. Grimsey,2 David Cusens,2,3 Sarah Coulter,2
Jason Cooper,1 Anne R. Bowden,2 Sandra M. Newton,5 Beate Kampmann,5 Jennifer Helm,6
Andrew Jones,6 Charles S. Haworth,3 Randall J. Basaraba,4 Mary Ann DeGroote,4
Diane J. Ordway,4 David C. Rubinsztein,1 and R. Andres Floto2,3
1Department of Medical Genetics and 2Department of Medicine, Cambridge Institute for Medical Research, University of Cambridge,
Cambridge, United Kingdom. 3Adult Cystic Fibrosis Centre, Papworth Hospital, Cambridge, United Kingdom. 4Department of Microbiology,
Immunology, and Pathology, Colorado State University, Fort Collins, Colorado, USA. 5Wellcome Centre for Clinical Tropical Medicine,
Imperial College London, London, United Kingdom. 6Adult Cystic Fibrosis Unit, Wythenshawe Hospital, Manchester, United Kingdom.
Azithromycin?is?a?potent?macrolide?antibiotic?with?poorly?understood?antiinflammatory?properties.?Long-
term?use?of?azithromycin?in?patients?with?chronic?inflammatory?lung?diseases,?such?as?cystic?fibrosis?(CF),?
results?in?improved?outcomes.?Paradoxically,?a?recent?study?reported?that?azithromycin?use?in?patients?with?
CF?is?associated?with?increased?infection?with?nontuberculous?mycobacteria?(NTM).?Here,?we?confirm?that?
long-term?azithromycin?use?by?adults?with?CF?is?associated?with?the?development?of?infection?with?NTM,?
particularly?the?multi-drug-resistant?species?Mycobacterium abscessus,?and?identify?an?underlying?mechanism.?
We?found?that?in?primary?human?macrophages,?concentrations?of?azithromycin?achieved?during?therapeutic?
dosing?blocked?autophagosome?clearance?by?preventing?lysosomal?acidification,?thereby?impairing?autopha-
gic?and?phagosomal?degradation.?As?a?consequence,?azithromycin?treatment?inhibited?intracellular?killing?of?
mycobacteria?within?macrophages?and?resulted?in?chronic?infection?with?NTM?in?mice.?Our?findings?empha-
size?the?essential?role?for?autophagy?in?the?host?response?to?infection?with?NTM,?reveal?why?chronic?use?of?
azithromycin?may?predispose?to?mycobacterial?disease,?and?highlight?the?dangers?of?inadvertent?pharmaco-
logical?blockade?of?autophagy?in?patients?at?risk?of?infection?with?drug-resistant?pathogens.
Introduction
Azithromycin is a potent antibiotic and is frequently used in pro-
phylaxis and treatment regimens for mycobacterial infections (1).
Long-term use of azithromycin and other macrolide antibiotics
has increased as a result of reports of improved clinical outcomes
in patients with cystic fibrosis (CF) (2), chronic obstructive pul-
monary disease (3), panbronchiolitis (4), asthma (5), and chronic
transplant rejection (6). These benefits are believed to result from
its antiinflammatory effects on macrophages and neutrophils
mediated through unknown mechanisms (7).
Worryingly, several studies have identified a synchronous
increase in mycobacterial infection of CF patients, predominant-
ly with the multi-drug-resistant, highly pathogenic nontubercu-
lous mycobacteria (NTM) Mycobacterium abscessus (8, 9), for which
there is no reliable treatment (1). We therefore wondered wheth-
er azithromycin paradoxically impairs host immunity against
mycobacteria. Macroautophagy (referred to herein as autophagy)
appears crucial for an effective cellular response against mycobac-
teria (10). Since the macrolide bafilomycin disrupts autophagy in
vitro (11), we examined whether azithromycin blocks autophagy
and thereby impairs intracellular killing of NTM. Here we pres-
ent clinical data suggestive of an association between long-term
azithromycin treatment and development of NTM infection in CF
patients. At concentrations achieved during therapeutic dosing,
azithromycin impaired lysosomal degradation of both autopha-
gosomes and phagosomes and led to failure of intracellular kill-
ing of mycobacteria and development of chronic infection with
M. abscessus in mouse models.
Results
Long-term azithromycin use may predispose CF patients to NTM infection.
Escalating azithromycin use over the last 5 years has mirrored an
increase in patients colonized or infected with NTM in our adult
CF center (Figure 1A), not explicable by changes in sputum sam-
pling (Supplemental Table 1; supplemental material available
online with this article; doi:10.1172/JCI46095DS1) or micro-
bial culture methods. Long-term azithromycin use was associ-
ated with developing infection with NTM, particularly M. abscessus
(P = 0.0009, 2-tailed uncorrected χ2; Figure 1B). When adjusted for
age — which was identified, as in previous studies (9), as a signifi-
cant covariate (P = 0.01; Supplemental Table 2) — azithromycin use
remained significantly associated with NTM disease (P = 0.00046;
odds ratio 9.80, 95% CI 2.09–45.87). Similar associations of chron-
ic azithromycin use with NTM infection have been reported for
an Israeli CF cohort (12), which suggests that azithromycin might
impair host immunity against mycobacteria.
Azithromycin blocks autophagosome degradation. Since autophagy
appears crucial for an effective cellular response against mycobac-
teria (10), we investigated whether azithromycin blocks autopha-
Authorship?note: Diane J. Ordway, David C. Rubinsztein, and R. Andres Floto are
co–senior authors.
Conflict?of?interest: The authors have declared that no conflict of interest exists.
Citation?for?this?article: J Clin Invest. 2011;121(9):3554–3563. doi:10.1172/JCI46095.

Page 2

research article
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
3555
gy, thereby preventing autophagy-dependent intracellular killing
of NTM. Autophagy is routinely assayed by quantifying autopha-
gosomes, achieved either by measuring levels of LC3-II, a protein
that specifically associates with autophagosomes, or by detection
of LC3-II+ vesicles (13). In cell lines, azithromycin caused a dose-
dependent increase in both the number of LC3+ vesicles and the
amount of LC3-II protein (EC50, 4.5 μg/ml; Figure 1, C and D).
This suggests an accumulation of autophagosomes, which could
result from either their increased synthesis or their decreased deg-
radation. We therefore examined the effect of azithromycin in the
presence of the lysosomal inhibitor bafilomycin A1 (BafA1), which
blocks LC3-II destruction (11). Cotreatment of azithromycin and
BafA1 resulted in no significant increase in LC3-II levels compared
with BafA1 alone (Figure 1E), which indicates that azithromycin
does not induce autophagosome synthesis, but rather blocks LC3-II
degradation. Moreover, the azithromycin-induced rise in LC3-II
was not accompanied by decreases in mTOR-dependent signaling
(monitored by phosphorylation of p70S6 kinase and S6), a major
negative regulator of autophagosome synthesis (14), but was
associated — in both HeLa cells and primary macrophages — with
Figure 1
Azithromycin blocks autophagy by impairing autophagic flux. (A) At Papworth Hospital, the numbers of adult CF patients on long-term azithro-
mycin (AZM; red) and with new sputum cultures positive for NTM (gray) rose over the last 5 years. (B) Analysis of the 2007–2008 CF patient
cohort. In patients currently or previously infected with NTM, long-term azithromycin use (current or within 6 months of first NTM culture) was
significantly more common that in uninfected patients (P = 0.0009; 2-tailed uncorrected χ2). (C) In GFP-LC3+ COS7 cells, 24-hour treatment
with azithromycin increased the number of cells containing 10 or more GFP-LC3+ vesicles more so than rapamycin (200 nM). (D) In HeLa cells,
azithromycin caused a dose-dependent increase in LC3-II levels (EC50, 4.5 μg/ml). exp., exposure. (E) In the presence of 400 nM BafA1, azithro-
mycin treatment of HeLa cells did not increase LC3-II compared with BafA1 alone. (F) Azithromycin treatment (16 μg/ml) of HeLa cells, while
increasing LC3-II, did not alter mTOR-dependent signaling (monitored by changes in phosphorylation of S6 and p70S6 kinase); however, it did
increase p62 levels. (G) Azithromycin dose-dependently increased LC3-II and p62 levels in primary human macrophages. (H) Similar effect of
azithromycin, assessed by LC3-II and p62 levels, in primary macrophages derived from clinically stable CF patients and healthy controls (n = 3
per group). *P < 0.05; **P < 0.005.

Page 3

research article
3556?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
elevations in the endogenous autophagy substrate p62 (ref. 15 and
Figure 1, F and G). Azithromycin treatment in vitro resulted in
accumulation of LC3-II and p62 in primary macrophages from CF
patients similar to that in cells from healthy controls (Figure 1H
and Supplemental Figure 1).
To confirm that azithromycin blocks autophagy-mediated
degradation at, or below, concentrations achieved in clinical
practice (between 12 μg/ml and 53 μg/ml in sputum, but several
hundred–fold higher in phagocytic cells; refs. 7, 16), we exam-
ined the clearance of 2 known autophagy substrates: mutant
(Q74) huntingtin (17) and A53T α-syncuclein (18). Azithromycin
treatment retarded the basal and rapamycin-enhanced degrada-
tion of A53T α-synuclein and led to enhanced accumulation of
intracellular aggregates of Q74 huntingtin (Figure 2, A and B).
We next examined whether inhibition of LC3-II clearance by
azithromycin was caused by failure of autophagosome-lysosome
fusion, as seen with long-term treatment with BafA1 treatment
(11), or failure of lysosomal function, a mechanism we recently
reported (19). Confocal microscopy of HeLa cells coexpressing
mCherry-LC3 and the lysosomal marker lgp120 revealed that
although the number of LC3+ vesicles per cell increased with rapa-
mycin, BafA1, or azithromycin treatment, the degree of LC3 colo-
calization with lgp120 was not significantly affected by azithro-
mycin, but was, as expected, enhanced by rapamycin and inhibited
by BafA1 (Figure 3A). However, azithromycin prevented acidifi-
cation of autophagosomes. In HeLa cells stably expressing the
mCherry-GFP-LC3 reporter, azithromycin treatment significantly
increased nonacidified autophagosomes (i.e., mCherry+GFP+) and
reduced acidified autophagosomes (mCherry+GFP–; Figure 3B).
Azithromycin disrupts autophagosome and phagosome function in
macrophages. Since intracellular killing of mycobacteria by macro-
phages is critically enhanced by IFN-γ–induced autophagy (20),
we examined the effect of azithromycin on autophagosomal flux
triggered by this cytokine. Azithromycin blocked IFN-γ–induced
as well as basal autophagosomal acidification in macrophages
(Figure 4A), suggestive of impaired lysosomal function. Direct
quantification of lysosomal pH, achieved by preloading human
primary macrophages with dextran double-labeled with acid-
sensitive (FITC) and resistant (TMR) fluorophores (21), indicat-
ed that 4 hours of treatment with azithromycin was sufficient
to trigger significant alkalinization of lysosomes (Figure 4B).
Azithromycin also blocked the acidification of phagosomes con-
taining mycobacteria (Figure 4C and Supplemental Figure 2)
or latex beads (Supplemental Figure 3). Using heat-killed
M. abscessus labeled with both acid-quenchable (FAM) and pH
stable (Alexa Fluor 633) fluorophores to report compartmental
pH, we found that azithromycin pretreatment resulted in signif-
icant alkalinization of mycobacterial phagosomes (Figure 4C),
to pH levels inhibitory to most lysosomal proteases. Similar
azithromycin-induced phagosomal alkalinization was observed for
cells ingesting PFA-killed M. tuberculosis (Supplemental Figure 2).
As expected, azithromycin significantly reduced phagosomal
protein degradation, as measured by time-dependent loss of sur-
face-bound OVA from internalized latex beads (Figure 5A and
Supplemental Figure 4). Although significant, phagosome-lyso-
some fusion was only impaired to a small extent by azithromycin
(Figure 5B). Thus, it is likely that the block in phagosomal degra-
dation results predominantly from a direct effect of azithromycin
on compartmental pH. Azithromycin also reduced TNF-α release
from macrophages in response to particulate but not soluble
Figure 2
Azithromycin blocks autophagic clearance and autophagosome acidifi-
cation. (A) Azithromycin reduced basal (black) and rapamycin-induced
(gray) clearance of mutant (A53T) α-synuclein in stable inducible PC12
cells. The A53T α-synuclein transgene was induced with doxycycline
for 48 hours and then switched off (by antibiotic removal) before cells
were treated with vehicle alone, rapamycin (200 nM), and/or azithro-
mycin for a further 24 hours. (B) Azithromycin reduced clearance of
mutant (Q74) huntingtin. Aggregates of GFP-Q74 huntingtin were
reduced by rapamycin treatment, but increased when autophagy was
blocked by either BafA1 or azithromycin (4–16 μg/ml). Con, control.
*P < 0.05; **P < 0.005. Scale bar: 10 μm.

Page 4

research article
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
3557
stimuli (Supplemental Figure 5), which suggests an inhibition
of phagosome-triggered cytokine release. These data suggest that
azithromycin treatment might compromise the host response
to mycobacterial infection both by preventing degradation of
mycobacteria within autophagosomes and phagosomes and by
disrupting release of cytokines, such as TNF-α, that promote
autophagy-dependent intracellular killing of mycobacteria.
Intracellular killing of mycobacteria is impaired by azithromycin. We
therefore examined the effect of azithromycin on the survival of
mycobacteria within primary human macrophages. Autophagy
is critical for intracellular killing of M. tuberculosis and M. bovis
BCG (10, 20), but also pathogenic NTM, including M. abscessus
(Supplemental Figure 6). Over a wide range of concentrations,
azithromycin blocked basal intracellular killing of M. bovis BCG
and M. abscessus, as well as preventing the ability of the autophagy
inducer rapamycin to enhance killing (Figure 6, A and B). The
control of M. tuberculosis and NTM infection in vivo is crucially
dependent on the inflammatory cytokines IFN-γ and TNF-α,
which enhance intracellular killing of mycobacteria through the
induction of autophagy (10). As expected, azithromycin treatment
Figure 3
Azithromycin blocks acidification of autophagosomes without impairing lysosomal function. (A) Azithromycin did not impair autophagosome-
lysosome fusion. Confocal microscopy of HeLa cells coexpressing LC3-mCherry with the lysosomal marker lgp120-GFP revealed that although
rapamycin, BafA1, and azithromycin (16 μg/ml) treatment increased the total number of LC3+ vesicles per cell, the colocalization of LC3+ vesicles
with lgp120 (LC3+/lgp120+) was not significantly different from controls after azithromycin treatment, but increased with rapamycin and decreased
with BafA1 treatment. (B) Azithromycin impaired autophagosome acidification. HeLa cells stably expressing the mCherry-GFP-LC3 construct
were treated with either 8–16 μg/ml azithromycin or BafA1 for 24 hours and analyzed by confocal microscopy. Compared with the control, azithro-
mycin caused a significant reduction in the number of acidified LC3+ vesicles, but increased total LC3+ vesicles per cell. Number of vesicles was
normalized to control cells. **P < 0.005. Scale bars: 10 μm.

Page 5

research article
3558?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011

Page 6

research article
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
3559
significantly blocked IFN-γ– and TNF-α–dependent killing of
M. abscessus in primary human macrophages (Figure 6C). Further-
more, inoculation of whole blood with M. abscessus revealed that
azithromycin pretreatment, while resulting in dose-dependent
killing of presumably extracellular bacteria, blocked the ability
of the autophagy inducers rapamycin and IFN-γ to enhance the
clearance of mycobacteria (Figure 6D). These data suggest that
the effect of azithromycin on NTM infection in vivo will depend
on a balance between its direct antibiotic effects and its ability to
enhance intracellular survival of mycobacteria through block of
autophagosomal and phagosomal degradation.
Azithromycin promotes chronic infec-
tion with M. abscessus in vivo. Although
frequently sensitive to azithromycin
in vitro, M. abscessus becomes highly
macrolide resistant in vivo (Figure 6E)
through inducible expression of the
ribosomal methyltransferase erm(41)
(22), equivalent to erm(37) found in
M. tuberculosis, as well as through for-
mation of a protective biofilm (23).
To examine the effect of azithromy-
cin treatment on the host response to
mycobacteria in vivo, we infected mice
via aerosol challenge with azithromy-
cin-resistant M. abscessus (Figure 7). As
previously described (24), lung infec-
tion in control mice peaked at day 15,
with clearance of mycobacteria and
resolution of peribronchiolar inflam-
matory infiltrates by day 30. In con-
trast, azithromycin treatment led to
persistent lung infection associated
with extensive granulomatous inflam-
mation and failure of macrophage
degradation of intracellular mycobac-
teria (Figure 7, A–C). As anticipated,
azithromycin also influenced pul-
monary cytokine levels during infec-
tion (Figure 7D). Lung macrophages
(defined as CD11bhi) and dendritic
cells (CD11b–CD11chiDEC205hi) pro-
duced less TNF-α and IL-12, whereas
CD4+ and CD8+ effector T cells gener-
ated more IL-4 and less IFN-γ; the rea-
sons for this are unclear, but may be associated with altered antigen
presentation. This cytokine environment within the lung would
be expected to further inhibit autophagic killing by macrophages
(25, 26). These findings indicate that the ability of azithromycin to
inhibit autophagy may outweigh any direct bactericidal properties
and lead to chronic infection with NTM in vivo, explaining our own
clinical observation and those of others (12).
Discussion
We found that azithromycin blocked lysosomal degradation of
autophagosomes and phagosomes, which, while potentially con-
Figure 4
Azithromycin blocks autophagosome and phagosome acidification by impairing lysosomal function. (A) Prevention of IFN-γ–induced autophagic
flux. RAW 264.7 cells stably expressing mRFP-GFP-LC3 were treated with vehicle alone (as control), BafA1 (200 nM), azithromycin (40 μg/ml),
IFN-γ (200 ng/ml), or azithromycin plus IFN-γ. Representative images are shown. Quantification of acidified (mCherry+GFP–; red) and nonacidi-
fied (mCherry+GFP+; green) vesicles revealed a significant reduction in basal and IFN-γ–induced autophagosomal acidification by azithromycin
treatment. (B) Reduced acidification of lysosomes in primary human macrophages. Double-labeled (FITC and TMR) dextran was loaded into
lysosomes before cells were treated for 4 hours with azithromycin (80 μg/ml) or BafA1 (400 nM). As determined by confocal fluorescence analy-
sis, azithromycin significantly reduced lysosomal acidification compared with controls, as did BafA1. Intracellular pH calibrations were performed
as described in Supplemental Methods. (C) Azithromycin decreased acidification of phagosomes containing M. abscessus. Patient-derived
M. abscessus strains were heat-killed, double-labeled with FAM and Alexa Fluor 633, and incubated for 24 hours with primary human mac-
rophages untreated or treated with azithromycin (20 μg/ml) or BafA1 (100 nM). Representative confocal/DIC images of macrophages with
intracellular labeled M. abscessus show increased mycobacterial FAM fluorescence (indicating reduced acidification) after treatment with
azithromycin or BafA1. Also shown is quantification by flow cytometry of FAM fluorescence of macrophages that have internalized M. abscessus
(i.e., Alexa Fluor 633+). Corresponding phagosomal pH values (see Supplemental Methods) demonstrated significant alkalinization of phago-
somes with azithromycin or BafA1 treatment. Scale bars: 10 μm.
Figure 5
Azithromycin blocks phagosomal degradation and phagosome-lysosome fusion. (A) OVA-coated
beads were incubated with primary human macrophages (1-hour pulse, 23-hour chase). Internal-
ized beads were released, and the amount of OVA coating was quantified by flow cytometry after
incubation with fluorescent anti-OVA antibody. Treatment with azithromycin (20 μg/ml) significantly
reduced OVA degradation compared with untreated cells, to levels close to those achieved by BafA1
(100 nM) or leupeptin/pepstatin. (B) Human primary macrophages were incubated (1-hour pulse, 4-hour
chase) with TMR- and biotin-conjugated dextran to load lysosomes and then fed with IgG-coated
streptavidin-conjugated fluorescent latex beads (30-minute pulse, 2-hour chase) with no treatment or
in the presence of azithromycin (80 μg/ml) or BafA1 (400 nM). Beads were recovered by cell disrup-
tion, and the degree of bound dextran fluorescence was quantified by flow cytometry. Shown are a
representative histogram and average geometric mean fluorescence of triplicate samples.

Page 7

research article
3560?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
tributing to its antiinflammatory effects seen clinically, impaired
intracellular killing of mycobacteria and promoted chronic infec-
tion with NTM in vivo. Infection of CF patients with the multi-
drug-resistant NTM M. abscessus is a growing clinical problem,
leading to accelerated deterioration in lung function despite
aggressive antibiotic therapy (8, 9). Previous epidemiological stud-
ies (12) and our current analysis suggest an association between
long-term azithromycin use and development of M. abscessus infec-
tion, although definitive proof will require a prospective multi-
center study with standardized sputum sampling and culture
methods. Our findings provide a plausible mechanism for how
chronic azithromycin use might predispose to NTM infection in
CF patients and raise potential concerns about long-term macro-
lide treatment for other inflammatory lung diseases, use of mac-
rolides in population-based eradication programs for trachoma
and other infectious diseases (27), and clinical introduction of
nonantibiotic macrolides as antiinflammatory agents.
The net effect of azithromycin during clinical NTM infection
will depend on the balance between its direct antibiotic effect and
its ability to block intracellular killing of mycobacteria. For slow-
growing NTM species, such as the M. avium-intracellulare complex
(MAC), which do not possess inducible macrolide resistance genes,
azithromycin therapy (as part of multi-drug regimens) is usually
highly effective. Whether it remains so during long-term, low-dose,
intermittent treatment is unclear, particularly when concentra-
tions of macrolides within MAC-containing phagosomes are likely
to be reduced in vivo to below bacteriostatic levels (28). For rapid-
growing NTM species, such as M. abscessus, that can acquire mac-
rolide resistance (Figure 6E) through rapid induction of erm(41),
single amino acid ribosomal mutation (1), and biofilm formation
(23), the likely effect of long-term azithromycin therapy will be to
promote infection, as seen in our mouse infection model. Azithro-
mycin treatment also influenced cytokine production in vitro and
in vivo, leading to reduced levels of macrophage- and dendritic
cell–derived IL-12 and TNF-α and skewing T cell responses away
from IFN-γ and toward IL-4. Although the mechanism underlying
these effects remains to be elucidated, the net result will likely be
further inhibition of autophagic killing (25, 26).
We demonstrated that azithromycin blocked autophagy in
primary macrophages from stable CF patients as effectively
as in those from healthy controls. Although basal levels of the
autophagy substrate p62 were lower in CF macrophages, poten-
tially as a result of transcriptional regulation (11), its levels were
increased in both control and CF cells by azithromycin, indicative
of blocked autophagic clearance.
Although the ability of azithromycin to block autophagy could
be clinically exploitable in anticancer therapy, where tumor suscep-
tibility to chemotherapy may be enhanced (29), our data suggest
Figure 6
Azithromycin blocks intracellular killing of mycobacteria. (A and B) Human macrophages were infected with luminescent mycobacteria —
M. bovis BCG (A) and M. abscessus (B) — for 2 hours, washed, and exposed to the indicated concentrations of azithromycin (μg/ml) and/or 200 nM
rapamycin for 24 hours. Viable intracellular mycobacteria were then assessed by measuring cell-associated luminescence after cell lysis. (C)
IFN-γ and TNF-α enhanced intracellular killing of M. abscessus in human macrophages, and this was blocked by azithromycin pretreatment.
Results were normalized to levels obtained without cytokine addition. (D) Patient-derived M. abscessus strains were incubated with peripheral
blood from healthy subjects pretreated for 24 hours with vehicle alone or increasing concentrations of azithromycin. After 72 hours of incuba-
tion at 37°C, viable mycobacteria were quantified after sample lysis by counting CFUs. Addition of azithromycin, while reducing mycobacterial
growth (white), prevented any additional autophagy-dependent killing achieved by coincubation with IFN-γ (200 U/ml) or rapamycin (100 nM).
(E) Induction of macrolide resistance in M. abscessus. M. abscessus-lux was grown for 6 days in liquid culture (i.e., Middlebrook 7H9 plus ADC
enrichment) in the presence of 0.1 μg/ml azithromycin or vehicle alone, washed, and resuspended in RPMI with 10% FCS with the indicated
concentrations of azithromycin. Viable mycobacteria were assessed by measuring luminescence after 24 hours. *P < 0.05; **P < 0.005.

Page 8

research article
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
3561
that its impairment of autophagic degradation in macrophages
will compromise host immunity to mycobacteria and potentially
block the efficacy of adjuvant IFN-γ therapy often used in patients
with difficult NTM or M. tuberculosis infection. Our findings have
profound implications for the treatment of multi-drug-resistant
mycobacteria, including M. tuberculosis, in which the ability of
azithromycin to block intracellular bacterial killing is not accom-
panied by a direct antibiotic effect.
Methods
Further information can be found in Supplemental Methods.
Culture of cell lines and generation of human macrophages
HeLa and COS-7 cells were grown as previously described. Human mono-
cyte-derived macrophages were isolated from peripheral blood as previ-
ously described (30), obtained from healthy consented subjects (approved
by Regional NHS Research Ethics Committee), or from patients with CF
(clinically stable and not taking azithromycin).
Lysosomal and phagosomal pH measurements
Human primary macrophages were loaded with TMR-FITC dextran (1-hour
pulse, 4-hour chase). Cells were then treated with compounds and visu-
alized by live cell fluorescence/differential interference contrast (fluores-
cence/DIC) confocal imaging. To determine the phagosomal pH, primary
macrophages were incubated (1-hour pulse, 23-hour chase) with albumin-
coated, FAM-conjugated 645-nm (Far Red) latex beads, patient-derived
heat-killed M. abscessus strains double-labeled with FAM and Alexa Fluor
633, or PFA-killed M. tuberculosis H37Rv double-labeled with FITC and
Alexa Fluor 633. Macrophages were pretreated for 24 hours with com-
pounds. Analysis was carried out by either live cell fluorescence/DIC con-
focal imaging or flow cytometry, in which the fluorescence of cells with
internalized particles (FarRed or Alexa Fluor 633 for beads and mycobac-
teria, respectively) was measured. Intracellular calibration was achieved as
detailed in Supplemental Methods.
Phagosomal degradation assays
Human macrophages were incubated (30-minute pulse, 2-hour chase)
with OVA-conjugated fluorescent latex beads at 4°C, 37°C, or 37°C with
15-minute protease inhibitor pretreatment. Macrophages were pretreated
with BafA1 or azithromycin for 24 hours prior to phagocytosis; treatment
continued throughout the experiment. Beads were then recovered by cell
disruption in the presence of proteases inhibitors, then incubated with
OVA-specific biotinylated antibody and fluorescent streptavidin, which
was quantified by flow cytometry.
Figure 7
Azithromycin promotes M. abscessus infection in vivo. (A) C57BL/6 mice were infected by aerosol challenge with azithromycin-resistant
M. abscessus. Treatment with azithromycin (100 mg/kg) by gavage for 5 days per week resulted in failure of M. abscessus clearance from lungs. (B)
Representative lung histology demonstrating that azithromycin treatment led to persistent lung infection associated with extensive granulomas (arrows)
and peribronchiolar inflammation at day 30. Ziehl-Neelsen staining confirmed the presence of large numbers of intracellular mycobacteria in azithromy-
cin-treated, but not control, animals. Scale bars: 200 μm (histology); 10 μm (Ziehl-Neelsen). (C) Extent and severity of lung lesions in M. abscessus–
infected mice at days 15 and 30 of treatment with vehicle alone or azithromycin. (D) Assessment of intracellular cytokine profiles for lung macro-
phages, dendritic cells, and CD4+ and CD8+ effector T cells during infection in the presence or absence of chronic azithromycin treatment.

Page 9

research article
3562?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
Phagosome-lysosome fusion assays
Human macrophages were incubated (1-hour pulse, 4-hour chase) with
TMR- and biotin-conjugated dextran to load lysosomes and then fed
IgG-coated, streptavidin-conjugated fluorescent latex beads (30-minute
pulse, 2-hour chase) with no treatment or in the presence of azithromycin
(80 μg/ml) or BafA1. Beads were recovered by cell disruption, the degree of
bound fluorescent dextran was quantified by flow cytometry, and average
geometric mean fluorescence was determined.
In vitro mycobacterial infection assays
Primary human macrophages. Macrophages were inoculated with myco-
bacteria (MOI 10:1) for 2 hours at 37°C, repeatedly washed, and then
incubated for 24 hours at 37°C in the presence or absence of azithro-
mycin or rapamycin, as indicated. Cells were finally lysed, and lumines-
cence was measured.
Whole blood mycobacterial infection assays. Patient-derived M. abscessus
strains were grown to log phase in Middlebrook 7H9 media. Heparinized
peripheral blood from healthy consented subjects was inoculated with
M. abscessus after pretreatment for 24 hours with vehicle alone or azithro-
mycin and incubated for 72 hours at 37°C. Viable mycobacteria were quan-
tified (CFU) on microbial plates. Where indicated, IFN-γ or rapamycin were
added to blood at the time of infection with mycobacteria.
Induction of macrolide resistance in vitro. M. abscessus-lux was grown in liquid
culture in the presence of 0.1 μg/ml azithromycin or vehicle alone, washed,
and resuspended in RPMI with 10% FCS with the indicated concentrations
of azithromycin. Viable mycobacteria were assessed by measuring lumines-
cence after 24 hours.
In vivo mycobacterial infection
Mice were challenged with M. abscessus as previously described (24). Control
(saline) and azithromycin groups (n = 5 per group) were treated 5 days per
week at a concentration of 100 mg/kg by gavage. On days 15 and 30 after
infection, bacterial loads in the lungs and spleen were determined by his-
tological and flow cytometry analysis (see Supplemental Methods). Bacte-
rial counts were determined by plating serial dilutions of homogenates of
lungs after 3–7 days.
Statistics
Densitometric analysis on the immunoblots was done by Image J soft-
ware. P values for the Cellomics array scan counting and for other experi-
ments requiring multiple comparisons were determined by factorial
ANOVA using STATVIEW software (version 4.53). P values for all other
assays were determined using 2-tailed Student’s t test. A P value of 0.05 or
less was considered significant. Experiments were performed on at least 3
separate occasions with at least triplicate samples for each condition and
represented as mean ± SD.
To test for an association between NTM status and azithromycin, we used
a logistic regression model adjusted for covariates. Forward logistic regres-
sion was used to select covariates associated with NTM status (P < 0.05) from
age, forced expiratory volume in 1 second, BMI, sex, colonization with Pseu-
domonas aeruginosa, CF-related liver disease, CF-related diabetes, CF-related
asthma, allergic bronchopulmonary aspergillosis, and gastroesophageal
reflux disease. Analyses were performed in R (http://www.r-project.org).
Study approval
All experimental protocols were approved by the Animal Care and Usage
Committee of Colorado State University and complied with NIH guidelines.
Acknowledgments
This work was supported by grants from the Wellcome Trust
(Senior Clinical Fellowships to R.A. Floto and D.C. Rubinsztein),
the NIHR Biomedical Research Centre at Addenbrooke’s Hos-
pital, Papworth Hospital NHS Trust, and the Jodi and Lucinda
Dunmore Foundation. The authors thank Juliet Foweraker and
Christian Laughton for help with microbiology and Jane Elliott
and Judy Ryan for help with clinical data collection.
Received for publication December 14, 2010, and accepted in
revised form June 8, 2011.
Address correspondence to: R. Andres Floto, Cambridge Insti-
tute for Medical Research, University of Cambridge, Wellcome
Trust/MRC Building, Hills Road, Cambridge CB2 0XY, United
Kingdom. Phone: 44.1223.768801; Fax: 44.1223.768801; E-mail:
arf27@cam.ac.uk. Or to: David C. Rubinsztein, Cambridge Insti-
tute for Medical Research, University of Cambridge, Wellcome
Trust/MRC Building, Hills Road, Cambridge CB2 0XY, United
Kingdom. Phone: 44.1223.762608; Fax: 44.1223.768801; E-mail:
dcr1000@cam.ac.uk. Or to: Diane Ordway, Colorado State
University, Mycobacteria Research Laboratory, Department
of Microbiology, Immunology, and Pathology, 1682 Campus
Delivery, 200 West St., Fort Collins, Colorado 80523-1682, USA.
Phone: 970.491.4117; Fax: 970.491.1815; E-mail: D.Ordway-
Rodriguez@colostate.edu.
1. Griffith DE, et al. An official ATS/IDSA statement:
diagnosis, treatment, and prevention of nontuber-
culous mycobacterial diseases. Am J Respir Crit Care
Med. 2007;175(4):367–416.
2. McArdle JR, Talwalkar JS. Macrolides in cystic
fibrosis. Clin Chest Med. 2007;28(2):347–360.
3. Martinez FJ, Curtis JL, Albert R. Role of macrolide
therapy in chronic obstructive pulmonary disease.
Int J Chron Obstruct Pulmon Dis. 2008;3(3):331–350.
4. Poletti V, Casoni G, Chilosi M, Zompatori M.
Diffuse panbronchiolitis. Eur Respir J. 2006;
28(4):862–871.
5. Piacentini GL, et al. Azithromycin reduces bron-
chial hyperresponsiveness and neutrophilic airway
inflammation in asthmatic children: a preliminary
report. Allergy Asthma Proc. 2007;28(2):194–198.
6. Yates B, et al. Azithromycin reverses airflow obstruc-
tion in established bronchiolitis obliterans syndrome.
Am J Respir Crit Care Med. 2005;172(6):657–659.
7. Idris SF, Chilvers ER, Haworth C, McKeon D,
Condliffe AM. Azithromycin therapy for neu-
trophilic airways disease: myth or magic? Thorax.
2009;64(3):186–189.
8. Esther CR Jr, Esserman DA, Gilligan P, Kerr A,
Noone PG. Chronic Mycobacterium abscessus infec-
tion and lung function decline in cystic fibrosis.
J Cyst Fibros. 2010;9(2):117–123.
9. Roux AL, et al. Multicenter study of prevalence
of nontuberculous mycobacteria in patients
with cystic fibrosis in France. J Clin Microbiol.
2009;47(12):4124–4128.
10. Vergne I, et al. Autophagy in immune defense
against Mycobacterium tuberculosis. Autophagy.
2006;2(3):175–178.
11. Klionsky DJ, et al. Guidelines for the use and inter-
pretation of assays for monitoring autophagy in
higher eukaryotes. Autophagy. 2008;4(2):151–175.
12. Levy I, et al. Multicenter cross-sectional study of
nontuberculous mycobacterial infections among
cystic fibrosis patients, Israel. Emerg Infect Dis.
2008;14(3):378–384.
13. Kabeya Y, et al. LC3, a mammalian homologue
of yeast Apg8p, is localized in autophagosome
membranes after processing. EMBO J. 2000;
19(21):5720–5728.
14. Ravikumar B, et al. Inhibition of mTOR induces
autophagy and reduces toxicity of polyglutamine
expansions in fly and mouse models of Hunting-
ton disease. Nat Genet. 2004;36(6):585–595.
15. Moscat J, Diaz-Meco MT. P62 at the crossroads
of autophagy, apoptosis, and cancer. Cell. 2009;
137(6):1001–1004.
16. Wilms EB, Touw DJ, Heijerman HG. Pharmaco-
kinetics and sputum penetration of azithromy-
cin during once weekly dosing in cystic fibrosis
patients. Ther Drug Monit. 2006;28(2):219–225.
17. Narain Y, Wyttenbach A, Rankin J, Furlong RA,
Rubinsztein DC. A molecular investigation of true
dominance in Huntington’s disease. J Med Genet. 1999;
36(10):739–746.
18. Webb JL, Ravikumar B, Atkins J, Skepper JN,
Rubinsztein DC. Alpha-Synuclein is degraded by
both autophagy and the proteasome. J Biol Chem.
2003;278(27):25009–25013.
19. Jahreiss L, Renna M, Bittman R, Arthur G, Rubin-
sztein DC. 1-O-Hexadecyl-2-O-methyl-3-O-(2’-acet-

Page 10

research article
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 9 September 2011
3563
amido-2’-deoxy-β-D-glucopyranosyl)-sn-glycerol
(Gln) induces cell death with more autophago-
somes which is autophagy-independent. Autophagy.
2009;5(6):835–846.
20. Gutierrez MG, Master SS, Singh SB, Taylor GA,
Colombo MI, Deretic V. Autophagy is a defense
mechanism inhibiting BCG and Mycobacterium
tuberculosis survival in infected macrophages. Cell.
2004;119(6):753–766.
21. Di A, et al. CFTR regulates phagosome acidifica-
tion in macrophages and alters bactericidal activity.
Nat Cell Biol. 2006;8(9):933–944.
22. Nash KA, Brown-Elliott BA, Wallace RJ. A novel
gene, erm(41), confers inducible macrolide resis-
tance to clinical isolates of Mycobacterium absces-
sus but is absent from Mycobacterium chelonae.
Antimicrob Agents Chemother. 2009;53(4):1367–1376.
23. Greendyke R, Byrd TF. Differential antibiotic sus-
ceptibility of Mycobacterium abscessus variants
in biofilms and macrophages compared to that of
planktonic bacteria. Antimicrob Agents Chemother.
2008;52(6):2019–2026.
24. Ordway D, et al. Animal model of Mycobacte-
rium abscessus lung infection. J Leukoc Biol. 2008;
83(6):1502–1511.
25. Singh SB, Davis AS, Taylor GA, Deretic V. Human
IRGM induces autophagy to eliminate intracellular
mycobacteria. Science. 2006;313(5792):1438–1441.
26. Harris J, et al. T Helper 2 cytokines inhibit autoph-
agic control of intracellular Mycobacterium tuber-
culosis. Immunity. 2007;27(3):505–17.
27. Mecaskey JW, Knirsch CA, Kumaresan JA, Cook JA.
The possibility of eliminating blinding trachoma.
Lancet Infect Dis. 2003;3(11):728–734.
28. Fréhel C, Offredo C, de Chastellier C. The phago-
somal environment protects virulent Mycobacte-
rium avium from killing and destruction by clar-
ithromycin. Infect Immun. 1997;65(7):2792–2802.
29. Kondo Y, Kanzawa T, Sawaya R, Kondo S. The role
of autophagy in cancer development and response
to therapy. Nat Rev Cancer. 2005;5(9):726–734.
30. Floto RA, et al. Loss of function of a lupus-
associated FcgammaRIIb polymorphism
through exclusion from lipid rafts. Nat Med.
2005;11(10):1056–1058.