DMM Advance Online Articles. Posted 14 December 2016 as doi: 10.1242/dmm.027086
Access the most recent version at http://dmm.biologists.org/lookup/doi/10.1242/dmm.027086
Modelling bronchopulmonary dysplasia in mice: how much
oxygen is enough?
Claudio Nardiello1,2, Ivana Mižíková1,2 Diogo M. Silva1,2, Jordi Ruiz-Camp1,2, Konstantin
Mayer2, István Vadász2, Susanne Herold2, Werner Seeger1,2, and Rory E. Morty1,2,*
1
Department of Lung Development and Remodelling, Max Planck Institute for Heart and
Lung Research, Bad Nauheim, Germany. 2Department of Internal Medicine (Pulmonology),
University of Giessen and Marburg Lung Center (UGMLC), member of the German Center
for Lung Research (DZL), Giessen, Germany
*Correspondence to:
Rory E. Morty
Department of Lung Development and Remodelling
Max Planck Institute for Heart and Lung Research
Parkstrasse 1
D-61231 Bad Nauheim
Germany
Tel: +49 6032 705 271
Fax: +49 6032 705 360
Keywords:
BPD, hyperoxia, alveolarisation, structure, animal model
Summary Statement:
A newborn mouse model of bronchopulmonary dysplasia is
presented, where oxygen injury was systematically standardised to recapitulate the two
pathological hallmarks of disease, while minimising animal stress insofar as possible.
© 2016. Published by The Company of Biologists Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction
in any medium provided that the original work is properly attributed.
Disease Models & Mechanisms • DMM • Advance article
e-mail: [email protected]
ABSTRACT
Progress in developing new therapies for bronchopulmonary dysplasia (BPD) is sometimes
complicated by the lack of a standardised animal model. Our objective was to develop a
robust hyperoxia-based mouse model of BPD that recapitulated the pathological perturbations
to lung structure noted in infants with BPD. Newborn mouse pups were exposed to a varying
fraction of oxygen in the inspired air (FiO2) and a varying window of hyperoxia exposure,
after which lung structure was assessed by design-based stereology with systemic uniform
random sampling. The efficacy of a candidate therapeutic intervention using parenteral
nutrition was evaluated to demonstrate the utility of the standardised BPD model for drug
discovery. An FiO2 0.85 for the first 14 days of life decreased total alveoli number and
concomitantly increased alveolar septal wall thickness, which are two key histopathological
characteristics of BPD. A reduction in FiO2 to 0.60 or 0.40 also caused a decrease in the total
alveoli number, but the septal wall thickness was not impacted. Neither a decreasing oxygen
gradient (from FiO2 0.85 to 0.21 over the first 14 days of life) nor an oscillation in FiO2
(between 0.85 and 0.40 on a 24 h:24 h cycle) had an appreciable impact on lung development.
The risk of missing beneficial effects of therapeutic interventions at FiO2 0.85, using
parenteral nutrition as an intervention in the model was also noted; highlighting the utility of
experimental intervention. Thus, a state-of-the-art BPD animal model that recapitulated the
two histopathological hallmark perturbations to lung architecture associated with BPD is
described. The model presented here, where injurious stimuli have been systematically
evaluated, provides the most promising approach for the development of new strategies to
drive post-natal lung maturation in affected infants.
Disease Models & Mechanisms • DMM • Advance article
lower FiO2 in selected studies, and underscoring the need to tailor the model employed to the
INTRODUCTION
Precise modelling of human disease using animal models is a major challenge in translating
bench science to the bedside, since (i) animal models must accurately recapitulate disease
processes to facilitate the identification of pathogenic pathways, and (ii) animal models
represent the limiting step in assessing which therapeutic interventions hold promise for
subsequent study. This is particularly evident in animal models of human diseases that are
characterized by perturbations to the architecture of an organ. Modelling disease pathogenesis
in experimental animals is problematic from multiple perspectives. Amongst these (i) the
injurious insult employed in the experimental model may not recapitulate key elements of
disease, thereby limiting the ability to evaluate the efficacy of candidate therapeutic agents.
Furthermore, (ii) the precision of the readout that is employed may be inadequate to detect
small changes in anatomical structures that are targeted by both the injurious insult and
candidate therapeutic intervention. A further confounding variable is the use of experimental
animals in medical research, as emphasis must be placed on “reduction, refinement, and
replacement” (the 3R concept) (Curzer et al., 2016), where the number of experimental
animals employed and the level of stress to which the animal is subjected must be maintained
at the minimum level possible, while still retaining the translational viability of the animal
model.
illustration of these concerns. BPD is the most common complication of preterm birth and
represents significant morbidity and mortality in the neonatal intensive care unit (Jobe, 2011;
Jobe and Tibboel, 2014). BPD is caused by a combination of the toxic effects of oxygen
supplementation used to manage respiratory failure in preterm infants, baro- and volu-trauma
from mechanical ventilation (Greenough et al., 2008), as well as other disease-modifying
variables such as infection and inflammation (Balany and Bhandari, 2015). BPD results in
long-term complications in respiratory function that persist into adulthood (Hilgendorff and
O'Reilly, 2015). The pathology of BPD has changed over time, where “old” BPD, which is
particularly characterized by fibrosis, thickened septa, and some alveolar simplification
results from aggressive mechanical ventilation with high oxygen levels. In contrast, “new”
BPD, which is the prevalent form today, is largely characterized by alveolar simplification,
where preterm infants are less aggressively ventilated, with lower oxygen levels (Jobe, 2011).
Due to improvements in the medical management of BPD, the incidence of BPD is increasing
because preterm infants delivered earlier have increasingly improved survival (Stoll et al.,
Disease Models & Mechanisms • DMM • Advance article
Modelling bronchopulmonary dysplasia (BPD) in experimental animals is a textbook
2015). This underscores a pressing need to develop new medical management strategies.
However, these efforts might be hampered by the lack of appropriate animal models of BPD.
In affected patients, two key elements of the lung structure are impacted: the oxygen
supplementation arrests lung development, which causes fewer alveoli of a larger size to be
generated. Concomitantly, the thickness of the delicate barrier between the alveolar airspaces
in the lung and the capillary network of the lung is thickened, which compromises gas
exchange, and is evident by thickened alveolar septal walls. In order to properly model BPD,
the injurious intervention (oxygen supplementation) must recapitulate both structural elements
of the pathology: a decreased number of alveoli and an increased alveolar septal wall
thickness.
To this end, a large number of methodologies have been reported which rely on the
exposure of newborn mouse pups to an increased fraction of oxygen in the inspired air (FiO2),
often reported as a percentage of oxygen in the inspired air, which ranges between 40% O2
(FiO2 0.4) and 100% O2 (FiO2 1.0) (Silva et al., 2015). Between 01 January 2013 and 30 June
2015, 41 different oxygen exposure protocols had been reported in BPD animal models (Silva
et al., 2015), sometimes with diametrically opposite findings concerning the effects of the
same intervention in different animal models of BPD [for example, compare the studies of
(Britt et al., 2013) and (Masood et al., 2014)]. This underscores the impact of the oxygen
exposure regimen selected on data analysis and study conclusions. Of course, it must be
has a multifactorial aetiology that cannot be fully modelled in a single, standardised animal
model.
The analysis of lung structure in material from animal models of BPD is also
problematic (Silva et al., 2015). Currently, a mixture of approaches are employed, often based
on the analysis of paraffin-embedded lung tissue with classical determinants of mean linear
intercept (MLI) and radial alveolar count (RAC) as surrogates of the size of the alveoli (Silva
et al., 2015). These determinations are made by direct measurement of the distance between
the adjacent walls of an alveolus. Similarly, the thickness of the alveolar wall is directly
assessed using a slide-rule and visual inspection, which is not unbiased. Furthermore, the
highly elastic structure of the lung results in substantial distortion of the lung structure during
the dehydration and rehydration of the lung during paraffin embedding (Schneider and Ochs,
2014). These are all important concerns in the analysis of the lung architecture, although these
approaches have been successfully used to identify potentially important pathogenic pathways
Disease Models & Mechanisms • DMM • Advance article
acknowledged that diversity in animal models is also a strength of the BPD field, where BPD
and for the pre-clinical evaluation of candidate therapeutic interventions (Liao et al., 2015;
Olave et al., 2015; Tibboel et al., 2013).
To address the concerns outlined above, important advances have been made replacing
paraffin embedding with plastic embedding, along with treatment of lung tissue with arsenic,
osmium, and uranium, which results in appreciable preservation of the lung structure
(Schneider and Ochs, 2014). Furthermore, a design-based stereological approach to the
analysis of organ structure has been developed, and continues to be refined (Schneider and
Ochs, 2013; Tschanz et al., 2014). This approach represents a substantial advance over the
MLI and RAC methods when applied to the lung, as the technique is both unbiased (thus not
subjective), and has a very high resolution (Mühlfeld and Ochs, 2013; Ochs and Mühlfeld,
2013). Design-based stereology has recently been applied to the analysis of the architecture of
developing lungs of newborn mouse pups by the authors, for the first time allowing the
assessment of the total number of alveoli in the lung, and the assessment of changes as <1 m
in magnitude during perturbations to newborn mouse lung development (Madurga et al.,
2015; Madurga et al., 2014; Mižíková et al., 2015).
These recent developments in tissue embedding and structural analysis represent an
important advance in our ability to study lung development. Here, we employed a
state-of-the-art tissue embedding methodology, together with state-of-the-art design-based
stereology to identify the magnitude of oxygen toxicity and the window of oxygen exposure
parameters selected were a decrease in total alveoli number in the lung, and an increase in the
alveolar septal wall thickness, which are the two anatomical hallmarks of BPD. We then
documented the extraordinary impact of using different oxygen levels in the experimental
model, on the observed efficacy of a candidate intervention; thus highlighting the critical
importance of selecting the correct model to evaluate experimental therapeutics.
RESULTS
To identify the optimal injurious levels of oxygen, as well as the optimal window of exposure
to injury, a series of oxygen exposure protocols were employed, which varied the FiO2 over
the first 14 days of life (Fig. 1A) in a variety of windows of exposure (Fig. 1B).
Disease Models & Mechanisms • DMM • Advance article
that is required to model BPD as perfectly as possible in mice. The primary threshold
Identification of the optimal level of oxygen required to model BPD in mice
Fourteen days after birth [post-natal day (P)14], the bulk of post-natal lung alveolarization has
been completed (Burri, 2006). Thus, maintaining newborn mouse pups (n=5 per experimental
group) under room-air (21% O2) conditions, starting on the day of birth, up to and including
P14, served as the control protocol and generated mouse lungs that exhibited an anatomically
normal alveolar structure (Fig. 2A,G). Increasing the concentration of oxygen in the inspired
air to 40% O2 over the first 14 days of life generated a less organised lung parenchymal
architecture (Fig. 2B,H vs. Fig, 2A,G) and decreased the total number of alveoli in the lung by
39% (Fig. 2M) in comparison to 21% O2 (control) group (complete data set in Table 1). This
was accompanied by a concomitant 25% decrease in the alveolar density (Fig. 2N) and a 20%
decrease in the gas exchange surface area (Fig. 2O), while no impact on alveolar septal wall
thickness was noted (Fig. 2Q). Increasing the concentration of oxygen in the inspired air to
60% O2 over the first 14 days of life did not further impact the visual appearance of the lung
structure (Fig. 2C,I), which exhibited a comparable number of alveoli, alveolar density, gas
exchange surface area, and alveolar septal wall thickness to the 40% O2 group (Fig. 2M-R),
however, the mean MLI was increased in the 60% O2 group (Fig. 2P). In contrast, newborn
mouse pups exposed to 85% O2 for the first 14 days of life exhibited further reductions of
69% in number of alveoli (Fig. 2M), 69% in alveolar density (Fig. 2N) and 37% in gas
alveolar septal wall thickness of 25% in comparison to 21% O2 controls was also noted (Fig.
2Q). Thus, 85% O2 was the only oxygen concentration that impacted both total number of
alveoli as well as alveolar septal wall thickness. For this reason, we performed two additional
studies: the exposure of newborn mouse pups to a gradient of decreasing oxygen
concentration from 85% O2 on the day of birth to 21% O2 on P14 (amounting to a step-wise
decrease of 5% O2 per day), as well as an oscillation between 85% O2 and 40% O2 on a
24 h:24 h oscillating cycle (Fig. 1A).
The exposure of newborn mouse pups to a decreasing gradient of oxygen between
85% O2 and 21% O2 over the first 14 days of life had no impact on any parameter of lung
architecture (Fig. 2E,K, Fig. 2M-R). In contrast, exposure of newborn mouse pups to an
oscillation of 85% O2 and 40% O2 over the first 14 days of life generated lungs that were
impacted by visual inspection, albeit lightly (Fig. 2F,L); and which exhibited a comparatively
moderate reduction of 25% in the total number of alveoli in the lung (Fig. 2M) and 20%
reduction in alveolar density (Fig. 2N), which was insufficient to significantly impact the gas
Disease Models & Mechanisms • DMM • Advance article
exchange surface area (Fig. 2O) compared to the 21% O2 (control) group. An increase in
exchange surface area, which remained unchanged (Fig. 2O). Similarly, alveolar septal wall
thickness (Fig. 2Q) was unaffected. Changes in the MLI parameter largely paralleled those
noted in other lung structure parameters (Fig. 2P). It is important to note that the MLI
reported here represents a stereologically-determined mean MLI over the whole lung, and not
the potentially biased MLI determined by visual inspection of selected lung sections, which is
widely reported in the literature. No change in lung volume was noted at P14 comparing all
experimental groups (Fig. 2R), and none of the parameters assessed exhibited any clustering
on the basis of the sex of the animals (Fig. 2M-R).
Together, these data demonstrated that exposure of newborn mouse pups to as little as
40% O2 over the first 14 days of life could recapitulate the changes seen in lung alveolar
number in patients with BPD, however, only exposure to 85% O2 could also recapitulate the
changes noted in lung alveolar septal wall thickness in patients with BPD.
Identification of the optimal window of oxygen exposure required to model BPD in mice
Given that only exposure of newborn mouse pups to 85% O2 over the first 14 days of life
could recapitulate the two hallmark perturbations to lung architecture seen in BPD, we then
set out to assess the critical (and minimal) window of exposure to 85% O2, comparing five
different windows of exposure (Fig. 1B) over the first 14 days of life (n=5 per experimental
group). Exposure of newborn mouse pups to 85% O2 for the first 24 h of life only (thus,
(Fig. 3B,H), or a design-based stereology analysis of the number of alveoli in the lung, the
alveolar density, the gas exchange surface area, the MLI, or the alveolar septal wall thickness
(Fig. 3M-R; complete data sets in Table 2). Increasing the window of exposure to include the
first 72 h of life, from P1 up to and including P3, followed by 11 days of 21% O2 (Fig. 3C,I)
generated a moderate (14%) decrease in alveolar density in comparison to the 21% O2 group
(Fig. 3N), without impacting any other parameter. When an alternative window of exposure,
where mouse pups were exposed to 85% O2 over a time-frame starting with and including P4,
up to and including P7 was selected, a 32% decrease in the number of alveoli (Fig. 3M) and a
39% decrease in the alveolar density (Fig. 3N) in comparison to the 21% O2 group was noted.
While these changes were not sufficient to impact the gas exchange surface area (Fig. 3O), an
increase in alveolar septal wall thickness of 19%, in comparison to 21% O2 was also noted
(Fig. 3Q).
Disease Models & Mechanisms • DMM • Advance article
followed by 13 days of 21% O2) had no impact on lung structure, either by visual inspection
Exposure of newborn mouse pups either to (i) 85% O2 from P1 to P7, followed by
seven days of 21% O2 (Fig. 3E,K) or (ii) to 85% O2 from P1 to P14 (Fig. 3F,L) generated the
most dramatic impact on the lung architecture. A decrease in number of alveoli, alveolar
density, and alveolar surface area, as well as an increase in MLI and alveolar septal wall
thickness was noted in both groups (Fig. 3M-R), suggesting that these two protocols best
model BPD in mice.
Effect of oxygen exposure on the performance of a candidate therapeutic intervention
Cottonseed oil was employed as an intervention, for parenteral nutrition, and exhibited no
impact on the structural development of newborn mouse lungs over the period P1 to P14 in
mouse pups (n=5 per experimental group) maintained under 21% O2 compared to the lungs of
control sham-treated mice maintained under 21% O2 (Fig. 4C,D vs. Fig. 4A,B; Fig. 4M-R;
complete data set in Table 3). Cottonseed oil tended to increase body mass (by up to 20%),
and had no impact on survival (data not shown). When cottonseed oil was applied to newborn
mouse pups exposed to 60% O2 over the period P1-P14, a pronounced recovery in the number
of alveoli (an increase of 78%) was noted compared to sham-treated pups in the same oxygenexposure protocol (Fig. 4G,H vs. Fig. 4E,F; Fig. 4M). Similarly, cottonseed oil application
increased the alveolar density by 31% (Fig. 4N) and increased the gas exchange surface area
lung volume of 33% (Fig. 4R). Since alveolar septal wall thickness was not impacted at
60% O2 (Fig. 2Q) no impact of cottonseed oil administration on alveolar septal wall thickness
was expected, nor was any effect noted (Fig. 4Q). These data highlight the therapeutic benefit
of cottonseed oil supplementation in this experimental animal model of BPD. However, when
the more severe hyperoxia exposure model was employed, namely continuous exposure of
newborn mouse pups to 85% O2 from P1 to P14, the beneficial impact of cottonseed oil
administration on alveolar number and alveolar density was lost (Fig. 4K,L vs. Fig. 4I,J; Fig.
4M,N). However, the impact of cottonseed oil on mean septal wall thickness was still noted
(Fig. 4Q). Taken together, these data highlight a key concern of the 85% O2 exposure
approach, where the extreme severity of the injurious stimulus may result in promising
candidate therapeutic interventions being missed or discounted.
Disease Models & Mechanisms • DMM • Advance article
by 55% (Fig. 4O) in the 60% O2 group. These changes were accompanied by an increase in
DISCUSSION
Given the increasing clinical burden of BPD (Jobe, 2011; Stoll et al., 2015), there is a
pressing need for new therapeutic options for this syndrome, which includes the identification
of novel pathogenic pathways and new therapeutic options. Progress in this regard has been
impeded by a lack of suitable animal models, which represent the first step in the translational
pipeline. While nonhuman primates are a better laboratory model for these studies than are
rodents (Herring et al., 2014), ethical considerations, and monetary and time costs support the
use of rodents as first-line models for drug discovery and studies on disease pathogenesis.
Thus, mice have found widespread application in BPD models (Berger and Bhandari, 2014),
particularly because of the availability of transgenic mouse lines. Most mouse BPD models
are based on exposure of newborn pups to elevated oxygen levels, since oxygen toxicity to the
developing lung is the medical basis of BPD in preterm infants (Jobe, 2011; Stoll et al., 2015);
and oxygen toxicity is known to disrupt elements of alveolar development, including the
maturation of the extracellular matrix (Mižíková and Morty, 2015), microRNA dynamics
(Nardiello and Morty, 2016), as well as stem cell plasticity (Domm et al., 2015; Yee et al.,
2016). However, there is a pronounced lack of standardisation of BPD mouse models, with
over 41 different oxygen exposure protocols reported in a 30-month period alone (Silva et al.,
2015). Although initially very high (90%-100%) oxygen concentrations have been employed
gradual shift in the degree of oxygen toxicity employed, with increased instances of oxygen
concentrations below 80% O2 becoming increasingly evident in recent published studies
(Bouch et al., 2016; Bouch et al., 2015; Ehrhardt et al., 2016; Maduekwe et al., 2015;
Reyburn et al., 2016; Sozo et al., 2015; Wang et al., 2014; Yee et al., 2016). In some
instances, the lack of model standardisation may have resulted in opposite findings that
evaluate the same candidate therapeutic intervention (Silva et al., 2015). These concerns
might suggest the benefits of a standardised model of BPD.
Recent developments in the stereological analysis of tissue and organ structure have
provided a tool with extraordinarily high precision that can be used to study pathological
changes – even of a very small magnitude – to the architecture of a target organ. Similarly,
these design-based stereology approaches may also be used to evaluate the efficacy of
candidate therapeutic interventions that can correct pathological perturbations to organ
structure. To this end, specific guidelines exist for the use of design-based stereology to study
lung structure (Hsia et al., 2010), and the design-based stereology approach has been refined
Disease Models & Mechanisms • DMM • Advance article
as an injurious stimulus in BPD models, and indeed, still are by many groups; there has been a
by the authors for the study of normal and aberrant lung development in the newborn mouse
(Madurga et al., 2015; Madurga et al., 2014; Mižíková et al., 2015).
Here, we report the use of design-based stereology to systematically explore the
impact of oxygen on post-natal mouse lung development, considering both the level of
oxygen in the inspired air, and the window of oxygen exposure. As in all organs, the structure
of the lung is intimately related to lung function (Hsia et al., 2016), and lung structure in
infants with BPD exhibit two pathological hallmarks: disturbance to the gas exchange
structure (cumulatively reflected by a change in total number of alveoli, alveolar density, and
gas exchange surface area), as well as thickening of the alveolar septal walls. Both lung
structural elements of BPD are recapitulated in the continuous exposure to 85% O2 for the
first 14 days of life.
We have selected P14 at the end-point for our studies as P14 is the time-point at which
the bulk of secondary septation is largely completed, and as such, perturbations – even subtle
perturbations – to secondary septation will be suitably amplified and (relatively) easy to detect
at P14. Given that secondary septation is broadly thought to extend at least to P10, P14
appeared to be a suitable time-point. Furthermore, reference to Silva et al. (2015) revealed
that P14 is the most commonly used end-point of all, for hyperoxia-based studies. As such,
this as the best time-point to terminate the studies presented here, to facilitate comparisons
with the work of others. Our data have revealed some interesting effects of oxygen exposure
40% O2 over the first 14 days of post-natal life did impact post-natal lung development, with
continuous exposure to 40% O2, 60% O2, and 85% O2 having an impact on the number of
alveoli (Fig. 2M), alveolar density (Fig. 2N), and gas exchange surface area (Fig. 2O), while
only continuous exposure to 85% O2 impact mean septal wall thickness (Fig. 2Q). In general,
the impact of 40% O2 and 60% O2 were largely comparable considering the effect of oxygen
exposure on number of alveoli and mean septal wall thickness. Some unexpected observations
were also made. Notably, a decreasing O2 gradient from 85% O2 at P1 to 21% O2 at P14 was
without any impact on number of alveoli (Fig. 2M) or mean septal wall thickness (Fig. 2Q),
although the cumulative oxygen exposure exceeded that of the continuous 40% O2 group
where both number of alveoli and mean septal wall thickness were impacted. It is speculated
that the immediate (after birth) and dramatic (85% O2) oxygen toxicity recruited lung
protective (anti-oxidant) pathways, which were progressively titrated down over time,
commensurate with progressively decreasing O2 levels, thus maintaining the lung protective
strategies at levels that were protective, but never damaging. Along similar lines, oscillating
Disease Models & Mechanisms • DMM • Advance article
protocols on the post-natal development of the mouse lung. Continuous exposure to as little as
on a 24 h:24 h cycle between 85% O2 and 40 % O2 also represented a cumulative oxygen
exposure that was greater than the continuous 40% O2 protocol, however, the degree of
damage to the lung was comparable, perhaps also indicating the recruitment and derecruitment of lung protective strategies by sudden exposure to dramatic changes in inspired
O2 levels.
The data presented here suggest that exposure of newborn mouse pups either to (i)
85% O2 from P1 to P7, followed by seven days of 21% O2 (Fig. 3E,K) or (ii) to 85% O2 from
P1 to P14 (Fig. 3F,L) generated the most dramatic impact on the lung architecture. Both
protocols recapitulated the two key pathological hallmarks of BPD: blunted secondary
septation (revealed by a decreased alveolar number) and an increased mean septal wall
thickness; with a more pronounced effect on alveolar number (Fig. 3M), alveolar density
(Fig. 3N), and MLI (Fig. 3P) in the P1-P14 85% O2-exposure group, suggesting that these two
protocols best model BPD in mice. While the continuous exposure to 85% O2 from P1 to P14
is a constant injurious insult, the impact of exposure to 85% O2 from P1 to P7, followed by
21% O2 between P7 and P14 is harder to define, since the latter protocol represents a period
of injury, followed by a period where the lung might repair itself during room-air exposure.
This might account for the less severe damage noted in the P1-P7 85% O2 group.
Alternatively, lungs in this group may simply have been less damaged due to the reduced
time-frame of exposure to the injurious insult. Although the degree of damage to the
(P1-P14) is proposed as the method of choice, since the lung repair mechanisms that may be
engaged after return to 21% O2 in the P1-P7 exposure group represent confounding variables
that are not present in the continuous exposure protocol (P1-P14), although it might be argued
that that “oxygen recovery” period is or translational significance. This “repair phase”
confounding variable may make data interpretation difficult, either when dissecting
pathogenic pathways modulated by oxygen injury or when investigating the efficacy of a
candidate drug to limit damage from oxygen injury. Of additional interest: the long-term
sequelae of hyperoxia exposure in the neonatal period remains of interest to model, since
alveolarization in humans continues into the teen years at least. However, the persistence of
alveolarization defects beyond P14 in mice was not considered in this study.
The P5.5 time-point is believed to represent the peak of secondary septation (Herriges
and Morrisey, 2014; Hogan et al., 2014), which is the driver of alveologenesis. Therefore,
mouse pups were also exposed to 85% O2 over the period P4-P7 (Fig. 3D,J), believed to be a
critical window of secondary septation. Mouse pups were also exposed to 85% O2 over the
Disease Models & Mechanisms • DMM • Advance article
developing lung in these two groups is largely comparable, the continuous exposure protocol
preceding period, P1-P3 (Fig. 3C,I). For both parameters: the number of alveoli (Fig. 3M) and
the mean septal wall thickness (Fig. 3Q), a dramatic impact was noted for the P4-P7 exposure
periods, but not for the P1-P3 exposure period. These data might confirm that the period P4P7 is a critical window of lung development that may be severely impacted by hyperoxia
exposure; however, it is also noteworthy that the cumulative oxygen dose for the period P4-P7
is greater (as this period is 24 h longer) that the P1-P3 period.
It is important to note that the only readout used in this study was a change in lung
structure, where interventions were examined for the ability to drive changes to the total
number of alveoli in the lung and in the thickness of the alveolar septal wall. In selected
instances, no impact of an oxygen exposure protocol on lung structure was noted. For
example, exposure of newborn mouse lungs to hyperoxia over the first three days of life
(Fig. 1B, Fig. 3C,I). However, it may well be that other important changes do occur in the
lung that are of physiological and pathological relevance, but which are not evident from an
examination of the lung structure alone. Indeed, exposure of newborn mice to hyperoxia over
the first three days of post-natal life has a dramatic impact on the airway
hyper-responsiveness in the long-term (at days 55-77) in adult mice (Regal et al., 2014).
Similarly, in the same model, an impact of early exposure to hyperoxia over the first three
days of post-natal life was also documented to de-regulate the expression of epithelial (Sftpc,
Abca3, Pdpn, Aqp5) as well as endothelial (Pecam) genes (Yee et al., 2014). Additionally,
in natural killer cell responses to influenza virus infection were noted in adult life (Reilly et
al., 2015). It should be noted that these three studies employed 100% O2 as an injurious
stimulus, but these reports do indicate that although alterations to lung structure were not
noted after hyperoxia exposure for the first three days of post-natal life in our standardised
BPD model presented here, other physiological changes clearly do occur in the lung.
A key concern in the use of experimental animal models of human disease is that often
the most injurious stimulus is employed, in an effort to obtain a clearly evident pathological
change in a parameter of interest. In these instances, changes are dramatic, and parameters are
better clustered, yielding a more favourable statistical comparison with controls. In the case of
the hyperoxia-based mouse model of BPD, the continuous exposure of newborn mouse pups
to 85% O2 from P1 to P14 yielded the most pronounced impact on the gas exchange structure
(cumulatively reflected by a change in total number of alveoli, alveolar density, and gas
exchange surface area; Fig. 2M-O) as well as disturbances to alveolar septal wall thickness
(Fig. 2Q). However, one danger of this severe model is that a highly injurious stimulus may
Disease Models & Mechanisms • DMM • Advance article
after slightly longer exposures (for the first four days of life) to hyperoxia, persistent changes
exert such a damaging effect that the positive (but moderate or weak) impact of a candidate
intervention may go unnoticed. Therefore, this idea was tested by providing nutritional
supplementation to newborn mouse pups (in the form of cottonseed oil) over the course of
oxygen exposure between P1 and P14, at both 60% O2 and 85% O2, and examined the
magnitude of improvement in lung structure in both oxygen exposure protocols. Indeed, a
beneficial impact of cottonseed oil administration was noted in the 60% O2 exposure protocol,
but not in the 85% O2 exposure protocol. Thus, if the 85% O2 model alone had been
employed, the potential utility of cottonseed oil as an intervention in BPD would have been
missed. The selection of the oxygen exposure is, therefore, critically important; and these
observations underscore the idea that if an experimental intervention proves unsuccessful in
severe models of disease, a less severe model may nevertheless highlight the potential utility
of the intervention. Further to this, however, the severe (85% O2) BPD model is the only
model that recapitulates both the disturbances to the gas exchange structure (cumulatively
reflected by a change in total number of alveoli, alveolar density, and gas exchange surface
area; Fig 2M-P,R) as well as disturbances to alveolar septal wall thickness that are seen in
human disease (Fig. 2Q). While an effect of cottonseed oil administration on alveoli number
in the 85% O2 model was missed, a pronounced positive impact of cottonseed oil on alveolar
septal wall thickness was still noted in the 85% O2 model, where cottonseed oil application
normalized the alveolar wall thickness (Fig. 4K,L, vs. Fig. 4I,J; Fig. 4Q). Taken together,
when assessing the potential utility of an intervention intended to drive post-natal lung
maturation in an experimental animal model of BPD. It is the recommendation of the authors
to employ 85% O2 to generate an impact of hyperoxia on both the number of alveoli and the
mean septal wall thickness. However, when an experimental intervention fails to blunt the
impact of 85% O2 on these two parameters of lung structure, it is recommended to evaluate
the same intervention using 60% O2 as an injurious stimulus.
This report addresses (exclusively) the use of hyperoxia as an injurious stimulus in the
arrested lung development associated with hyperoxia exposure, as a model for BPD.
Clinically, hyperoxia is one of many contributors to BPD, with additional important disease
modifiers including, inter alia baro- and volu-trauma from mechanical ventilation (Greenough
et al., 2008), and the background of infection (Balany and Bhandari, 2015). Future efforts to
further optimize the mouse (and other) models of BPD should address hyperoxia in
combination with mechanical ventilation and/or infection. Keeping animal welfare in mind
(Curzer et al., 2016), where the number of experimental animals employed and the level of
Disease Models & Mechanisms • DMM • Advance article
these data reveal that the oxygen exposure protocol requires tailoring and broad consideration
stress to which the animal is subjected must be maintained at the minimum level possible,
while still retaining the translational viability of the animal model, this model represents the
best compromise between keeping the injurious insult as mild as possible, while maintaining
the translational relevance of the model.
MATERIALS AND METHODS
Approvals for studies with experimental animals
All animal procedures were approved by the local authorities, the Regierungspräsidium
Darmstadt (approval numbers B2/344, B2/1051, and B2/1108).
Mouse models of bronchopulmonary dysplasia
The normobaric hyperoxia-based model of BPD in mice was conducted essentially as
described previously, with the modifications outlined below (Alejandre-Alcázar et al., 2007;
Madurga et al., 2014; Mižíková et al., 2015). Newborn C57Bl/6 mice (Mus musculus
Linnaeus) were randomised to equal-sized litters (average seven mice per litter), and placed
into either a normoxic or hyperoxic environment within two hours of birth. The
hyperoxia-exposure protocols are illustrated in Fig. 1A,B. For the 40% O2, 60% O2, and 85%
O2 oxygen exposure protocols, mouse pups were exposed to the appropriate oxygen
additional oxygen level protocols were also performed: (i) a decreasing gradient of O2 from
85% on P1 to 21% on P14 (a reduction in oxygen concentration of 5% per day); and (ii) an
oscillation between 85% O2 and 40% O2, for a 24-h period, on a 24 h:24 h oscillation cycle.
To determine the necessary window of oxygen exposure over the first 14 days of life,
newborn mouse pups were exposed to 85% O2 for discrete “windows”, which included: (i) the
first 24 h of life (P1), (ii) the first three days of life, starting at P1, up to and including P3; (iii)
starting at the beginning of P4 and continuing to (and including) P7; (iv) the first seven days
of life, starting at P1, and continuing to (and including) P7; and (v) the entire first 14 days of
life, ending with (and including) P14. All experiments were terminated at P14.
For all oxygen-exposure protocols, nursing dams were rotated every 24 h, to ensure at
least one 24-h period of 21% O2 every two days. This addresses the oxygen toxicity issues in
adult mice, which are highly susceptible to prolonged periods of hyperoxia. Nursing dams
received food ad libitum. Mice were maintained in a 12 h:12 h dark/light cycle. All pups were
euthanized at the end of P14, with an overdose of pentobarbital (500 mg/kg, intraperitoneal;
Disease Models & Mechanisms • DMM • Advance article
concentration starting on the day of birth (P1), continuously up to and including P14. Two
Euthoadorm®, CP-Pharma, Burgdorf, Germany), followed by thoracotomy, followed by lung
extraction and processing for design-based stereology (Madurga et al., 2014; Mižíková et al.,
2015).
Design-based stereology
All methods employed for the analysis of lung structure were based on American Thoracic
Society/European Respiratory Society recommendations for quantitative assessment of lung
structure (Hsia et al., 2010). The protocol employed for the design-based stereological
analysis of neonatal mouse lungs has been described in the detail previously (Madurga et al.,
2015; Madurga et al., 2014; Mižíková et al., 2015), based on state-of-the-art methodology
(Mühlfeld et al., 2015; Mühlfeld et al., 2013; Muhlfeld and Ochs, 2014; Mühlfeld and Ochs,
2013; Ochs and Mühlfeld, 2013; Schneider and Ochs, 2013; Schneider and Ochs, 2014).
Briefly, mouse lungs were instillation-fixed through a tracheal cannula at a hydrostatic
pressure of 20 cmH2O with 1.5% (wt./vol.) paraformaldehyde (Sigma, Darmstadt, Germany;
catalogue number P6148), 1.5% (wt./vol.) glutaraldehyde (Serva, Heidelberg, Germany,
catalogue number 23116.02) in 150 mM HEPES (Sigma, Darmstadt, Germany; catalogue
number H0887), pH 7.4, for 24 h at 4 °C, after which lung tissue blocks were collected
according to systematic uniform random sampling for stereological analysis (Schneider and
Ochs, 2014).
catalogue number 05039) and cut into 3-mm sections. The total volume of the lungs was
measured by Cavalieri’s principle (Madurga et al., 2015; Madurga et al., 2014; Mižíková et
al., 2015). Whole lungs were treated with sodium cacodylate (Serva, Heidelberg, Germany;
catalogue number 15540.03), osmium tetroxide (Roth, Karlsruhe, Germany; catalogue
number 8371.3), and uranyl acetate (Serva, Darmstadt, Germany; catalogue number
77870.01) and embedded in glycol methacrylate (Technovit 7100; Heareus Kulzer, Hanau,
Germany; catalogue number 64709003). For the determination of alveoli number, each tissue
block was cut into sections of 2 µm, and every first and third section of a consecutive series of
sections throughout the block was stained with Richardson’s stain. For all other parameters,
every tenth section of a consecutive series throughout the block was similarly prepared (four
sections per block were selected).
All slides were scanned using a NanoZoomer-XR C12000 Digital slide scanner
(Hamamatsu, Herrsching am Ammersee, Germany). Analyses were performed using the
Visiopharm NewCast computer-assisted stereology system (Visiopharm, Hoersholm,
Disease Models & Mechanisms • DMM • Advance article
Lungs were embedded in toto in 2% (wt./vol.) agar (Sigma, Darmstadt, Germany;
Denmark). Parameters analysed included the mean linear intercept (MLI), alveolar septal wall
thickness, total surface area, as well as alveolar number and alveolar density, as described
previously (Madurga et al., 2015; Madurga et al., 2014; Mižíková et al., 2015). Intrinsic to
this analysis is the separate scoring of parenchymal elements, which are discriminated from
vessels and airways. A total of approximately 40 tissue sections were evaluated per animal,
for all parameters, except the determination of alveolar number, in which case total of 10
sections per animal were evaluated. In each case, 2–5% of each section was analyzed. The
coefficient of error (CE), the coefficient of variation (CV), as well as the squared ratio
between both (CE2/CV2) were measured for each stereological parameter (Tables 1,2,3), and
the quotient threshold was set at 0.5 to validate the precision of the measurements.
Therapeutic intervention with generic parenteral nutrition
Cottonseed oil (Sigma-Aldrich, Darmstadt, Germany; catalogue number C7767) was applied
by daily intraperitoneal injection to pups, where the cottonseed oil dose was de-escalated over
a range starting at 20 ml.kg-1day-1 (P1, P2, P3), followed by 15 ml.kg-1day-1 (P4, P5, P6),
followed by 10 ml.kg-1day-1 (P7, P8, P9) and finally to 5 ml.kg-1day-1 (P10, P11, P12, and
P13), to avoid the injection of a large oil bolus. The experiment was terminated at P14. The
cottonseed oil application was undertaken in the BPD model using either continuous exposure
to 60% O2 (FiO2 0.60) or 85% O2 (FiO2 0.85). The application of cottonseed oil was in
with or at risk for BPD in a neonatal intensive care setting (Beken et al., 2014).
Sex genotyping of mice
Sex determination of mouse pups was undertaken exactly as described previously (Lambert et
al., 2000). Essentially, genomic DNA was isolated from tail biopsies, and screened by
polymerase chain reaction using forward primer 5-TGGGACTGGTGACAATTGTC-3 and
reverse primer 5-GAGTACAGGTGTGCAGCTCT-3 to detect the male-specific Sry locus;
together with forward primer 5-GGGACTCCAAGCTTCAATCA-3 and reverse primer
5-TGGAGGAGGAAGAAAAGCAA-3 to detect the Il3 gene present in both the male and
female sex. Amplicons were resolved on a 1.5% (wt./vol.) agarose gel, and visualized by
ethidium bromide staining.
Disease Models & Mechanisms • DMM • Advance article
analogy with generic oil-based nutritional supplementation that is provided to pre-term infants
Statistical analysis
All data are presented as mean ± SD. Differences between groups were evaluated by one-way
ANOVA with Tukey’s post hoc test for all experiments. The P values ˂ 0.05 were regarded as
significant. All statistical analyses were performed with GraphPad Prism 6.0. The presence of
statistical outliers was tested by Grubbs’ test, and none were found. All data sets are small for
proper determination of normal distribution; however, a corrected Anderson-Darling statistic
demonstrated normal distribution of data sets.
Competing Interests
The authors declare no competing or financial interests.
Contributors CN, IM, DMG, JR-C and REM: designed the research; CN, IM, DMG, and
JR-C: performed the hyperoxia studies, tissue extraction and embedding, and design-based
stereology studies; KM, IV, SH and WS: provided infrastructure and critical advice; CN and
IM prepared the figures. CN, IM and REM wrote and edited the manuscript.
Funding
This study was financially supported by the Max Planck Society; Rhön Klinikum AG grant
“LOEWE Programme”, the German Center for Lung Research (Deutsches Zentrum für
Lungenforschung),
and
by
the
German
Research
Foundation
(Deutsche
Forschungsgemeinschaft) through Excellence Cluster EXC147, Collaborative Research
Center SFB1213/1, Clinical Research Unit KFO309/1, and individual research grant Mo
1789/1.
Disease Models & Mechanisms • DMM • Advance article
Fl_66; the Federal Ministry of Higher Education, Research and the Arts of the State of Hessen
REFERENCES
Alejandre-Alcázar, M. A., Kwapiszewska, G., Reiss, I., Amarie, O. V., Marsh, L.
M., Sevilla-Pérez, J., Wygrecka, M., Eul, B., Köbrich, S., Hesse, M. et al. (2007).
Hyperoxia modulates TGF-beta/BMP signaling in a mouse model of bronchopulmonary
dysplasia. Am J Physiol Lung Cell Mol Physiol 292, L537-49.
Balany, J. and Bhandari, V. (2015). Understanding the impact of infection,
inflammation, and their persistence in the pathogenesis of bronchopulmonary dysplasia. Front
Med (Lausanne) 2, 90.
Beken, S., Dilli, D., Fettah, N. D., Kabatas, E. U., Zenciroglu, A. and Okumus, N.
(2014). The influence of fish-oil lipid emulsions on retinopathy of prematurity in very low
birth weight infants: a randomized controlled trial. Early Hum Dev 90, 27-31.
Berger, J. and Bhandari, V. (2014). Animal models of bronchopulmonary dysplasia.
The term mouse models. Am J Physiol Lung Cell Mol Physiol 307, L936-47.
Bouch, S., Harding, R., O'Reilly, M., Wood, L. G. and Sozo, F. (2016). Impact of
dietary tomato juice on changes in pulmonary oxidative stress, inflammation and structure
induced by neonatal hyperoxia in mice (Mus musculus). PLoS One 11, e0159633.
Bouch, S., O'Reilly, M., Harding, R. and Sozo, F. (2015). Neonatal exposure to mild
mice without altering lung structure. Am J Physiol Lung Cell Mol Physiol, ajplung 00359
2014.
Britt, R. D., Jr., Velten, M., Tipple, T. E., Nelin, L. D. and Rogers, L. K. (2013).
Cyclooxygenase-2 in newborn hyperoxic lung injury. Free Radic Biol Med 61, 502-11.
Burri, P. H. (2006). Structural aspects of postnatal lung development - alveolar
formation and growth. Biol Neonate 89, 313-22.
Curzer, H. J., Perry, G., Wallace, M. C. and Perry, D. (2016). The Three Rs of
animal research: what they mean for the institutional animal care and use committee and why.
Sci Eng Ethics 22, 549-65.
Domm, W., Misra, R. S. and O'Reilly, M. A. (2015). Affect of early life oxygen
exposure on proper lung development and response to respiratory viral infections. Front Med
(Lausanne) 2, 55.
Ehrhardt, H., Pritzke, T., Oak, P., Kossert, M., Biebach, L., Forster, K., Koschlig,
M., Alvira, C. M. and Hilgendorff, A. (2016). Absence of TNF-alpha enhances
Disease Models & Mechanisms • DMM • Advance article
hyperoxia causes persistent increases in oxidative stress and immune cells in the lungs of
inflammatory response in the newborn lung undergoing mechanical ventilation. Am J Physiol
Lung Cell Mol Physiol 310, L909-18.
Greenough, A., Premkumar, M. and Patel, D. (2008). Ventilatory strategies for the
extremely premature infant. Paediatr Anaesth 18, 371-7.
Herriges, M. and Morrisey, E. E. (2014). Lung development: orchestrating the
generation and regeneration of a complex organ. Development 141, 502-13.
Herring, M. J., Putney, L. F., Wyatt, G., Finkbeiner, W. E. and Hyde, D. M.
(2014). Growth of alveoli during postnatal development in humans based on stereological
estimation. Am J Physiol Lung Cell Mol Physiol 307, L338-44.
Hilgendorff, A. and O'Reilly, M. A. (2015). Bronchopulmonary dysplasia early
changes leading to long-term consequences. Front Med (Lausanne) 2, 2.
Hogan, B. L., Barkauskas, C. E., Chapman, H. A., Epstein, J. A., Jain, R., Hsia,
C. C., Niklason, L., Calle, E., Le, A., Randell, S. H. et al. (2014). Repair and regeneration
of the respiratory system: complexity, plasticity, and mechanisms of lung stem cell function.
Cell Stem Cell 15, 123-38.
Hsia, C. C., Hyde, D. M., Ochs, M., Weibel, E. R. and Structure, A. E. J. T. F. o.
Q. A. o. L. (2010). An official research policy statement of the American Thoracic
Society/European Respiratory Society: standards for quantitative assessment of lung structure.
Am J Respir Crit Care Med 181, 394-418.
Challenges of Gas Exchange. Compr Physiol 6, 827-95.
Jobe, A. H. (2011). The new bronchopulmonary dysplasia. Curr Opin Pediatr 23,
167-72.
Jobe, A. H. and Tibboel, D. (2014). Update in pediatric lung disease 2013. Am J
Respir Crit Care Med 189, 1031-6.
Lambert, J. F., Benoit, B. O., Colvin, G. A., Carlson, J., Delville, Y. and
Quesenberry, P. J. (2000). Quick sex determination of mouse fetuses. J Neurosci Methods
95, 127-32.
Liao, J., Kapadia, V. S., Brown, L. S., Cheong, N., Longoria, C., Mija, D.,
Ramgopal, M., Mirpuri, J., McCurnin, D. C. and Savani, R. C. (2015). The NLRP3
inflammasome is critically involved in the development of bronchopulmonary dysplasia. Nat
Commun 6, 8977.
Disease Models & Mechanisms • DMM • Advance article
Hsia, C. C., Hyde, D. M. and Weibel, E. R. (2016). Lung Structure and the Intrinsic
Maduekwe, E. T., Buczynski, B. W., Yee, M., Rangasamy, T., Stevens, T. P.,
Lawrence, B. P. and O'Reilly, M. A. (2015). Cumulative neonatal oxygen exposure predicts
response of adult mice infected with influenza A virus. Pediatr Pulmonol 50.
Madurga, A., Golec, A., Pozarska, A., Ishii, I., Mižíková, I., Nardiello, C., Vadasz,
I., Herold, S., Mayer, K., Reichenberger, F. et al. (2015). The H2S-generating enzymes
cystathionine beta-synthase and cystathionine gamma-lyase play a role in vascular
development during normal lung alveolarization. Am J Physiol Lung Cell Mol Physiol 309,
L710-724.
Madurga, A., Mižíková, I., Ruiz-Camp, J., Vadász, I., Herold, S., Mayer, K.,
Fehrenbach, H., Seeger, W. and Morty, R. E. (2014). Systemic hydrogen sulfide
administration partially restores normal alveolarization in an experimental animal model of
bronchopulmonary dysplasia. Am J Physiol Lung Cell Mol Physiol 306, L684-97.
Masood, A., Yi, M., Lau, M., Belcastro, R., Li, J., Kantores, C., Pace-Asciak, C.
R., Jankov, R. P. and Tanswell, A. K. (2014). Cyclooxygenase-2 inhibition partially
protects against 60% O2 -mediated lung injury in neonatal rats. Pediatr Pulmonol 49, 9911002.
Mižíková, I. and Morty, R. E. (2015). The extracellular matrix in bronchopulmonary
dysplasia: Target and source. Front Med (Lausanne) 2, 91.
Mižíková, I., Ruiz-Camp, J., Steenbock, H., Madurga, A., Vadász, I., Herold, S.,
cross-linking is altered during aberrant late lung development associated with hyperoxia. Am J
Physiol Lung Cell Mol Physiol 308, L1145-58.
Mühlfeld, C., Hegermann, J., Wrede, C. and Ochs, M. (2015). A review of recent
developments and applications of morphometry/stereology in lung research. Am J Physiol
Lung Cell Mol Physiol 309, L526-L536.
Mühlfeld, C., Knudsen, L. and Ochs, M. (2013). Stereology and morphometry of
lung tissue. Methods Mol Biol 931, 367-90.
Muhlfeld, C. and Ochs, M. (2014). Measuring structure - What's the point in
counting? Ann Anat 196, 1-2.
Mühlfeld, C. and Ochs, M. (2013). Quantitative microscopy of the lung: a problembased approach. Part 2: stereological parameters and study designs in various diseases of the
respiratory tract. Am J Physiol Lung Cell Mol Physiol 305, L205-21.
Nardiello, C. and Morty, R. E. (2016). MicroRNA in late lung development and
bronchopulmonary dysplasia: the need to demonstrate causality. Mol Cell Pediatr 3, 19.
Disease Models & Mechanisms • DMM • Advance article
Mayer, K., Seeger, W., Brinckmann, J. and Morty, R. E. (2015). Collagen and elastin
Ochs, M. and Mühlfeld, C. (2013). Quantitative microscopy of the lung: a problembased approach. Part 1: basic principles of lung stereology. Am J Physiol Lung Cell Mol
Physiol 305, L15-22.
Olave, N., Lal, C. V., Halloran, B., Pandit, K., Cuna, A. C., Faye-Petersen, O. M.,
Kelly, D. R., Nicola, T., Benos, P., Kaminski, N. et al. (2015). Regulation of alveolar
septation by microRNA-489. Am J Physiol Lung Cell Mol Physiol 310, L476-L487.
Regal, J. F., Lawrence, B. P., Johnson, A. C., Lojovich, S. J. and O'Reilly, M. A.
(2014). Neonatal oxygen exposure alters airway hyper-responsiveness but not the response to
allergen challenge in adult mice. Pediatr Allergy Immunol 25, 180-6.
Reilly, E. C., Martin, K. C., Jin, G. B., Yee, M., O'Reilly, M. A. and Lawrence, B.
P. (2015). Neonatal hyperoxia leads to persistent alterations in NK responses to influenza A
virus infection. Am J Physiol Lung Cell Mol Physiol 308, L76-85.
Reyburn, B., Di Fiore, J. M., Raffay, T., Martin, R. J., Prakash, Y. S., Jafri, A.
and MacFarlane, P. M. (2016). The effect of continuous positive airway pressure in a mouse
model of hyperoxic neonatal lung injury. Neonatology 109, 6-13.
Schneider, J. P. and Ochs, M. (2013). Stereology of the lung. Methods Cell Biol 113,
257-94.
Schneider, J. P. and Ochs, M. (2014). Alterations of mouse lung tissue dimensions
during processing for morphometry: a comparison of methods. Am J Physiol Lung Cell Mol
Silva, D. M., Nardiello, C., Pozarska, A. and Morty, R. E. (2015). Recent advances
in the mechanisms of lung alveolarization and the pathogenesis of bronchopulmonary
dysplasia. Am J Physiol Lung Cell Mol Physiol 309, L1239-72.
Sozo, F., Horvat, J. C., Essilfie, A. T., O'Reilly, M., Hansbro, P. M. and Harding,
R. (2015). Altered lung function at mid-adulthood in mice following neonatal exposure to
hyperoxia. Respir Physiol Neurobiol.
Stoll, B. J., Hansen, N. I., Bell, E. F., Walsh, M. C., Carlo, W. A., Shankaran, S.,
Laptook, A. R., Sanchez, P. J., Van Meurs, K. P., Wyckoff, M. et al. (2015). Trends in
care practices, morbidity, and mortality of extremely preterm neonates, 1993-2012. JAMA
314, 1039-51.
Tibboel, J., Joza, S., Reiss, I., de Jongste, J. C. and Post, M. (2013). Amelioration
of hyperoxia-induced lung injury using a sphingolipid-based intervention. Eur Respir J 42,
776-84.
Disease Models & Mechanisms • DMM • Advance article
Physiol 306, L341-50.
Tschanz, S., Schneider, J. P. and Knudsen, L. (2014). Design-based stereology:
Planning, volumetry and sampling are crucial steps for a successful study. Ann Anat 196, 311.
Wang, H., Jafri, A., Martin, R. J., Nnanabu, J., Farver, C., Prakash, Y. S. and
MacFarlane, P. M. (2014). Severity of neonatal hyperoxia determines structural and
functional changes in developing mouse airway. Am J Physiol Lung Cell Mol Physiol 307,
L295-301.
Yee, M., Buczynski, B. W. and O'Reilly, M. A. (2014). Neonatal hyperoxia
stimulates the expansion of alveolar epithelial type II cells. Am J Respir Cell Mol Biol 50,
757-66.
Yee, M., Gelein, R., Mariani, T. J., Lawrence, B. P. and O'Reilly, M. A. (2016).
The oxygen environment at birth specifies the population of alveolar epithelial stem cells in
Disease Models & Mechanisms • DMM • Advance article
the adult lung. Stem Cells 34, 1396-406.
Figures
Fig. 1. Schematic illustration of the different oxygen exposure protocols employed. (A)
Determination of the effects of the level of oxygen employed as an injurious stimulus.
day (P)1 (the day of birth) to the end of P14, to (i) ambient room-air (21% O2), (ii) 40% O2,
(iii) 60% O2, and (iv) 85% O2. Alternatively, newborn mouse pups were exposed to (v) a
decreasing gradient of O2 concentration from 85% O2 at P1 to 21% O2 at P14; or (vi) to an
oscillation between 85% O2 and 40% O2 in an oscillation cycle of 24 h: 24 h, starting with
85% O2 on P1, and ending with 21% O2 on P14. (B) Determination of the effects of the timeframe (window) of exposure to injurious oxygen levels. The exposure protocols included the
continuous exposure of newborn mouse pups from P1 to the end of P14 to (i) ambient
room-air (21% O2); or (ii) to 85% O2 for the first 24 h of life, followed by 21% O2 from the
start of P2 to the end of P14; or (iii) to 85% O2 from P1 to the end of P3 followed by 21% O2
from the start of P4 to the end of P14; or (iv) to 21% O2 from P1 to the end of P3, followed by
85% O2 from the start of P4 to the end of P7, followed by 21% O2 to the end of P14; or (v) to
85% O2 from P1 to the end of P7, followed by 21% O2 from the start of P8 to the end of P14;
or (vi) to 85% O2 from P1 to the end of P14 .
Disease Models & Mechanisms • DMM • Advance article
Exposure protocols include the continuous exposure of newborn mouse pups from post-natal
newborn mice. Histological images were obtained from the lungs of mice subjected to the six
oxygen exposure protocols illustrated in Fig. 1A. (A)-(F) are low magnification images from
the lungs, which were embedded in glycol methylacrylate plastic after fixation with buffered
paraformaldehyde/glutaraldehyde and treatment with sodium cacodylate, osmium tetroxide,
and uranyl acetate, and stained with Richardson’s stain. Images (G) to (L) are higher
magnification images derived from part of the corresponding images at left (demarcated by
the red box), to highlight changes in alveolar septal wall thickness. Each image is
representative of lung sections obtained from four other mouse pups within each experimental
Disease Models & Mechanisms • DMM • Advance article
Fig. 2. Optimising oxygen exposure levels to model bronchopulmonary dysplasia in
group (n=5, per group). Scale bars in both sets of photomicrographs represent 200 m.
Design-based stereology was employed to assess (M) total number of alveoli in the lung, (N)
alveolar density, (O) gas exchange surface area, (P) mean linear intercept, and (Q) alveolar
septal wall thickness. (R) The lung volume was estimated by the Cavalieri method. In panels
(M)-(R),  denotes male animals, while ▲ denotes female animals. Data represent mean 
S.D. Data comparisons were made by one-way ANOVA with Tukey’s post hoc test. n.s., not
Disease Models & Mechanisms • DMM • Advance article
significant (P  0.05).
in newborn mice. Histological images were obtained from the lungs of mice subjected to the
six oxygen exposure protocols illustrated in Fig. 1B. (A)-(F) are low magnification images
from the lungs, which were embedded in glycol methylacrylate plastic after fixation with
buffered paraformaldehyde/glutaraldehyde and treatment with sodium cacodylate, osmium
tetroxide, and uranyl acetate, and stained with Richardson’s stain. Images (G) to (L) are
higher magnification images derived from part of the corresponding images at left
(demarcated by the red box), to highlight changes in alveolar septal wall thickness. Each
image is representative of lung sections obtained from four other mouse pups within each
experimental group (n=5, per group). Scale bars in both sets of photomicrographs represent
200 m. Design-based stereology was employed to assess (M) total number of alveoli in the
Disease Models & Mechanisms • DMM • Advance article
Fig. 3. Optimising the oxygen exposure window to model bronchopulmonary dysplasia
lung, (N) alveolar density, (O) gas exchange surface area, (P) mean linear intercept, and (Q)
alveolar septal wall thickness. (R) The lung volume was estimated by the Cavalieri method. In
panels (M)-(R),  denotes male animals, while ▲ denotes female animals. Data represent
Disease Models & Mechanisms • DMM • Advance article
mean  S.D. Data comparisons were made by one-way ANOVA with Tukey’s post hoc test.
Fig. 4. Comparison of the efficacy of a candidate therapeutic intervention in a severe vs.
a less severe hyperoxia-based model of bronchopulmonary dysplasia. Mouse pups within
O2, or 85% O2; presented in rows] were either sham-treated (control) or treated with
cottonseed oil (average of 12 ml.kg-1day-1; see Materials and Methods for precise dosing
protocol) via daily intraperitoneal injection, to provide supplementary parenteral nutrition.
(A)-(L) Images are sections from lungs embedded in glycol methylacrylate plastic after
fixation with buffered paraformaldehyde/glutaraldehyde and treatment with sodium
cacodylate, osmium tetroxide, and uranyl acetate, and stained with Richardson’s stain. Each
image is representative of lung sections obtained from four other mouse pups within each
experimental group (n=5, per group). Within each treatment group (control vs. cottonseed oil),
the column of images at left represents low magnification images, while higher magnification
images derived from part of the corresponding images at left (demarcated by the red box) are
presented at right, to highlight changes in alveolar septal wall thickness. Scale bars in
photomicrographs represent 200 m. Design-based stereology was employed to assess (M)
total number of alveoli in the lung, (N) alveolar density, (O) gas exchange surface area, (P)
mean linear intercept, and (Q) alveolar septal wall thickness. (R) The lung volume was
Disease Models & Mechanisms • DMM • Advance article
each of the three oxygen exposure groups [continuous ambient (21% O2) room air; or 60%
estimated by the Cavalieri method. In panels (G)-(L),  denotes male animals, while ▲
denotes female animals. Data represent mean  S.D. Data comparisons were made by oneway ANOVA with Tukey’s post hoc test. CSO, cottonseed oil-treated group; Ctrl., control
Disease Models & Mechanisms • DMM • Advance article
(sham-treated) group; n.s., not significant (P  0.05).
Table 1. Assessment of stereology parameters for the developing mouse lung in response to different levels of hyperoxia exposure. Values
are presented as mean  S.D (n=5 lungs per group). Data were compared by one-way ANOVA with Tukey’s post-hoc analysis.
O2
40%
60%
85%
Gradient
Oscillation
mean  S.D.
P value
versus
21% O2
mean  S.D.
P value
versus
21% O2
mean  S.D.
P value
versus
21% O2
mean  S.D.
P value
versus
21% O2
mean  S.D.
P value
versus
21% O2
V (lung) [cm3]
0.2680 ± 0.0165
0.2275 ± 0.0381
0.1928
0.2253 ± 0.0313
0.1505
0.2626 ± 0.0262
0.9995
0.2912 ± 0.0257
0.7359
0.2577 ± 0.0135
0.9888
CV [V (lung)]
0.0616
Parameter
0.1676
0.1389
0.0997
0.0882
0.0524
VV (par/lung) [%]
90.83 ± 2.304
88.96 ± 5.560
0.9573
92.37 ± 3.806
0.9813
91.78 ± 3.256
0.9980
88.88 ± 2.578
0.9489
89.44 ± 2.530
0.9883
N (alv, lung) 106
3.998 ± 0.4302
2.445 ± 0.2340
<0.0001
2.276 ± 0.3700
<0.0001
1.252 ± 0.2825
<0.0001
3.977 ± 0.3184
>0.9999
3.02 ± 0.2990
0.0011
NV (alv/par) 107 [cm-3]
1.641 ± 0.1031
1.224 ± 0.1179
<0.0001
1.092 ± 0.0560
<0.0001
0.5162 ± 0.0743
<0.0001
1.543 ± 0.1420
0.6998
1.312 ± 0.1208
0.0007
CV [N (alv/lung)]
SV [cm-1]
S (alv epi, lung)
[cm2]
0.1076
0.0957
0.1626
0.2256
0.0801
0.0990
795.9 ± 33.88
772.8 ± 48.20
0.9640
637.1 ± 70.96
0.0001
505.9 ± 22.01
<0.0001
848.8 ± 47.07
0.4539
768.7 ± 32.88
0.9293
193.9 ± 17.26
154.8 ± 16.36
0.0090
132.1 ± 19.85
<0.0001
121.8 ± 12.58
<0.0001
219.4 ± 19.79
0.1600
176.9 ± 6.36
0.5625
CV [S (alv epi, lung)]
0.0890
V (alv air, lung) [cm3]
0.1285 ± 0.0254
0.1257 ± 0.0253
>0.9999
0.1415 ± 0.0293
0.9343
0.1676 ± 0.0119
0.0918
0.1654 ± 0.0212
0.1254
0.1368 ± 0.0138
0.9903
τ (sep) [µm]
9.612 ± 0.4921
9.837 ± 1.0740
0.9946
10.18 ± 0.4249
0.7647
11.97 ± 0.8598
0.0002
8.559 ± 0.3606
0.1784
10.60 ± 0.5666
0.2342
CV [τ (sep)]
0.0512
MLI [µm]
31.11 ± 1.428
CV [MLI]
0.0459
0.1057
0.1503
0.1092
32.25 ± 4.035
0.1251
0.1033
0.0417
0.9974
43.04 ± 7.220
0.1678
0.0902
0.0718
0.0011
55.25 ± 3.686
0.0667
0.0360
0.0421
<0.0001
30.13 ± 2.508
0.0832
0.0535
0.9987
30.92 ± 2.475
>0.9999
0.0800
Abbreviations: alv, alveoli; alv air, alveolar airspaces; alv epi, alveolar epithelium; CV, coefficient of variation; MLI, mean linear intercept; N,
number, NV, numerical density; par, parenchyma; S, surface area; SV, surface density; τ (sep), arithmetic mean septal thickness; V, volume; VV,
volume density.
Disease Models & Mechanisms • DMM • Advance article
21%
mean  S.D.
Table 2. Assessment of stereology parameters for the developing mouse lung in response to different windows of hyperoxia exposure.
Values are presented as mean  S.D (n=5 lungs per group). Data were compared by one-way ANOVA with Tukey’s post-hoc analysis.
O2
85%
P1
P1-P4
P4-P7
P1-P7
P1-P14
mean  S.D.
mean  S.D.
P value
versus
21% O2
P1-14
mean  S.D.
P value
versus
21% O2
P1-14
mean  S.D.
P value
versus
21% O2
P1-14
mean  S.D.
P value
versus
21% O2
P1-P14
mean  S.D.
P value
versus
21% O2
P1-P14
V (lung) [cm3]
0.2680 ± 0.0165
0.2448 ± 0.0340
0.7323
0.2620 ± 0.0372
0.9991
0.2557 ± 0.0239
0.9750
0.2164 ± 0.0090
0.0503
0.2626 ± 0.0262
0.9994
CV [V (lung)]
0.0616
Parameter
VV (par/lung) [%]
N (alv, lung)
106
NV (alv/par)
107
[cm-3]
CV [N (alv/lung)]
SV
[cm-1]
S (alv epi, lung)
[cm2]
CV [S (alv epi, lung)]
V (alv air, lung)
[cm3]
0.1414
0.1421
0.0935
0.0415
0.0997
90.83 ± 2.304
91.05 ± 2.973
>0.9999
90.40 ± 3.401
0.9999
88.54 ± 1.917
0.7878
86.91 ± 2.701
0.2715
91.78 ± 3.256
0.9941
3.998 ± 0.4302
3.418 ± 0.5320
0.3242
3.372 ± 0.7381
0.2491
2.723 ± 0.2282
0.0014
1.701 ± 0.0890
<0.0001
1.252 ± 0.2825
<0.0001
1.641 ± 0.1031
1.532 ± 0.0485
0.3879
1.413 ± 0.1359
0.0045
1.206 ± 0.0721
<0.0001
0.9062 ± 0.0569
<0.0001
0.5162 ± 0.0743
< 0.0001
0.1076
0.1556
0.2189
0.0838
0.0523
0.2256
795.9 ± 33.88
798.9 ± 23.40
>0.9999
820.1 ± 59.95
0.9161
751.0 ± 38.72
0.4637
605.4 ± 40.95
<0.0001
505.9 ± 22.01
<0.0001
193.9 ± 17.26
177.9 ± 24.43
0.8517
195.1 ± 38.20
>0.9999
169.7 ± 15.76
0.5177
113.9 ± 10.41
<0.0001
121.8 ± 12.58
0.0003
0.0890
0.1373
0.1958
0.0929
0.0914
0.1033
0.1503 ± 0.0078
0.1285 ± 0.0254
0.4399
0.1476 ± 0.0271
0.9999
0.1300 ± 0.0208
0.5165
0.1147 ± 0.0037
0.0531
0.1676 ± 0.0119
0.6753
τ (sep) [µm]
9.612 ± 0.4921
10.690 ± 0.5728
0.2503
9.255 ± 0.9632
0.9734
11.430 ± 0.8255
0.0098
12.880 ± 0.7049
<0.0001
11.97 ± 0.8598
0.0006
CV [τ (sep)]
0.0512
MLI [µm]
31.11 ± 1.428
CV [MLI]
0.0459
0.0536
28.73 ± 1.941
0.0676
0.1041
0.8262
30.48 ± 3.750
0.1230
0.0722
0.9995
30.51 ± 2.787
0.0913
0.0547
0.9996
40.57 ± 4.062
0.1001
0.0718
0.0008
55.25 ± 3.686
<0.0001
0.0832
Abbreviations: alv, alveoli; alv air, alveolar airspaces; alv epi, alveolar epithelium; CV, coefficient of variation; MLI, mean linear intercept; N,
number, NV, numerical density; par, parenchyma; S, surface area; SV, surface density; τ (sep), arithmetic mean septal thickness; V, volume; VV,
volume density.
Disease Models & Mechanisms • DMM • Advance article
21%
P1-P14
Table 3. Assessment of stereology parameters for the developing mouse lung in response to hyperoxia exposure with concomitant
cottonseed oil administration (or sham treatment). Values are presented as mean  S.D (n=5 lungs per group). Data were compared by
one-way ANOVA with Tukey’s post-hoc analysis.
O2
Control
mean  S.D.
Parameter
V (lung) [cm3]
0.2646 ± 0.0144
CV [V (lung)]
0.0542
VV (par/lung) [%]
N (alv, lung)
106
NV (alv/par)
107
[cm-3]
CV [N (alv/lung)]
SV
[cm-1]
S (alv epi, lung)
[cm2]
CV [S (alv epi, lung)]
V (alv air, lung)
[cm3]
60%
Cottonseed Oil
P value
mean  S.D.
versus
21% O2
Ctrl.
0.2338 ± 0.0266
0.5682
0.01138
Control
mean  S.D.
P value
versus
21% O2
Ctrl.
0.2226 ± 0.0380
0.2462
0.1707
85%
Cottonseed Oil
P value
mean  S.D.
versus
21% O2
Ctrl.
0.2950 ± 0.0159
0.5815
0.0538
Control
mean  S.D.
P value
versus
21% O2
Ctrl.
0.2734 ± 0.0331
0.9966
0.1210
Cottonseed Oil
P value
mean  S.D.
versus
21% O2
Ctrl.
0.2756 ± 0.0381
0.9905
0.1381
92.49 ± 2.311
88.59 ± 3.097
0.1518
88.71 ± 3.125
0.1745
88.39 ± 1.602
0.1188
92.24 ± 1.074
>0.9999
90.16 ± 2.635
0.6564
4.067 ± 0.7105
3.232 ± 0.6314
0.0880
2.058 ± 0.4149
<0.0001
3.653 ± 0.2792
0.7281
1.536 ± 0.3534
<0.0001
1.993 ± 0.1918
<0.0001
1.659 ± 0.2324
1.589 ± 0.1374
0.9671
1.075 ± 0.0935
<0.0001
1.405 ± 0.1332
0.0845
0.6054 ± 0.0934
<0.0001
0.8110 ± 0.1120
<0.0001
0.1747
0.1954
0.2016
0.0764
0.2301
0.0962
824.3 ± 43.12
890.3 ± 25.74
0.3016
693.7 ± 76.37
0.0035
807.7 ± 8.23
0.9940
486.3 ± 25.51
<0.0001
607.3 ± 72.14
<0.0001
201.7 ± 18.41
184.7 ± 24.80
0.6597
135.8 ± 19.28
<0.0001
210.6 ± 12.21
0.9651
124.2 ± 18.46
<0.0001
149.1 ± 8.77
0.0012
0.0913
0.1343
0.1420
0.0580
0.0914
0.1033
0.1595 ± 0.0115
0.1370 ± 0.0224
0.6392
0.1357 ± 0.0268
0.5856
0.1762 ± 0.0081
0.8535
0.1885 ± 0.0188
0.3724
0.1875 ± 0.0374
0.4089
τ (sep) [µm]
8.450 ± 0.5961
7.281 ± 0.8976
0.1338
8.324 ± 0.8499
0.9997
8.016 ± 0.3676
0.9239
10.27 ± 0.7922
0.0055
8.229 ± 0.6041
0.9960
CV [τ (sep)]
0.0705
MLI [µm]
31.73 ± 2.262
CV [MLI]
0.0713
0.1233
30.04 ± 2.320
0.0763
0.1021
0.9987
41.59 ± 7.983
0.1919
0.0459
0.0771
33.49 ± 0.651
0.0194
0.0771
0.9950
61.91 ± 4.760
0.0769
0.0734
<0.0001
50.22 ± 8.810
0.0002
0.1754
Abbreviations: alv, alveoli; alv air, alveolar airspaces; alv epi, alveolar epithelium; CV, coefficient of variation; MLI, mean linear intercept; N,
number, NV, numerical density; par, parenchyma; S, surface area; SV, surface density; τ (sep), arithmetic mean septal thickness; V, volume; VV,
volume density.
Disease Models & Mechanisms • DMM • Advance article
21%