The mice used were of the outbred NIHS and outbred ICR (CD-1) strains (Harlan UK, Bicester, UK). Late-term pregnant female mice of both strains were bought, and both male and female pups were used from all litters. Postnatal Day 1 (P1) was defined as the day of birth, and therefore a P4 mouse pup was 72 hours old. Animals were allowed food and water ad libitum, and maintained on a 12:12 hour light-dark cycle.

NIHS mice were divided equally into three groups: Control, Dex, and RA. ICR mice were divided equally into four groups: Control, Dex, RA, and 5× RA. Dex, RA, and 5× RA group animals received Dex (Sigma-Aldrich, Gillingham, UK) dissolved in PBS at a concentration of 40 μg/ml, given as a daily 10-μl subcutaneous injection for 10 days from P4 to P15, with a 2-day break on P9 and P10. Control group animals received the vehicle only. Four weeks later, all animal groups received further treatment for 10 days from P42 to P53, with a 2-day break on P47 and P48. RA group animals received atRA (Sigma) dissolved in dimethyl sulfoxide (DMSO) (Sigma) and peanut oil (Sigma) at a concentration of 2 mg/ml, administered intraperitoneally daily as 1 μl per g body weight. 5× RA group animals received atRA in DMSO and peanut oil at a concentration of 5 mg/ml, given intraperitoneally daily as 2 μl per g body weight. Dex and Control group animals received vehicle only, given intraperitoneally daily as 1 μl per g body weight. The 2-day break in dosing was an attempt to prevent full induction of the CYP enzymes, which catabolize retinoids and would therefore reduce the efficacy of the administered atRA. All animals were killed at P90.

The animals were weighed. The lungs were then carefully removed from the thorax and the trachea intubated with a 22-gauge cannula. The lungs were inflated to a standard pressure of 20 cm H₂O with 4% paraformaldehyde for 10 minutes, after which the trachea was ligated and the tissue placed in 4% paraformaldehyde for a minimum of 48 hours. The lungs were then washed in distilled water, placed in 30% alcohol to equilibrate with the hilum open, and the lung volumes (LV) measured by fluid displacement. The lungs were then processed, embedded in paraffin wax, and 8-μm sections were stained with hematoxylin and eosin.

Sections were analyzed on a light microscope linked to a digital imaging system. Images were acquired at 10× objective and processed using Image ProPlus software (Data-Cell Ltd, Finchampstead, UK) to calculate the alveolar mean chord length (Lm). Five nonoverlapping fields per section from six sections were analyzed for each animal. Following Weibel (24), knowledge of the Lm and LV allows gas-exchanging surface area (SA) and the surface-to-volume ratio (SA/LV) to be calculated. All parameters are expressed as group mean values ± SEM. Statistical significance of differences between groups was assessed by one-way ANOVA with Bonferroni correction. The % Lm recovery was calculated compared with the Dex group mean value as a proportion of the possible range of response by [(Dex Lm − RA Lm) ÷ (Dex Lm − Control Lm)] × 100. Similarly, % SA/LV recovery was calculated as [(RA SA/LV − Dex SA/LV) ÷ (Control SA/LV − Dex SA/LV)] × 100.

Using the protocol previously developed in TO strain mice to generate an experimental model of emphysema and alveolar regeneration (5), NIHS strain mice were divided into three groups (RA, Dex, and Control), and received Dex followed by atRA, Dex followed by vehicle, or PBS followed by vehicle, respectively. No significant differences in group mean LV, mean body weight (Wt), or mean LV/Wt ratio were seen between groups (Table 1) at P90.

The 11 Control animals showed alveolar histology and morphometry that were indistinguishable from normal uninjected mice (not shown), confirming that the vehicles used had no biological effect. A typical Control lung is shown in Figure 1A.

Figure 1. Typical histologic images of lungs from NIHS strain mice. (A) Control group: PBS- and dimethyl sulfoxide (DMSO)/oil-treated lung. (B) Dex group: Dex- and DMSO/oil-treated lung. (C) Retinoic acid (RA) group: Dex- and all-trans-RA (atRA)-treated lung. Morphologic differences between A, B, and C are clear and in accord with quantitative data in Table 1 and Figure 2.

Download Figure | Download Powerpoint

[More] [Minimize]

Figure 2. Graphs of (A) Lm (in μm) and (B) SA/LV (in cm²/cm³) in NIHS strain mice for Control (dark gray bars), Dex (light gray bars), and RA (hatched bars) groups. Values are group means. Error bars = SEM.

Download Figure | Download Powerpoint

[More] [Minimize]

Eight animals were included in the Dex group and showed markedly different alveolar morphology from Control animals, with fewer, larger alveoli (Figure 1B). Mean alveolar Lm was increased by 24% from 67.9 to 84.4 μm (P < 0.001) (Table 1, Figures 1 and 2A). Mean alveolar SA was reduced by 12% (P = 0.012) and the mean alveolar SA/LV ratio was reduced by 19% from 589.8 to 476.0 cm²/cm³ (P < 0.001) (Table 1, Figure 2B).

Figure 3. Typical histologic appearances of lungs from ICR strain mice. (A) Control group: PBS- and DMSO/oil-treated lung. (B) Dex group: Dex- and DMSO/oil-treated lung. (C) RA group: Dex- and 2 mg/kg atRA-treated lung. (D) 5× RA group: Dex- and 10 mg/kg atRA-treated lung. Morphologic differences between A and B, B and D, and C and D are clear and in accord with quantitative data in Table 2 and Figure 4.

Download Figure | Download Powerpoint

[More] [Minimize]

A group of eight animals received atRA and displayed alveolar histology more similar to Control than to Dex group animals (Figure 1). Mean Lm for the RA group was 75.1 μm, which represents a 56% recovery from the Dex group mean of 84.4 μm (P = 0.004) (Table 1, Figure 2A). This average value hides the variation of response to atRA between individual animals. Of the eight animals in the group, two regenerated completely (defined as a Lm of 65–71 μm), three showed a partial response (defined as Lm 71–76 μm), and three failed to show any morphologic response (defined as Lm > 76 μm). This gives a 25% regeneration rate, 38% partial response rate, and 63% overall response rate in the NIHS strain in response to atRA.

Mean alveolar SA was increased by 30% to 370.4 cm² in the RA group compared with the Dex group mean SA of 356.1 cm² (Table 1), but this improvement failed to reach statistical significance (P > 0.05). However, mean SA/LV was significantly increased to 535.3 cm²/cm³ in the RA group compared with the Dex group mean of 476.0 cm²/cm³ (P = 0.005) (Table 1, Figure 2B). This represents a 48% recovery of mean SA/LV in atRA-treated NIHS strain mice.

ICR strain mice were divided into four groups: Control, Dex, RA, and 5× RA. At P90, no significant difference in LV was seen between groups. Mean Wt was similar for Control, Dex, and 5× RA groups, but was lower in the RA group (P < 0.005 for comparisons with all other groups). LV/Wt ratio was therefore higher in the RA group than in all other groups (Table 2).

Lung histology and morphometry of the 12 Control group animals was indistinguishable from normal uninjected mice (not shown), confirming that both vehicles used had no biological effect in ICR strain mice (Figures 3A, 4A, and 4B).

Figure 4. Graphs of (A) Lm (in μm) and (B) SA/LV (in cm²/cm³) in ICR strain mice for Control (dark gray bars), Dex (light gray bars), RA (hatched bars), and 5× RA (solid bars) groups. Values are group means. Error bars = SEM.

Download Figure | Download Powerpoint

[More] [Minimize]

The Dex group contained 13 animals, and histology showed evidence of inhibited postnatal alveolar septation, with fewer, larger alveoli than Controls (Figure 3B). The group mean Lm was increased by 54% from 65.1 to 100.4 μm (P < 0.001) by Dex treatment, with a concomitant 43% fall in mean SA from 560.8 to 319.8 cm² (P < 0.001), and 34% reduction in alveolar mean SA/LV from 615.2 to 407.1 cm²/cm³ (P < 0.001) compared with the Control group (Table 2, Figures 4A and 4B).

Eleven animals received a standard 2 mg/kg daily dose of atRA in the RA group. Histology was indistinguishable from Dex group animals (Figure 3C) and morphometric analysis gave a mean Lm of 110.5 μm, a mean SA of 316.8 cm², and a mean SA/LV of 364.7 cm²/cm³ (Figures 4A and 4B). For each parameter, comparison with the Dex group mean failed to reach statistical significance. However, comparison with Control group means showed highly statistically significant differences in Lm (70% increase), SA (44% reduction), and SA/LV (41% reduction) (P < 0.001 for all parameters). These RA group results are consistent with successful inhibition of postnatal alveolar septation by Dex treatment, but failure of subsequent atRA 2 mg/kg treatment to induce alveolar regeneration.

A fourth group of 11 animals received an increased daily dose of 10 mg/kg atRA (i.e., a 5-fold increase of the standard dose). Histology in this 5× RA group typically showed a more complex alveolar structure, with more numerous, smaller alveoli than either the Dex or RA group animals, and similar to appearances in the Control group (Figure 3). Compared with the Dex group, the 5× RA group mean Lm was recovered by 52% from 100.4 to 81.9 μm (P < 0.001), SA recovered by 37% from 319.8 to 409.5 cm² (P < 0.001), and SA/LV recovered by 41% from 407.1 to 492.5 cm²/cm³ (P < 0.001) (Table 2, Figures 4A and 4B).

Again, these group mean values hide the range of variable individual response to 5× RA. When individual values of Lm are considered, three animals regenerated completely (defined as Lm 65–75 μm), five showed a partial response (defined as Lm 75–86 μm), and three failed to show any morphologic change (Lm > 86 μm). (The boundary values for these definitions have been adjusted to take into account the greater magnitude of change in Lm between Control and Dex group ICR strain mice.) This gives a 27% regeneration rate, 45% partial response rate, and an overall combined response rate of 72%.

Male and female mice were included in all experimental groups for both outbred mouse strains. Since group sizes were larger in the ICR study, male and female results can be analyzed separately to determine whether the response to treatment differs between sexes.

The mean alveolar Lm in Control animals at P90 was almost identical in male and female animals (Table 3). There was also a similar magnitude of increase in Lm in the Dex group for both sexes. Neither sex showed a significant recovery of Lm in response to standard dose atRA treatment. Both male and female 5× RA group mean Lm values fell in comparison to Dex group means, representing a 50% recovery in males and a 55% recovery in females. While this reduction was statistically significant for male animals (P = 0.001), it failed to reach statistical significance for females (P = 0.229). Analysis of response rates (using Lm values as above) for females showed 40% regeneration (n = 2) and 40% partial response (n = 2), giving an overall response rate of 80%. Male animals showed 17% regeneration (n = 1) and 50% partial response (n = 3), giving an overall response rate of 67%.

Absolute values of mean SA and mean SA/LV in each treatment group differed between males and females (Table 3). However, in both sexes Dex treatment was seen to reduce mean SA and mean SA/LV, standard atRA treatment produced similar values to the Dex group, and 5× RA treatment yielded sizeable increases in both SA and SA/LV. While this improvement in SA was statistically significant for females (P = 0.004), male values failed to reach statistical significance (P = 0.058). The 5× RA group mean SA/LV was 40% recovered in males and 43% recovered in females. This recovery of SA/LV was statistically significant in males (P < 0.001), but nonsignificant in females (P = 0.154).

We have used the Dex-treated mouse model to determine whether RA-induced alveolar regeneration is a strain-dependent phenomenon. When administered to rats and mice in the developmentally critical first 2 weeks after birth, Dex inhibits alveologenesis, and this deleterious effect cannot be spontaneously corrected later in life (5, 25, 26). As a result, animals have fewer, larger alveoli and reduced gas-exchanging surface area. Dex acts via nuclear receptors and is likely to invoke multiple mechanisms to inhibit alveologenesis. Microarray data from postnatal rat lungs has shown that Dex down-regulates vascular endothelial growth factor receptor-2 (VEGFR-2) expression (27). Dex may also interfere with RA signaling during this critical period, since it reduces expression of RA receptor β (RARβ) (28), cellular retinol binding protein I (CRBPI), and cellular retinoic acid–binding protein I (CRABPI) (29) in vivo, and halves the amount of RA released by lipid-laden fibroblasts in vitro (30).

We have previously developed a robust and repeatable model of disrupted alveolar development in outbred TO strain mice (5). In the current study, use of our established protocol for Dex treatment in outbred NIHS and ICR strain mice resulted in typical appearances of inhibited alveologenesis, with fewer, larger alveoli, thin alveolar walls, increased Lm value, and reduced alveolar SA and SA/LV (Tables 1 and 2; Figures 1B, 2, 3B, and 4). The degree of Dex effect in ICR mice was comparable with that previously seen in TO mice (5, 15). NIHS mice showed a smaller absolute morphologic effect of Dex (Table 1, Figure 2A), but statistically the difference between Dex and Control group animals was clear-cut.

atRA (2 mg/kg) successfully induced alveolar regeneration in NIHS strain mice, with a 56% recovery of Lm and 48% recovery of SA/LV (Table 1; Figures 1 and 2). These figures and the animal response rate of 63% (based on individual Lm values) are similar to those recorded in TO mice in response to atRA (15).

By contrast, ICR strain mice failed to regenerate when treated with a standard 2 mg/kg dose of atRA (Table 2; Figures 3C and 4A). If the experiment had stopped at this point, we would have reported a negative result for this strain. However, when the atRA dose was increased 5-fold to 10 mg/kg, alveolar regeneration was successfully induced, with a 52% recovery of Lm and 41% recovery of SA/LV. The response rate, based on individual Lm values, was 72%. Again, these quantitative measures of alveolar regeneration compare favorably with previously reported TO strain figures (15). The 5-fold increase in atRA dose was chosen based on our understanding of in vivo retinoid pharmacokinetics and our experience of frequent toxic events with a 10-fold increase in dose (data not shown).

We can conclude from these results that RA-induced alveolar regeneration is not an isolated, strain-dependent phenomenon. Regeneration has now been demonstrated using the Dex-treated mouse model in four mouse strains (NIHS, ICR, TO [5, 15], and C57BL/6 [14–16]).

It is also apparent from our results that the dose threshold for regeneration can differ between strains. ICR mice required a significantly higher dose of atRA for treatment effect to be seen (Tables 1 and 2). This is likely to reflect a difference in retinoid pharmacokinetics. ICR mice have a high incidence of ocular retinal degeneration. Although this observation has not been fully explained in the literature, since retinoid metabolism plays an integral role in photoreceptor function, and human retinal degeneration can result from accumulation of retinoid metabolic products (31), it is tempting to suggest that outbred ICR strain mice have an inherited abnormality of retinoid metabolism. If so, this may also explain our results.

The pharmacokinetics of retinoids can vary considerably among species, due to differences in bioavailability, volume of distribution, clearance rate, and metabolism. This can result in very different systemic exposures from seemingly equivalent doses. Such differences between animal strains might explain the previous reports of failed alveolar regeneration in response to atRA, since these studies all employed similar doses within a narrow range. With this in mind, it would be wise for future studies to incorporate a broader range of atRA doses to definitively determine whether alveolar regeneration can be induced in the model used.

Recent studies of alveolar sexual dimorphism indicate that regulation of alveolar formation and maintenance is more complex in females than males (32–36). A number of investigators have therefore chosen to use a single sex of animals in trials of RA-induced alveolar regeneration (3, 13, 14, 16–20, 22, 23). When male rats or mice were used, regeneration succeeded (3, 13, 16, 17) with the exception of one rat study conducted at altitude (20). In contrast, both female-only mouse studies failed (18, 19). This might suggest that female rodents are unable to regenerate in response to RA. However, two of the published positive studies included both sexes (5, 15).

To examine any sex-related difference in response to RA, ICR results were analyzed separately for males and females. Observation of individual mice in the 5× RA group showed clear histologic evidence of alveolar regeneration in several mice of both sexes. When these changes were quantified, the degree of response to atRA was similar for both sexes (Table 3). However, the difference between the mean values of Lm and SA/LV for female Dex and 5× RA groups failed to reach statistical significance, unlike the male groups. This appears to be due to a greater variance of individual values in these female groups (Table 3), and may simply reflect the small group sizes. It might also raise the possibility that the estrogen axis interacts with Dex and atRA treatment effects.

NIHS strain female mice in the RA group showed a statistically significant reduction of mean Lm compared with female Dex group mice (P < 0.001), a regeneration rate of 40%, a partial response rate of 60%, and overall response rate of 100% (n = 5, data not shown).

These results indicate that RA-induced alveolar regeneration can occur in both male and female mice.

Other variables may also explain the published failures of alveolar regeneration, such as age at time of atRA treatment. Alveolar regeneration may represent a recapitulation of postnatal developmental alveolar septation, and as such, could be an age-restricted phenomenon. Animal age at the time of atRA treatment was not stated in all studies. However, animals in successful mouse studies (5, 14–16) were significantly younger than those in failed studies (18, 21). Further experiments at increasing age will be required to resolve this issue. It is also noteworthy that two of the negative studies were conducted in centers at high altitude (18, 21). Hypoxia inhibits alveologenesis in newborn rats (37–40), and so relative hypoxia may explain why C57BL/6 mice failed to regenerate in Denver (18) but succeeded at sea level in London (15), Washington (16), and Sendai, Japan (14). However, the fact that developmental alveologenesis was not impaired in these mice before treatment could argue against this hypothesis.

It could be argued that the Dex-treated mouse model is more directly applicable to BPD than emphysema, since the “emphysematous” lung phenotype is generated by inhibition of normal development rather than alveolar destruction. However, Dex-treated mice develop more homogeneous enlargement of alveoli than elastase-treated animals, and therefore could be said to provide a better substrate to assess the morphologic treatment effects of RA. Rather than attaching a parallel clinical label, it may be seen as a robust and repeatable model to investigate the basic mechanisms of alveolar septation.

In summary, we have shown that RA-induced alveolar regeneration is neither strain- nor sex-dependent. We have also provided a possible explanation for previously reported failures of alveolar regeneration. Our results support the theory that mechanisms of alveolar formation, maintenance, and regeneration are conserved between strains and species. Further work, to uncover the underlying cellular and molecular mechanisms involved, promises to unlock the potential for therapeutic induction of alveolar regeneration or rescue of failed alveologenesis in humans in the future.