Within pulmonary research, the development of mouse models has provided insight into disease development, progression, and treatment. Structural phenotypes of the lung in healthy inbred mouse strains are necessary for comparison to disease models. To date, progress in the assessment of lung function in these small animals using whole lung function tests has been made. However, assessment of in vivo lung structure of inbred mouse strains has yet to be well defined. Therefore, the link between the structure and function phenotypes is still unclear. With advancements in small animal imaging it is now possible to investigate lung structures such as the central and peripheral airways, whole lung, and lobar volumes of mice in vivo, through the use of micro-CT imaging. In this study, we performed in vivo micro-CT imaging of the C57BL/6, A/J, and BALB/c mouse strains using the intermittent iso-pressure breath hold (IIBH) technique. The resulting high-resolution images were used to extract lung structure phenotypes. The three-dimensional lobar structures and airways were defined and a meaningful mouse airway nomenclature was developed. In addition, using these techniques we have uncovered significant differences in the airway structures between inbred mouse strains in vivo.
- micro-computed tomography
- 3-dimensional mouse airway
- airway nomenclature
- mouse lung phenotype
within pulmonary research, the development of mouse models has provided insight into disease development, progression, and treatment. Structural phenotypes of the lung in healthy inbred mouse strains are necessary for comparison to disease models. To date, progress in the assessment of lung function in these small animals using whole lung function tests has been made; however, assessment of in vivo lung structure of inbred mouse strains has yet to be well defined. With advancements in small animal imaging it is now possible to investigate lung structures such as the central and peripheral airways, whole lung, and lobar volumes of mice in vivo, through the use of micro-CT imaging.
In this study we used an in vivo micro-computed tomography (micro-CT) imaging technique to evaluate the phenotypic differences in three commonly used inbred mice strains: the C57BL/6, A/J, and BALB/c. In vivo imaging of the mouse lung is still in its early stages and a comprehensive characterization of the lung from in vivo micro-CT data sets has not been completed. This has been predominantly limited due to the complexity of imaging such small structures that are dynamically in motion. This unique requirement parallels some of the same challenges faced in measuring lung function in mice (1, 2, 13).
In this study we present a strain comparison of the airway tree structure in vivo using micro-CT. The results include high-resolution images of the in vivo mouse lung, development of an airway nomenclature for the monopodial structure of the mouse airway, and image analysis through extraction of airway luminal dimensions using the newly developed nomenclature to compare differences across C57BL/6, A/J, and BALB/c mouse strains. The results show that there is a discernible difference in the airway structures between different strains of mice when appropriately compared using a systematic approach. We believe these findings that display lung structure variations between common mouse strains will affect future development of mouse models of pulmonary disease.
MATERIALS AND METHODS
A total of 9 normal male mice: 3 C57BL/6, 3 A/J, and 3 BALB/c aged 20 wk were used for this study. The body weights were 29.5 ± 0.5 g, 25.1 ± 0.7 g, and 30.9 ± 1.53 g, respectively, for the C57BL/6, A/J, and BALB/c mice. For each mouse studied, the anesthesia protocol proceeded with an initial sedation of 4–5% isofluorane followed by an intraperitoneal injection of ketamine (87.5 mg/kg) and xylazine (12.5 mg/kg). Following a nonresponsive pedal reflex test, a tracheotomy was performed and the mouse was connected to a computer-controlled ventilator, which consisted of a modified Scireq (Montreal, Quebec, Canada) Flexivent system controlled via Labview (National Instruments, Austin, TX) as previously described (16). To prevent uncontrolled motion artifacts in the micro-CT images, respiratory paralysis was induced through administration of 0.1 mg/kg pancuronium. Throughout the duration of the imaging protocols a concentration of 1–2% isofluorane was used to maintain anesthesia. Although anesthesia reduces respiratory rate and depth of ventilation, using breath hold imaging at user defined pressures enables quantitative intra- and interstrain comparisons. All animals were housed and treated in accordance with the University of Iowa Animal Research Committee guidelines.
In vivo micro-CT image acquisition.
An Imtek Micro-CAT II scanner (Siemens Pre-Clinical Solutions) was used for in vivo micro-CT imaging in this study. Each mouse was placed supine onto a polysterene bed and mounted on the micro-CT carbon fiber stage. During imaging, the mouse was monitored via ECG and temperature sensors using a BioVet C1 data acquisition system (Supertron Technologies, Newark, NJ).
During in vivo image acquisition, a trade-off was made between the acquisition time to perform each scan, the reconstructed image resolution, and the radiation dosage. From these three requirements all scans were performed using the following settings: 60 kVp, 500 ua, 500-ms exposure, and 720 projections over 200 degrees (half-scan). The raw projection files were reconstructed at 1,536 × 1,536 × 1,024 pixel with a 28 μm isotropic voxel size. The field of view that was represented in this image was thus 43 × 43 × 28 mm.
The x-ray projections were acquired using a custom gating and breath hold technique called the intermittent iso-pressure breath hold (IIBH) (16), which enables high-resolution images for in vivo mouse lung imaging. This was made possible through the use of a computer-controlled ventilator and the development of a gating program that could trigger the scanner to take projections while the ventilator was stopped and a fixed airway pressure was set. IIBH, more specifically, consists of a unique breathing sequence that involves three phases: 1) slightly hyperventilated breathing at 90 breaths/min and 20 ml/kg for a total of 4–5 s, 2) two deep breaths, and 3) apnea for 4–5 s while a forced airway pressure is induced and the micro-CT scanner is triggered to capture multiple angles of view. Since a total of 720 projections are required for a complete in vivo lung scan and four projections are captured during each complete IIBH breathing sequence, the gating process where a breath hold is induced occurs ∼180 times. For this study, the breath hold was maintained at 20 cmH20 airway pressure for all mice. The total radiation dosage that was administered to each animal per scan was ∼85 Rad (cGy).
Image processing and analysis.
Manual segmentation of the visible airways was completed using an in-house written software program for lung image analysis, Pulmonary Analysis Software Suite (PASS). The segmentation was completed in each transverse image slice, ∼700 slices per micro-CT scan. The resulting segmentations were saved as binary image files as regions of interest. In addition to segmenting the visible airways from the micro-CT datasets, the bifurcations were further identified and five serial segmented slices at 50% of the branch length were isolated and extracted for regional quantitative analysis. The segmented data was imported into AMIRA (Mercury Computer Systems, Chelmsford, MA) for three-dimensional (3D) visualization of the airway structures to aid in the correct identification of the branching structures in the inbred mouse strains and development of a common nomenclature.
To calculate the lung volume, defined as air plus tissue volume, the lung was segmented using a region growing technique where a seed pixel was specified and voxels within a specified threshold were labeled as lung. The resulting voxels were summed to give a total pixel volume and multiplied by the image resolution of 28 μm in all three dimensions. The same threshold was used for all lungs segmented for total lung volume calculation.
The lungs were also segmented into the four right lobes and single left lung. Currently no automated or semi-automated tools exist for segmentation of the lobes due to the difficulty in automatically detecting fissures in the mouse lung. Therefore, the individual lobe segmentation was completed via manual tracing of each slice with aid of a live wire tool that “locked” to the contrasting boundary features. Following segmentation, the lobes were rendered in 3D for visualization purposes. Each lobe was given a unique label and color and visualized through a transparent 3D voxel intensity projection.
Airway measurements were completed on the manual airway tracing results obtained from the in vivo micro-CT datasets. An ellipse was fitted to the perimeter of each airway segment and the area, major and minor diameters, as well as the centroid were recorded with respect to the inner airway wall. The area and major and minor diameters were measured at 50% of the branch length for each of the identified airway branches. Five segments were averaged to give a mean value of these airway metrics relative to a position within the branch. In addition, the airway branch angles were measured (see Table 4). The measurements were based on the angle between the daughter branch and parent branch.
Mouse airway nomenclature.
Through visualization of the 3D airway tree from the three strains of mice, it was clear that the mouse airway tree required a nomenclature to allow for meaningful cross-correlative studies.
Therefore, we developed a nomenclature specifically for the mouse airway tree, taking into account the accepted airway nomenclatures that exist and developed an anatomically significant nomenclature that is a hybrid of the previous techniques (3, 11, 18, 30).
The labels for the airway segment of interest begin with an indicator of the lobe in which they occupy, as specified by the Nomina Anatomica Veterinaria, namely: left (L), cranial (Cr), intermediate (I), caudal (Ca), and middle (Mi) (28). Due to the monopodial branching of the mouse airways a main bronchus (MB) stem was assigned to traverse to the base of each lobe. For instance, in the cranial lobe the cranial right main bronchus (CrRMB) traverses to the base of the lobe (Fig. 1). The first generation of bronchi (B) that offshoot from the lobar main bronchus are split into divisions denoted by capital letters such as A, B, and C, and preceded by the letter B indicating a bronchi as opposed to MB for the main bronchus. For example, the first offshoot from CrRMB2, would be labeled as CrRBA3, where this segment is from the right cranial lobe (CrR), it is a bronchi (B) and is the first offshoot (A). In addition, it is the third generation from the trachea. Divisions off of these bronchi are further distinguished by assignment of a lower case letter a, b, c. For example, the first division off CrRBA3 would be labeled CrRBA4a, where this refers to the CrR, it is the first offshoot bronchi (BA) and the first division (a). Here we can see that the generation number is four, and its location in the label is now fixed after the first offshoot (BA). The next generation is labeled using roman numerals such as i, ii, and iii. For example, the first sub-division off CrRBA4a would be labeled CrRBA5ai. In all segments, the assigned number represents the generation of that segment with respect to the pathway from the trachea, which is defined as generation 0. An example of the nomenclature applied to a mouse airway schematic is shown in Fig. 1. Table 1 serves as a key for the nomenclature used throughout this study and includes the label, anatomical name, and generation number, through the 4th generation, for the labels that are depicted in the airway tree schematic. In addition, Table 1 references the binary labeling developed by Phalen et al. (18).
High-resolution micro-CT lung imaging.
Using the IIBH technique, high-resolution in vivo images have been obtained providing clear depiction of the mouse lung anatomy, including airway and lobar structures in the C57BL/6, A/J, and BALB/c mice (Fig. 2). Here in Fig. 2, we can clearly discern the central and peripheral airways, airway wall, parenchyma micro-architecture and, most impressive, the fissure lines from the transverse and coronal views taken at the anatomical position where the middle and accessory branches split on the right side. Also evident from the representative images in Fig. 2 is the visual difference in the central airway size of the C57BL/6 compared with the A/J and BALB/c mice.
These datasets have been further processed as detailed in the methods through manual and semi-automated segmentation for quantitative analysis and qualitative 3D visualization.
Three-dimensional mouse lung and airway tree visualization.
Segmentation of the lobes in the right and left lung was completed to verify the number and position of lobes that exist in the mouse lung for the three strains of mice used in this work. It was found that the C57BL/6, A/J, and BALB/c mice each had four lobes in the right lung consisting of the cranial, accessory, caudal, middle and a single left lung. The reconstruction of the mouse lung anatomy for the C57BL/6, A/J, and BALB/c mice is depicted with both the segmented lobe structures overlaid onto the segmented airways (Fig. 3). In addition to integrating the airways with the lobes as in Fig. 3, they have also been independently presented in Fig. 4.
Lung volume analysis.
The mean total lung volume for the C57BL/6, A/J, and BALB/c mouse strains were calculated at 20 cmH2O positive airway pressure. The total mean and standard error (SE) volume of the lungs of the C57BL/6, A/J, and BALB/c mice was 992.5 (±39.7) μl, 965.5 (±50.2) μl, and 1,152.5 (±62.5) μl. A statistical comparison of the mean lung volumes among the strains was completed using a one-way ANOVA followed by Tukey's test for the pairwise comparison of means between the strains and it was found that with the current sample size there were no significant differences in calculated lung volume between the three strains of mice at a positive airway pressure of 20 cmH2O. Comparable values for inter-strain lung volumes and intra-strain variation for lung volume and have been reported for the C57BL/6 and A/J strains (15, 23, 27).
An airway analysis approach was completed at the midpoint of the branches that were identified. An ellipse fitting algorithm was applied and airway measurements were computed at 50% of the branch length. The resulting metrics included major and minor airway diameters as well as lumen area (Tables 2 and 3).
Pairwise mean comparisons between strains were used to compare the area, major, and minor diameters among the three strains. This analysis was performed for the main bronchus pathways of each lobe. Significant differences were found for multiple main bronchi pathways in certain lung regions between the C57BL/6 inbred strain and both the A/J and BALB/c strains. For instance, the MiRMB pathway of the C57BL/6 was significantly different (P < 0.001) for area, major, and minor diameters compared with both the A/J and BALB/c strains. In addition, the RMB pathway of the C57BL/6 was significantly different (P < 0.0002) for area and major diameter compared with the A/J and BALB/c strains. However, there were no significant differences found for the RMB pathway of the C57BL/6 compared with the A/J and BALB/c strains based on the minor diameters. This indicates that the major diameter measure is more sensitive to the “bulging” region, which is noncircular and visually evident from the 3D depiction of the airways. The CaRMB and LMB pathways for the C57BL/6 were also found to be significantly different (P < 0.05) compared with the A/J and BALB/c strains. No significant differences (P > 0.99) were found between the A/J and BALB/c strains for the airway measures of area, major, or minor diameters at any of the measured branch points. Also, there were no significant differences detected between the three strains in the CrRMB and AcRMB pathways.
Phenotyping the lung function in the mouse has revealed strain specific differences between parameters such as lung volumes, compliance of the lung, respiratory system resistance, and ventilatory responses (10, 15, 19, 21, 25–27), to name a few. However, as reported by Reinhard et al. (19), there is no direct evidence that any one strain exhibits directly correlated relationships between lung function parameters, which suggests that they are not necessarily coinherited. In addition, there has been an increased interest in the hypothesis that inter-strain variations in lung function are linked to inter-strain variation in lung morphology.
Previous studies on mouse lung structure have included investigation of alveolar structures at the light microscopy level and found substantial differences in alveolar size between inbred mouse strains (24). However, to date, no comparison of the size and shape of the airways in these mice in vivo has been completed. Studies evaluating the morphology of the airway structures in the mouse have used lung casting techniques where manual caliper measurements of airway length, diameter, and branching angle have been made (4, 17). More recently, micro-CT imaging of these airway casts has also been completed for a more rigorous quantitative analysis and development of an airway tree data structure (4). Casting methods have shed light on the morphology of the mouse lung, yet they are intensive, inherent to processing and subjective analysis errors, and are not from the native in vivo state. Additionally, the process of casting may induce a distortion in the airway shape and due to its destructive method limits the study between the airways and the surrounding lung tissue. Casting also limits the ability to image over time to evaluate airway dynamics.
On the basis of the limitations of mouse airway casts, the development of noninvasive imaging systems for these small animals (4–6, 12, 16, 20, 22) is essential in creating airway descriptions. In vivo imaging provides a noninvasive method for assessment of lung structure in 3D with a high resolution. Using the micro-CT imaging technique outlined in the methods we are able to extract airway dimensions of 100 μm and consistently track the airways down to the 6th generation (∼400 μm) based on our nomenclature scheme. Previous micro-CT imaging has been used for in vivo assessment of the mouse lung, with very limited measurements of the airways (only the left and right main bronchi) due to resolution limitations (7). Ex vivo micro-MRI scans from fixed lungs have been previously performed with a voxel size of 63 μm, but without a detailed report on the actual resolving capacity and no extracted airway measurements it is difficult to quantitatively compare techniques (9). In general micro-MRI imaging does not provide the same signal-to-noise that we can obtain with x-ray CT, and from such previous studies the resolution appears to be several times worse than the presented micro-CT results.
We have, through our developed techniques for imaging the mouse lung, uncovered significant differences in lung structure in vivo that may play an important role in understanding the lung function in the normal and diseased mouse. Use of in vivo micro-CT imaging to aid in phenotyping the lung structure within the C57BL/6, A/J, BALB/c inbred mouse strains has further confirmed that the “normal mouse lung” consists of a spectrum of characteristics. We have in this body of work focused on the structural differences in three commonly used mouse strains. Our findings provide evidence that just as lung function varies between strains lung structure also varies.
Qualitative assessment from our 3D airway trees (Fig. 4) in the C57BL/6, A/J, and BALB/c mice clearly demonstrates the differences that exist between mouse strains. The most notable phenotype can be seen in the 3D airway reconstructions, with the significant bulging pattern or inverted cone occurring in the C57BL/6, which does not appear in the A/J and BALB/c mice. The A/J and BALB/c mice exhibit a more linear cylindrical airway shape. From previous observations of mouse airway casts and the 3D renderings of the mouse airway tree from micro-CT datasets of the C57BL/6, A/J, and BALB/c inbred mouse strains in this study, it can be seen that the major airways in the mouse have a near zero bifurcation angle at the carina that is almost perpendicular to the central axis of the imaging plane (14). However, due to the strain-dependent unique geometrical properties of the mouse airway tree revealed through this work, it was found that the airways could not be modeled as an object with a circular cross-section due to the unique geometrical properties. This was vital in the implementation of our quantitative analysis to verify the bulging pattern of the C57BL/6 airways and not incorrectly identify the bulges as circular cross sections that were ellipsoids due to the airway angle and imaging plane. Through our imaging technique we have not only identified a bulging pattern of the C57BL/6 airways, but we have also verified quantitatively (Tables 2 and 3) that the C57BL/6 has statistically significant larger airways in several of the main pathways.
To quantitatively analyze the airway structures between strains, we developed a nomenclature for the mouse airways. Historically, the bronchi have been labeled anatomically to aid clinicians in identifying and locating disease conditions that are localized to certain regions within the lung. The nomenclature systems developed for the human airways include that of Jackson and Huber and Boyden (3, 8). Both systems are based on division of the lobes into sublobes and classifying the bronchial segments into the sublobe regions in which they occupy. Subsequently, labeling techniques for analysis were developed by Weibel, Phalen, and Horsfield (11, 18, 30) for the dichotomous branching structure of the human airways. However, the monopodial branching structure of the mouse airway tree would not be labeled appropriately using the techniques previously developed by Weibel and Horsfield due to the assumption that both make regarding a dichotomous system. As research continues to use animal models, in particular mice, for studying human disease conditions in the lung, it is also vital to have a useful nomenclature customized for the monopodial mouse airway tree. Wallau et al. (29) created a nomenclature for the entire rodent mammalian order. The goal was to develop a nomenclature that described and classified rodent lungs to assess relationships between species using biological systematics (i.e., phylogenetic analysis). Application of this nomenclature to the mouse airway tree for comprehensive morphometric analysis as undertaken here, revealed many drawbacks. In particular, the inability to track generation number and group similar size branches for comparative analysis of peripheral airways was problematic.
Using our developed nomenclature we can group the main bronchi branches and side branches, which follow the monopodial system, according to anatomical location as well as generation number for comparative analysis to reveal intra-strain and inter-strain variablity. According to our method, based on the monopodial system, a generation is defined at the point where a new branch is found to split off of the branch of origin. Therefore, using our imaging technique we have been able to reliably image up to the 6th generation on the monopodial lung structure.
The observed differences in structure between the C57BL/6, A/J, and BALB/c strains have important implications in understanding normal mouse lung function, inhalation studies of flow, and the development of mouse models for respiratory disease research. Lung function measures including airway resistance and lung compliance could be affected by the observed inter-strain differences in airway morphology. On the basis of the bulging pattern seen within airway segments of the C57BL/6 strain it would be anticipated that the airway resistance would be lower compared with the A/J and BALB/c strains. However, the overall airway resistance reported for these inbred strains has not been found to be significantly different (21). Additionally the overall dead airspace in the C57BL/6 (97 μl) was not significantly different than the A/J (102 μl) and BALB/c (111 μl) strains. It is interesting to note that the bulging occurs between bifurcations, while the airway dimension is narrower at the bifurcations. This could be related to differences in the tissue composition, such as the collagen matrix, at these different regions throughout the airways. Ultimately, these regional differences in structure affecting airway resistance and airflow may not be discerned through the current techniques used to measure global airway resistance. Lung compliance could also be affected by airway morphology due to the interdependence of forces between the parenchyma and airway tree during respiration. Significant differences in lung compliance have been reported in the C57BL/6, A/J, and BALB/c strains as 62, 130, and 89.7 μl/cm, respectively (21). One hypothetical cause of the difference in airway structure could be due to the significantly stiffer lungs of the C57BL/6 strain placing an outward tension on the airways, resulting in the bulging regions that have been visualized in this study.
Due to the differences in airway structures observed in the normal lung of the C57BL/6 and A/J and BALB/c strains we believe that there are potential pathologic implications in the fields of inhalation toxicology and obstructive lung disease. In the field of inhalation toxicology, the differences in airway size and shape between strains will likely affect airflow patterns, differing distributions of inhaled toxins, and differences in the regional function of the lung based on airway geometry. Asthma modeling in the mouse may also change based on our findings of the bulging and enlarged airways of the C57BL/6. The impact on the structural differences between strains on asthma modeling remains to be seen.
Ultimately, further investigation into the relationship between structural and functional measurements for evaluation of normal and diseased conditions in the mouse lung needs to be undertaken using both new imaging approaches presented here and lung function techniques that have been previously reported.
Our research has uncovered new knowledge of the phenotypes that exist in the normal lung of the C57BL/6, A/J, and BALB/c mouse strains in vivo. We have, through the use of the breath hold technique, been able to clearly extract features of the normal anatomy of the mouse lung using in vivo micro-CT imaging. This includes the ability to segment the central and peripheral airways as well as segmenting the lobar boundaries.
We believe that the variability seen between these three strains of mice has the potential to have important implications on the respiratory mechanics, physiology, and pathology of the mouse lung. Inevitably, these differences will need to be assessed carefully for future research investigating and developing mouse models of pulmonary disease conditions. Table 4
This project was supported in part through an National Institutes of Health shared instrumentation Grant S1RR019242 and National Heart, Lung, and Blood Institute HL-080285.
No conflicts of interest, financial or otherwise, are declared by the authors.
The authors thank Jered Sieren for scanner facility support.
- Copyright © 2010 the American Physiological Society