The Phenotyping Workflow
    Our skeletal phenotyping program is only affordable because the existing KOMP program at Jackson Laboratory has absorbed the cost of confirming the genomic structure of the inactivating insertion and the animation of the mouse lines from hES cells.  Their production pipeline provides breeders that have been genotyped and their initial experience with the breeding performance and vitality of the mouse line.  In addition, their phenotyping includes a full body skeletal X-ray and DEXA for body composition, and that information is incorporated into our skeletal dataset.  Where possible, we choose to characterize homozygous KO lines to avoid genotyping costs inherent to a heterozygous line, but even this cost is somewhat mitigated by the existing PCR probes and facility for genotyping that is in place with the KOMP program.   Because there are two other production sites in the US, the potential for capturing a larger volume of KO mice for skeletal phenotyping has not yet been fully realized.   Our workflow has been designed to be able to flexibly handle the potential volume that these production sites could provide. 
1. Breeding and sample collection:

    Two harem breeding cages are usually sufficient to produce our target of 10 male and female mice that are grown to 12 weeks of age.   At 11 weeks of age, the mice are injected with calcein, and the day prior to sacrifice they are injected with alizarin complexone.  The tissues from the adult animals are harvested in batches until a full complement of 8 animals per sex.   Both lower limbs, the spine, head and tail snip are place in 10% neutral formalin and shipped on ice to our skeletal phenotyping center (UCHC).

    Upon receipt at UCHC, the staff confirms that it has received the animals that were sent, and they check to tail snips to ensure that they did incorporate the green and red fluorescent label.  Samples that did not get labeled are replace with another animal from the reserved of 2 animal per sex that obtained from the breeding process.  The processing site accumulates the tissues from each KO line until the full 8 samples per sex are received.

 

2. Data generation:

     After 3 days of formalin fixations, the sample processing site distributes the dissected tissues into bar coded vial containing cold 70% ethanol (µCT) or 30% sucrose/PBS (histology) at -80ºC until all samples have been received.  The 8 femur and vertebral (V4-6) samples per sex are processed first for µCT analysis and the resulting information deposited into the LIMS system.  The process for scanning the bone samples and the subsequent steps identifying the ROI, threshoding the sample, calculating the measurements and depositing the results into the LIMS has been significantly upgraded to reduce the time and technician effort because every KO line must go through the µCT step.  At our weekly video-conference meeting, the decision is made (dashed arrow), based on the DEXA and µCT data if performing the histomorphological analysis is warranted. 

      The histomorphometic analysis is performed on the distal femur and vertebral (V1-3) bodies.  The sample set of 16 long bones or vertebrae are processed as a single batch using the cryohistological protocol that has been developed in our laboratory.  Four samples are positioned and frozen into a single sample mold.  The tape transfer process captures the section with the 4 bones, and two tapes containing all 8 samples from the same sex are place on the microscope slide.  Three sections per each bone, sampled at 100µ intervals, are used for the analysis.  Thus for a complete study, 6 slides are generated and all the subsequent steps are performed on these 6 slides.  A crucial step is the placement of fluorescent beads adjacent to each tissue section that are utilized for registration of the images that will be generated.  Additional replicas of the slides are made and held in reserve in case a repeat analysis is required.

    The slide set then enters our imaging -> staining -> reimaging process to produce the fluorescent signals that are used to compute the histomorphometry.  The slides are loaded into the Axioscan Z1 microscope for imaging the accumulated mineral and mineralization lines that are used to compute the static and dynamic measurements.  The computer software discriminates each sample and provides the appropriate file name for each sample and imaging step.  Three files, mineral, green label and red label are generated in the first imaging step.  The slides are removed from the scanner and stained for TRAP activity under acid conditions, which removes the fluorescent mineral signals.  The slides are rescanned for yellow (TRAP), red and green (registration beads).  The third step is the AP stain which generates a red fluorescent signal as well as DAPI (blue) that is used for identifying osteocytes in the cortical bone and red/green for the registration beads.  The last step is a chromogenic stain using toluidine blue the currently is performed to produce a familiar image of bone but eventually will be used for phenotyping of articular cartilage.  The individual files from each round of imaging steps (approximately 4G of digital data), each with its unique name, is ported to a sample server located on the UCONN Storrs campus.

    The image analysis algorithms that are used in our study were developed by Dr. Seung-Hyun (Sean) Hong over a 3 year period of testing and evaluation. The details of the method have been published (ref). In the past year, the analysis has been adapted to a 24 node computer cluster that dramatically increases the throughput of the analysis. Prior to running the study, the program checks the submitted files for problems that would require a sample resubmission. Issues such as a weak enzyme stain, improper stitching or inadequate number of registration beads are problems that might require further attention by the histology staff. The analysis generates the histomorphometry measurements, identifies values that fall outside 1.5 SD of the group mean and calculates the group means, SD and p values. These calculations plus the computer-generated images on which the study was based are deposited into the LIMS database.
  3. Data analysis and digital publication:
    The LIMS is designed to capture all the individual steps of the analysis as well as the collected data from a study. This information is displayed in a variety of formats to assess the quality control of the process and to identify problems that is slowing the workflow or adding technical variation that can confound the analysis. However the major purpose of the LIMS it to summate the information into an easily understood overview of the phenotype of each KO line. Inherent in the analysis is determining the stability of the control (background) group. Each month a new control line is entered into the analysis pipeline and the results are contrasted with previous runs. It is still to early to know if there are seasonal variations that can be detected, but definite problems with certain histological step have been identified and corrected by this process. The PIs of this project meet weekly by 4 way video conference to review the control data and to interpreted the information gained from the analysis of a KO line. The 8 samples per sex provides sufficient power to discriminate a 20-30% difference in most measures, and it also allows for some failed measures without compromising the entire analysis. Various visual modalities for summating the analysis in terms that are meaningful to the bone biologist are being developed. The interpretation of the bone phenotype is categorized into computer searchable terms that should be useful for an external user to identify specific types of KO lines. Once the PI group agrees with the interpretation and classification of the KO line, key elements of the analysis will be uploaded to the searchable public database for skeletal phenotyping that is house on this website.