A review of Bioinformatics 

resources on Limb Dysplasia

 

Sandra Rodriguez-Zas and Heather Grisco

Department of Animal Sciences

University of Illinois at Urbana-Champaign

Questions can be directed to Dr. Sandra Rodriguez-Zas  (rodrgzzs@uiuc.edu) Email




Outline

Introduction
Limb Dysplasia Genotypes and Phenotypes
Laboratory Animal Studies
Objectives
Biology Workbench
Getting started with Biology Workbench
Alignment of sequences using CLUSTALW
Post alignment comparison of sequences
Database Search Exercises
Pattern-based Protein Databases
Nucleic Tools 
Other Genes that Affect Skeletal Development
Phylogenetic Tree
Quaternary Structure
Technical Summary
Commercialization of Molecular and Bioinformatic Discoveries
Is Bioinformatics in your future?
References and other Links



Introduction


Skeletal development in vertebrates is controlled by proteins such as the Bone Morphogenetic Proteins (BMPs) and the closely related Growth and Differentiation Factors (GDFs), members of the TGF beta superfamily . In vitro (not using a living organism) and in vivo (using living organisms) studies concluded that BMP modulates development of limb or chondrogenesis (Roark and Greer, 1994). Francis-West and coworkers (1999, Development 126, 1305-1315) studied how GDF-5 acts during skeletal development.  These authors showed that show that GDF-5 affects the two stages of skeletal development through two different mechanisms. Firstly, GDF-5 promotes the initial condensation of mesenchymal cells (undifferentiated cells that originate bone and cartilage) by increasing cell adhesion. Secondly, GDF-5 signals to the epiphyses of the adjacent skeletal elements controlling chondrocyte (differentiated cell responsible for extracellular matrix of cartilage) proliferation in the joints. 

A dictionary on many biomedical concepts can be found at the CancerWeb (http://www.graylab.ac.uk/omd/index.html).


Limb Dysplasia Genotypes and Phenotypes

A hand Two examples of abnormal skeletal development are Hunter-Thompson and Grebe's disorders.  These disorders are characterized by the shortening of the appendicular skeleton and loss or abnormal development of some joints, with more severe effects distally and the loss of one or more joints. Photos depicting skeletal abnormalities and tissue development can be found in Francis-West et al. (1999) paper.

Single chromosome The protein GDF-5 (also known as CDMP) has a major role in the development of the skeleton extremities, particularly at the early stages of chondrogenesis. Two good resources on GDF5 and Dysplasia are the papers by Manouvrier-Hanu et al. (1999) and by Francis-West et al. (1999). The variety of symptoms observed is due to the fact that the level of GDF-5 expression and activity influence the size of skeletal elements. Researchers found that a mutation in the GDF gene leads to structural changes in the coded GDF protein. The altered protein cannot combine with BMP properly, preventing the formation of this functional heterodimer and the secretion of signaling molecules. Recessive mutations (mutations detrimental in homozygotes) in the GDF-5 gene are responsible for Hunter-Thompson and Grebe's type chondrodysplasias in humans and for brachypody in mice. 

A hand In Hunter-Thompson Dysplasia cases, feet are more severely affected followed by hands.  Hunter-Thompson's syndrome and brachypodism are caused by a missense mutation that changes an amino acid in GDF-5 resulting in the loss of GDF-5 function (Storm et al., 1994).  The mutation (addition or deletion of a nucleotide) causes a frameshift in the reading frame of the nucleotide sequence, altering the highly conserved 7-cysteine pattern that conditions the protein structure and function.

A foot A person carrying two copies of the allele that causes Grebe's Dysplasia has missing or greatly shortened limbs. The bones in the hands are abnormal and fused, and some may be absent. Carriers of one copy of the mutated gene may have mild skeletal abnormalities such as shortening of the first metacarpal and middle phalanges of fingers two and five.  Grebe's disorder is due to a mutation in the conserved cysteine in the functional domain of the protein CDMP.  The effects of Grebe's mutation are more severe than those associated with Hunter-Thompson's mutation and may be due to loss-of-function of other protein members of the TGF beta family with which GDF-5 combines forming an heterodimer. 

Question mark Is the gene coding for TGF beta in a large or small chromosome?  How many chromosomes does a human have? How large are they? 
Magnifying glass The Genome Database at NCBI provides answers to these questions. NCBI, the National Center for Biotechnology Information, is part of the National Library of Medicine, National Institutes of Health. This source comprises a wide range of molecular biology information including genome and biomedical information.
Question markAre there other skeletal development disorders caused by mutations in GDF-5 or CDMP?

Magnifying glass Visit the PubMed web page and search for information on Dysplasia disorders.  PubMed is search service offered by the National Library of Medicine with access to over 11 million citations. How many did you find? Compare your findings with other resources available in the Reference Section.

Inheritance of Limb Dysplasia and Mendel's Law

Limb Dysplasia is an heritable trait.  The more extreme cases are observed in homozygous individuals that carry two copies of the mutated gene allele.  These individuals received one mutated allele from their mother  and one from their father.  The probabilities of an individual to be homozygous for the mutation are: 25% if both  parents are carriers of one copy of the mutation; 50% if one parent is homozygous and the other is an heterozygous  carrier; and 100% if both parents are homozygous for the mutation.  These probabilities are computed following  Mendel's First Law.  Genetic tests are available to assess if an individual with no Limb Dysplasia phenotype is a  carrier of the mutated allele.



Laboratory Animal Studies

A henAlthough the main interest in skeletal development is associated with human well being, the study of laboratory animals has helped to elaborate hypothesis on how the GDF and BMP proteins influence the skeleton. One example of this is the research on the effect of loss or gain-of-function of proteins in the mouse and chicken on defective skeletal development (Kingsley et al., 1992). Also, subcutaneous implants of BMPs in adult rats were proved to cause skeletogenesis (Carrington and Reddi, 1991; Rosen and Thies, 1992).  A mouse


Objectives

Our aim is to better understand the genes, proteins and signal pathways influencing Limb Dysplasia in humans and in other animals using Bioinformatic tools and public resources. 

A detectiveWe will be bioinformatic detectives, searching for gene and protein clues. We will search public databases, analyze sequences and visualize protein structures related with the growth dependent factor 5 (GDF-5 or CDMP1) and the transforming growth factor-beta (TGF-beta or TGF beta). Both factors influence growth and mutation of their sequences could potentially cause Limb Dysplasia.



Biology Workbench Biology Workbench logo


A computer The Biology Workbench 3.2 (http://workbench.sdsc.edu) will be our virtual detective's office, a place in the internet where we can gather and analyze the gene and protein information.  The Biology Workbench integrates a myriad of bioinformatic resources and tools facilitating individual and guided investigation and discovery.  Through the Biology Workbench we can also access the resources available at NCBI and PubMed that you have already used to find more information on Limb Dysplasia. 

Tool preferences
 

Although most of the Bioinformatic tools can be used with Netscape Navigator or Microsoft Internet Explorer, a few are only compatible with Netscape, so this browser is preferred over Explorer. The screenshots used in this tutorial were obtained using Netscape 4.7 running under Windows 98 so your screen may be slightly different. We recommend to open two web browser simultaneously. To do this, select the "New" option under the "File" pull-down menu in your browser and then select "Window". This will open a second window. Then in one of the windows, open the Biology Workbench for Students . In the other, window open the Biology Workbench (http://workbench.sdsc.edu). You can size the windows so that each occupies half of the screen (if you want). Having both windows open, one next to the other, permits quick and easy move between the tutorial directions and actually doing the Bioinformatic tasks. Alternatively, print the tutorial and follow the exercise using one web browser window.

HINT: If you are having problems viewing or printing the results, it may be helpful to change the background color (say no color). In Netscape this is done using the 'Colors' option in the Preferences from the Edit menu. In Internet Explorer this can done using the 'Colors' options under Internet Options from the Tools menu. You may also need to select or select a box to force your browser to use your new colors. Unfortunately, the actual wording varies between browsers and versions - in Netscape 4.6 select "Always use my colors, overriding document", in Netscape 6 select "use my chosen colors, ignoring the colors specified", and in Internet Explorer you must deselect the "Use Windows colors" box before you can change the colors.


Getting started with Biology Workbench

Step intoThe first step is to open the web page of Biology Workbench (http://workbench.sdsc.edu) and click in the link "Enter the Biology Workbench 3.2".  You will be prompt to give a "User Name" and a "Password". First time users need to set up an account in the Biology Workbench. Once this is done you can immediately use the Biology Workbench.
 

Type your "User Name" and "Password" and click the button "OK".  This will send you to the main page of the Biology Workbench.  The first set of Bioinformatic tools we will be using are the protein tools.  Click the button "Protein Tools".  This will send us to a window with various protein tools.  We are interested in searching information on specific molecules (Gdf5, TGF beta, etc.) so, click the button "Ndjinn-Multiple Database Search". These options hosts a miriad of databases that can be queried on a topic of interest. Then click the "Run" button.
 

Select Ndjinn
 

In the next window, check the boxes ( A checked box) next to the databases we want to query.  These are PIR, PDBFINDER, and SWISSPROT. The protein information resource (PIR), the protein database finder (PDBFINDER) and an European protein database (SWISSPROT) have structural information and references for proteins and related molecules. Other databases can be chosen for this project, but these three are the most commonly used because they are very comprehensive, constantly updated.
 
Type "Gdf5" in the white query space. Click on the "Search" button to submit your query (the default, "Show 10 Hits" is sufficient for our example). Results should be available within a few seconds. Once the search is completed a few matches will appear in the next window. Searches may take more or less time depending on a variety of factors such as databases included and internet connections. Stopwatch gif

A computer To bring a few records up to your workbench check ( A checked box) the SWISSPROT Gdf5 human and SWISSPROT Gdf5 mouse records and then click the "Import Sequences" button. This will bring you back to the screen that you originally started with. Re-check the two boxes that say SWISSPROT gdf5: human and mouse, in preparation to align the sequences.



Alignment of sequences using CLUSTALW

CLUSTALW aligns two or more sequences maximizing the similarity between paired positions (equal or similar amino acids are more likely to be paired than amino acids with different properties).  In order to obtain the best alignment, sometimes it is necessary to introduce gaps in any one of the sequences. 

We wish to align the protein sequences that we have just imported. The method chosen, CLUSTALW needs to be highlighted (click on it in the scroll menu).  We have already selected both Gdf5 sequences, so the last step is to click the "Run" button. 
 

Select Clustalw
 

The next screen indicate the default settings that CLUSTALW will use.These default options are suitable for our alignment, so just click the "Submit" button. The results of the alignment are: 
 

Alignment of the GDF5 human and mouse sequences
 

In the aligned sequences, blue (asterisk) denotes that the position is completely conserved, green (:)  denotes conservation of strong groups and, dark blue (.) denotes conservation of weak groups.
  Magnifying glass Examine this alignment and identify what areas are completely or highly conserved.
  Question mark Are there many similarities between the human and mouse sequences? How similar do you think the nucleotide sequences of the genes that coded for these two proteins are?  Remember that three nucleotides (triplet or codon) code for one amino acid but of the 64 possible combinations, only 20 amino acids are formed.


Post Alignment Comparison of Sequences

Another way to compare the already aligned sequences is by using the BOXSHADE tool within the "Alignment Tools".  Once you reviewed the CLUSTALW alignment, click on the "Import Sequences" button. This will send you back to the initial screen. 

Click on the "Alignment Tools" button. Check the box marked CLUSTALW-Protein that contains the GDF alignment. Now highlight the BOX SHADE option from the scroll menu and click the "Run" button.  The defaults of BOXSHADE will be presented. As with the CLUSTALW, we will accept the defaults so the only thing left to do is to click on the "Submit" button. The BOXSHADE output colors the aligned sequences based on the degree of similarity between amino acids at every position. Different colors are assigned to positions that are fully conserved in all proteins (e.g. same amino acid), nearly conserved (amino acids with similar properties), and not conserved.  The conserved terminology related to the assumption originally there was one GDF5 sequence, and that across generations there have been mutations in some positions, leading to the slight differences between the human and mouse sequences.
  Magnifying glass Examine the BOXSHADE alignment.
  Question mark How similar are the amino acid sequences? How similar would the tertiary structure be? Do you expect that the amino acid differences would influence the number and arrangement of the alpha-helices and beta-strands in the tertiary structure? 



Database Search Exercises

Magnifying glass Search for Hunter-Thompson type of Acromesomelic Dysplasia in the OMNI database. What information did you get?

Magnifying glass Search for Grebe's type of Chondrodysplasia in the OMNI database.
What information did you get?

Magnifying glass Search for growth differentiation factor 5; GDF5 in the OMNI database. Did you find information complementary to that found in SWISSPROT?

Magnifying glass Search for transforming growth factor beta; TGF beta in the OMNI database. Is it a common peptide?

Magnifying glass Search for in the bone morphogenetic protein 5; BMP5 in the OMNI database. How different is BMP5 from BMP4t?



Pattern-based Protein Databases

Jump into Search for TGF beta in the PFAM protein database.

Question markWhat information did you get? How does this result differs from the SWISSPROT one?  PFAM is a database of protein domain family alignments based on Hidden Markov Models.  Note that together with TGF beta you get information of other proteins with the same or similar patterns (domain).

Magnifying glass Look at the results from PFAM. By October 2000, PFAM reported 215 sequences sharing the TGF beta pattern. 
Question mark How many proteins were assigned to this group in your search?  What other proteins were found?
Magnifying glass Search for TGF beta in the PROSITE protein database.

How do these results differ from the PFAM ones?  The database PROSITE contains protein families and domains and offers information on profiles that can help in assigning proteins to families. 
Magnifying glass Review the PROSITE record for TGF beta. 

By October 2000, 117 proteins contain the TGF-beta family signature. 
Question mark How many proteins were assigned to this group in your search? 
Question mark What other proteins are part of this family? (BMP?, IHA?, DVR?, MIS?,...). 

Note that PFAM found 215 matches meanwhile PROSITE reported 117.  What proteins did not show in the PROSITE search?  Why do the number of proteins differ?  The different approaches used to group proteins influence the results from both databases.

Create your own Bioinformatics Tutorial around IHA or MIS.


Nucleic Tools
 

DNA iconLet's compare the nucleic sequences of the GDF5 human and mouse genes and see how close your prediction is to the Bioinformatics approach. To do this, go to the main page of the Biology Workbench. You can reach this page from other Biology Workbench pages (sections) by clicking in the "Return" button. Once in the main page, select the "Nucleic Tools" button.

Select NDJINN-Multiple Database Search and click in the "Run" button.  In the next page, type GDF5 in the query box and check the Genetic Databases GBPRI, GBROD and GBVRT and and click in the "Search" button. These three GeneBank databases concentrate in PRImates (Human), RODents (mouse) and other VERtebrae (e.g. chicken).  This time select the GBVRT, GBROD, and GBPRI databases. Are you having "flash-backs"?  Yes, the nucleotide database search resembles the protein one you just did on GDF5. 

A few matches will appear. Select (check)from the matches one human GDF5 (GBPRI), one mouse GDF5 (GBROD) and one chicken GDF5 (GBVRT) matches.  Click the "Import Sequences" button to bring these sequence to your workbench and compare them.  This will bring you back to the main page and now you can perform a CLUSTALW alignment and then obtain the BOXSHADE comparison of these nucleotide sequences, following the same procedures used for proteins.

  Magnifying glass Examine the CLUSTALW and BOXSHADE results.

  Question mark How similar are these sequences? 
Are there two sequences that seem to be closer to each other than to the third one?  Which sequence is more different from the rest. 
 

GDF5 human, mouse and chicken nucleic sequence alignment 



Other Genes that Affect Skeletal Development

TGF-Beta

TGF beta is expressed in the prechondrogenic mesenchyme (like GDF5) and can induce or promote the formation of the initial condensations in micromass cultures (Heine et al., 1987).

The above searches can also be done for TGF beta.  Remember, TGF beta is the protein that heterdimerizes with GDF5, providing the signals necessary for adequate skeletal development. 

Jump into Apply to TGF beta the same Bioinformatic concepts and tools that you used to know more about GDF5 (type in TGF beta instead of GDF5).

 
Magnifying glass Search for TGF beta in the OMNI database. What information did you get?
Magnifying glass Search for TGF beta in the Genetic Database TRANSFAC. What additional information did you get?

If there is a mutation in one of the genest coding for GDF5 or TGF beta, the heterodimer formation can still occur, but it will be nonfunctional, the signaling will be distorted and Limb Dysplasia may occur. 

Other Genes
BMP-2, BMP-3: act at the later steps of chondrogenesis (Carrington et al., 1991; Jiang et al., 1993).

Activin: acts at older ages and promotes chondrogenesis by increasing cell adhesion and recruitment of mesenchymal cells into condensations (Jiang et al., 1993).

BMP-4: overexpression increases chondrogenesis.
Magnifying glass Find more information on these genes following the same steps as for GDF5 and TGF beta. 
Magnifying glass Search for Noggin, a gene similar to the Nog gene in mouse in the OMNI database. How is this gene related to skeletal development?


Phylogenetic Tree

Within the Alignment Tools, in addition to BOXSHADE, there are various methods to estimate the "genetic distance" between nucleotide (or protein) sequences and create a tree shaped output providing the relative distances between the sequences.  Once you have "Imported" an alignment (e.g. run CLUSTALW), click on "Alignment Tools". Highlight (click) the "DRAWTREE-Draw Unrooted Phylogenetic Tree" from "Alignment" option from the scroll menu and click the "Run" button. The defaults of DRAWTREE will be presented. We will accept the defaults so the only thing left to do is to click on the "Submit" button. The DRAWTREE output is: 

 

Animated treeTree (Note: tree is in PDF format. You must have Adobe Reader Installed to view it)



Quaternary Structure

The quaternary structure is the total structure of a protein, and is based on primary, secondary and tertiary structures. The primary structure is the nucleotide sequence, the secondary structure is the amino acid sequence and the tertiary structure is composed of alpha helices and beta strands coils and formations that allow the protein to perform its functions.

To get a record containing information on the 3D structure of TGF beta, search the Protein Databases (Ndjinn) for this protein.  Repeat the Protein search steps done for TGF beta but now use (check) the PDBFINDER protein database because it contains information on the 2D and 3D structure.

From the PDBFINDER TGF beta matches, select (check) PDBFINDER:1KLC.  Click on the button "Show Record(s)". A new web page with information on the protein 1KLC will be opened. 
Magnifying glass Browse the record and assess the number and position of the alpha helices and beta strands in the sequence.
Question mark How many twists and convolutions do you think TGF beta has? 

There are two ways to visualize the 3D structure of TGF beta.  One, click the "Protein Explorer" link provided in the records web page.  We will follow a different route.  We will bring the 3D record to our computer and visualize TGF beta using RASMOL.  To bring the record to our computer, click in the button "View 1klc structure".  A window will open asking you to open or save the file.  Check the option "Save the file", click the "OK" button, select a folder where you want to place this file and click the "Save" button.

 
RASMOL

We can visualize the quaternary structure of a TGF beta protein record retrieved in the Biology Workbench using a program called RASMOL (Sayle, 1996). If RASMOL is not already installed in the computer you are using, you can download it by clicking in the "Helper Application" link provided in the Biology Workbench main page.  Alternatively, go to RASMOL web page (http://www.umass.edu/microbio/rasmol).  The installation requires you to 1) chose the RASMOL version compatible with your computer (e.g. PC) and, 2) chose the folder in your computer where RASMOL will be saved.  To save RASMOL in the My Computer folder, highlight C: and click save.  The file name will automatically be rw32b2a.exe. Tutorials on using RASMOL can be found at http://www.umass.edu/microbio/rasmol/rastut.htm.
 
3D structure of TGF beta

Open RASMOL by clicking on the rw32b2a.exe file in your computer. In the black screen that appears click on "File", then click on "Open", and find the folder where the 1klc.pdb file was saved (it is usually convenient to save the molecule in the same folder where RASMOL is).

A wire frame structure of TGF beta 1 is displayed. There are also many other ways to view this structure.

Try other representations by clicking in the "Display" option of the toolbar (e.g. ball and stick). The "Cartoon" display shows the helices, strands and how TGF beta coils.
Magnifying glass Observe the TGF beta 1 molecule from various angles. You can rotate the molecule by holding it with the cursor and moving it.

TGF beta can only form functional heterodimers when available in this from.  Alteration of this "shape" would influence the capability to combine with BMP and emit signals for cartilage growth.
  Certain amino acids tend to be in the surface of the molecule because of their physical and chemical properties (e.g. hydrophilic).  Select in the toolbar, "Options" and within the menu, "Labels".  This will place labels in all the amino acids of TGF beta1.
Question mark Can you tell what amino acids tend to appear in the surface? 

Find a segment with a sequence of cysteines, common to humans, mice and chickens.  This region is highly conserved because it is critical to the shape and function of TGF beta. Changes in the DNA sequence, such as the frameshift mutation disrupts this region.  The TGF beta can still bond with other molecules but forming a nonfunctional heterodimer and causing Limb Dysplasia. 

 
TGF beta + BMP

A useful approach to understand the molecular complexity of the biological process is to visualize the formation of the heterodimer.


Magnifying glass Search in the protein database PDBFINDER for the molecule 1BMP.  Save this molecule in your computer as done with 1KLC.  Start another RASMOL session and open the 1BMP.pdb file.  Place the RASMOL windows containing 1KLC and 1BMP side by side.

1BMP (BMP1)


Question markHow do you think these two molecules would bind together forming a functional heterodimer? Changes in the molecules will cause the formation of a nonfunctional heterodimer.


Technical Summary

A frameshift mutation in the genes coding for TGF beta and GDF5 affects the function of these proteins and their possibility to bind to other BMP molecules forming functional heterodimers, critical in signaling cartilage growth. We have used various Bioinformatic tools such database search, sequence alignment, phylogenetic reconstruction and visualization of 3D protein structures to enhance the understanding of this complex biological process. 



MoneyCommercialization of Molecular and Bioinformatic Discoveries

Various companies are successfully marketing medicine and treatments based on the effect of GDF (CDMP) on skeletal development.  For example, Genzyme (a private company) has patented an approach that by removing cartilage from a healthy joint, they can multiply these cartilage and inject it back to an injured or arthritic (knee or jaw) joint.  To know more about this application of Bioinformatic results visit the National Institute of Dental and Craniofacial Research Digest web page on the topic. Physical Therapy


Is Bioinformatics in your future?

Fortune CookieWhat about doing an undergraduate project or a Master's Degree in the genetic factors influencing skeletal growth using Bioinformatic tools?  Learn more about two graduate students doing research in GDF5.


 

Acknowledgements 

 

We wish to thank the Biology Student Workbench team (B. Southey, C. Bruce, K. Engelsen, E. Jakobsson, D. Raineri, U. Thakkar, J. Snow and M. Won) for their invaluable contributions to this educational material.




References and other Links  Books

Acromesomelic Dysplasia, Maroteaux Type. 2000. OMNI
 

Cancer-Web online medical dictionary. 2000. http://www.graylab.ac.uk/omd/index.html.
 

Carrington, J.L. and Reddi, A.H. 1991. Parallels between development of embryonic and matrix-induced endochondral bone. Bioessays, 13:403-8. 
 

Carrington, J.L., Chen, P., Yanagishita, M. and Reddi, A.H. 1991. Osteogenin (bone morphogenetic protein-3) stimulates cartilage formation by chick limb bud cells in vitro.Dev Biol. 146::406-15.
 

Genetic tests for Skeletal Dysplasia. 2000.  http://www.kispi.unizh.ch/stomol_main/Stomol_Skeldys.html
 

Grim, M. 2000. Sickle-Cell Anemia tutorial. http://glycine.ncsa.uiuc.edu/educwb/tutorials/Sickle_Cell_Anemia/.
 

Heine, U., Munoz, E.F., Flanders, K.C., Ellingsworth, L.R., Lam, H.Y., Thompson, N.L., Roberts, A.B. and Sporn, M.B. 1987. Role of transforming growth factor-beta in the development of the mouse embryo. J. Cell Biol., 105:2861-7286.
 

Human Genome Map Project. 2000.  www.hgmp.mrc.ac.uk
 

Information on Dysplasia at Johns Hopkins University. 2000.  http://www.med.jhu.edu/Greenberg.Center/Greenbrg.htm#clinical.
 

International Skeletal Dysplasia registry at Cedar-Sinai Medical Center. 2000.  http://www.csmc.edu/genetics/skeldys/.
 

Jiang, W., Kahn, S.M., Zhou, P., Zhang, Y.J., Cacac,e A.M., Infante, A.S., Doi, S., Santella R.M. and Weinstein, I.B. 1993. Overexpression of cyclin D1 in rat fibroblasts causes abnormalities in growth control, cell cycle progression and gene expression.Oncogene, 8:3447-3457.
 

Kingsley, D.M., Bland, A.E., Grubber, J.M., Marker, P.C., Russell, L.B., Copeland, N.G. and Jenkins, N.A. 1992. The mouse short ear skeletal morphogenesis locus is associated with defects in a bone morphogenetic member of the TGF beta superfamily. Cell, 71:399-410.
 

Limb Dysplasia photos (Pediatric Research).2000. http://lww.com/PDR/0031-39983-99p291.html
 

Manouvrier-Hanu, S., Holder-Espinasse, M. and Lyonnet, S. 1999. Genetics of limb anomalies in humans. Trends Genet., 15:409-417.
 

Office of rare diseases.  1997. http://rarediseases.info.nih.gov/ord/news-reports/FY97annual/NICHD.htm
 

Roark, E.F. and Greer, K. 1994.Transforming growth factor-beta and bone morphogenetic protein-2 act by distinct mechanisms to promote chick limb cartilage differentiation in vitro. Dev. Dyn., 200:123-116.
 

Rose, V. and Thies, R.S. 1992.The BMP proteins in bone formation and repair. Trends Genet., 8:97-102. 
 

Storm, E.E., Huynh, T.V., Copeland, N.G., Jenkins N.A., Kingsley, D.M. and Lee, S.J. 1994.Limb alterations in brachypodism mice due to mutations in a new member of the TGF beta-superfamily. Nature. 14:639-643.