Holding It Together:a Closer Look at Collagen

Using the MacDNAsis program (version 3.5), several different analyses of the cDNA for collagen were performed:

• the largest open reading frame (ORF) was obtained from human collagen cDNA,
• the nucleotide sequence from the largest ORF was translated and a predicted molecular weight for human collagen was determined,
• a Kyte and Doolittle analysis was performed on the translated ORF to create a hydropathy plot in hopes of determining if collagen is an integral membrane protein,
• a Hopp and Wood analysis was performed on the translated ORF to create an antigenicity plot to determine the portions of the protein against which a monoclonal antibody could be developed,
• based on the ORF from human collagen, the predicted secondary structure of collagen was determined and compared to the three dimensional Rasmol image obtained from the NCBI archives,
• a multiple sequence alignment was performed on the primary protein structure of collagen from five different organisms: Gallus gallus (chicken), Caenorhabditis elegans (worm), Drosophila melanogaster (fly), Mus musculus (mouse), and Homo sapiens (human),
• a phylogenetic tree was constructed to determine the degree of amino acid conservation between these five species over time.

Part I: Open Reading Frame (ORF)

The DNA analyzed was the human cDNA for collagen. The largest reading frame of this DNA was chosen. This segment of the DNA began at nucleotide number 235 and terminated at nucleotide 5271. The image below shows the entire segment of cDNA. The largest reading frame, designated by the black box was the segment of DNA used for the rest of the DNA analysis.

The entire collagen cDNA sequence (along with protein translation) was obtained from the NCBI (National Center for Biotechnology Information. 14 November 1997. Entrez Protein Search. <http://www.ncbi.nlm.nih.gov> Accessed: 20 March 1998) and can be viewed by clicking below:

Part II: Translation and Determination of the Predicted Molecular Weight

The selected ORF (from above) was translated (DNA to protein) and its predicted molecular weight was then calculated (in daltons):

Part III: Hydropathy Plot (Kyte and Doolittle)

Using the algorithm developed by Kyte and Doolittle, a hydropathy plot was made to determine if collagen is an integral membrane protein. Below is the plot:

Positive values on this plot indicate areas of the protein that are hydrophobic. To be a candidate for a transmembrane domain, a segment of the protein must have a hydrophobicity reading greater than or equal to +1.8. As indicated in this plot, there are no portions of collagen with a hydrophobicity greater than or equal to +1.8. There are, however, several peaks that come close to the +1.8 hydrophobic threshold, including a peak at the very beginning of the plot.

Collagen is primarily an extracellular protein. To be secreted, it must be translated into the ER and then modified in the Golgi apparatus. In order to find its way into the ER, collagen carries a hydrophobic signal sequence at its 5' end. This sequence binds the ribosome and the collagen mRNA (that is being translated) to the ER membrane so that the mRNA can be translated directly into the lumen of the ER. The hydrophobic peak at the left-most part of Figure #2, therefore, may correspond to a signal sequence.

With the exception of this peak, there are no other peaks with hydrophobicity readings great enough to be transmembrane domains suggesting that collagen is not an integral membrane protein.

Part IV: Antigenicity Plot (Hopp and Wood)

Using the algorithm developed by Hopp and Wood, an antigenicity plot was constructed. The following are the results of the antigenicty study:

Antigenic plots are used to determine areas of the protein that are charged and therefore hydrophilic. Charged areas of the protein cannot be associated with the phospholipid bilayer (because it is hydrophobic); therefore, these areas of the protein point away from the membrane. In this configuration, these segments of the protein can be recognized and bound by immunoglobulins (antibodies). To make a monoclonal antibody against collagen (to be able to detect it), one of these areas would be used. The antigenicity plot above reveals numerous antigenic (hydrophilic) areas that could be used for monoclonal antibody production.

Part V: Predicted Secondary Structure

The translated ORF of the human collagen cDNA was further studied by constructing a predicted secondary structure map:

This predicted secondary structure can be compared to the actual, crystallized structure of type VI collagen. Click below to see the Rasmol image of type VI collagen. Once the image has appeared, click on Display and then drag down to Ribbons. This is the best format to see the secondary structure (alpha helices and beta-pleated sheets) of collagen to which the above map can be compared.

• Rasmol Image of Collagen (Type VI) (source: National Center for Biotechnology Information. 14 November 1997. Entrez Structure Query. <http://www.ncbi.nlm.nih.gov/Structure> Accessed: 20 March 1998.)

The predicted secondary structure from the MacDNAsis agrees, in part, with the actual three dimensional structure depicted in the Rasmol image. The Rasmol image is not, however, the complete type VI collagen protein. Instead, it is only a fragment. Through a comparison with the predicted secondary structure determined by the MacDNAsis program, it appears that the Rasmol image shows the first third of the total collagen protein. Both the predicted secondary structure and the Rasmol image both start with an alpha helix followed by beta sheets and interspresed coiled sections of protein. The Rasmol image terminates with a coiled section followed immediately by a long alpha helix. This same pattern appears in the predicted secondary structure nearly a third of the way through the map.

Furthermore, by clicking on Windows and then Command Line (of the Rasmac program), some facts are given about the Rasmol image of type VI collagen. Some of the relevant information with respect to this discussion is:

• Number of alpha helices: 2
• Number of strands: 5
• Number of turns: 7

These numbers give further credence to the idea that the Rasmol image graphically represents the first third of the predicted secondary structure. The first seven turns of the Rasmol image correspond (in terms of the postion and number of alpha helices, strands, beta pleated sheets, coils, and turns) almost exactly with the first third of the structure predicted by the MacDNAsis.

Finally, there are many types of collagen (thus far at least twelve different types of collagen have been reported in the human body). Therefore, some minor discrepancies between the Rasmol image (type VI collagen) and the MacDNAsis predicted secondary structure (type IV collagen) may be present because the human collagen DNA used in the MacDNAsis is not be the same type of collagen that appears in the Rasmol image.

Part VI: Multiple Sequence Alignment

This portion of the analysis consisted of comparing the collagen protein sequences (primary structure or order of amino acid residues) from each of the five organisms listed above (Gallus gallus, Caenorhabditis elegans, Drosophila melanogaster, Mus musculus, and Homo sapiens). A short sequence of the comparison appears below. It is clear that the collagen samples contain a limited amount of homology as the amino acid sequences had to be manipulated (spaces were added or removed) quite extensively in order to obtain the best alignment. Normally, this would suggest poor homology between the protein samples (i.e., the proteins most likely do not have a common protein origin or "ancestor"). In this case, however, the protein from Gallus gallus (chicken) was only a small piece or fragment of the entire protein. Because the entire protein was not available for analysis, the entire alignment procedure was altered. The poor alignment, therefore, might be explained in this manner. Furthemore, the collagen protein sequences from worm and fly were also quite short making an accurate (or exact) alignment very difficult.

The cDNA (and translated protein) sequences for each organism's collagen can be viewed by clicking next to the appropriate image below:

Gallus gallusDrosophila melanogasterMus musculusCaenorhabditis elegansHomo sapiens

Part VII: Phylogenetic Tree

In order to obtain a quantitative description of the alignment (above), a comparative, phylogenetic analysis was performed between the five types of collagen. The goal of this analysis was to determine the degree of amino acid conservation. The percentages listed in the tree tell the likelihood that the observed overlap between two sequences was due to a common protein origin versus chance.

This image supports the conclusion obtained above through the alignment study (Part VI: Multiple Sequence Alignment): very little homology exists between the different collagens (from different organsisms). This conclusion is not definative; because a fragment from the chicken was used and because the fly and the worm collagen samples were so short, the alignment/homology analysis is probably not perfect.

The cDNA (and translated protein) sequences for each organism's collagen can be viewed by clicking next to the appropriate image below:

Gallus gallusDrosophila melanogasterMus musculusCaenorhabditis elegansHomo sapiens