Here is a model of SP-A after weeks of trying to figure out which species (of the 7 looked at) to use, and where they are different and why the two programs I have used fail to predict the structure of the collagen segment… so i confess i am not a protein scientist, but seeing the really poor diagrams presented for this molecule (rather this octadecamer) by some very hefty scientists right here in this same University (no names mentioned here, though it would be my misguided pleasure to do so) I was determined to craft something that really might look like the real thing.  Thus… the diagram below. The whole top part (mostly the carbohydrate recognition domain and coiled coil neck) was easily found as a pdb file and rotatable as a ribbon diagram  too, and worked using the sequences for mouse, ferret, guinea pig and dog…. they are very similar, with a few pointy and craggy places along the top (presumably those places for recognition of organisms in the lung) and other immune functions.  Coiled coil regions modeled particularly nicely, almost like rhythmic segments , and aptly called a neck…three or so bumps. These were also pretty similar among species. No problem there. But where the neck region meets the collagen-like stretch it just didn’t model the way (literally ALL) the diagrams represented it– except when i used it in the protein structure databases without the neck and CRD domains and part of the N terminal.  Even then, most often that stretch of collagen-like sequence, even with the N terminal region added or subtracted, did not align with the suggested diagrams.  Where, and how kinky the kink is IN THIS DIAGRAM is “speculation, and information from other diagrams” used here rather than the protein modeling (which didn’t model a “kink”).

So I pieced together a separate prediction for the collagen-like and N terminal sequences, and then “cut and pasted” the single protein into trimers and octadecamers, giving them the perspective that I was hoping to get by modeling the molecule as a “whole” and using the 3D options to rotate.  Take this model as something synthesized from available models protein modeling software and available published diagrams.

The octadecamer is likely about 25-35nm across the top, and a similar height (which apparently changes with the spread of the CRD depending on open or closed conformation.  These measurements are according to references found in the literature.

Top diagram, the perspective provided for the three angles that the trimers might be found in if they are as orderly (as indicated by the transmission electron micrographs i have posted. (Remember this is a partial-reconstruction and a partial diagram). Left view, kink facing forward, middle view, kink slightly rotated to side, right view is SP-A CRD top, neck slightly wider than the collagen and N terminal segments below.

And here is the whole octadecamer,  ordered with the three braided collagen-like domains aligned at the N terminal (as the literature suggests) then mirrored (from three above images) to make a total of 18 individual molecules of my generic SP-A).

No laugh here,  but I bet this little amino acid has a lot to do with some of the function of the carbohydrate recognition domain.  Here is a Pymol view of guinea pig SP-A from residue 51 to 347, rotated to show this AA.  Someone out there knows exactly which on it is… i will try to find out.  In addition, I will post the ferret sequence from the same location (also residue 51 to end) to see if there is a similarity. RED arrow points to the spike.