Category Archives: Bovine conglutinin

Bovine conglutinin: Did image processing improve information

Bovine conglutinin, surfactant proteins A and D

Bovine conglutinin: Did image processing improve information in this image of a multimer (Andersen et al)? In one sense i think it did increase the ease with which one can find the N termini (which in this case seem to be side by side) and also it increases the ability to see what seems like an acute folding of all four arms of the bovine conglutinin in to an arrow shape, with the point being the N termini junction. This was both an exercise in using Gwyddion to perform data processing (the 2D continuous wavelet transform (7 px gaussian).

Top left image (original image Figure 8b, Andersen et al, 1992) tantalum shadowing. Top middle, photoshop layered background contrast reduced, top right, image inverted and flattened and contrast enhanced. Bottom left, image imported into Gwyddion, bottom middle 2D continuous wavelet transform, bottom right, saved and highlights, shadows and midtones adjusted in CorelDRAWx5.

It is amusing to see peaks along the arms of the bovine conglutinin collagen-like-domain… three or four little luminance peaks, reminds me of SP-D. Upon measuring them in imageJ, it would be difficult to distinguish these peaks from background. I will figure out a way.

The N termini junction has the distinct look of dodecamers bent at an acute angle and nestled together side by side (nb the bottom group of four N termini junctions and approximately 16 independent arms radiating.

Andersen et al report that the arm length (in their figure 10 and legend show an arm length (without the CRD as 40nm, and the CRD as contributing 4.5 to the length and add to that another 3.25 nm (half the N termini group) comes out to 47.75nm, which is like their 48nm reported as half he 96nm CRD to CRD total that they report.

Individual arms were measured with a segmented line but in this sample the length of the arms did not differ from what they measured as the “whole multimer” (see below). That said, my diameter measures 101.19 touching many of the outer borders of the CRD, so there is a 4nm gain thee somewhere. A measure of a circle that is not best way to measure this shows a 10nm gap (which might be less or more depending upon the multimer). I wish I could say that there is a ring concentric to the N termini lumps that looks like an elevation for a glycosylation site – but i dont see anything that resembles it here.

Pretty amazing result from 2D CWT using Gwyddion

Bovine conglutinin, surfactant proteins A and D

In trying to find the actual magnification for some early publications using negative staining and rotary shadowing to delineate the structure of surfactant protein D and also a closely-resembling member of the C type lectin group I ran into some images of bovine conglutinin (Strang et al, in the mid 1980s) that had images which were highly pixelated probably from the type of printing used in that era.  I did a screen print of one of the negatively stained images (screen print of original here) and opened it in photoshop to play with different filters to reduce the noise.  Gaussian blur and smart blur in photoshop did provide some improvement in the image but I decided to see what Gwyddion could do with the image in terms of processing out the noise.

Much to my surprise, a marvelous image emerged after applying “integral transform> 2D CWT with 10px gaussian blur.  The imported image in gwyddion, and the transform are shown here.  Upper right, screen print (on the computer keyboard) of a screen-printed (the technique for printing – likely and less expensive than offset printing) from the original article. (the pixels are not from the screen print of the image, as i did it at full screen capacity, they are from the screen used in the printing process).  The image is poor quality at best. Image to the right of that is opened in Gwyddion as RGB, and image below on the left is a 2Dcontinuous wavelet transform of the image above, with gaussian blur set at 10px.  The image on the bottom right is smart blur performed in photoshop.  Each of these has had ImageJ used to mark a line through the molecule (by chance this bovine conglutinin molecule had just one arm that could be marked with a segmented line), and beside it is the plot generated in ImageJ.  There is a tremendous difference in the information obtained from the 2D CWT processed image.  Plots show the Ntermini junction, the CRD at either end and several plots along the collagen-like domain, so similar, yet so different from surfactant protein D.  A great comparison.  Nothing short of amazing.