There is quite a bit of variation in the heights of the LUT peaks plotted in SP-D dodecamers (same is true visually with other oligomers, but i have not measured them yet). In the summary sheet I created with 80+ images of SP-D visualized by AFM in publications just by Arroyo et al, it seems pretty clear that not all the molecules in all the trimers are glycosylated.

It is easy to see some distinct brightness on either side of the Ntermini junction by eye, after processing, and easily seen on LUT plots (in fact to miss it is not really possible), but it is not consistent enough in height but is consistent enough in place) to allow some speculation that each of the trimers might have a different amounts of glycosylation.

This is also apparent in the single image from these manuscripts that shows a de-glycosylated dodecamer and there is inconsistent “brightness” in those first peaks on either side of the Ntermini peak.

Below is the same multi-image paste up or layout with about as many dodecamers (there is one pentadecamer and one nonamer excluded from the calculation) as i could easily dissect from the figures. Each has my quick assessment of brightness (peak height) at the site that they call the glycosylation site.  If there was no peak, then 0, if there was an “iffy peak”, then 0.5  if clearly all four trimers in a dodecamer had a bright peak, that was a 4.

Mean 2.6 arms glycosylated,  median 2 peaks, mode 2 peaks n=83.  I have not yet found a specific statement that says one or two molecules in a trimer can be glycosylated, or even just one.  Larger image is two or three posts back.

LUT plots of SP-D hexamer-arms from SP-D dodecamers. These, as mentioned many times, are from publications by Arroyo et al, and my purpose has not changed. Look for consistent peaks along the collagen like domain, and to sort out which processing algorithms are useful for analyzing AFM and TEM images.  At this point I think that image processing is cosmetic. It does make plots more pleasing but really doesn’t change the peaks much. (I am not talking about the fantastic and artistic filters that something like CorelPhotopaint and Photoshop2021 can offer….  but the simple filters that remove scars, noise, pixelation but don’t really change data.

Plots and image below: two relatively matching dodecamers of SP-D found hunting through the sometimes crowded and pixelated image that are found in publications. the left hand image is purported to have a glycosylation peak (next to the brightest peak in the center of each hexamer (the N termini). I am not convinced that all arms of a dodecamer are glycosylated to the same extent, surely it is easy to see that peak height is not the same on the four trimeric arms.  The image on the right is a de-glycosylated SP-D that just happens to conform to the same X style as the image on the right.  THis is not the most typical form since angles between the trimers are most often acute and obtuse, not right angles as seen here…. but this is a good spread and can show the peaks along the arms nicely. Image on the righ has diminished peaks on the horizontal plot, but does have a peak on the vertical plot lower half (see white arrows where peaks should not occur in completely de-glycosylated SP-D but do seem to show some peak).  All four arms of the SP-D have differing size peaks (red and green arrows).

The filter used here is a “highpass” (a filter option present in the four programs that i have most often used to look at image processing as an aid to molecular modeling (CorelDRAW x5 with attending CorelDRAW photopaint, CorelDRAW19 and attending CorelDRAW Photopaint as well as Photoshop 6, and Photoshop 2021. These programs are touted to be about the best around.  I have used Gwyddion as well, but find the ease with which the former can be used is a time saver.  ImageJ is an easy program (and i have not found an equivalent in photoshop or CorelDRAW or Gwyddion) for making LUT plots.

Red plots (most horizontal arms, green plots, more vertical arms — these formations are by chance). At least this pair would suggest that level of glycosylation is graduated, and maybe not present through out a dodecamer the same on each arm.

Just in general…. looks like a minimum (including the glycosylation peak) of four bright areas occur along the collagen like domain.  This is so consistent that I am hopeful to get statistics to verify this.

All images from two papers by Arroyo et al (2018 and 2020). These are among the images that have planted the idea in my mind’s eye that there are very discrete and nicely spaced and sized peaks between the N termini joining and the CRD.  There is often a glycosylation peak (described by Arroyo) but it is not consistently found on all arms, that provides a huge burden of proof of the number of peaks along the collagen like domain (which is where I first became interested in this domain….as it has been slow in being modeled on the protein database websites).  I  can use this summary set to approximate what percentage of the arms (trimers) that have a peak sufficiently high (on a scale either of 0-255 or 1-100) to be called glycosylated. It might help determine how many peaks are along the collagen like segment if a quantifiable height is found (or not found) for those glycosylation sites.

The following set of 80+ images all come from different figures. Please note that sometimes error bars in the figures are off, sometimes the measurement of error bars by me is off and this all translates into a need for a solid consensus measurement of dodecamer arm (hexamer) width.

I have done my best to size all images using available bar markers.  And also helps figure out whether the N termini are joined end to end or side to side (the latter being my personal preference based on the width of the center N termini composite peaks. (again, I acknowledge Arroyo et al for these images with which I have worked)