Same original image from Arroyo in Friday’s post, has been trimmed and cut and centered and I though it didn’t have a CRD (or at least not a complete trimeric CRD component and the LUT plot clearly shows that something is not right with the CRD on the left hand side of the plot.  Right hand peak first peak to the right of the Ntermini, the alleged glycosylation site which was de-glycosylated either has a greatly reduced peak height (compared to untreated dodecamers, or has flattened out and become part of peak 2 in the collagen like domain.  It doesn’t seem likely that the peak itself is “gone” but rather has lost peak height. I will try to remember to measure the peak heights (and distances between them) to see if the those distances are at all like the intermittent peaks found in type  … more plots required to determine which. CRD clearly present on the right hand side.

ASIDE:  it seems that the alleged glycosylation site in the collagen-like domain gets really “high” or “bright” when two from each hexamer (in a dodecamer) are close together (or joined) which changes the shape and nm distance of the center (Ntermini)…  sometimes this happens, sometimes not. A series of un-glycosylated images could be counted for such bright peaks where two are in close approximation vs those no treatment has occurred.

Arroyo et al published a plot (height) of a particular SP-D hexamer and I have used this image to see whether some processing will increase the amount of information available. The image below sums up the comparison. Firstly, using the micron marker they provided and using the node (crux) rich vector line they used for their “height” plot, I could not replicate how they arrived at the dimensions of the molecule (the plot says about 150 nm but my analysis comes up with just 116nm (using their line and their micron marker). My line and their micron marker comes up with a slightly higher nm assignment of the molecule (about 123.8nm). In addition the diameter of a circle (which works pretty well at finding the arm length of dodecamers) doesn’t come up with a good estimate of the size of the hexamer (here is comes in at about 133nm).

I exported their image as a whole, cut it into 1nm slices and centered the slices then exported as a tiff and plotted the LUT tables in ImageJ and arrived at a similar plot to theirs. I also made plots of just the Ntermini with the first collagen-like domain peaks on either side, and each arm independently. The plots are basically similar, but my plot centered gives a more informative analysis of potentially important peaks in the collagen-like domain on either side of the Ntermini.
Plots and comparisons are in the image. My plot (orange, bottom figure, with a red plot overlying where each arm of the hexamer is adjusted to half of the total width of the molecule.

Their plot is replicated from their publication, and a my red plot with cut, centered, plotted and arm length evened up is overlying their original plot. It is obvious that the Ntermini in the “center” of the molecule are somewhat flattened…  that peak should in fact be higher than the first peak in the collagen-like domains on either side.  Maybe another hexamer will show better results. HOWEVER, that said, there are still three obvious peaks in both arms of the collagen-like domains as well as something that might be a fourth peak.

haha… here is a thought about why the Ntermini peak is smaller than the collagen-like domain peak 1…..  in comparison to the dodecamer…. half as many Ntermini – that is just TWO overlapping so it is understandable that this peak does not reach as high.

I think the short answer is that the width of peaks do not change, but the height of peaks (brightness, not unexpectedly) changes, and since contrast and brightness are so variable, then a way to create an internal control is necessary.  This could probably be done by setting a mean (calculated using many plots) for the background and for the peaks of the Ntermini.  The latter could be set at 250, and the background would be best set around 50 (which is where I find that the images generally sort to.

Looking at the AFM images of this particular (very nice) image from Arroyo et al, I made a test of whether changing the contrast changed the number of plots visible in the LUT tables (the stretch from the center of the Nterminus to the CRD – that is one arm of the dodecamer. The actual dodecamer is pictures above, and the preparation of the single trimeric arm used with the different contrast applications appear below. The whole image (top) is as is from Arroyo et al, but just converted to rgb with no saturation and exported as a tif for ImageJ. Middle image has the original opened in photoshop and contrast was enhanced manually without losing light and dark areas (subjectively). Lowest pair of images is using the original image and autobalance in photoshop. Each of the dissected portions (one trimeric arm, all Ntermini are on the right hand side) was cut into 1nm slices, then those slices were centered horizontally (to compensate for the bend in the arms of SP-D.  Almost all multimers of SP-D and many single trimers do show an arc (which i calculated a long time ago thus need to be straightened and realigned to the concensus magnification (in nm). This is actually the same image as posted in the previous update.

The LUT plots for the contrast enhanced (2) and the original image (1) are compared.  In truth, while the peaks and valleys in the high contrast image are more separate than uncontrast enhanced, or autocontrast adjusted, they all show plots very close in alignment and shape.  (See a composite with the red plot (gray scale as is), green plot (autocontrast adjusted in photoshop) and blue, with very high contrast. Peak heights change considerably, width of peaks did not, neither did relative size of neighboring peaks. This is particularly important in the collagen-like domain where three peaks along that stretch of amino acids occur.  There is a problematic fourth peak, lateral to the Nterminus, where the coiled coil neck is, and while i have not measured it, the angle of deviation in the globular units should match the angle in RCSB models of the neck and CRD regions.