Category Archives: Ultimate order, the cell

The beauty and order of life is astounding.

One image, SP-D sizes by diameter

surfactant proteins A and D, Ultimate order, the cell

Just trying to get a common size for SP-D multimers and dodecamers. These are data using molecule diameter, a single AFM image from Hartshorn et al.
You can see that these images are just slightly smaller than the means found for Arroyo et al. Whether prep, or differences in the actual SP-D are responsible is a question.

Some species just cannot be determined others are covered up, and so it is an estimate, i understand that. For trimers smallest diameter was used, for hexamers as well, for all other multimers at least half the CRD had to be on the rim of the circle to determine the diameter.

SP-D dimensions

surfactant proteins A and D, Ultimate order, the cell

7 dodecamers, each arm measured with a vector line each node measured separately. Images are AFM from Arroyo et al., randomly selected.

n=14 (two arms each); 144.49nm +/- 7.02nm

Diameter circle;  n=7, 139.79 +/- 3.14

USING an online t-test calculator the two methods are not significantly different at the 0.05 level, but close, with the diameter always underestimating arm length using the three CRD touching methods.  p-value is .062519

3 SP-D dodecamers: vector length of arms

surfactant proteins A and D, Ultimate order, the cell

There is such huge variation in the size of these molecules and the size of reporting that it becomes difficult to measure where the height-peaks (LUT) and bright areas are. I have used an un-glycosylated SP-D molecule image, and 3 molecules at pH 7.4 from oft cited paper by Arroyo et al to see how much variation there is.  Vector lines, then broken and measured in parallel dimension, summed and using the bar micron marker in the publication the nm diameter and arm lengths were determined.

Mean of 12 measures (one for each arm and one for each diameter) to arrive at: n=8 arms of 4 dodecamers, plus 4 diameters: 143.8nm+/- 6.97nm. I am more than sure that adding more will change this value. But for now, anyone who wants to examine the process… here is the image.

 

Un-glycosylated arm of dodecamer: LUT plot

surfactant proteins A and D, Ultimate order, the cell

This is a pretty nice image from Arroyo et al, and I wanted to compare the width of the LUT peaks and number of peaks to see whether the bright spot on a single arm of this dodecamer was in the right position of they proposed glycosylation site. It is not conclusive because on the left arm… clearly peaks are disrupted, on the right portion of the arm (and right portion of the plot) there are still two large and one very small peak between the Ntermini peak and the right hand CRD peak. The Ntermini peak looks a wider than usual and measures (using their bar marker) at about 32nm wide….  wider than other measurements.

Original image of theirs with a line with 10 nodes to determine actual length, and a diameter (which came out slightly smaller than the line method.  Image was cut into 153 vertical slices, ungrouped, centered horizontally, converted to greyscale, exported as RGB tif and plotted using ImageJ.  Plot (at 153nm width) is above the images.

SP-D CRD to CRD measurements

Ultimate order, the cell

Arroyo et al is perhaps the gold standard for measuring SP-D.  I keep coming back to this publication as the most apt to be correct in terms of measurements of surfactant protein D and while my intent was to go back and measure all the other images against each publications’ bar marker I am pretty convinced that the measurements from Arroyo et al are probably the best to use.

Comparing my analysis of a single dodecamer that appears in that publication and their analysis is interesting. I have used their micron marker as a scale bar for my analyses.

1) I have used a circle to estimate the size of the dodecamer (three out of four of the CRD are touching the circumference.  Arroyo calls peak 2, which is likely the peak that i call collagen-like domain peak 1 (because it is closest to the Nterminus) is variable in width but almost always less than the Ntermini peak of a dodecamer (as seen in Arroyo’s LUT (or height) plot and in almost all of the LUT plots that I have made as well. This peak is the brightest in shadowed images using TEM.

2) From peak 1 of the collagen-like domain the measurements (using 148 as the diameter of this molecule) is about 40nm and mean valley to valley of the two individual collagen 1 peaks in this particular molecule is about 20nm. The width of the peaks in Arroyo’s graph for CRD is about 28nm (using my 148nm diameter based on their 30nm bar marker).

3) Since this molecule is made up of identical monomers (12 of them) and since the Ntermini peak is NOT in the center of their plot (nor is it in the center of my LUT plots), I think that it is perfectly reasonable to “force” the half plots into equity. It is easy to see that the collagen 1 peak in the collagen like domain is wider on the side where the arm length is greater so i think that normalizing each side to “half” the distance would be a good (and fair) way to minimize the variability in size (which is most likely due to random falling onto the substrate during preparation (that is simplistic, yes).

4) the segmented line measurements are very close to the diameter measurements (case here where diameter of 148nm and segmented line measurements are 148.5 for the single arm measured is quite close. It is the same arm measured by Arroyo et al, who came up with a measurement of 160nm  which might be a little long.

5) the third image (set of images) on the top row (left to right) show the same arm of the dodecamer cropped out, then sliced into 100nm (approx) segments, and those segments are aligned horizontally (technique described before on this blog) with the resulting image exported as tif for LUT tables using ImageJ. (that straightened arm is shown as the right-most image).

6) of all the dodecamers I have plotted, this happens to be what I called #43, labeled as such, and the two sets of arms are labeled a,b  and c,d  so that plots of each can be compared. I derived LUT tables from a) a single line (as per Arroyo) and as a rectangle in ImageJ.  There is some smoothing obtained by using a rectangle, but both methods produce pretty much the same results, just like the diameter vs the segmented line (measured twice, once in the vertical image, and once in the original analysis (red segmented lines at the base of the plot) to measure arm length are very close. The former is way easier however.

7) background was subtracted from the peak height (brightness or lightness, formerly lumanance) at the lowest point in the LUT tables within the molecule.

8) sq nm were determined for the  CRD. and the plot height and area as well.  All in all, i think there are some additional facts (namely two additional regularly occurring peaks in the collagen-like domains) and relative peak heights and areas as well as a simplified method for calculating the arm length of multimers of any kind (using both shadowed and AFM images, particularly relevant to those which form multimers like SP-A, SP-D, conglutinin C1Q etc.

9) comparing shadowed and AFM methods, and information derived from each (which is better? or more informative? only careful comparisons will tell. IT seems to me however that using the AFM for comfirmation, that the shadowed image show almost as much about the SP-D molecule as the AFM does. Maybe it is a little harder to ferret out.

I will say again, Arroyo did some nice morphometry…. for that i am appreciative. I dont think the resolution of their LUT plots (nm height but likely equivalent to the grayscale or LUT plots or relative sq nm in some measurements I have made) the are very conservative in choosing only one peak in the collagen-like domain…. I think consensus would say there are more.

Molecular models and micrographs: do they really equate

Ultimate order, the cell

Atomic force microscopy images of fibrinogen, fibrin oligomers, and protofibrils and reconstruction of a protofibril model. (a-i). Images by high-resolution atomic force microscopy (used from a second hand image posted from original  of Drs. Anna D. Protopopova, Nikolay Barinov, Dmitry Klinov) from a post on Research Gate.

Sometimes putting the molecular model with the AFM image is a little difficult. Using the measure of 45 nm for the molecular model at the same magnification as the micrograph is marked (bar=30nm) something just looks out of proportion, however, the length of what would show up as between-peak brightness probably does equate to the model, but what is puzzling is the thickness of the two strands…  model vs micrograph where just a tiny portion of the micrograph (about the middle of the original (shown below) was cropped and enlarged to scale of the molecular model diagrams.

 

Rotary shadowed background grains in this image are about 4.5nm

surfactant proteins A and D, Ultimate order, the cell

Confirming the background grain size in a particular image (so I can do LUT plots on a few SP-D dodecamers and type 1 collagen molecules for comparisoin with AFM images) is pretty uniform… varying about 1.5 nm  (n=79 x=4.63+/-1.45 nm.  This means that it might be possible to establish areas of greater luminance (brightness, lightness) on a 1-255 gray scale that correspond with different AA sequences in either (both) these molecules.

Individual lines and their nm thick peak-valleys, and a mean of all 4 lines together. Same plots as last week, same resource, just with the measurements of peak width (from lowest point in valley one side to the other.

and the two LUT plots made with rectangles, and two more plots made with diagonals. It seems that the rectangle might be the least good way to measure the nm of the shadowed background.  But all in all, it seems that a line, anyway you draw it will work just fine.

The next step is to see the width and variation in the type 1 collagen collagen and SP-D from the same micrograph.

Rotary shadowed SP-D and collagen – is there a difference

surfactant proteins A and D, Ultimate order, the cell

The obvious answer is yes, but it is not too clear still whether the plots of rotary shadowed type 1 collagen and the collagen-like domain of SP-D can be used to determine peaks and valleys along the relative paths of each, respectively.

SP-D has some unique characteristics which show up pretty well with AFM but there are apparently not a lot of images for “share” to get sufficient statistical information on how the width and height of the peaks along the SP-D dodecamers line up with similar sequences in other proteins.

There are quite a few rotary shadowed images of SP-D (particularly dodecamers) that have been published and there are some obvious, and not so obvious areas of similar organization that is seen with AFM). So I am trying to figure out whether these data can be used along with the AFM measurements to come up with some relevant anatomical points along the SP-D molecule, especially the collagen-like domain. Some years back there was an article that showed SP-D (dodecamers mostly) shadowed along with type 1 collagen molecules.  I have used these images initially to see whether the shadowing of the background is too coarse to allow accurate measures of the the smaller (less than 10nm) bright (light – high) areas to correlate with those seen with AFM.  Below is the original image from which a hexamer (from a dodecamer) has been cropped out, and a collagen molecule has also been cropped out along with their adjacent “background” shadowing. When needed the molecules have been straightened and aligned to the 135nm diameter that the dodecamers seems most likely to be…and LUT plots determined. (TOP image)


The image below is the LUT plot summary (blogged yesterday without the images and nm measures).  ONLY the SP-D dodecamer has peaks around or above 200 (brightness – presumably relating to peak height), all other peaks are less (green plot is end to end (2) trimers of SP-D) and red fill is in areas above 200 – in a grayscale of 0-255). They are, not coincidentally, the positions of the N termini joining at the center of the dodecamer and the CRD at either ends…. a perfectly obvious fact. Collagen itself did not have any peak heights (brightnesses) which were above 200.  Similarly, backgrounds adjacent to the SP-D and the collagen do not show peaks above 200.  The collagen-like domain peaks in AFM are somewhat similar to the collagen LUT tables below, and only a few peaks seen in AFM of the collagen-like domain reach the brightness of the N termini or the CRD. In a sense, this is supportive of the fact that shadowed images can provide similar results to AFM images of SP-D.  What has to be determined is what impact specific AA sequences have in the peak/valley plots of the collagen-like domain (as similar peaks to show up with AFM) that are different than the shadowing of type 1 collagen used by  Crouch et al.

Can LUT tables provide meaningful data from shadowed images?

surfactant proteins A and D, Ultimate order, the cell

LUT tables are collected from AFM images, and provide detail that is interesting and likely informative. I wondered the same about shadowed TEMs. Here is clue as to whether any information about SP-D dodecamers (or other multimers or trimers or hexamers) can be derived from shadowed images. It is pretty clear that a background LUT plot of the same micrograph in the same area (blue lines) does have a repeating pattern of the deposited shadowing material.  The Ntermini that meet in the middle of the dodecamer and the CRD (of which there are 4) do show a different dimension than the regular repeating 8-10nm background.  How well the shadowed images predict the position of the presumed peaks between the Ntermini and the CRD could probably be worked out.

These particular images are derived from a micrograph published by E. Crouch et al, so further LUT analyses might show something about the full structure of SP-D which is not yet available on RCSB. (i am not sure why since it is pretty clear that the collagen-like domain presents at a slighly straight to curved portion of the dodecamer.   What also might happen is that the Ntermini might bind side to side, either in addition to or instead of end to end.