MIT Class Site
Laura Maria Gonzalez
April 6, 2021
It was a beautiful weekend in Cambridge and the perfect time to take out the foldscope and go on a Microscopy adventure! This week in HTG(A)A we
explored the world of microscopy. To begin we assembled a foldscope. See the video of above for how mine came together. The foldscope is an ultra-
low cost, DIY brightfield microscope intended for hands-on science education with the hopes of empowering a worldwide community of amateur
microscopists. I had a ton of fun.
To collect the samples I went for a walk by the Charles river with my partner who documented the process. I collected samples from the river, the
shore, flowers and seeds from blooming trees, as well as moss on tree bark.
What started as empty bags quickly filled up! With bags and tubes full of samples and a few extra
ones given to us at home, it was time to look through the foldscope. I quickly learned that thick
samples which blocked light weren't as interesting as transparent thin samples. In the end I chose
mostly petal and water samples from my haul.
The images above are of the Spirulina Major sample provided by the TAs. I was fascinated by my lab pet since it was handed down to me and I was
amazed by the structures I saw through the foldscope. As I zoomed in I could start to see the spiral structure

With a field of view of .51mm (510μm) and an indices of refraction for borosilicate glass of n=1.517 I estimated that the approximate resolution of my
microscope is 1.52 μm. This means that the smallest distance between two points on a specimen that can still be captured is 1.52 μm. But note that
because of the spherical ball lens not all regions can be simultaneously in focus at this resolution.

One difference between a typical microscope and the Foldscope is that the Foldscope anchors the sample at a fixed location, while the optics and
illumination stages are moved in sync.

One difference in the optical design is that a Foldscope uses a spherical ball lens which is advantageous for high-volume manufacturing vs using an
objective lens or several lenses as in compound lens systems. A compound lens is typically use to correct chromatic and spherical aberrations. I
believe I came across the spherical aberration issue since I was never able to have an image fully in focus without zooming in to the center with the
phone camera.
100 μm
100 μm
I took a pass at trying to look at some wet samples through the foldscope. To do so I used a slide and squished some of the collected river water
with a cover glass above. After combining through 6 different slides over a couple hours from three different locations I spotted movement! Could this
be the elusive tardigrade? See video below for more evidence:

Spirulina Major (Blue-Green Algae)
140x and 280x with 2x zoom on phone

A Charles River Tardigrade?
I also looked at the various other samples I collected/ were provided. The most interesting images had a transparent quality to them such as flower
petals or stems. From top left to bottom right the images are of: Fern Rhizome, Hibiscus Pollen, Female Pine Cone Cross Section, Blue Flower Petal,
0.5mm Guide, and Maple Tree Flower Petal. All images were captured at 140x magnification
The second part of this week involved designing a FISH or spatial transcriptomics analysis assay. To start off FISH stands for Fluorescent In Situ
Hybridization and it is a molecular cytogenetic technique that uses fluorescent probes that bind to only those parts of the chromosome with a high
degree of sequence complementarity. FISH can be used to detect and localize the presence or absence of specific sequences. Probes have a
fluorescent label that are complementary to the sequence that is targeted and when the strands denature they hybridize.

To do so we needed to choose a problem. As a beginner I used the bio-techne Guide for RNAscope in order to guide me through to make sure I
understood the concepts along the way. The problem I picked is using imaging to visualize a homogeneous expression of MICA and MICB in human
ovarian cancer tissue. MICA/B are polymorphic proteins that are induced upon stress, damage, or transformation of cells. They are expressed in
most tumor types. The gene I chose is MICB and the assay that could be used is the RNAscope homogeneous expression assay. This assay is
chromogenic and would only require a single plex. The assay enables highly specific and sensitive detection of target RNA that can be visualized as
a dot and quantified based on ACD scoring criteria.
Next we needed to design the homology domains (target sequence domains) for 5 probes. To start we used a Colab Notebook and picked 5 random
sequences each 25 bases long. Evaluating the melting temperature and secondary structure of each:
The following random probes are not all good. They very in primer temperature range with a few below the 60-65C ideal range. All probes flagged a
hairpin, but several were at a much lower tm. However, probe 5 is certainly not an ideal probe because the hairpin tm is greater/close to the primer
tm and could likely result in binding problems.

Next using the "PROBE-O-MATIC" I attempted to pick less random probes. The script finds ideal probes within the given parameters. To start of I went
with the default settings of target length 25, min_tm = 60, max_tm = 70 and hairpin max tm = 45. This yielded over many results so I could get a little
more picky. I reduced the temperature to 65 and still received many possible results. Then I reduced the hairpin max. I tried a hairpin tm of 0 and got
no results. At first I thought this meant I always had a hairpin but looking at the code I realized I needed to set it to 1 and not 0 and perfect! I found
sequences, over a hundred of them, that did not have any hairpins. Next I chose 5 intentional sequences from different starting base regions.
Finally I used BLAST to see if there is a good match to MICB without a high amount of off-target hits. Testing the first probe yielded promising
results. The high alignment scores were associated with MICB.
For this final part we needed to analyze the smFISH images generated by Erkin in the lab using Fiji. To start we imported one of the Z stack images
which was saved as an nd2 file. The image is of a Fluorescence Multiplex V2 assay on an FFPE pellet of mouse 3T3 cells. The gene panel is a three
plex of housekeeping genes Polr2A, PPIB, and BC. They are all homogeneously expressed in the cells.
Next, to count the number of molecules we used the threshold function combined with watershed process and the analyze particles command. By
setting the particle size we were able to get counts overlaid on to the image which we could compare our analog counts. For these image counts I
used the Otsu threshold method, manually adjusting till it looked about right. And then set the particle size to 10-300 microns for the nuclei
and .01-1.00 microns for the subsequent gene channels to count the RNA.
100 μm
100 μm
100 μm
100 μm
100 μm
100 μm
100 μm
RNA Sequence for MICB from UCSC Genome Database
Semi-Quantitative Scoring, Bio-Techne
BLAST results for Probe 1 for MICB
Channel 1 - Nuclear Stain, Visual Count: 150
Channel 1 - Nuclear Stain, Visual Count: 150
Channel 3 Polr2A, Visual Count: 960
Channel 3 Polr2A, Visual Count: 960
Channel 2 - UBC, Visual Count: 15,360
Channel 2 - UBC, Visual Count: 15,360
Channel 4 - PPIB, Visual Count: 10,240
Channel 4 - PPIB, Visual Count: 10,240
tm: 63C
hairpin: Yes, but tm at 48C so maybe alright.
tm: 61C
hairpin: No
tm: 59C
hairpin: Yes, tm at 54C which is quite close to the primer tm. May not be ideal.
tm: 63C
hairpin: No
tm: 61C
hairpin: Yes, but tm low at 37C
tm: 61C
hairpin: No
tm: 52C
hairpin: Yes, but tm low at 26C
tm: 61C
hairpin: No
tm: 53C
hairpin: Yes, and tm is 54C which is greater than primer tm making this an unfit sequence
tm: 63C
hairpin: No
Visual Estimate: 150
Visual Estimate: 15,360
Visual Estimate: 960
Visual Estimate: 10,240