The overarching experimental goal for this project was to be able to install the designed molecular beacon into a protein crystal and ensure its functionality in the presence of target strands.  To accomplish this, several intermediate goals were in place:

  • Preparing protein crystals and verifying the functionality of their cysteine mutation

  • Using short ssDNA strands as “practice” to assess the following:

-Ability to load ssDNA into the protein crystal (not covalent installation)

-Ability to covalently install ssDNA within the crystal 

-Fluorescing and quenching capabilities of the fluorophore and quencher

  • Testing the molecular beacon in-solution to verify increase in fluorescence in the presence of target strand

  • Loading the beacon into the protein crystal

The experiments for this project can be compartmentalized into seven main parts:




The first step in the experimental process was to express the desired protein needed to make protein crystals.  The protein of interest was a Campylobacter jejuni protein with a cysteine mutation.  For expression, prepared competent Escherichia coli cultures from premade glycerol stocks were plated and selected for using antibacterial selection (Kanamycin) in a 37°C incubator.  Colonies were then swabbed and added to a starter culture containing Terrific Broth (tryptone, yeast extract, glycerol, phosphate buffer), and grown until the OD600 read over .5 absorbance.  Expression was induced with IPTG and cells were grown in 2 L shake flasks for 16 hours in a 25 °C shaker. Next, cells were lysed and purified using nickel column chromatography. Dialysis was then performed in a 4°C cold room to enrich protein concentration.  The protein was concentrated via centrifugation, and a Bradford Assay was used to calculate the final protein concentration.


The protein used for a majority of experiments was obtained from an expression and purification trial of CJ Greg (Campylobacter jejuni with a cysteine modification) that yielded eleven 30 µL aliquots of protein with a concentration of 12.58 mg/mL.  These aliquots were stored for the duration of the summer in a -30°C freezer.




CJ Greg protein was crystallized via sitting drop vapor diffusion.  In sitting drop vapor diffusion, crystallization buffers are loaded into a surrounding reservoir and a protein-buffer mixture is loaded into the center well.  The entire well is sealed with tape and diffusion occurs, leaving crystals in the center well.   

The ideal buffer reservoir for CJ Greg crystallization consists of a 400 µL mixture of the following: RO water, 1 M Bis Tris (pH 5.5 or pH 6), and 4 M Ammonium Sulfate (pH 5.5).

sitting drop plate photo.jpeg

Methods (continued):

The method for sitting drop vapor diffusion was applied upon 24 well plates were used to screen for a variety of crystallization conditions (see figure for crystallization conditions of each well).  To set up a plate, the reservoir buffers were added first. After adding all of the contents to the reservoir, the plate was mixed by gently swirling it around on the counter while keeping it flat.  Working one row at a time, 1 µL of protein (12.58 mg/mL) was pipetted into the center of each well. Using a different pipette tip each time, 1 µL of the reservoir was taken from each well and mixed with the protein in the center well by pipetting up and down three times.  The row was neatly sealed off with tape before starting on the next row. The plates were left overnight at room temperature and analyzed the next day to see if crystals had formed.


Over the course of the summer, about a dozen crystal plates were set up.  Observations were recorded for each well of each plate. In general, about 25% of the wells in each plate yielded decent sized, loopable, hexagonal crystals.  The other 75% of the wells typically displayed aggregate, air bubbles, or crystals too small for the purpose of this experiment.


CJ Greg crystal with aggregate present



On their own, CJ Greg protein crystals are sensitive and cannot withstand many solution conditions other than the buffer in which they were crystallized.  A technique known as crosslinking is used to create covalent bonds that increase the versatility of these crystals. To preserve the cysteines of the CJ Greg crystals, crosslinking is conducted using 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC).

Crosslinking was achieved by washing the crystals in three different solutions in a siliconized 9-well glass plate.  First, the crystals were looped from a well of the sitting drop plate into 500 µL of 4.2 M TMAO, pH 7.5 on the glass plate. The well was sealed and the crystals were kept in this solution for approximately 30 minutes.  The purpose of this step was to “wash” the crystals to remove any residual monomers. Next, the crystals were looped into a well containing a solution of 20mg of 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) in 500 µL of 4.2M TMAO and 50mM imidazole, pH of 7.5.  The crystals remained in this solution for two hours. This step is where the actual crosslinking reaction occurs. Finally, the crystals were looped into 500 µL of 50 mM borate buffer, pH 10. This solution works as a “quencher” to stop any further reaction. After 20 minutes, the crystals were transferred to a well of 4.2 M TMAO, pH 7.5 to be stored for future use.


The success of crosslinking was evident when the crystals did not dissolve when submerged in other solutions such as borate buffer or RO water.  The preservation of thiols in the crystal was assessed in the next experiment involving Ellman’s reagent.



Ellman’s Reagent (DTNB) is a chemical containing a disulfide bond that is frequently used for thiol verification.  When Ellman’s reagent reacts with a scaffold (i.e. the protein crystal) with a thiol side chain, an exchange reaction occurs that cleaves the disulfide bond of the Ellman’s reagent chemical (see figure for mechanism).  The cleaved molecule exhibits a bright yellow color on its own. Seeing this bright yellow color can help confirm that thiols were present in the scaffold and that the reaction did occur.












For a visual, quantitative verification of thiol presence in the CJ Greg crystals:


Several crystals that had been crosslinked with EDC were transferred into a well of a 9-well glass plate containing 100 µL of Ellman’s Reaction Buffer (0.1 M Sodium phosphate, 1 mM EDTA, pH 8.0).  An initial image was taken with the microscope camera. Then, 8.0 mg of Ellman’s was combined with 1.000 mL of Ellman’s Reaction Buffer in a 1.5 mL tube. Approximately 100 µL of Ellman’s Reagent solution was pipetted onto the crystals, and a video was recorded to analyze any color changes.  After, the crystals were looped into a new well containing 180 µL of 1x TE Buffer , and another initial image was captured. A video recording was started and 20 µL of 50 mM TCEP was pipetted into the well. The crystals quickly turned an intense gold color and then the color flowed out of the crystal into the rest of the solution.

Consult the Ellman’s reagent user guide for more information: https://www.thermofisher.com/document-connect/document-connect.html?url=https%3A%2F%2Fassets.thermofisher.com%2FTFS-Assets%2FLSG%2Fmanuals%2FMAN0011216_Ellmans_Reag_UG.pdf&title=VXNlciBHdWlkZTogIEVsbG1hbiZyc3F1bztzIFJlYWdlbnQ=


Several variations of the Ellman’s reagent experiments were conducted, including a comparison to a recreated standard cysteine curve.  Overall, the release of yellow color suggests that the protein crystals did contain thiols that were able to react appropriately with Ellman’s reagent.  The crystals were washed in TCEP and then in 1x TE buffer in order to reduce the thiols and prepare them for further experimentation.



Before jumping straight to loading the beacon into the crystal, loading experiments were conducted with various strands of single-stranded DNA (ssDNA) with a length of eight base pairs.  This was done to minimize cost and have some practice assessing the functionality of loading DNA into a CJ Greg crystal. Loading experiments were performed under a confocal microscope, with time lapses frequently recorded.


The first step was to ensure that ssDNA could be loaded into a crystal.  Non-thiolated ssDNA with a fluorophore (CG CTG GCG with 5’ FAM). It was hypothesized that because the ssDNA did not have an attached thiol, it would not covalently bond to the crystal and therefore could be easily washed out with ATP.  This experiment would serve as a “control” for the future covalent installment of single-stranded DNA with a thiol attached.

One trial of the control experiment was conducted as follows:


Two crystals were looped onto a chip with a well containing 2  µL of 1xTE Buffer. Images prior to ssDNA loading were taken with the confocal microscope.  25 µM ssDNA was loaded into the well (total of 10 µL of the solution in the well: 7.5 µL 1x TE Buffer, 2.5  µL 100 µM ssDNA). A 30 minute time-lapse was taken, which showed the crystal gradually gaining fluorescence over the course of 30 minutes.  The fluorescent crystals were then looped into a well of 100 µL of 1x TE Buffer and washed for two hours. After two hours, the crystals were looped back onto the chip in 5 µL of 1x TE Buffer and imaged again.  At the same lighting conditions, the crystal appeared to have maintained fluorescence after being washed. Then 5 µL of 40 mM ATP in 1xTE was loaded onto the chip. A thirty minute time-lapse was recorded. A distinct loss of fluorescence in the crystal was noticeable over the course of thirty minutes

Control loading non-thiolated ssDNA

Methods (continued):

The next step was to test for covalent installation.  The principle was that thiolated ssDNA could be covalently installed in a CJ Greg crystal via formation of disulfide bonds between the thiol side chain of the cysteine mutation in the protein crystal and the thiol side chain of the ssDNA.  To test this out, a crystal was incubated with 25 µM thiolated ssDNA (CG CTG GCG with 5’ FAM and 3’ thiol) in a similar fashion to the control experiment. A timelapse was recorded for two hours, showing a distinct increase in fluorescence throughout the crystal.  The crystal was washed twice in 100 µL of 1x TE for 30 minutes each. The crystal was transferred into the well of the chip with 3 µL of 1x TE for an initial image before adding 3 µL of 40 mM ATP in 1x TE Buffer, pH 7.5 (to make a total volume concentration of 20 mM).  A two hour time lapse was recorded. The crystal maintained fluorescence throughout the time-lapse.

Loading thiolated ssDNA

Methods (continued):

The other main test was to ensure that fluorescence from the ssDNA with a fluorophore could be quenched by a complementary strand with a quencher (CGC CAG CG with 3’ DAB).  The covalent installation experiment was repeated. After attempting to unload with ATP, the crystal was then washed twice in 100 µL of 1x TE for thirty minutes each. The crystal was then imaged under the confocal microscope in the presence of 25 µM of ssDNA with the quencher.  Over the span of two hours, the crystal went “dark.”

Unlike non-thiolated ssDNA, thiolated ssDNA did not unload in the presence of ATP


These experiments give just a general gist of the ssDNA loading experiments conducted.  Several variations were conducted to analyze things like deprotection of thiols, the effect of different buffers, and variation in incubation/washing time. 

Check out some of our additional loading experiments here!

Control unloading non-thiolated ssDNA



Before loading the beacon into the crystal, it was needed to make sure that the beacon did what it was designed to do: increase in fluorescence in the presence of the desired target strand.


In-solution testing was performed using a plate reader and a 384-well black plate (to prevent photobleaching of the fluorescent DNA).  First, tests were run to determine the ideal working concentration of beacon in the plate reader to prevent maxing out the fluorescence reader as the beacon has slight fluorescence on its own.  The ideal concentration was found to be 10 nM. Then a perfect target strand (CTG CAG TGA AGA TGG AAA AAT TGC AG) was added in varying concentrations (between 0.25 and 10 µM).


On average, the fluorescence of the well at least tripled after introducing the target strands (see table below).  Not many trials of in-solution beacon verification were conducted due to time constraints. However, it does seem that there is an apparent increase in fluorescence once the target strands are added (results that were sufficient enough to be able to start beacon loading experiments).  In the future, variations in target strands will be tested and more trials will be conducted to provide a thorough investigation of the beacon’s functionality.



After running through all of the other experiments, it was time to start working on loading the beacon.  This beacon was non-thiolated, so covalent installation was not expected. The goal was simply to make sure that the beacon could make its way into the crystal.

One CJ Greg crystal was washed twice in 100 µL of 1x TE Buffer for 30 minutes each.  The crystal was placed in the well of a chip with 25 µM DNA beacon (CT GCA ATT TTT CCA TCT TCA CTG CAG with 5’ FluorT and 3’ DAB).  A two-hour timelapse with images taken every 5 minutes at 5% intensity, 488 nm was recorded. Because the crystal had only slightly increased in fluorescence over the two hour loading period, the chip was wrapped in tin foil and the crystal was allowed to incubate with the beacon solution overnight.  The crystal was analyzed the next day.


The next day, the crystal was a bright yellow color visible to the eye.  Analysis was done by taking another image with the confocal and using ImageJ to compare intensity.  While more trials are needed (working with crystals can be tricky!), initial testing does suggest that the beacon is able to load into the crystal and remains in the crystal even after being washed (as indicated by fluorescence).

Loading molecular beacon into crystal (trial 1)

Loading molecular beacon into crystal (trial 2)


All data collected suggests that covalent installation of ssDNA via disulfide bonds within a protein crystal is possible.  A control experiment showed that non-thiolated ssDNA could easily load into the crystal, but also easily unload when ATP was introduced to the system.  A thiolated version of the same strand did not unload from the crystal in the presence of ATP.  Instead, the thiolated strand could be “turned off” by adding a quenching strand that quenched the fluorescence of the thiolated ssDNA.  Covalent installation was further proved in an additional experiment where a combination of ATP and TCEP was used to displace the crystal-ssDNA disulfide bond, as a combination of the two could work together to reduce the disulfide bonds (TCEP) and “push” the ssDNA out of the crystal (ATP).  This covalent installation experiment served as a simpler proof-of-concept to showcase (a) the ability of ssDNA to load into the crystal (suggesting that loading a ssDNA beacon is likely), (b) covalent installation of ssDNA via disulfide bonds (to show likelihood of being able to permanently install a thiolated molecular beacon), and (c) the fluorescing and quenching abilities incorporated into the design of the beacon.  In-solution tests with a plate reader helped to show that the beacon did function as expected, with a distinct increase in fluorescence occurring when an ideal target strand was added to the beacon. Lastly, preliminary experiments suggest that the molecular beacon is able to load into the protein crystal. In general, additional trials are needed so that eventually statistical analysis can be performed to truly analyze the significance of the results obtained.  Specifically, trials are needed that more carefully preserve the integrity of the crystal throughout the experiment (crystals were often badly damaged after being looped several times). Additionally, more trials of in-solution beacon verification must be conducted as there were errors with serial dilution and pipetting such small quantities. Overall, the experimental results obtained thus far give promising hopes for the future of this project.



While experimental results obtained to-date are exciting, further experimentation is needed.  Additional target strands for the beacon were ordered that contain slight variations from the ideal perfect target strand (i.e., off by 2 base pairs, off by 4 base pairs, etc.).  These target strands should be tested with the same in-solution verification procedure in order to assess the specificity of the beacon. Other in-solution tests to be done include testing the effect of temperature and time on fluorescence of the beacon and target strands.  Beacon loading is in its baby stages right now. The next steps would be to test whether the beacon unloads in the presence of ATP, and whether the fluorescence of the crystal increases when target strands are introduced to a crystal that has already been loaded with the molecular beacon.  Once the functionality of a non-thiolated beacon has been fully tested, a thiolated molecular beacon will be ordered to test covalent installation via disulfide bonds. Wayyyyyyyy further down the road, our group has our eye on testing the beacon with actual West Nile virus samples (retrieving target strands from a natural sample is a challenge in itself), and designing beacons for other diseases!



Hartje, L. F.; Bui, H. T.; Andales, D. A.; James, S. P.; Huber, T. R.; Snow, C. D. Characterizing the Cytocompatibility of Various Cross-Linking Chemistries for the Production of Biostable Large-Pore Protein Crystal Materials. ACS Biomaterials Science & Engineering 2018, 4(3), 826–831.


©2019 by CSU BIOMOD

Cell pellets after centrifugation