<!DOCTYPE html> Template



Our team tried to reduce the damage done by frost on apricot trees. In order to do this, we designed three different approaches.

The first is to protect the sensitive tissues by using Antifreeze Proteins (AFPs). The second and third approaches aim at targeting the pathogen P. syringae syringae, a species of bacteria that produces Ice Nucleating Proteins (INP) that cause the worsening of frost damage. To prevent the production of INPs, we pursued two strategies: killing the pathogen using specific protein complexes, tailocins, or inactivating its INP gene using a CRISPR/Cas9 system delivered by a phagemid.

To find out if our approaches would work as expected, we performed various assays to measure the potential efficacy of our different treatments to protect plants from frost damages.

Antifreeze protein

We first focused on developing ways to quantify the effect of antifreeze proteins on ice formation. To do so, we created a specific assay: "I Said Freeze'' or ISF. AFPs can be characterised in multiple ways, the most common one is to measure the Thermal Hysteresis (TH) of an AFP solution with a given concentration.

Thermal hysteresis is the difference of the freezing and melting temperature of existing ice crystals in solution. In other words, TH represents a difference between freezing and melting point of a solution. Below the freezing point temperature, the ice crystal grows, above the melting point the ice crystal shrinks.
For example, if we monitor a melting temperature at 0 ºC and a freezing temperature at -2 ºC, the TH is 2 °C. Between these temperatures, ice crystals do not grow or shrink (Fig. 1).

Figure 1 | Schematic view of the thermal hysteresis process.
In this example, ice water, when heated, will melt at 0ºC. When water is cooled down it will freeze at -2ºC. The freezing point and the melting point are different, and this difference is the TH and it is about 2ºC.

Deionised Water has a TH of 0 ºC. When supplemented with AFPs the TH of this solution will increase depending on the activity and the concentration of these proteins. We have used this characteristic to check if our solution have tailocins and if they work as expected.

To estimate the TH, we first needed to measure melting and freezing temperatures. We defined the melting and freezing temperatures as the temperatures measured when there is only one ice crystal left and when the single ice crystal is growing steadily, respectively. To see these ice crystals, we needed to be able to tightly control the temperature of liquid samples kept under a microscope. For this purpose we created: FROZONE, a highly precise cooling device that controls the temperature of a copper plate, allowing us to measure the thermal hysteresis of our different sample solutions.

Figure 2 | Image of FROZONE under a microscope being loaded with a pipette.

Put simply, we created a machine that controls the temperature of a copper plate in order to measure the thermal hysteresis of our different sample solutions.

For more information about FROZONE and the hardware check the Hardware page.

For more information about the experimental details check the thermal hysteresis measurement protocol.

Thermal Hysteresis - AFP

Using our ISF assay, we collected data on our AFPs respective thermal hysteresis values at two concentrations (Fig. 3).

Figure 3 | Thermal hysteresis of RiAFP and FfIBP increase with their concentration.
Connected scatter plot graph showing the mean of biological triplicate +/- standard deviation. Thermal hysteresis is measured with two different AFPs, RiAFP and FfIBP, at different concentrations (µM).

The obtained data show that the TH of our proteins increases as their concentrations in solution rise. The TH, or difference between freezing and melting points, is used to characterise our antifreeze proteins and depends on their concentrations. The fact that we can visualise a TH shows that our AFPs are functional. Without any AFPs this graph shows a TH of 0.5. TH of the Buffer (protein concentration = 0µM) is at 0.5 ºC, whereas we excepcted a TH of 0 ºC. This difference is most likely due to a lack of human error when controlling FROZONE. This graph also illustrates that RiAFP is more effective than FfIBP at a concentration of 50µM.

Thanks to the model we created, we can project the TH values we could obtain with higher AFP concentrations and compare our values to the ones found in literature.

For more information about the model we used click here.

Frost Damage Treatment

To measure how our treatments will affect/reduce frost damage on plants, we developed a “Frost Damage Treatment” assay or FDT (Fig. 1). We immersed the plant Arabidopsis thaliana (Wild-type) in our frost-protecting solution (i.e. in a buffer containing antifreeze proteins) and placed them in a climatic chamber at -5ºC overnight. We then stained the leaves with Trypan blue [1]. This dye is routinely used to stain dead cells with compromised membranes. In our case, we wished to stain the cells damaged by ice crystals.

Figure 4 | Schematic of the workflow employed for the FDT assay.

After staining, we cut all the leaves of our plants and placed them on microscope slides in order to observe them under a microscope. We took pictures of all the leaves. We then analyse the obtained images with our algorithm, VISION, in order to determine the percentage of damaged area on every leaf.

Measurements - Frost damages

Thanks to the FDT assay, we were able to quantify the damages caused by frost on plants submerged in different solutions and therefore assess the potential protective effect of our AFPs.

Figure 5 | Our AFP solutions protect plants from frost damage.
The graph shows mean +/- standard deviation. Percentage of Damage on Arabidopsis thaliana leaves measured after overnight incubation in a climatic chamber at -5ºC. Plants were immersed in phosphate-buffered saline (PBS) liquid solution supplemented with two different AFPs: FfIBP [200 µM] (n=13) and RiAFP [50 µM] (n=17) and a mix of them (RiAFP + FfIBP) (n=15) at a concentration of 25 µM and 100 µM, respectively. These solutions were compared with the control in which the plants were immersed in the PBS buffer without any AFP (buffer) (n=17).
The damage ratio is significantly different between the buffer and all AFPs. ANOVA, * significant at p<0.05.Damage ratio between the different AFPs are not significant.

The data obtained from our FDT assays show that our solutions containing AFPs significantly decreased the amount of tissue damage of our plants compared to the buffer alone (Fig. 2). Indeed, while we observed about 12% of damage due to frost on plants submerged in Buffer solution , the measured damages on plants treated with our AFPs was on average ten times lower (0.9%, 1.47% and 0.66% of damage when treated with FfiBP, RiAFP and both, respectively). Interestingly, we did not observe significant differences in the percentage of damage between plants treated with the different AFPs. Overall, our data clearly demonstrate that our purified AFPs have a strong protective effect against frost damage on plants, and could therefore potentially be a good solution to safeguard apricot trees from freezing spells.

Pseudomonas syringae pv. syringae

P. syringae syringae theoretically facilitates the formation of ice thanks to a membrane protein called Ice Nucleation Protein (INP). To test this hypothesis, we performed an assay named “Time to freeze” or “TTF”, using FROZONE.

The idea of our TTF assay is to monitor the freezing time of our different solutions. Thanks to FROZONE, we can precisely set a temperature, in our case -10 ºC, and time when our drop freezes. If P. syringae syringae does facilitate the formation of ice, we should observe a decrease in the freezing time.

Measurements - Time To Freeze

Using the TTF assay, we measured the time before freezing of multiple solutions in order to quantify the effect of P. syringae syringae.

Figure 6 | Pseudomonas accelerates the freezing of liquid solutions.
The graph shows a barplot with the mean of biological triplicate +/- standard deviation. Time measured before observed freezing of NYB liquid solution supplemented with five different samples:
P. syringae syringae with an OD of 0.7 (P. syringae syringae 1); P. syringae syringae diluted twice with an OD of 0.39 (P. syringae syringae 0.5); P. syringae syringae diluted four time with an OD of 0.155 (P. syringae syringae 0.25); NYB buffer (Buffer); Escherichie coli with and OD of 0.72.

Data in figure 6 show that adding some P. syringae syringae accelerates freezing. We used two different controls: the same Buffer (NYB) we used with P. syringae syringae and a culture of E. coli BL21 with the same concentration as P. syringae syringae (OD=0.7). NYB is freezing on average after 300 seconds and the E. coli solution is freezing after 1,500 seconds. The time before freezing observed for the sample containing E. coli is higher than for the buffer alone, most likely because the E. coli cells physically disturb ice formation. By comparing the three different concentrations of P. syringae syringae tested, we can assume that this phenomenon does not appear to be concentration dependent. It is likely that only a small amount of P. syringae syringae will affect the time before freezing. Nevertheless, data show a significant difference in the speed of freezing between the solution containing E. coli and P. syringae syringae of 1397 seconds and a difference between Buffer and P. syringae syringae of 197 seconds. These results support the conclusion that P. syringae syringae accelerates freezing of solutions compared to other bacteria.

As we want to protect apricot trees from frost damage, we aim to increase the time before freezing. To avoid the negative effect of P. syringae syringae, we came up with an original solution: Tailocins.


As our tailocins can specifically kill P. syringae syringae, they theoretically should reduce the impact of this pathogen on freezing temperature. To verify this hypothesis, we developed a new assay, called “versus”. In this assay, we placed two drops of liquid solutions that we wish to compare using our FROZONE device Thanks to EDNA (the FROZONE's controller), we set the temperature at -15 ºC and observed which drop is freezing first. We compared the freezing time of two 10 µL drops of buffer both supplemented with P. syringae syringae (OD = 1), where one sample was first incubated for one hour with tailocins. If tailocin does reduce the effect of P. syringae syringae, we should observe a difference in the freezing time between the two drops.

Figure 7 | Image of two drops of solutions observed under the microscope.
Left drop contains P. syringae syringae and right drop contains P. syringae syringae that was exposed to tailocins). Right panel shows the FROZONE controller software . The temperature of FROZONE was at -11.18 ºC and only the left drop (with P. syringae syringae) was frozen at this temperature.

Data obtained from our experiment showed that tailocins could successfully prevent freezing of liquid solutions containing P. syringae syringae (Fig 7). Indeed, we observed that at a temperature of -11.18 ºC, non-treated solutions of P. syringae syringae froze, while identical samples exposed to tailocins remained liquid.

Measurements - Freezing Time

To further obtain quantitative measures of this phenomenon, we measured the actual freezing time of our samples (Fig 6).

Figure 8 | Pseudomonas death induced by tailocins increases time before freezing.
Graph shows mean of biological triplicate +/- standard deviation. Time measured before observed freezing of NYB liquid solution supplemented with Pseudomonas syringae pv. syringae (P. syringae syringae) or P. syringae syringae and tailocin proteins (Tailocin + P. syringae syringae). Times before freezing are significantly different. T-test, * significant at p<0.05, p=0.0114.

We analysed the data we obtained using the assay versus and concluded that there is a significant difference in the time before freezing between the non-treated and tailocin-treated liquid samples. This confirms that tailocins affect the freezing time of a P. syringae syringae solution and can be used to protect apricot trees against Pseuodmonas’s pathogenic effect.

Electron microscopy

After the purification of our tailocin (LINK), we performed a lot of After the purification of our tailocin, we performed many tests in order to confirm we had mostly tailocins and no phages or other bactericides in our solution. For instance, we used electron microscopy to visualize the actual content of our liquid solution and detect the presence of tailocins alone. We were also able to measure the size and width of the tailocin. Measurements being: 168 nm long and 25 nm wide.

Figure 9 | Electron microscopy photography of our tailocin.
This image of a tailocin solution, we can see the structure of the tailocin and its tailfibres. The tailocin measures are: 168 nm long and 25 nm wide.

To our knowledge, this type of tailocin was never accurately measured before, and this represents a valuable characterization of the protein complexes we produced.


Thanks to these measurements, we were able to:

  • Determine the thermal hysteresis of our different AFPs, and therefore prove that they are functional.
  • Demonstrate that our purified AFPs have a strong protective effect against frost damage on plants.
  • Determine the effect of P. syringae syringae on the freezing time.
  • Prove that our tailocins affect the freezing time of a solution containing P. syringae syringae.
  • Prove that we successfully produced tailocins.
  • Measure the size of our tailocins

  1. [1] Group FM, Staining of Arabidopsis thaliana leaves with trypan blue and aniline blue, Vol. 314, pp.1-2, 2010
  2. [2] II H, M R, AG H, Mechanisms of frost resistance in Arabidopsis thaliana, In: Plantam Vol. 248, No. 4, pp. 827-835, October 2018

    Available from:
  3. Figure 4 created with

Follow Us

Instagram logo Twitter logo linkedin logo


logo mail