Introduction

With Binanox, we aimed to establish a biomanufacturing system for the production of metallic nanoparticles optimized for photothermal therapy (PTT). Upon conducting a thorough literature review, we established an experimental foundation for our project. On this page, we describe the most important results that portray our journey while designing Binanox.

Previous research findings revealed that 800 nm is the optimal absorbance wavelength for nanoparticles for photothermal therapy¹. To find the synthesis conditions that optimize the absorbance at 800 nm we applied a statistical model called Box-Behnken Design (BBD). The details on the theoretical background can be found on the Modeling page. According to literature, pH, temperature, and gold and silver salts concentration have been shown to have a significant influence on nanoparticle formation^2,3,4. Therefore, these conditions were chosen to be varied.

Experiments were performed in two iterations (find out more on the modeling page). In the first iteration, where we tested ascorbic acid nanoparticle synthesis, gold and silver salts concentration and pH were varied. Identifying the optimum of these parameters for a simple chemical system allowed us to obtain a solid basis before moving on to more complex experiments with a bioreducing system. Thus, the model was used to find the optimal HAuCl₄:AgNO₃ ratio at the pH of the MH broth, which was identified to be optimal for nanoparticle production.

Prior to starting the modeling experiment, a preliminary experiment was performed in order to establish a suitable HAuCl₄:AgNO_3. Ascorbic acid was added to varying concentrations of HAuCl₄ and AgNO₃, varying concentrations were obtained by diluting these salts in ddH₂O. Fig. 2 shows how this affected absorbance at 800 nm. The highest average absorbance was measured with 0.20 mM HAuCl₄ and 0.10 mM AgNO₃; therefore, these values were chosen as the central values for Box-Behnken Design. At the same time, the absorbance was high at 0.05 mM and 0.15 mM AgNO₃. Therefore, we decided to make a range of 0.20±0.05 mM for HAuCl₄ and 0.10±0.05 mM for AgNO₃.

The first step in establishing a bioreduction system for producing PTT-optimized nanoparticles was to identify how the growth medium could influence the production of nanoparticles. To investigate this matter, we compared four media: LB, LB low salt, MH, and TSBS. These experiments were done with the wild-type (WT) strain Escherichia coli (E. coli) BL21, which was grown in four different broths overnight at 37°C.

Supernatant was used to create a cell-free system for nanoparticle production. Additionally, it was tested whether including the cell lysate, which may contain reducing agents, in the supernatant had a beneficial effect on nanoparticle synthesis. A full description of the experimental set-up can be found on the Notebook page.

The absorption spectra were measured after 24h (Fig. 6). The graphs indicate nanoparticle formation is less optimal when using lysed cells. A possible clarification is that the content of the cells may interfere with nanoparticle formation or proteases released during lysis may have denatured proteins capable of metal ion reduction. Furthermore, the supernatant of E. coli grown in MH broth showed a higher absorbance, indicating that this medium is beneficial for the production of nanoparticles. Notably, the medium alone also resulted in high absorbance, which suggests that the medium could also reduce the silver and gold salts to some extent.

After the absorbance measurements were taken, TEM was used to study the formed nanoparticles. TEM pictures for all media can be found on the Notebook page. For synthesis in LB and MH broth, TEM micrographs showed the formation of nanoparticles of which some contained spikes (Fig. 7). The TEM pictures show a zoomed-out overview of the formed nanoparticles (Fig. 7), as well as isolated urchin-like nanoparticles for both LB (Fig. 7A) and MH (Fig. 7B). The sample of nanoparticles produced by LB supernatant (Fig. 7A) showed a higher degree of aggregation. Moreover, the spikes seen on these nanoparticles were not as prominent as in MH broth (Fig. 7B). Additionally, the size observed for the nanoparticles was around 100 nm, which is a desirable size for PTT. Therefore, based on the absorbance results and the TEM pictures, we proceeded with MH broth in subsequent experiments.

After obtaining the optimal medium for nanoparticle production with the supernatant of WT E. coli BL21, we decided to employ modeling to further optimize the Binanox system. For this experiment, we combined the knowledge obtained from iteration 1, described above in experiment 1, to find the optimal pH, temperature, and metal salt concentration for the production of nanoparticles with a high A_{800 nm.}

Our previous experiments revealed the optimal gold:silver salts concentrations (experiment 1), and that the non-lysed supernatant of WT E.coli BL21 grown on MH broth performed best as a bioreduction system (experiment 4). This obtained knowledge was used for the following experiment. This experiment was performed using WT E. coli BL21 supernatant with E. coli BL21 lysate. The hypothesis was that the overexpressed proteins would be present in higher quantities after lysing the cells, increasing the reducing potential. Before starting the experiment, it was necessary to identify an adequate range for the optimizable parameters: pH, temperature, and concentration of gold salts. More information about the conditions and the ranges can be found below.

pH
Assuming that better protein function will result in better nanoparticle formation, the optimal growth pH for E. coli was used as the central value: pH 7⁹. The adequate range was further restricted by the biologically relevant pH values for E. coli: growth is only possible between pH 5 and 9⁷. Therefore, our tested pH values were 5, 7, and 9.

Temperature
Since 37°C is the optimal temperature for E. coli growth it was chosen as the central value for the temperature range¹⁰. In addition, literature showed that nanoparticle formation benefited from higher temperatures⁸. Therefore, the upper boundary was set to 49°C, which was the highest the shakers in our lab could achieve. Finally, according to Box-Behnken Design, an equidistant point was selected, which made the final temperatures tested: 25, 37 and 49°C.

Gold concentration
After obtaining the optimal gold:silver salts ratio in the first iteration (1.85:1), it was necessary to establish which concentration of gold and silver was optimal for PTT-optimized nanoparticle production using E.coli. To identify an adequate concentration range to be tested in the modeling, it was necessary to conduct a preliminary experiment. The preliminarily tested range for the concentration of golden salts was 0-8 mM. The results showed that A_{800 nm} decreased as the concentration increased (Fig. 8). Therefore, it was decided to test 2, 4, and 6 mM for modeling.

The defined adequate ranges for pH, temperature, and [HAuCl₄] are summarized in Table 1. The absorption spectra of the different combinations of conditions are represented in Fig 9. To find the optimal conditions, the A_{800 nm} of the supernatant with lysate was maximized in the software Design Expert by Stat Ease. The maximization yielded pH 5.2, gold concentration 5.75 mM, and temperature 49 °C as optimal conditions.

Similar results have been found in other studies. Divya et al., performed a study in which pH and temperature were independently varied to optimize for AgNO₃ nanoparticle synthesis with E. coli². pH 5, 7, and 9 were tested and pH 5 was found to be optimal, which is in agreement with the optimal pH of 5.2 that we found. For temperature, they tested 25°C and 37°C, of which 37°C performed better. In our experiments too, higher temperatures were found to be more optimal. Work by Gou et al. has found similar findings⁸. They performed silver nanoparticle synthesis using Lysinibacillus sphaericus and temperature from 20°C to 90°C. Between 20°C and 70°C, the absorbance peak increased as the temperature increased.

Given that the efforts to transform the metal ion reducing genes from C. metallidurans and C. albicans were unsuccessful (See experiment 10 below, under supplementary Experiments), biological production of bimetallic nanoparticles was attempted using the ASKA collection, which is explained under experiment 8. The genes selected from this collection were napA, melA, cueO and copA.

NapA
NapA is a nitrate reductase often released by E. coli and is capable of forming nanoparticles biologically. The process involves NapA converting nitrate to nitrite, which leads to the oxidation of Ag⁺ to Ag²⁺, which is easier reduced to its elemental form. This results in the reduction of silver ions to silver nanoparticles¹¹.

MelA
MelA is a tyrosinase that is capable of converting L-DOPA to melanin. Melanin contains functional groups such as quinone and semiquinone, that work as effective reducing agents to convert gold ions to gold¹².

CopA and CueO
CopA and CueO are proteins associated with copper ion regulation in E.coli cells. These proteins are often involved in the conversion of copper ions to its elemental form. However, research has shown that upon exposure to gold ions, CopA is over-expressed in C. metallidurans. Thus, we attempted to check if the protein functions to reduce silver and gold ions in E.coli^6,13.

We approached this experiment by using the supernatant from BL21-pET16b overnight liquid culture, which was then supplemented with concentrated protein extract from the ASKA strains. An overview of the experiment is briefly explained below; for a more detailed overview, read our Notebook.

Liquid cultures for BL21-pET16b and all ASKA strains were cultured overnight in MH broth to an approximate OD₆₀₀ of 0.6-0.8. The liquid culture for BL21-pET16b was spun down to obtain the supernatant, which was further sterilized using filter sterilization. In order to induce the production of proteins in the ASKA strain, liquid cultures were re-inoculated in fresh MH broth to an OD₆₀₀ of 0.4 and induced with IPTG. These liquid cultures were centrifuged to pellet the bacterial cells, which were then sonicated to break open the cells to release the expressed proteins. Finally, the experiment was executed in a 24-well plate, where the supernatant was separately supplemented with each protein extract, AgNO₃ and HAuCl₄ salts in duplo. These plates were left in a shaking incubator for 24 hours before the absorption measurements were taken.

The graph below (Fig. 10) shows absorbance obtained for ASKA CopA at 800 nm. High absorbance was observed for the samples containing medium with salts. This can be attributed to the presence of tryptone in the medium which acts as a strong reducing agent. Read more on the role of each media component under the experiment section, testing media for nanoparticle synthesis. The highest absorbance was observed when silver and gold salts were added to the WT supernatant. However, upon the addition of lysate to supernatant with gold and silver salts, the absorbance value dropped. An even lower absorbance was recorded for lysate with silver and gold. These results were in contrary to the hypothesis, where a higher protein concentration resulted in increased nanoparticle synthesis.

The graphs in Fig. 11 & 12 show A_{800 nm} obtained for ASKA CueO and NapA. The results obtained are quite similar to absorbance observed for CopA, as explained above. Thus, it is highly likely that cellular components released upon lysis interfere with protein function or nanoparticle synthesis.

The graph below (Fig. 13) shows absorbance obtained for ASKA MelA at 800 nm. As discussed earlier, MelA uses L-DOPA for metal ion reduction. Thus, a final concentration of 1 mM L-DOPA was used to test the effect on nanoparticle production in the presence of MelA.

Silver and gold salts were added to either MH broth, supernatant WT, and lysate from WT. These experiments showed similar results to the data discussed for CopA, CueO and NapA. However, the addition of L-DOPA to WT supernatant containing MelA lysate and to MH with either lysate from WT or ASKA MelA strains, causes an increase in absorbance at A_{800 nm.} The wells containing L-DOPA developed an orange color within a few minutes of addition, which may explain the higher absorbance values. This color change may be due to the conversion of L-DOPA to melanin¹².

Our experiments have indicated that the addition of cell lysates containing metal reducing proteins did not lead to increased production of nanoparticles as hypothesized. It is highly likely that the cell lysate contains components which interfere with nanoparticle synthesis or protein function such as proteases. Thus, it may be ideal to obtain purified aliquots of the proteins of interest for testing nanoparticle production or using a cell system for biological production of nanoparticle synthesis.

This experiment was inspired by S. Gurunathan et al. where they used LB nitrate for the production of silver nanoparticles. These experiments aimed to identify whether nitrate in the MH broth, identified as an optimal medium through our experiments, has an impact on the production of nanoparticles⁷.

An identical approach as detailed under ASKA- Effect of adding pellet lysate was used, however, BL21-pET16b was cultured overnight in MH broth supplemented with nitrate salts. Read more about the experimental design and protocol in our Notebook.

The graph shown in Fig. 16 compares the A_{800 nm} for samples containing CopA lysate for nanoparticle production in MH broth and MH broth supplemented with nitrate (MH broth nitrate). For samples that contain medium with silver and gold salts, the absorbance remains consistent. However, the addition of salts to the MH broth nitrate supernatant causes an increase in the absorbance compared to MH broth supernatant. It is likely that the nitrate in the medium promotes the formation of nanoparticles. Unfortunately, the addition of lysates to MH broth nitrate results in an even lower absorbance when compared to MH broth with lysates. It is possible that these differences are due to variations in execution of the experiments or that the nitrate in the medium interacts with the components of the cell lysate and causes further interference in nanoparticle production.

The graph shown in Fig. 17 is for A_{800 nm} for samples containing CueO lysate for nanoparticle production in MH broth and MH broth nitrate. For samples that contain medium with silver and gold salts, the absorbance remains consistent, whereas the absorbance for the supernatant increases. Interestingly, the absorbance increases upon addition of lysate to supernatant from MH broth nitrate. This perhaps suggests that CueO works better in MH-nitrate and results in increased production of nanoparticles as evidenced by a higher absorbance.

For samples containing NapA, the medium with silver and gold salts shows again a consistent absorbance, as indicated in Fig.18. However, the addition of lysate to the supernatant obtained from MH broth nitrate shows a slight drop in absorbance. This may suggest that the production of NapA may not be largely affected by changes in media conditions. However, take note that the error bars are quite large for MH broth nitrate supernatant with lysate.

For samples prepared with MelA and L-DOPA, the medium with silver and gold salts shows a similar absorbance, as indicated in Fig. 19. However, the supernatants that are supplemented with either WT lysate or MelA containing lysates, salts and L-DOPA show differences in absorbance. However, the error bars recorded for these experiments are quite broad, thus we cannot clearly state whether changes in medium impact the performance of MelA in nanoparticle production.

To design our constructs we chose 3 different genes based on literature¹⁴. The Cup1 gene from Candida albicans encodes for a metallothionein, which is a metal binding protein and was proven to be important in the formation of nanoparticles¹⁵. The genes CopA and NapA from Cupriavidus metallidurans were chosen as it was previously shown that they are involved in the production of gold and silver nanoparticles. The two genes were taken from C. metallidurans as it is a very metal resistant strain, especially to high concentrations of gold¹⁶.

The DNA sequences were ordered from IDT in the form of gBlocks. In order to simplify the ligation into our vector pET-16b the sequences were first cloned into pJET1.2/blunt and plated on LB plates. We then isolated the plasmid by Miniprep and digested it with two restriction enzymes BamHI and NdeI, and checked the result by gel electrophoresis. Next, the inserts were purified from gel and the sequences were cloned into pET-16b by ligation with the ligase T4. The ligated product was transformed into E. coli BL21 and plated on ampicillin LB plates, as the vector contains an ampicillin resistance gene. Eventually, the sequences were checked by diagnostic restriction digestion and Sanger sequencing (Fig. 24-26). The sequencing results showed an alignment for Cup1, but unfortunately the ligation for CopA and NapA failed twice and could not be repeated due to time constraints.

Additionally, four different genes from the ASKA collection were used¹⁷. The selected genes were copA, napA, cueO, and melA. The ASKA plasmids were isolated by Miniprep, and they were then transformed into E. coli BL21. This was done to have comparable and reproducible data to our constructs, as they are also expressed in the BL21 strains.

In order to confirm that these genes were actually expressed, the strains were induced with IPTG (end concentration 1mM) for 2h and the expression was confirmed with an SDS gel (Fig. 27). As a control the non-induced strains and BL21 were also added. The size of the different proteins was looked up on Uniprot¹⁸. The sizes of the proteins were 51kDA for MelA, 57kDA for CueO, 88kDA for CopA, and 93kDa for NapA. As seen on the SDS gel, the proteins were clearly expressed.

After the results of the experiment with 4 different media were obtained, we noticed absorbance peaks for samples in which growth media were used to reduce metallic salts. This was an unexpected result, as we did not come across such a phenomenon during the preliminary literature review. Therefore, when these absorbance peaks were observed, we set up an experiment in which the influence of the various media constituents on nanoparticle formation was studied.

We tested the ingredients of LB, LB low salt, LB-Nitrate, MH and TSBS: deionized (DI) Water, Tryptone (substitutes Casein Hydrolysate), Yeast Extract (substitutes Soy Extract in TSBS), Starch, NaCl, Glucose and KHPO^19,20,21 (Soy Extract and Casein Hydrolysate were substituted with ingredients that fulfill the same function and have a comparable composition). The used concentrations were achieved in DI water, according to the respective recipes. We also tested the interaction between the different constituents of MH broth, since this medium was the most relevant for the scope of the project. We only show this plot on this page, but the obtained spectra for all other media are shown in Fig. 39-41 on the Notebook page. They yielded similar results.

The absorption spectra for MH, supernatant of MH broth and the individual components of the broth, including their interactions are shown in Fig. 28. The observed peaks at ~550 nm are only present for tryptone-containing samples. All other compounds seem to not have any reducing effect.

In order to achieve a cell-free system, it was important to ensure that the supernatant used was free of all bacterial cells. Filter sterilization was used to achieve this, however, it is a more expensive process and is rather inefficient with respect to time. Hence, this experiment was designed to check whether adding antibiotics to the supernatant may work as a suitable alternative.

A BL21-pET16b culture having an ampicillin resistance gene, was treated either with 20 ul of Kanamycin or Chloramphenicol. These antibiotics are expected to kill the bacteria given they do not have the respective resistance gene. The supernatant of the treated BL21 cultures was then plated on LB ampicillin plates. As a control, we added a BL21 culture, where the supernatant had been filter sterilized. The cultures were left to grow overnight in an incubator set to 37°C.

The image shown (Fig. 28) below indicates that supernatants treated with antibiotics resulted in growth of bacterial colonies overnight, whereas no growth is observed for the control.

This data suggests that filter-sterilization is the optimal technique to obtain a cell free system, rather than antibiotic treatment of supernatants.

In this experiment the impact of shaking on nanoparticle production was tested. In all the previous experiments the flasks were shaken elliptically at 200 rpm. For this experiment, E. coli BL21 and three different ASKA genes (CueO, NapA, and MalE) were grown in MH broth with a gold and silver ratio of 1.85:1. These flasks were shaken in a parallel movement at three different speeds. First, they were left on the bench without shaking for 48h. The same strains were further tested at a parallel shaking speed of 20 rpm and 60 rpm over the course of 48h.

The absorption spectrum was measured for all different shaking speeds for the four different strains. E. coli BL21 seems to perform best at 0 and 20 rpm, while 60 rpm seems to have the least effect (Fig. 30). The ASKA strains MalE and NapA, show the highest peak when not being shaken, while for CueO the highest peak can be observed at a speed of 60 rpm (Fig.30). In general, for the three ASKA strains the absorbance values are very close to each other and it seems like the shaking speed does not have a substantial effect on the yield of nanoparticles.

TEM pictures were taken to see what effect the shaking has on the morphology of the nanoparticles. The pictures were taken for all four different strains at shaking speeds 0 rpm and 60 rpm. The pictures for E. coli BL21 at a shaking speed of 0 rpm (Fig. 31 A) show a lot of aggregates and no clear nanoparticle formation, while at 60 rpm (Fig. 31 B) we can observe less clumping and more distinctively sphere-like nanoparticles. The single nanoparticles had a size around 100 nm, while the aggregates were observed at a scale of 500 nm.

For the ASKA strains only pictures of CueO and NapA were taken, this was simply due to time constraints and the availability of the TEM. ASKA-CueO shows the same behavior as BL21. At 0 rpm, we can observe a lot of aggregates and no clear formation of nanoparticles (Fig. 32 A), while at a shaking speed of 60 rpm the aggregates seem to break down and we can again observe sphere-like nanoparticles (Fig. 32 B). The single nanoparticles had a size around 200 nm, while the aggregates were observed at a scale of 500 nm.

Interesting results were observed for ASKA-NapA. At 0 rpm we did not see the agglomerates as for the 2 previous strains, but could already see sphere-like nanoparticles (Fig. 32 A). At 60 rpm, we could observe additional growth on the nanoparticles, indicating the formation of spikes (Fig. 32 B). Additionally, the size observed for both shaking speeds was around 100 nm.

In general, shaking seems to have a positive effect on the formation of nanoparticles, especially to prevent agglomerates.

In this experiment the WT was used with silver and gold salts over 24h to create the nanoparticles. Furthermore, visual inspection of the absorption spectra of the nanoparticles with different percentages of sucrose (Fig. 34), showed a peak at 600 nm for all samples. The height of the peak differed among the samples, however, no correlation between the percentage of sucrose and absorbance was observed. We were also interested in the size of our nanoparticles, therefore we used a sucrose density gradient ultracentrifugation technique. Our previous research findings revealed nanoparticles of different sizes. We aimed to separate the nanoparticles based on size using the ultracentrifuge. This is based on the principle that the differently sized nanoparticles would layer on the surface of a gradient whose density increases linearly from top to bottom.

Unfortunately, after centrifugation the nanoparticles were not separated but were all pushed to the side of the tube (Fig. 35), making it impossible to make a conclusion about size distribution.

In order to improve the outcome of this experiment a different rotor insert is needed, called a “swing bucket rotor” that would prevent the nanoparticles from being pushed to the sides of the tubes. Additionally, our nanoparticles were not stabilized by a coating and aggregates formed, this could also be a reason that they were not properly separated in the centrifuge.

Conclusion

With Binanox, we aimed to establish a biomanufacturing system for the production of bimetallic nanoparticles optimized for application in PTT. For the nanoparticles to be optimal for PTT; an urchin-like shape, A_{800 nm} , and a size smaller than 150 nm are desired.

The optimal conditions for nanoparticle synthesis were established using the modeling experiments. The first iteration revealed an optimal gold:silver ratio of 1.85:1 and the second iteration showed an optimal pH of 5 and temperature of 49 °C. In addition, E.coli BL21 supernatant showed to perform best in MH broth since it yielded the best A_{800 nm} and TEM pictures confirmed the formation of spiky nanoparticles of a size of around 100 nm. These defined optimal conditions were used in further experiments. Cell lysates containing metal reducing proteins were introduced to the supernatant WT, however, showed no improvement in nanoparticle production. This could be caused by the cell lysates containing components that interfere with protein function or nanoparticle synthesis or by the degradation of the metal reducing proteins. The addition of L-DOPA to the MelA lysate, positively influenced the A_{800 nm}. Furthermore, the addition of nitrate to the MH broth seemed to not have improved the A_{800 nm}. Finally, our construct has not been tested since all the transformations, except for metallothionein, failed and due to lack of time no further experiments were performed.

For future experiments, our constructs should be tested in combination with a protease inhibitor to prevent the degradation of the metal reducing proteins. Furthermore, surfactants should be added to improve the stability of the nanoparticles and prevent the formation of aggregates. Finally, the ultracentrifuge could be a promising technique to separate the nanoparticles based on size. Therefore, we would like to use the ultracentrifuge in combination with a swing bucket to prevent the nanoparticles from being pushed toward the side.

All in all, our research findings contribute to a better understanding into the biosynthesis of bimetallic nanoparticles using a microbial factory. Further future experiments are required to optimize and validate our biomanufacturing system, however the results are promising and lay a good foundation for the biosynthesis of nanoparticles and their application in PTT.