Rubber results

Did we indeed make rubber?

Now that we control the expression of the genes it is time to characterize our central genes (dxs (B. subtilis) and HRT2). We invite you along on a journey through our attempts to obtain proof of concept; to show that Bacteriorganic Rubber is a true possibility.

Characterization of dxs (B. subtilis)

Functionality assay
To optimize the flow through the MEP pathway, the dxs gene was overexpressed, the expectation being increased levels of IPP and DMAPP (see Specification). To examine if overexpression indeed resultats in an increase in substrate, we attempted to assay the levels of DMAPP using a headspace gas chromatography (GC)-technique.

DMAPP was hydrolyzed in acid to the volatile hydrocarbon gas isoprene. The gas was subsequently analyzed with headspace GC. A linear relationship between amount of detected isoprene and DMAPP concentration has previously been established. Source: Fisher AJ, Rosenstiel TN, Shirk MC, Fall R. Nonradioactive assay for cellular dimethylallyl diphosphate. Anal Biochem. 2001 May 15;292(2):272-9. (Link) We were capable of producing a standard curve by reacting DMAPP with acid for 2 min (instead of the 60 min specified in the previous study) (Fig. 1). At this time, we obtained optimal peak detection for standard solutions. We were, however, incapable of detecting isoprene, even in high concentrations of bacterial samples treated with acid. The test was expanded to include acid hydrolyzation for 2, 30, 60 or 90 min, yet we could not detect isoprene, and therefore not detect DMAPP.

Figure 1. Optimization of the procedure is needed before characterization of the dxs bricks can be completed using this approach. We suspect that the complexity of the bacterial samples is too high, and thus the reaction does not take place as fast as might be necessary for detection in our setup. Sonication of bacterial samples with and without addition of standard DMAPP, and subsequent measurements might shed some light on this hypothesis. However, it should be noted that the GC wasn’t fully functional during the test period, consequently leading to broader peaks and thus lowered the sensitivity of the instrument. On the 3rd of October, we received a mail from Professor Lars Porskjær Christensen, Department of Chemistry-, Bio- and Environmental Technology, University of Southern Denmark:

"...The GC has now been repaired and the sensitivity has been improved considerably. The GC-peaks should be very sharp now. This may be the reason that you have not observed any release of isoprene from your samples...MailTranslated from danish".

Unfortunately, with only 2 days left to wiki-freeze, there was no time for another round of testing. Since the European Jamboree in Lyon we have performed the experiment multiple times, but it still needs further optimization, as our results are inconclusive.

Growth Experiment
Figure 2. Carrying the Dxs devices and expressing the gene could impair the growth, and hence be important in production purpose. To test if the growth of MG1655 bacteria is impaired when carrying and expressing our Dxs devices, we measured the growth rate with OD600 measurements.

Two triplicates of MG1655 carrying no vector, empty pSB1C3 vector, pSB1C3-Plac-dxs (B. subtilis), pSB1C3-Pcon-lacI-Plac-dxs (B. subtilis), or pSB1C3-Pcon-lacI:LVA-Plac-dxs (B. subtilis) were started from ONC at time 0 hours, OD600=0.005, and grown at 37ºC and 180 rpm. OD600 measurements were done every half an hour, and 1 of each triplicates was induced at time 2.5 hours. All strains grew at the same pace and induction didn’t impair growth rate (Fig. 2).

Characterization of HRT2

Rubber purification
To discover rubber, it is useful to isolate the rubber from the rest of the bacterial cells. This will allow us to remove as many variables in the detection assays as possible. The rubber was purified according to self-written SOP0030 - Rubber purification.
The method of purification was based on a literature search and chemical evaluation of solubility of the polyisoprene. Cells were sonicated in ethanol suspension, washed in acetone, and extracted in n-hexane (both steps were overnight).

Two different methods were tested to evaluate the most efficient extraction method. In the first method, we washed with acetone and extract with n-hexane in a soxhlet extractor. The second method excluded the time-consuming soxhlet steps and washed with acetone for a shorter duration of time (15 min shake at 37ºC). Rubber was extracted by adding hexane to the cell suspension, then centrifuging the sample. The supernatent was saved (hexane solution). We tested the methods of rubber extraction on a positive control made from MG1655, mixed with polyisoprene. Both the soxhlet and the non-soxhlet method were evaluated using H1-NMR. We deemed the results to be similar, and since the required time for non-soxhlet extraction was significantly less, we chose to use the non-soxleth method.

The H1 NMR test

We now seemed able to detect rubber that was extracted from a mixture of bacterial debris and polyisoprene, and felt ready for a test of our production system. H1 NMR allowed us a certain specificity and although a significant amount of polyisoprene was needed to enable detection, it was preferred over C13NMR. All samples were run an a bruker 400 mHz NMR spectrometer. We also knew that we could control the expression of the genes in both of our plasmids. We were ready to put our abilities to extract rubber through the non-soxhlet method and detect it using H1 NMR to the test. In other words, we were ready to find the rubber present in our strain! Source: Donald L. Pavia and Gary M. Lampman. Introduction to Spectroscopy, International Edition 4e (Book) ISBN-13: 9780538734189 / ISBN-10: 0538734183 (Link)

During the first test of our CPS bacteria (MG1655 transformed with both plasmids - see Design), bacteria were grown to late exponential phase at 37ºC, then induced with 1mM IPTG, 0.2% Ara, and 1mM MgCl2. It was then transfered to room temperature for another 4 hours of growth. Non-soxhlet extraction was used to purify any rubber present in the sample, which was then compared to a positive control (WT mixed with polyisopren - also undergone non-soxhlet extraction) and a standard isoprene solution.

Figure 3 illustrates the peaks given by pure polyisoprene are found at 5.12 (A), 2.04(B) and 1.68(C)in the ratio 1(A):4(B):3(C). Additionally, a peak is visible at 1.56 indicating water (the standard was not dried in a vacuum oven ON like the rubber extraction was). The peak at 0.00 ppm is the defining peak of the ppm axis and represents TMSTMSTetramethylsilane which is a calibrating standard.

Our rubber extraction of WT + polyisoprene (Fig. 4) gives the same peak placement as the pure polyisoprene ((A), (B) and (C)). However, the integration of the 3 peaks shows a relationship of approximately 1:5:4. The discrepancy can be explained by the impurities in the area between 0-2.5 ppm. Such impurities might add additional integration value to the peaks assigned to the polyisoprene peaks (B) and (C) - hereby causing a disruption of the true relationship. Some peaks from the solvents used to purify the rubber (Acetone, Ethanol, and Hexane), as well as a small amount of water (1.56 ppm) are seen, too.

The conclusion arrives in figure 5, named DXS+PT, which is our double plasmid CPS bacteria, and displays the same peaks (A), (B) and (C) as both the WT + polyisoprene and pure polyisoprene. This is a strong indication of the presence of rubber - specifically, this is an indication that our CPS bacteria produces rubber: Bacteriorganic Rubber. Peak distortion of the spectrum due to solvents are the same as seen in the rubber purification from WT + polyisoprene.

A second round of testing was done to validate the first experiment. We wanted to include a negative test (WT), in order to exclude the possibility of a naturally occurring polyisoprenoid compounds in E. coli. We performed rubber purification on WT, our CPS bacteria as well as a strain containing only the pSB1K3-araC-Para-HRT2 device. Bacteria were grown and rubber extracted as described above. The three samples where unfortunately not dried properly in the vacuum oven due to apparatus malfunction.

The test provided the following results: Fig 6 lacks the characteristic peaks of polyisoprene at 5.12, 2.04 and 1.68. It can be concluded from their absence that there is not rubber (nor any other compound that might provide a similar chemical shift values) in the wildtype. In the CPS (Fig. 7), we observe very a slight peak at 5.12, indicating the presence of (A) hydrogen. This peak has the same splitting pattern as the first round of H1-NMR, but it has a very low intensity. The (B) and (C) peaks are hidden in the background noise, which is most likely due to cell debris and solvents, which did not evaporate appropriately. We suspect that the machinery has a decreased sensitivity towards the isoprene peaks due to the high amount of solvent seen from the assigned peaks, but the peak at 5.12 (with the recognizable splitting pattern) is a strong indication that rubber is once again present in our CPS bacteria (containing both HRT2 and Dxs plasmids).

When we advanced to the finals in Boston, we wanted to verify that our previous experiments where replicable as well as investigate the quantity and yield of rubber. To assess this we did a third round of H1-NMR on MG1655 (WT), and MG1655 carrying pSB1C3-Pcon-araC-term-Para-HRT2 (HRT2), or pSB1C3-Pcon-lacI(N)-term-dxs (B. subtilis) and pSB1K3-Pcon-araC-term-Para-HRT2 (Dxs + HRT2). 1 L LB media was inoculated with the strains, starting from OD600=0.05. The cultures were grown at 37ºC with shaking for 5 hours; the temperature was then lowered to 20ºC and cultures were induced with 1mM IPTG, 0,2% arabinose, and 1mM MgCl2 ON. Rubber purification was commenced the following day, and the final yield was solubilized in 2 mL d-chloroform and analyzed by H1-NMR.

From the spectrum reflecting the WT strain, none of the peaks originating from polyisoprene were observed (fig. 8). From the spectrum reflecting the Dxs + HRT2 strain, all three peaks originating from polyisoprene were observed (fig. 9). Lastly, from the spectrum reflecting the HRT2 strain (fig. 10), the peaks potentially originating from polyisoprene could not be conclusively identified indicating the decreased rubber production performance of this strain compared to the Dxs + HRT2 strain. This emphasizes the importance of excess IPP and DMAPP (rubber precursors) available as a consequence of Dxs overexpression, and serves as an indirect proof of the effect of MEP pathway optimization.

In order to determine rubber yield, we compared the yield from 1L of our Dxs + HRT2 strain with 5 rubber purifications on WT + a selected amount of standard polyisoprene. Yields from all 5 purifications were solubilized in equal volumes of d-chloroform (2 mL), and the resulting TMS peaks could consequently be used for normalization of the polyisoprene peaks at 5.12 ppm. This peak was selected for quantification due to its localization in a region without background noise. From these 5 purifications, we calculated a standard curve to which we compared the integral value of the 5.12 ppm peak originating from our Dxs + HRT2 strain.

From the sample with the lowest amount of polyisoprene added (0.2 mg), no peaks could be identified, indicating the detection limit of the H1-NMR setup. Together with normalized peak integrals from the other 4 samples, we calculated the standard curve depicted in fig. 13. Based on the standard curve and the integral originating from the sample containing 3.2 mg polyisoprene (Fig. 13, 12), we roughly estimate our polyisoprene yields to be in the range of 0.5-5 mg polyisoprene/L (fig. 11). We realize that this yield estimation is far from optimal, however, only the sensitivity of H1-NMR allowed us to give a rough estimate of our yield.

When we tried to weigh our yield, we observed a many-fold increased yield compared to our input amount, and we assume that this conflicting reading is due to the presence of cell debris and other impurities adding to the output reading. This situation forced us to seek other yield calculation strategies, such as H1-NMR.


We thoroughly studied the literature concerning polyisoprene on MALDI-ToF and found that the formation of adducts by adding AgNO3- would make it possible to ionize the long alkene chain, despite its lack of functional groups that can be ionized. be ionized. Source: Hyuneui L, Yeonhee L, Seunghee H, Youngsook Y, and Kang-Jin K. Investigation of Polystyrene, Polyisoprene, and Poly(2-vinylpyridine) using Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry. Bull. Korean Chem. Soc. 1999, Vol. 20, No. 7 p. 853-856. (Link)

We tried several matrixes including MBTMBT2-mercaptobenzothiazole, DHBDHB2,5-dihydroxybenzoic acid, CHCACHCAalpha-cyano-4-hydroxycinnamic acid, SASAsinapinic acid and DTDTdithranol, but didn't have the time to perform the tests on anything but our standard polyisoprene (Mw 38 kDa). Unfortunately, it appeared to be too large a molecule to be detected by the bruker MALDI-ToF machine. It was concluded that the machine's hardware settings could not match the requirements for such large molecules. However, the expected length that we expect HRT2 produces is in the 2-10 kDa range. This is within the limits of the machine, though we have yet to test this hypothesis as the machine is frequently occupied by other research groups.

Composite production system

Growth experiments

Carrying the HRT2, dxs and both (CPS) devices and expressing the genes could impair the growth, and hence be important in production purpose. To test if the growth of MG1655 bacteria is impaired when carrying and expressing our devices, we measured the growth rate with OD600 measurements.

Four different MG1655 strains carrying either no plasmid (WT), pSB1C3-Pcon-araC-term-Para-HRT2 (HRT2), pSB1C3-Pcon-lacI(N)-term-Plac-dxs (B.subtilis) (Dxs), or pSB1K3-Pcon-araC-term-Para-HRT2 and pSB1C3-Pcon-lacI(N)-term-Plac-dxs (B.subtilis) (CPS), were grown from OD600=0.05 and induced with 1 mM IPTG and/or 0.2% arabinose during growth. The first experiment showed impaired growth of the bacteria only carrying construct expressing HRT2. We thought that the reason why the strains expressing HRT2 as well as Dxs didn’t show impaired growth rate could be due to a phenomenon called inclusion bodies. When overexpressing too many proteins, there is a risk of misfolding and hence loss of function. A way to lower this risk is to lower the temperature. Therefore we set up a growth experiment where the temperature was lowered from 37 to 20ºC at time of induction. This did not prove to make a difference, and CPS strains with and without IPTG and/or arabinose induction grew at same pace as WT. However, we want to verify these results by a more thorough investigation and consequently setup a new growth experiment including OD600 measurements, counts of colony forming units (CFU) and finally scanning electron microscopy (fig. 14).

Figure 14.

Cell viability

Three different MG1655 strains; WT, HRT2 and CPS (similar to above), were grown under equal conditions as when cultures were grown for rubber purification. 1 L of bacterial cultures were initiated at OD600=0.05, from overnight cultures and grown at 37°C and shaking at 180 rpm. After 5 hours of growth, cultures were induced with 1 mM IPTG, 0.2% arabinose and 1 mM MgCl2 and the temperature was lowered to 20°C to prevent the formation of inclusion bodies. The OD600 was measured each hour throughout the experiment and sequential dilution series were plated out at time 4, 5, 6, 7 and 8 hours after initiation. Furthermore, samples for electron microscopy were taken from the induced ON cultures.

OD600 measurements illustrate a slight impairment of growth for the HRT2 strain compared with the WT strain, whereas the CPS strain shows heavily decreased Growth (Fig. 15A) . CFU counts reflect the number of viable cells in a culture. We observed a steady amount of CFU counts for the WT and HRT2 strains throughout the experiment similar to the CPS strain, although this strain presents with a lower amount of CFU counts (Fig. 15B).

We notice that the growth rate of the CPS strains is controversial to the growth experiment presented earlier, which we cannot directly explain. Importantly, the viability of the CPS strain does not decrease upon induction in both growth experiments, which indicates that the amount of produced rubber is not toxic to the bacteria. This might be due to the low amounts of rubber produced in the cells as validated by H1-NMR.

Figure 15.

Additionally, images from an equal dilution and time point in the CFU assay are displayed below (Fig. 16). The difference in size between the colonies from the three different strains confirm the decreased growth rate of the CPS strain compared to the WT and HRT2 strains.

Figure 16.

Lastly, scanning electron microcopy indicate no noticeable difference in cell morphology between the three strains upon induction (Fig. 17). This emphasizes our prior conclusion that the bacteria tolerate the amount of rubber produced in our molecular biology production system.

Figure 17.

Now that you have viewed our most valuable results describing our constructs ability to produce rubber, we invite you to dig further on, to see a comprehensive list of the parts we have submitted to Parts Registry.