From 2012.igem.org

Revision as of 22:55, 26 October 2012

When is meat rotten?

To make a rotting-meat sensor, we have to have a definition of rotten meat. For this, we used the guidelines of the European Union (2006, source (in Dutch)) and did a simple Total Aerobic Microbial Count test. With this test, one can estimate the amount of colony forming units (CFU) per gram of meat. See our food safety page for more in depth answer to this question. Our meat of choice was 70% pork, 30 % beef minced meat from our local supermarket. This type of meat is often bought in large amounts with leftovers stored in the fridge, making it the ideal candidate for our Food Warden system. Minced meat is also easy to handle when it is placed in a jar, simplifying lab work. Most importantly, as a meat lover it is hard to sacrifice a very nice expensive steak for science. We incubated the meat in closed airtight jars, in portions of 1 gram at room temperature, and tested the TAMC at time points 0, 3, 5, 7 and 24 hours. The test has been done in triplo.

Results of TAMC counting. TAMC (colony forming units/gram meat, y axis) of meat incubated at room temperature for indicated time (x axis). Red area indicates the critical values where the meat is not allowed to be distributed for consumption according to the EU.

To see the working of our own inbuilt rotting sensor, Elbrich bravely tested the smell and appearance of the meat for 5 hours. According to these tests, we humans can smell bad meat rather well. Side note: the meat has been exposed to air many times so it could be smelled. The color of the meat changed a bit: it turned greyer.

Smelling test. Minced meat was left at room temperature for 6 hours. The “nastiness” of the meat smell according to the tester was recorded.

Now that we defined and smelled rotten meat, we want to know what the volatiles are. The rotting of the meat is caused by bacteria proliferating in the meat. These bacteria produce a lot of volatiles and we can only smell a few. To form a complete picture of the aromatic and non-aromatic volatiles we analyzed the rotten meat volatiles with GC-MS.

Volatiles from rotten meat

With Gas Chromatography-Mass Spectrometry one can separate and identify different volatiles present in meat that is starting to spoil. We thought that the identification of these volatiles by GC-MS would point out the exact compounds that influence the behavior of our identified sensors, but we were surprised by what we found…

The University of Groningen has a lot of GC-MS equipment available and a large commercial database with compounds that we could use to identify the substrates we found in the GC-MS data. So far so good. However, one of the drawbacks of the GC-MS is that the compounds that we might identify from meat that starts to spoil, will be destroyed during the measurements. No further analysis of these compounds is possible then. But if the GC-MS measurements succeed, reliable qualitative data can be obtained. But we discovered that the data was hard to analyze due to the large diversity of the volatiles present.

Arjan and Tom at work with the GC-MS.	Sample bottles with rotten minced meat.

GC-MS setup: All our samples, ready to be measured. The volatiles will be taken up by the syringe in the left corner (red, to the left)

Introduction to the GC-MS technique

Substrates in the gas chromatograph are separated by the amount of time needed to pass through a capillary column in the machine. The volatiles, in a gas phase, pass through the column which has a liquid carrier. Because of the constant flow, the equilibrium between the gas phase and a interaction with the liquid carrier is constantly redefined. Interaction with the carrier will slow down the substrate that is traveling through the column. For the HP-1 and HP-5 columns, used in our experimental setup, the interaction between substrate and carrier is based on the boiling point of the substrate. Each substrate has an unique amount of interactions with the carrier and thus a different amount of time is needed to pass through the column. Based on this principle, our rotten-meat volatiles are separated and later identified with mass spectrometry.

We used the headspace method to search for compounds from meat that was left to rot in a bottle (see picture): after incubation at a high temperature the vapors were extracted from the bottle and injected into the GC-MS. Note: the rotting process of the meat was controlled as in microarray experiment for the identification of pBAD-meat sensors.

As stated we used the HP-5 and HP-1 for GC experiments. There are a broad range of columns commercially available, and all are capable of separating substrates bases on different properties. The HP-1 and HP-5 are good columns for general use. Most GC setups at RUG utilize these columns, but it has to be noted that these columns cannot identify amino compounds. Because of this we will miss some of the volatiles in the rotten meat and unfortunately we did not have the budget to buy more specific columns.

HP-5:
Dimensions: 30 m x 0,25 mm x 0,25 um
Manufacturer: Agilent
Column number: 19091J-433
Stationary phase: (5%-Phenyl)-95%methylpolysiloxane

HP-1:
Dimensions: 30 m x 0,25 mm x 0,25 um
Manufacturer: Agilent
Column number: 19091Z-433
Stationary phase: 100% polysiloxane


After this separation method based on boiling point, the measurement continues with Mass Spectrometry. This is technique enables you to identify compounds based on their differences in mass-to-charge ratio. Our machine uses electron impact ionization in a vacuum with quadruple separation. This means that the separate substrates in the machine will be bombarded with electrons in a vaccuum. Because of this, they will receive a charge and are most likely to break into pieces. These flying bits and pieces of substrates, each with their own charge, will be prevented from crashing into the wall by the quadruple poles. One can adjust the changing charge of the poles and thereby choose the range of spectrum of the identifiable compounds with their different mass-to-charge ratios.

The reliability of this measurement depends on the range of the spectrum, the reaction(s) with other molecules (which might cause redundant mass charge ratios), and how small the differences in molecule structure are (like isomers) that are difficult to separate.

The first GC-MS experiments with rotten meat resulted in blank spectra. Unfortunately, the concentration of volatiles was probably too low in the extracted sample for the equipment to measure. However, during the microarray we saw that the bacteria were already able to detect small amounts of volatiles. In the machine, it might have happened that some volatiles were not able to escape from the meat. To obtain more volatiles from the meat we used brine as solvent. It can extract the volatiles, but due to the high salt concentration, it does not allow the volatiles to stay in solution with a higher temperature. Hence, during the incubation most volatiles will evaporated and be extracted from the bottle, before being injected into the GC machine.

But also this did not work out, because we got blank spectra again. We soon came to the conclusion that bacteria seem to be able to sense volatiles better than a state-of-the-art equipment like a GC-MS, cannot detect. A very impressive thought! But we did not give up, though.

Liquid injection with organic solvents

The first measurements with only volatiles gave no reliable results. Therefore, we decided to dissolve the rotten meat into organic solvents to ensure that more compounds are available for measurement. Furthermore, we used liquid injection instead of the headspace method. Our meat samples were left to rot for more than a day at room temperature before adding the organic solvent. After addition of the solvent, the meat was left to incubate in the solvent for several hours, while frequently vortexing. The solvent was extracted and filtered before injection in to the GC. But we did not use only one solvent; we needed to study this thoroughly, and to cover all the polar and non-polar volatiles, we used different organic solvents.

For an apolar solvent we used toluene and hexane. Toluene gave an emulsion after it was extracted from the meat. In order to prevent damage to the columns we decided to only use hexane as the apolar solvent. Dichloromethane was used as the mid-polar solvent, and methanol as the polar solvent.

Results

Finally, this method produced interesting results from both the HP-1 and HP-5 column spectra. After the library search; only compounds with a quality over 80% were considered reliable. We took the compounds found in the spectra of rotten meat and subtracted the compounds found in the spectra of fresh meat and the solvent blank. Subtracting the solvent blank removes the background noise, while subtracting the fresh meat highlights those spectra up- and downregulated in rotten-meat volatiles. This approach identified the following compounds in rotten meat, with a quality that matches with approximately 80% to the reference library.

We discussed our results with Prof. Dr. ir. Minnaard, an organic chemist. He mentioned that tridecanoic acid was an interesting find because this is a C30 fatty acid, which is expected not to come out of the columns easily. This substrate is also found on both columns and with different solvents.

He also noted that the overall reference quality was low because results with less than 80% quality are considered as unreliable. This is probably caused by compounds that lie close together in the same region of the MS-spectra, making it harder to match them to compounds in the library and thus resulting in a lower quality match.

Some of the compounds found contain a fluoro group, which is rarely found in nature. It’s most likely that these library matches are not correct, thus we excluded these compounds. Also nitro compounds are not likely to occur in these conditions, so these substrates are also not likely to be correct. An explanation for this lies in the library search. It wants to fit spectra to known spectra in the database, but since our spectra show very uncommon things, our spectra are fitted to known results with the best fit. Due to this overlap, a good fit is not guaranteed and the computer ‘blindly’ decides what the best fitting substrate spectra is. These fits can be incorrect, so we should take care when extracting conclusions from these results.

Besides the tridecanoic acid, there were several more interesting compounds. Benzenecarboxylic acid was found with multiple solvents and on both columns. Other interesting compounds noted by Prof. Minnaard were Bicyclo[4.3.1]decan-10-one, 1-Hexadecanol and beta.-Phenylpropiophenone.

Ideally, we would like to verify that these compounds are correct. The normal procedure is to acquire the pure substrate and inject into the GC-MS. The spectra are compared and a conclusion on the reliability of the data is made. However, due to shortage of time and funds, we chose not do this. Hopefully we are able to peform more experiments in the future, but for now the GC-MS is a interesting part of our iGEM project, not the major part. We decided to spend our time on other research although Prof. Minnaard suggested that we should do another microarray experiment using these compounds instead of the rotten meat.

GC-MS spectra using HP-5 column and the organic solvents: hexane, dichloromethane and methanol.

GC-MS spectra using HP-1 column and the organic solvents: hexane, dichloromethane and methanol.

Future plans

If time allows, we will try to set up an experiment where we can use the exact compounds to trigger the promoter in the construct and activate the pigment. Instead of using rotted meat during the growth experiment we can inoculate the media with these compounds or pump the fumes into the culture during the growth of B. subtilis with our construct.

Reference and acknowledgement

We were very lucky that we could borrow these tools during the summer holiday. Therefore, we would like to thank Monique Smith from Bio Organic Chemistry for her valuable help during the measurements and her explanations about the technique.

Our sponsors:

Retrieved from "http://2012.igem.org/Team:Groningen/volatiles"