by Christopher Altamirano, James Moore
In 1825 the human population was estimated to have exceeded the 1 billion mark; a true indication of how technology and innovation was allowing us to grow in number and with longevity. Farms were becoming more industrialized and methods for preserving food and preventing spoilage became paramount for transporting food over great distances. By the early 20th century, methods such as canning, pasteurizing and vacuum sealing were being utilized to prolong food products until they reached their destination. However these preservation methods have proven to be imperfect with many consumers suffering from food-borne illness throughout the past century. Food-borne illnesses are caused by microbial contaminants that include a vast array of parasitic, fungal, bacterial and even viral agents. When any one of these microbial agents is consumed, the agent will replicate quickly inside the body causing a number of medical ailments and in severe cases even death.
Today, we have more stringent government oversight and continuous microbial testing to ensure food safety. These tests include selective media culturing, various forms of microscopy (or spectroscopy), enzyme linked assays, and PCR (or genetic) screenings of food samples prior to being released to the general public. However, even with the multiple detection assays at our disposal, food-borne illnesses are still a public health risk. Reports of bacterial contamination and food recalls of chicken meat, raw spinach and pet treats still resonate in our collective consciousness and have prompted food industries and government agencies to find better ways of detecting the presence of microbial contamination.
With the exception of viruses, most food-borne microbial agents have one common life process; they eat, they grow (and/or replicate), and they produce waste. In fact, all living things do this using a process known as respiration.  Although some microbes can do this in the absence of oxygen (known as anaerobic respiration or fermentation), aerobic respiration describes the process in which energy is released from complex organic molecules (i.e. sugars, fats or proteins) in the presence of oxygen. When this occurs gaseous carbon dioxide and water vapor are released as by-products and the organism loses mass (or weight). The loss of mass not only comes from water vapor being released but more so by the release of carbon from a solid state into a gaseous state in the form of CO2.
To test this hypothesis, we employed the use of loss-on-drying moisture analyzer (Computrac® MAX® 4000XL) with an affixed humidity air pump to keep samples at a controlled humidity. The Computrac® MAX® 4000XL was used due to its ability to maintain a stable temperature, give accurate weight loss readings over an extended period of time, allow for downloadable graphs and data, and its internal plumbing for gas flow (humidity vapor). Although most types of food-borne pathogens consist of bacteria (e.g. Escherichia coli, Salmonella typhi-murium, Clostridium botulinum), we, at Arizona Instrument, chose to use a common Baker’s Yeast, Saccharomyces cerevisiae as our model organism. S. cerevisiae is a relatively fast growing (90 min. doubling time) unicellular yeast that is commonly used as a leavening agent in bread and pastries. It is also innocuous in nature and NOT a human pathogen, but shares a similar physiology to some pathogenic yeasts such as Candida albicans (an opportunistic human pathogen).  In this experiment, we first used an in vitro approach to demonstrate proof-of-concept by measuring weight loss on yeast inoculated potato dextrose agar plates and later tested ‘real’ food (sliced bread) using similar methods.
The outcome of this investigation will offer new insights on a novel loss-on-drying approach to bioactivity measurement for food industries who wish to expand their breadth of microbial testing. This method does not intend to replace any single traditional detection assay, but hopes to add yet another checkpoint in the quality assurance of food safety. Because Baker’s Yeast was specifically used in this study, this approach might be particularly useful in determining the viability of certain yeast cultures for bread or for brewing industries where traditional spectroscopy driven methods (optical density) may only tell of a yeast population density, not necessarily their carbon emission capacity.
Media Preparation and Yeast Culturing
Potato Dextrose Agar (PDA)
Potato dextrose agar (PDA) is a semi-solid media which is commonly used for fungal cultivation in the lab. Agar is a seaweed extract that has consistency of a stiff gelatin mold, but is composed of cellulose in place of proteins (i.e. gelatin).  Because protein is not used as scaffolding, it can be heat sterilized at 121° C and still congeal unlike gelatin which would denature at these temperatures. It is important to note that the agar itself is not consumed when microbes are grown on it, it is simply scaffolding that retains the liquid broth that is mix in with it (in this case PDB).
Potato Dextrose Broth (PDB)
Potato dextrose broth (PDB) is created by stewing peeled potatoes and adding dextrose (which is a simple sugar) in the correct proportions to create a nutrient broth that can be consumed by fungi (including S. Cerevisiae). Unicellular yeasts can grow in this liquid broth in a single cell suspension and is referred to as a liquid culture. PDA or PDB can be purchased through any major biomed distributor but we chose to make it ourselves using a recipe from ‘The mushroom cultivator’ (see reference page for more details)
Pouring Agar Plates
For this experiment, aluminum deep-well plates (5 inch diameter) were used with the outer rim removed (traditional petri plates made from polystyrene would also work) and sterilized at 121° C for 30 minutes. When the molten PDA mixture came out of the oven, it was allowed to cool inside a loosely capped 500 mL Erlenmeyer flask until it reached ~50° C. Once at this cooler temperature, ~30 mL of molten PDA was directly poured into the sterile empty plates in a sterile hood where then each of the plates were covered with a heat sterilized aluminum ‘lid’ (AZI 4KXL waffle pans) to prevent contamination. Once the PDA solidified, the lids and plates were sealed with strips of Parafilm® to prevent excessive moisture loss and stored in a refrigerator set at 4° C.
Liquid culture of S. Cerevisiae
A portion of the PDB was withheld to cultivate a growing stock from a 7 gram packet of freeze dried Fast Rising Instant Yeast S. Cerevisiae (Bakipan®). 100 mL of PDB and ~3.5 gram of freeze dried yeast was added to a heat sterilized 500 mL beaker and mixed with a sterile loop. The culture was periodically agitated every 30 minutes for 2 hours and either used for the experiment or stored in a refrigerator set at 4° C for later cell culturing. 
Inoculating PDA plates with S. Cerevisiae
Once a robust culture of yeast had been created, a flamed loop was dipped into the liquid culture and spread evenly over all of the available surface area on each plate to create a ‘lawn’ of yeast. 3 ‘loop fulls’ were used on each plate in order to supply a sufficient amount of starting population of yeast. All inoculated plates were then inverted to prevent condensation from the lid from falling onto the lawn and disrupting cellular growth (common practice).  See diagram below for a visual representation.
Inoculating Bread slices with S. Cerevisiae
The bread slices used for this experiment were from a Nature’s Own Honey Wheat® loaf. The sliced bread loaf was stored in its original packaging in a refrigerator set at 4° C, but was taken out one hour prior to the experiment to allow the whole loaf to warm up to room temperature. For the control, a slice of bread was placed onto a waffle pan and 5 drops of sterile PDB was placed onto the bread slice (4 drops being placed on the 4 corners of the bread and 1 in the center using a sterile transfer pipette). For the experimental, the same drop placement was performed on a slice of bread but with a homogenous mixture of liquid yeast culture suspended in PDB. (See diagram below drop placement)
Computrac® MAX® 4000XL Programing and Procedure
In order to track weight loss in real-time, twelve 2-hour tests had to be linked to provide an accurate and continuous depiction of weight loss over a 24 hour period. Be sure to connect an Ethernet cable to the appropriate port in the back of the instrument and enable the ‘Web Server’. Enabling the webserver allows the analyst to download the raw data onto a Microsoft® Excel® Spreadsheet at a later point in time for data analysis. Once you enable the optional ‘Web Server’, feature, turn the instrument on and off again to allow the instrument to recognize the new connection. To observe the newly established instrument IP address, select ‘Menu’ à ‘Set Up Menu’à ‘Ethernet Set Up’. Record this IP address for later data analysis (see manual for further details).
Because S. Cerevisae is not a true human pathogen, 37° C (the temperature of human body) is too warm for optimal yeast growth; therefore the temperature setting was dropped to 30° C.  The purpose of using a moisture analyzer was to use it as an incubator to grow the yeast at optimal temperature and measure the weight loss over a fixed amount of time. Test parameters for each individual 2-hour test are listed below:
Sample Size: 25.0 +/- 15.0 g
Idle Temperature: 30° C
Testing Temperature: 30° C
Pan Tare: Standard
Sample Tare: Standard
Ending Criteria: Time, 120 Minutes (2 hour maximum limit per run)
Link Test: Yes
Because each individual test will automatically read as an individual ‘% Moisture’ result, a custom equation was created to track the total percent of weight Loss. The first 2-hour link does not require this custom equation, but the following links will in order to give an accurate real-time reading of weight loss.
With WI1 being the initial weight of the first test in the series and WF being the final weight of the current test.
Computrac® Dry Air Generator (DAG) Humidity Flow
Most food products have a substantial amount of water and because of that have a unique evaporation profile dependent on sample size, distribution, surface area and the environment it is in (humid or dry). To minimize the evaporation for both agar plates and bread slices, a humidifier was created out of a repurposed Computrac® Dry Air Generator (DAG) with ambient air being bubbled through distilled water held in a 500 mL spouted Erlenmeyer flask, and carried past the sample chamber of the Computrac® MAX® 4000XL. Without the modified DAG apparatus, the internal chamber of the instrument read at relative humidity of 10.6%. After a sufficient amount of time (~30 minutes) of humid air being pumped through the chamber, the relative humidity read between 26.5% – 30.1% with the instrument holding temperature at 30°C.
Originally, the MAX® 4000XL was designed for a nitrogen purge function in which highly volatile and potentially explosive samples could be tested in an oxygen free chamber (preventing combustion). However, this set up was repurposed to provide a semi humid environment for the sample and for the yeast growing on it. For further information on the Nitrogen Purge setup, consult the manual. A diagram below shows the humidified MAX 4000XL connected by Tygon® tubing.
MAX® 4000XL Testing
Each test was started when the first link in the 12 linked series was selected and allowed to reach 30° C. Once the temperature was attained, the analyst pushed start, the instrument tared the pan and prompted analyst to add sample within the prescribed sample window (10-40 g). For the PDA samples (both control and inoculated alike), pans were uncovered from their original ‘lids’ and placed upside down onto the waffle pan/pan support on the 4KXL. The instrument recognized the change in sample weight and was within the acceptable weight limits and urged analyst to close lid. Once the lid was closed, the test began and continued for 24 hours.
Data was stored on the MAX® 4000XL instrument as well as continuously being reported to the network. The IP address, determined at an earlier stage, was entered into the address bar of an Internet Explorer 10 Interface and the twelve linked tests were highlighted and ‘downloaded with graphs’. This gives you the raw data in the form of an Excel® spreadsheet. The raw data had to be reoriented by cutting and pasting into a new spreadsheet since the series was reported in the opposite order. The ‘total % weight loss’ equation allows you to observe in real-time on the display screen but is not captured from the download; so a similar custom equation must be used in Excel® to determine the continuous % weight loss for each data point. The equation used was as follows:
The MAX®4000XL reports the sample weight and the rate every ~ 30 seconds. However when a new test begins, the instrument does not immediately recognize that it is a link test and the % weight loss and rate drops back to zero. If the first 30 second data point was left in the series the graphs produced would have drastic dips every two hours upon the onset of each test and would affect the average and therefore was omitted.
Graphing in Excel®
The graphs presented in the Results section of this this paper were created using standard a ‘Scatter Plot’ graph in Excel® taking the averaged ‘total % weight loss’ (calculated from the previous equation) from 3 individual runs and plotted against time (in hours). The 4 dependent variables observed were:
1) PDA (not inoculated) – Control
2) PDA w/ yeast inoculation
3) Bread slice with 5 drops sterile PDB – Control
4) Bread slice with 5 drops liquid culture
The error bars represented in each graph corresponds to the standard deviation taken at each 2-hour interval among the 3 run of each variable.
PDA Weight Loss
The first hurdle in our investigation was to support the theory that ‘bioactivity’ can be measured by weight loss. It was understood from the beginning that whatever food product was placed in the chamber would lose weight from evaporation alone. Therefore, we needed to get a baseline evaporation curve for the nutrient rich PDA plates specifically made as an optimal media for S. Cerevisiae. The linear ‘blue’ line of the graph represented below corresponds to three separate PDA plates void of any S. Cerivisae.
Once the baseline rate of evaporation (more accurately the change in total % weight loss) was determined, it was predicted that if a living organism (i.e. S. Cerevisae) began consuming the dextrose in the media, that carbon dioxide and water vapor would be produced and the agar plate would begin to lose weight at a faster rate that the control plate. Although we cannot necessarily prove that carbon dioxide or water vapors are the culprits of this loss, we can still show that a greater percentage of weight loss occurred at ~8 hrs. into the incubation as compared to the control. As time proceeds, a higher % weight loss is observed in the experimental group. After a 24 incubation period, it was observed that ~ 6.88% weight loss was achieved by the control, whereas 11.21% weight loss was achieved by the inoculated experimental group; a difference of ~ 4.33%. Both variables were each performed in triplicate and averaged. Error bars represent the standard deviation at every 2 hour interval.
Bread Weight Loss
After confirming that weight loss due to bioactivity can be measured under ideal conditions, we turned our focus to a simple yet practical food item; bread. Because yeast is used in the process of bread making, bread inherently has nutrients that yeast can readily use. As mentioned before, each food type has a different water loss profile, and must first be determined to set a baseline to compare against weight loss due to bioactivity (or respiration). Because bread has high moisture content and has a greater surface area than an agar plate, the rate of evaporation proved to be much greater, but eventually plateaus. With the PDA experiment, 3 ‘loop fulls’ of liquid yeast culture was used to inoculate the experimental group which does not add a substantial amount of moisture on to the PDA plate (certainly not 4.33%). However for the bread slice experiment, a loop of liquid culture would not be a sufficient amount of starting culture considering the surface area of all of the pores of the bread. In order to yield a higher number of yeast onto the bread, 5 drops of liquid culture suspended in PDB was added to the bread slice (this also gives the yeast ample moisture to grow in the start). However, 5 drops of liquid culture certainly adds a great deal of moisture to the experimental group so to even the starting moisture levels between the control and experimental, 5 drops of sterile PDB was added to the control slice (no yeast). Over a 24 hour incubation time, the control slice of bread (blue line) losses 26.76% from its initial starting weight while the Yeast inoculated bread slice (red line) losses 28.39%; a difference of 1.63%. This is not as big of a difference in weight loss as compared to the PDA experiment, however considering how much moisture is lost by evaporation alone, we still see a statistically significant trend between the two groups by hour 14 in a 24 hour period. Both variables were each performed in triplicate and averaged. Error bars represent the standard deviation at every 2 hour interval.
Incubation based assays for determining the presence of microbial contamination on food products is not a new technique in food safety screening. In fact, selective media culturing and spectroscopy methods often rely on growing any number of potential forms of contamination in greater number so as to be visibly perceived or detected using light spectroscopy. Using loss-of-weight as an indication for the microbial presence on food however is a novel approach but requires a baseline evaporation curve of each product in question. This approach allows the analyst freedom from using expensive selective media or specialized vessels for optical density (as in some spectroscopy methods). The food product in question can be directly placed into the chamber of the MAX® 4000XL without any processing or culturing techniques. Moreover, this approach allows the analyst to view the rate of growth (via loss of weight) in real time. It is important to note that a 24 hour incubation period (held at 30° C) was selected due to the growth profile of S. Cerevisae (doubling rate of 90 minutes). As mentioned before, S. Cerevisiae is not normally a human pathogen which tend to be bacterial in nature. One prime example of a food-borne bacteria is Escherichia coli (E. coli). Under ideal conditions (37° C), E. coli has a doubling rate of only 20 minutes and if bacterial contamination is suspected a shorter incubation time may be more useful for the analyst than a 24 hour period, however some preliminary work would be required by the analyst to determine this to be true. 
This loss-of-weight detection assay is not a complete assay for microbial detection, but offers another method in the process of microbial determination. This approach assumes there is a high enough microbial load already present in the food to be measured under standard incubation conditions. This loss-of-weight approach will only detect the presence of living organisms on the food in question, it will not differentiate between innocuous and pathogenic organisms. However, if an organism is growing on a food product at 37° C it may be a good initial indication of the presence of a potential pathogen in which other detection assays can be utilized to elucidate.
Although the repurposed Computrac® Dry Air Generator was used as a humidifier, further investigation is required to determine if this apparatus is essential in applying moisture to a growing organism, but was used as a precautionary measure to resist excessive desiccation during testing. This brings into light the possibility of other gasses being used to fill the growth chamber. In this experiment humidified air was primarily utilized, but humidified nitrogen was also used in a separate experiment to observe a change in % weight loss under anaerobic conditions (void of oxygen). However, after several tests, it was determined that the % weight loss of yeast on PDA under anaerobic conditions was not statistically different to the aerobic conditions (data not shown). This may have been caused by the strain of yeast used in this experiment, since many commercial strains have been selected for their resiliency in anaerobic conditions (i.e. within a loaf of bread). It is interesting to note that a mild ‘alcohol’ smell was detected by analysts after these anaerobic tests had been completed. Although this is purely speculation, it is possible for the yeast to have undergone fermentation in which carbon dioxide and ethyl alcohol may have been released instead of carbon dioxide and water.
This data may be particularly useful for members within the baking and brewing industries. Testing the viability of yeast using other methods besides cellular enumeration might be useful since the number or density of a culture is not necessarily an indication of a strong yeast culture. Determining the rate of growth based on % weight loss of different yeast strains might help bakers or brew masters decide on the best strain to use for their products.
Although there may be drawbacks to having a single plate incubator (MAX® 4000XL), the data collected from the growth of an organism offers great insight on how it is interacting with your food product. Moreover, having a stable and controlled environment means that the lid never has to be opened once the test has begun, which prevents fluctuations in humidity and temperature. If it is in the interest of analyst, it is even possible to link a final ‘sterilization link’ at the end of the 24-hour series (depending on the sample being tested). The MAX® 4000XL can attain temperatures as high as 275° C, though a final link set to 121° C would be sufficient to sterilize most food products (NOTE: if an inverted agar plate is being tested, it would be important for the analyst to turn the plate right side up prior to running this test, in which case a linked option would not be recommended.)
Overall, the Computrac® MAX® 4000XL proved to be useful in detecting the presence of microbial growth based on the loss-of-weight of respiration. This method adds one more assay for detecting potential food-borne contaminants in food and gives us insight on how real-time growth occurs on solid and semi-solid food types.
- Kennelly, John J. Feeding 9 billion People, Ninth Annual AIA Conference, Banff March 2013, University of Alberta link http://www.albertaagrologists.ca/files/documents/5662_John_Kennelly_Feeding_9_Billion_People.pdf
- Tauxe R.V. Emerging foodborne diseases: an evolving public heath challenge. Emerg Infect Dis 1997: 3:425-34
- World Health Organization. “Chapter 2 Foodborne Hazards in Basic Food Safety for Health Workers <internet>” (PDF). Retrieved Dec 25 2013.
- American Public Health Association (1992) Compendium of Methods for the Microbiological Examination of Foods 3rd Edition APHA Inc. Washington DC. Retrieved Dec 27, 2013
- Stryer, Lubert(1995). Biochemistry (fourth ed.). New York – Basingstoke:W.H. Freeman and CompanyISBN 978-0716720096
- Fratamico PM and Bayles DO (editor). (2005). Foodborne Pathogens: Microbiology and Biology. Caister Academic Press ISBN 978-1-904455-00-4.
- T. Goekhout, V. Robert, ed. (2003). Yeasts in Food: Beneficial and Detrimental aspects. Behr’s Verlag. P. 322 ISBN 978-3-86022-961-3.
- “Bacterial nutrition”. Microbiology Laboratories, University of Wisconsin. Retrieved Dec 27, 2013.
- Stamets, P. The Mushroom Cultivator: A Practical Guide to Growing Mushrooms at Home. http://www.ehow.com/how_5872701_make-potato-dextrose-agar.html