The State of Michigan is in the process of scrutinizing molecular methods and culture based methods for the detection of Total Yeast and Mold (TYM) on cannabis.

They chose to run this study on DiChloro Rose Bengal with Chloramphenicol (DRBC).

The chloramphenicol is an antibiotic. This can be helpful to eliminate the bacteria which can form obscuring colonies in 24 hours while yeast and molds can take 5-10 days.

Unfortunately, chloramphenicol also inhibits the growth of fungi and it inhibits the most dangerous fungi on cannabis (Aspergillus, Penicillium and Fusarium).

To confirm the peer reviewed literature, we acquired organisms from ATCC and plated them on PDA and DRBC.

ATCC glycerol stocks performed better than Microbiologic pellets.
Microbiologic desiccated pellets required more time to grow on DRBC (12 days)

As expected the Aspergillus species either did not grow or took 12 days to grow on DRBC. PDA was able to detect them all. PCR failed to detect Rhizopus oryzae.

Aniger 1-10 dilution 4 days-1

These are some of the most commonly found pathogens on Cannabis. Aspergillus and Penicillium are the most commonly cited.

External AOAC study on PDA vs PDA with CAMP vs DRBC with CAMP

After sharing these data with AOAC, they suggested an independent review. This study evaluated the same cannabis matrix plated on 3 different plates. This was performed by an independent laboratory under AOAC guidance.

1)DRBC with Chloramphenicol (CAMP)

2)PDA with Chloramphenicol (CAMP)

3)PDA alone

You can see a LOG scale more colonies on PDA with CAMP than with DRBC. This demonstrates the carbon sources in DRBC which are known to starve yeast and molds are in fact starving yeast and molds and therefor under count microbial risks.

You can also see a LOG scale more organisms on PDA without selection. This is 100 fold more CFU. Some of these may be bacteria or yeast and molds that are sensitive to Chloramphenicol.

This is important to consider in light of ITS qPCR. ITS qPCR can also vary by organism due to copy number variation in the ITS regions but this copy number variation is not 100 fold variant as you will observe by simply shifting the carbon source used for plating. ITS qPCR can also amplify any organism while only a sub percentage of the organisms can culture. This is known as The Great Plate Count Anomaly (Staley and Konopka, 1985). Even high throughput systems that attempt to culture organisms in multiple different culture broths, fail to exceed culturing more than 14% of the microbial content present in a sample (Connon and Giovannoni, 2002). These highly parallel culture systems are too cumbersome to deploy in routine food testing so a single carbon source or plating type is often used instead. The choice of that plating method will greatly vary making this as a suggested “gold standard” that is undefined.

We performed additional platings of Botrytis cinerea on the 3 different medias (PDA vs PDA-CAMP vs DRBC-CAMP. In the past, we have noticed this very common cannabis pathogen did not plate in 3 days on Rapid YM plates. As a result, MGC designed PathoSEEK TYM test to have independent Botrytis primers to afford better benchmarking for medias which fail to grow this. These primer can be run independently for Botrytis confirmation. We are working on methods to pool these primers into the TYM FAM channel foe use in states that use media that can grow Botrytis. One can see low to no growth on DRBC but excessive growth on PDA. These were grown for 5 days. No colonies were observed on DRBC in 3 days (72 hours). The same volumes were plated on all medias.

Conclusions

These data demonstrate that migration to DRBC as a ‘gold standard’ for benchmarking other TYM detection technologies is flawed. The failure to detect the most dangerous human pathogenic organisms found on cannabis was noted with organisms ordered directly from ATCC. This problem is further exacerbated in real world cannabis samples as these organisms are usually endophytes which live inside the plant. It is difficult to extract Aspergillus from within the cannabis plant and still maintain its viability. In this case, even if the Aspergillus were viable, it wouldn’t grow on DRBC.

The use of chloramphenicol selection in cannabis microbial testing presents clinical risk to cannabis patients and patient safety.

Supplementary Data

AOAC Matrix Study on PathoSEEK demonstrates overcalling with qPCR compared to DRBC (qPCR fails things when DRBC passes them). The same study demonstrates concordance with qPCR and PDA. We also would like draw attention to the Standard Deviation (Sr) seen in plating. It is often 10X higher than qPCR. It is difficult to achieve statistical concordance when the reference methods are this variable. This directly contradicts the concerns voiced regarding qPCR passing moldy cannabis. DRBC will in fact pass more moldy samples that and is now being used as the standard in the state of Michigan.

Below are the PDA with CAMP plates from the comparison study. These grew more 10X yeast and molds than DRBC. The colonies do not look to be bacteria. Whole genome sequencing is under way.

These colonies were all whole genome sequenced and the results of that work can be found here:

PDF version can be found here:

10.5281/zenodo.4759883