In 2015 we published a microbiome survey of medicinal cannabis flowers. We followed up this study with another publication in 2016 in the open access and open peer review platform F1000. Both of these studies demonstrated widespread prevalence of Penicillium fungi on cannabis. These studies focused on cannabis available in Massachusetts and it was unclear if they would be representative of cannabis grown in other jurisdictions. Punja et al. settled this debate detecting many of the same microbes on cannabis in Canada and went on to catalog the frequency of the pathogens. The highest frequency pathogens were penicillium species. Both Plating and PCR/DNA Sequencing were cornerstones to all of these studies.
In 2020, a laboratory in Michigan was concerned our TYM qPCR assays was failing to detect visible mold on their cannabis. Many of the most visible molds on cannabis do not culture (Powdery Mildew is the most common) and many of the most dangerous molds in cannabis are endophytes and usually not visible (Aspergillus and Penicillium).
We have not focused our efforts on detecting mold that one can detect visually but instead, we have focused our efforts on detecting the most clinically relevant molds that have reported fatalities in the clinical literature. Endophytes like Aspergillus, Penicillium and Fusarium are at the top of this list. It is worth noting that when cannabis has visible mold, it usually has invisible mold as well as Powdery mildew infections tend to invite other secondary infections into the plant. So it is not uncommon to have Powdery mildew infected plants with visible mold on the surface and to detect colonies with plating but sequencing those colonies reveals a different microbe than what was visibly seen on the plant. It is also quite common to have visible mold on the plant and not get colonies on plates due to the mold being uncultureable. Over 95% of microbes are believed to be unculturable.
A Michigan lab detected this mold on a 3M Rapid TYM Petri film and since they cannot ship us the flower it was derived from, they shipped us the colonies. We proceeded to sequence the genomes of the colonies with Illumina Whole Genome Sequencing. We have made this data public for full transparency.
SPADES Assembly >NODE_2_length_8345_cov_84.091074 has a BLAST hit to an ITS region.
The ITS regions are multi-copy in the genome and provide the high coverage contigs (84X) that match Penicillium brevicompactum seen in the above BLAST alignment. It is important to recognize that we were shipped a culture. We were not shipped the actual flower samples (federal law prohibits this) and the culture may have changed the representation or sampling of the original flower sample that had failed in Michigan. Penicillium brevicompactum is an oil producing endophytic filamentous fungi. These properties can complicate the sampling as the organism is rarely uniformly distributed in aqeuous culture like a mono dispersed bacterial culture.
In our experience the largest source of discordance between culture and molecular methods for TYM testing is related to homogenization and sampling filamentous fungi. Filamentous fungi clump in solution and have multiple nuclei within a cell where each nuclei is the infective unit and can form a colony when homogenized. Excessive homogenization is required to access endophytes but also may alter viability of the organisms. qPCR counts nuclei and each nuclei usually has 10-30 targets (ITS regions) making the detection very sensitive and resistant to sheared DNA. ITS assays can often reach sub CFU sensitivities.
The sampling technique used for plating is usually not the ideal sampling technique used for qPCR. Plating cannot be used to quantitatively survey endophytes. Accessing endophytes requires one lyse open plant cell walls. These conditions also lyse open bacteria and fungal cell membranes and thus destroys their viability but leaves their DNA intact. In cases where plating is picking up microbes that qPCR is missing and you have confirmed sequence and evidence of your primers amplifying the organism, the culprit is usually the lysis conditions or sampling ratio used. The lysing conditions can easily monitored with Internal plant DNA controls which monitor the effective of cannabis cell wall lysis.
The sampling requires a bit more attention. For example, imagine you grow 10ml of culture and you place 1ml of that culture onto a Petri dish. You have subsampled 1:10 and will have a hard time counting samples at 1 CFU/sample as your sampling rate will make it likely that you missed the 1 CFU in 10 mls. 9 out of 10 times you sample 1ml from 10mls, you will miss the 1 CFU present.
With qPCR, this ratio is often larger (1:200 subsampling common) as smaller volumes are used in qPCR. This can be optimized if labs centrifuge or enrich their samples to compensate for the subsampling. DNA purification also acts as a mechanism to concentrate DNA into smaller volumes than can be stuffed into small qPCR reactions.
This is one line of ongoing investigation at MGC to improve qPCR to TYM plating concordance. Increasing the volume being surveyed in qPCR is likely to result in higher concordance with clumping filamentous fungi.
There are other sources of discrepancies with qPCR and Plating that we have address with Live-Dead PCR techniques. One of the deficits of plating is that one cannot aggressively homogenize samples without destroying the viability of the microbes, thus endophytes are under-surveyed. This deficit is often thrown at PCR as being too sensitive as it can detect DNA from even irradiated samples and often fails too much cannabis product producing signal from sterilized products. By leveraging DNAses, we can eliminate DNA that is free floating from DNA that is internal to a cell. We have presented on this work at CannMed and released products to assist in Live-Dead qPCR in the cannabis field. These methods help to tighten qPCR to TYM concordance but are only as reliable as the sampling methods utilized.
In summary, we have published methods that demonstrate qPCR and TYM plating concordance but these studies were limited to samples derived in Massachusetts. We are expanding these studies to investigate the impact of sampling and homogenization and the impaction Robustness.