A whole genome atlas of 81 Psilocybe genomes as a resource for psilocybin production. (2024)

Version 2 Per recommendation of the Moderator/Editor.

Response to Dentinger and Bradshaw

We appreciate the many helpful comments from Dentinger and Bradshaw. Many of them are on point and the manuscript will be adjusted accordingly. A few comments appear to be a non-addressable given the legal status of live tissue of Psilocybe cubensis. As the reviewers themselves point out, this is not unique to this manuscript. We will address these points as well as we can, given the current legal climate.

1) The potential therapeutic value of psilocybin for SARs-CoV-2 seems a bit hand-wavy.

Fair point. The SARs connections have since strengthened and could in fact use more detail and citations to support. The 9K person clinical trial for Fluvoxamine just completed this month and is showing better efficacy than the vaccines. We have added more references and language to back this association and we have itemized the link as being associative to address the reviewers concerns over this nascent field.

2) This is incorrect. None of these studies proposed either HGT nor convergent evolution for psilocybin biosynthesis INPsilocybespp. HGT has been proposed as the mechanism for the gain of psilocybin biosynthesis in other genera (e.g.,Panaeolus,Pluteus) and convergent evolution has been proposed for the gain of psilocybin biosynthesis in the genus Inocybe. Thus, these statements are inaccurate representations of the published literature and need to be corrected.

This has been corrected. This is an important and valid critique. A prior draft of this manuscript referred to this as ”hallucinogenic mushrooms”. In attempting to be more specific, we erroneously replaced this with Psilocybe spp. Which is too narrow and excludes the relevant families raised by the reviewers.

3) “Given the broad geographic distribution of psilocybin-producing mushrooms and the history of human cultivation and selection for psilocybin-producing strains, these models may not be mutually exclusive. “

We have removed the discussion of human selection.

4) Do the authors mean “non-canonical”? N.B. the term “conical” appears throughout the manuscript where I suspect the term “canonical” is intended.

We have changed the conical/canonical language to “gene cluster” as this appears to be the more agreed upon language describing this recently described region of interest.

5) First, “described” is not the correct term. Second, the sequence in GenBank was derived from a specimen identified asP. tampanensis. Third, the percent identities cited are not sufficient --- percent identity over what portion of the sequence? And was this one sequence the top hit for all three of the ITS sequences derived from the WGS?

We have changed this to “reported to NCBI by”. We agree and clearly state the Rockefeller sequence is from P.tampanensis as you suggest. The listed percentages are across the full-length accession described by Rockefeller (946bp). We have annotated the length of the alignments and included an additional accession from Wesselink et al. from P.mexicana.

6) The degree to which the spore samples apparently consisted of contaminants and in at least one clear case a gross misidentification calls into question the identity, reliability and reproducibility of the results from the other “strains.”

We believe this is an unfair critique as our transparency is being weaponized against us. We could have easily omitted the contaminated sample from the manuscript but believe negative results are informative. It is important for readers to know that contamination does exist in the circulating spore preps and it is important to build tools to detect and screen for them. We should also be careful not to blame the spore providers for this contamination given endofungal bacteria are known to exist in many fungi and we can’t be certain of the source of the contamination (Aspergillus is ubiquitous and our lab handles it). To attack the other sample veracity, because the authors were honest about one obvious contamination event out of 81 genomes is counter-productive and leads to more scientists hiding their negative results to avoid critiques like this.

7) This is one way to determine endogenous content, but other methods to detect contaminant DNAs in metagenomic datasets (e.g., taxon-annotated GC plots) should be explored to corroborate the indirect inference from sequencing coverage (esp. b/c this assumes that target and contaminant sequencing coverage is bimodal, which is not necessarily the case if there are multiple contaminants and/or contaminants are at varying proportions relative to target).

We provided read mapping rates for all libraries. Highly contaminated samples don’t produce 90% read mapping rates unless the contamination is of the same species. This is evident in the assembly statistics we provided for every genome sequenced. Furthermore, assemblies are filtered with tBLASTn to evaluate genes with high amino acid sequence hom*ology to P.cubensis specific genes (PsiK,PsiM,PsiD etc). Any additional minor contamination is thus filtered from this part of the analysis.

Nevertheless, we have added the suggested Taxon-annotated GC plots for the 5 of the P.cubensis libraries and 5 of the non-cubensis libraries per the reviewers recommendations (supplementary figure 1-4). There is some bacterial contamination evident in the P.galindoi sample but as mentioned before, we have filtered the contigs of interest out using tBLASTn for the relevant genes of interest. This isn’t a relevant critique related to the assessment of the contiguity of the Psilocybin gene cluster in question.

8) The authors should be more cautious about this interpretation, as there is no evidence to support the identifications of the strains supplied nor their identity with authentic specimens (e.g., types or others). Who knows if these spores labeled as these species/strains are correct?

We have added additional disclaimers to address the concerns raised over the spore providers veracity. In many ways these 81 genomes best address the very concerns these reviewers present. We were careful to sequence many strains with identical names from multiple vendors so this question could be addressed. There is no other biobank of P.cubensis that has this level of public sequence verification on the nomenclature system.

We’d like to emphasize to the editors and readership that the reviewers correctly point out that the psilocybin legality issue plagues all studies in P.cubensis but this should not be used as a club to discredit manuscripts generating data that might address the very issue. It can even be seen in the reviewers own excellent work which we cite. Below is the methods section from Awan et al.

"Psilocybe cyanescens genome Illumina sequencing
DNA was extracted from fresh whole mushrooms of P. cyanescens by first freezing with
liquid nitrogen and then grinding the tissue into a fine powder with a mortar and pestle in 2X
CTAB buffer following a modified protocol from Doyle (1991) DNA extracts were then
cleaned through a CsCl density-gradient centrifugation and resuspended in TE buffer. DNAs
were sent to Eurofins Scientific (Ebersberg, Germany) for 2 x 150 paired-end sequencing on
an Illumina HiSeq 2500 platform."

No information is provided on how to acquire the samples Awan et al. sequenced. We have at least provided a vendor name and we were careful to collect samples with the same name from multiple vendors in replicates to test the veracity of the spore providers nomenclature. We believe our manuscript does a more thorough job to address such concerns than manuscripts prior as most previous studies simply sequence a few samples without replicates or guidance on spore providers.

While this is not the core focus of our manuscript, we would turn the reviewer’s attention to the sequence and the SNP metrics provided for the samples with the same name but from multiple different spore providers as these metrics are informative to this question. This is better sample veracity than we observe in the cannabis market and we are unaware of any study that has this level of sequence data to authenticate the nomenclature system in Psilocybe.

We will reiterate that the nomenclature in the field is still very nascent and unvalidated. Sequencing studies like this are the first steps required to catalog it.

9) It should be made clear that in Awanet al., it was proposed that convergent evolution explains psilocybin biosynthesis inInocybe, a distant relative ofPsilocybe. On the other hand, Awanet al.did suggest there may be an alternative pathway present in thePsilocybegenomes. Which of these are the authors referring to here?

We have clarified Awan et al. description of convergent evolution.

10) “PsiK exists in the conical cluster on scaffold_7 but also has a close hom*olog on scaffold_1.”

While this may be true, the high potential for contaminant DNAs in the samples sequenced suggests this may be the result of a metagenome assembly. This alternative needs to be ruled out.

We believe this is a misunderstanding. Scaffold_1 is from a peer reviewed Pacific Biosciences P.envy HiFi assembly. This genome is an annotation grade, 32 contig assembly with a N50 over 3.3Mb. We have labelled each scaffold in that paragraph with P.envy scaffold # to help clarify which references we are referring to. P.envy has been annotated and is in NCBI. NCBI filters for such contaminant when importing reference grade genome assemblies and rarely to do bacterial contaminants consist of a 4.5Mb HiFi chromosome (with a telomere) that shares equal coverage to all of the other Psilocybe contigs. Scaffold_1 and Scaffold_7 in the P.envy reference genome have already been screened for contaminants (source).

11) Replicates? And what evidence is there that these are truly biological replicates rather than the same spores being sourced from multiple places? (what’s the likelihood that these companies are all deriving their spores independently?)

Biological replicates from the same vendor are critical to understand how much the sequencing process varies on the same biological sample. The study was careful to select inter-vendor and intra-vendor replicates. This question is best addressed with the data now public with this manuscript. Of note is the very high genotyping concordance between samples with the same name from the same vendor (Malabar). This genotyping concordance is shared with Malabar samples from different vendors (SW and PS). Contrary to this, B-plus samples from different vendors have lower genotyping concordance. This suggests the spore providers are not clones of each other but do have some strains that appear to be clonal with other providers while also hosting some unique genetics under the same name. This data is in Figure 1. A complete genomic audit trail on the vendor veracity is outside the scope of this preliminary analysis but data does currently exist in Figure 1 to address these concerns. All of the FastQ files are in the SRA for people to explore these other questions.

12) Please be more specific wrt what in these previous studies is being cited. At the beginning the authors incorrectly cite these papers as evidence of HGT and convergent evolution of psilocybin biosynthesis inPsilocybe. So what exactly is being cited here?

We have clarified the Family/Genus regarding Reynolds and Awan.

13) Why did the authors not bother to attempt to align these to the other existingPsilocybegenomes? Two other (potentially closely related) genomes are available (P. cyanescensandP. serbica), yet the authors made no attempt to align their reads from the putativeP. tampanensis,P. galindoi, orP. azurescensto them (the latter is a potential synonym ofP. cyanescens). This could help further clarify the source of the low mapping rates to "P. envy" and may provide further insight into the non-canonical hom*ologs of PsiK and PsiM that they report here.

We have already compared the P.envy references to these submissions in McKernan et al.(link). We will reiterate that the gene cluster is intact in those species. Neither of these suggested references (P.serbica and P.cyanescens) have raw reads in the SRA. As a result, they cannot be run through blobtools for Taxon-annotated GC plots (coverage required from a mapped BAM file) to address the contamination problems on the same footing as the other data you are asking for comparisons to. Nevertheless we have mapped the reads from 5 non-cubensis strains to the above references and added a new table to demonstrate these results (Table 1). We also included 3 cubensis strains as a comparison. You can see from this work that P.azurescens is indeed more closely related to P.cyanascens, as the reviewers suggest, but P.tampanensis and P.galindoi are more closely related to each other but more distant to P.cyanescens.

14) Methods: generally thorough enough to be reproducible, but in some parts it reads like a laboratory protocol rather than text written for a manuscript.

We have reformatted this section to read more as a journal method than a lab protocol.

15) Can the authors display the phylogenetic tree differently? As it currently is, it is difficult to interpret clustering because the branch lengths have been transformed in a way that is awkward to read. I would also suggest using an unrooted tree and provide branch support values.

We have added a link to the newick file so readers can redisplay the phylogenetic tree multiple different ways.

We’d like to thank the reviewers for their careful review of the manuscript and for pointing out critical taxonomy considerations.