Discussion

Hup A，a natural acetylcholinesterase inhibitor for treatment of Alzheimer's disease，is the SM of both Huperziaceae plants and endophytic fungi.After induction by the extracts of H.serrata，Hup A production of hy4 and Cg01 increased.Relevant DEGs were mostly clustered in carbon and nitrogen metabolism and in nucleolus，ribosome，and rRNA processing.It is reported that primary carbon and nitrogen metabolism could influence secondary metabolism(Drew and Demain，1977).Generally，the presence of rapidly utilized carbon and nitrogen sources inhibits secondary metabolism(Drew and Demain，1977).Some SMs are produced when the rapidly used carbon and nitrogen in the medium are exhausted(Ward and Packter，1974；Kennel，1977).In this study，most upregulated genes were in the DEGs enrichment KEGG pathway“Glycolysis/Gluconeogenesis”.Therefore，the rapid use of glucose may be of benefit for the biosynthesis of the SM Hup A.In addition，the action of primary metabolism on secondary metabolism might be the result of precursor activity or an even more important role，i.e.an inducer for biosynthesis of the SMs(Drew and Demain，1977).One example involves tryptophan as a stimulatory precursor of ergoline alkaloid in Claviceps(Robbers and Floss，1970；Vining，1970).In the DEGs enrichment KEGG pathway，most genes involved in nitrogen metabolism were downregulated.It is possible that the nitrogen metabolites were the precursors or inducers for Hup A biosynthesis，so it is repressed in the amino acid metabolism pathways“Methionine metabolism”，“Tryptophan metabolism”，and“Glutamate metabolism”.

Ribosomes，serving as the site of biological protein synthesis，are important cellular organelles that can respond to nutrition level and regulate growth rate(Gutteridge et al.，2010).Ribosomes also play a pivotal role in SM biosynthesis by regulating relevant gene expression.Therefore，the rate of ribosome processing could influence the protein synthesis ability and the synthesis of SMs(Ochi et al.，2004).“Ribosome engineering”，proposed by Ochi，could activate or enhance the production of SMs by targeting S12，RNA polymerase，and other ribosomal proteins and translation factors(Ochi，2017).In this study，the addition of the inducing extracts of H.serrata may act on ribosomes or RNA polymerase，which results in a similar ribosomal stress mechanism，and consequently synthesis of the SM Hup A.

The biosynthetic pathway of Hup A was ambiguous not only in plants but also in fungi.Most of the proposed biosynthetic genes in endophytic fungi were only characterized and expressed in vitro，such as LDC，CAO，and PKS genes(Sun et al.，2012；Yang et al.，2016；Xu et al.，2017).Previous studies could not establish the direct relationship between these genes and Hup A biosynthesis.Zhang et al.over-expressed LDC and CAO genes in the endophytic fungus C.gloeosporioides ES026 and some of the over-expressed transformants produced higher Hup A than the wild type(Zhang et al.，2015).However，there is no repeat in the Hup A analysis of these transformants fermentation.In the current study，there was no significant effect on Hup A production in the mutant when the single LDCgenes，CAO genes，and PKSgenes of C.gloeosporioides Cg01 were knocked-out.However，when we knocked-out the Cg LDC1 and Cg LDC2 genes simultaneously，Hup A production increased significantly.In P.polonicum hy4，the expression of the LDC gene Pp LDC，the sole LDC gene discovered，decreased after induction.In addition，we found an interesting phenomenon in analyzing the expression of homologous genes between P.polonicum hy4 and C.gloeosporioides Cg01.For a pair of homologous genes that encoded lysine biosynthetic regulatory proteins or the fungal specific transcription factor(hy4_05232-RA Cg01_07009-RA)，expression was down-regulated in both P.polonicum hy4 and C.gloeosporioides Cg01 after induction.We suspect that Hup A synthesized in endophytic fungi is likely different from the mechanism in plants.Hup A and lysine may be end products of a branched biosynthetic pathway，with some shared pathway or precursors，like penicillin formation in Aspergillus nidulans(Busch et al.，2003).Moreover，a recent paper reported on the biosynthesis of pipecolic acid(Pip)of H.serrata，a non-protein amino acid containing a piperidine ring，which is the biosynthesis precursor of piperidine alkaloids(Xu et al.，2018).This report demonstrated in vitro that lysine aminotransferase HsAld1 can transfer amino groups from L-lysine to pyruvate.The intermediate 1，2-dehydropipecolic acid(1，2-DP)then isomerized to 2，3-dehydropipecolic acid(2，3-DP)and was reduced to Pip by lysine cyclization deamination enzyme HsSard4.Therefore，it is most likely that Hup A could be biosynthesized via other pathways.(https://www.daowen.com)

The origins of the SMs in endophytes have been the source of debate.Some investigators suggested the source of fungal SMs as the host plant due to inconsistent and unsustainable production over several generations(Venugopalan and Srivastava，2015).However the discovery of the SM-synthesizing fungi isolated from non-SM producing plants indicates that fungi could be the source(Flores-Bustamante et al.，2010).We also found that the endophytes isolated from non-Hup A producing plants could produce Hup A with extracts of H.serrata，contradicting the idea of fungal SM being a host plant adduct.Regarding the SM-biosynthesis pathway in endophytes，some studies suggest that HGT may have taken place between plants and endophytes during their long period of association，while the others hypothesize that plants and endophytes may have co-evolved parallel pathways to produce SM(Venugopalan and Srivastava，2015).Most of the identified fungal Taxol-biosynthesis genes were found to be homologous to their plant counterparts(with＞96% sequence similarity)，which strongly supports the HGT theory(Staniek et al.，2009；Zhang et al.，2009).However，the sheer complexity of the Taxol biosynthetic pathway，which involves more than 20 steps localized to different subcellular compartments，with genes possibly scattered over different plant chromosomes，has raised doubts over the possibility of transfer of the entire pathway from the host plant to the endophytes.The case of gibberellin biosynthetic pathways which show several differences in fungi and higher plants might support the co-evolution theory rather than HGT theory(Hedden et al.，2001；Tudzynski，2005).Using our methods of homology and DEGanalysis with the knock-out validation，it was difficult to find the genuine Hup A-biosynthesis gene in fungi.Therefore，we supposed that the Hup A-biosynthesis genes in fungi might co-evolve with the plant rather than being acquired through HGT.

The epigenetic regulation of chromatin is a complex process and certain histone modifications can play dual roles as they can have different effects depending on the physical location of the chromatin stretch.For example，in Saccharomyces cerevisiae，H3K4 tri-methylation is a permissive mark for gene transcription in euchromatic regions but is also required for gene silencing at mating type loci and sub-telomeric regions(Bryk et al.，2002；Mueller et al.，2006).Deletion of ccl A，a yeast Bre2 ortholog involved in H3K4 methylation，activated the expression of cryptic SM clusters in A.nidulans.The deletion strain produced several additional metabolites，among them monodictyphenone(MDP)，emodin，and four emodin derivatives not previously observed in A.nidulans(Bok et al.，2009).In general，histone deacetylation is associated with gene silencing(Bannister and Kouzarides，2011).However，there is increasing evidence that HDACs are also required for gene activation(Shwab et al.，2007；Lee et al.，2009；Li et al.，2011).Deletion of the A.fumigatus HDAC hdaA gene increased the production of several SMs but decreased production of gliotoxin whereas over-expression hda A increased production of gliotoxin(Lee et al.，2009).In this study，the expression of HDAC gene CgSir2～6 was upregulated after induction.When it was knocked-out，Hup A production of the mutants decreased as expected.However，expression of the HMT gene CgClr4 and the HDAC gene CgClr3 was down-regulated in DEG analysis.Surprisingly，deletion of CgClr4 and CgClr3 did not result in the increased accumulation of Hup A but in a decrease.A similar effect was observed for Ff-Hda1 in Fusarium fujikuroi and cphos2 in Claviceps purpurea：both deletion and overexpression of the HDAC genes led to the same accumulation of SMs(Studt et al.，2013).These results are representative of the complexity of the epigenetic regulation in SM biosynthesis.HDAC deacetylates not only the histone but also transcription factor，where deacetylation is closely related with recruit HAT.When HDACs were knocked out，the transcription factor could not be deacetylated，and consequently could not recruit HAT to acetylate the histone located adjacent to the Hup A biosynthesis gene，leading to the decoupling of pathway-specific transcription factor and biosynthesis genes(Studt et al.，2013).When HDACs were downregulated，the histone adjacent to the Hup A biosynthesis gene was acetylated，and Hup A production increased.In addition，HDAC Sir2 and Clr3 are reported to be required for Clr4 activity with telomeric or centromeric heterochromatin assembly(Alper et al.，2013).In regulation of Hup A biosynthesis，whether CgClr4，CgClr3，and CgSir2-6 function together remains unknown.