*3.2. Genetic and Phylogenetic Analysis, and Species Identification*

As trichothecenes may well characterize the mycotoxin profile of FIESC members, the presence in A1021B of genes involved in biosynthesis of these compounds was investigated by PCR as previously described [15]. Amplicons of all the selected regions (*TRI-1*, *TRI-4*, *TRI-5*, *TRI-8*, *TRI-11*) were detected (data not shown), indicating that strain A1021B actually holds the genetic features to produce these mycotoxins. Nevertheless, the related SMs were not detected in any of the culture conditions used in this study.

Even if the genetic data matched with the hypothesis that A1021B might belong to the FIESC, a different indication resulted from the phylogenetic analysis, conducted using concatamers of ITS, *TEF1* and *CAL1* sequences previously employed in the characterization of this species complex. In this experiment, a strain of *F. concolor* was used as the outgroup [41]. Interestingly, A1021B clustered with the latter instead of any of the several identified or unidentified FIESC members (Figure 3). Nevertheless, a new BLAST search in the NCBI database based on *TEF1* sequences did not yield a consistent homology with the available strains of *F. concolor*.

**Figure 3.** Maximum likelihood tree inferred from ITS-TEF1-CAL1 concatamers. Phylogenetic analysis including A1021B, FIESC members and *F. concolor* as outgroup inferred using the maximum likelihood method (MEGA7). The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed.

The hypothesis that strain A1021B represented a novel taxon could explain such discrepancy. However, a subsequent BLAST search carried out in December 2018 revealed an unexpected 100% homology with a series of *TEF1* sequences from the species *F. babinda* [49], which were made available in October 2018 after another notable taxonomic revision [45]. Following this finding, another phylogenetic tree including isolates of *F. concolor*, *F. babinda* and FIESC was generated where A1021B clearly clustered with the strains of *F. babinda* (Figure 4).

**Figure 4.** Maximum likelihood tree inferred from TEF1. Phylogenetic analysis including A1021B, strains of FIESC, *F. babinda* and *F. concolor* inferred using the maximum likelihood method (MEGA7). The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed.

#### **4. Discussion**

A recent study [45] demonstrated that *F. babinda* represents the correct identification for many strains previously ascribed to *F. polyphialidicum*. This species, on the other hand, has now been reported as a synonym of *F. concolor*, deserving priority in taxonomy as an older accepted species name [50]. While *F. concolor*/*polyphialidicum* is known as a typical trichothecene producer [28,45], the mycotoxin profile of *F. babinda* seems to be centered on FA, and no clues of trichothecene biosynthetic abilities were detected in the limited assessments carried out so far. An analysis concerning the genetic basis for trichothecene synthesis in a single strain of this species (NRRL 25539) also provided negative results [45]. The same study reported that strain NRRL 25539 has the gene clusters for the production of some compounds (enniatins, fusarin, fusarubin), which, however, were absent in our cultures.

In this work, metabolomic analysis confirmed that in axenic cultures strain A1021B basically produced FA and some known compounds. Interestingly, bikaverin was found to accumulate mainly in CDB cultures, where the carbon source is represented by sucrose, in consistency with a previous report that the availability of this sugar stimulates bikaverin production in vitro [51]. Molecular data indicated the presence of trichothecene biosynthetic gene clusters, but they were not expressed under the culture conditions we tested, thus making A1021B divergent from strain NRRL 25539 [45]. Considering that *F. babinda*, which formerly had been reported only from Australia, turned out to have a worldwide diffusion [45], and that *F. polyphialidicum* was described as a typical producer of type-A trichothecenes [28], our finding highlights the need for more exhaustive investigations on the mycotoxin profile of this emerging species. In this respect, an assessment concerning occurrence of

*TRI-5* in *F. equiseti* detected this gene sequence in 50% of the examined strains only [36], confirming previous evidence of uneven production of trichothecenes in this species [52].

Recent evaluations of the mycotoxin-producing ability indicate that *Fusarium* phylogenetic relationship may vary, and non-conforming strains, new species or lineages often result after the exploration of new ecological contexts, particularly those involving endophytic fungi [53–55]. In fact, an intriguing ability to synthesize unexpected SMs can be ascribed to endophytes, which are able to establish physical contacts and eventually interact through horizontal gene transfer (HGT) with both plants and other microorganisms living in this particular ecological niche [56,57]. Indeed, ecological proximity has been considered to favor HGT [57].

In fungi, genes coding for the synthesis of SMs are typically adjacent to one another in clusters of co-expressed genes, including a core gene responsible for the synthesis of a basic structure, and side genes which control chemical modifications, transport, and regulation [58]. Biosynthesis of FA, bikaverin, culmorin and trichothecenes is governed by polyketide synthases, large multi-domain enzymes that catalyze sequential condensation of simple carboxylic acids. A few hundreds of gene sequences involved in the biosynthesis of polyketides have been detected in *Fusarium* spp., which corresponded to 67 clades in a phylogenetic analysis, where each clade refers to distinct products. This analysis also pointed out a genetic potential to synthesize compounds which are the same or similar to those known to be produced from other fungi, but not reported in *Fusarium* so far [59].

From an evolutionary viewpoint, HGT of gene clusters regulating mycotoxin biosynthesis is theoretically supported by the reasonable inference that clustering confers a selective advantage to the cluster itself [56,60]. In addition, the hypothesis that the *TRI-5* gene cluster may have spread among unrelated fungal species through HGT has already been advanced in the past [61]. HGT was also indicated as the means of transmission of a 5-gene cluster presiding over the synthesis of bikaverin from *Fusarium* to *Botrytis cinerea* [51], and as a more general evolutionary mechanism in *Fusarium* [62]. Moreover, it has been demonstrated that transfer of lineage-specific genomic regions occurred in *Fusarium*, including even entire chromosomes up to more than one-quarter of the genome, and involving genes related to pathogenicity. These were effective in converting pathogenic strains into non-pathogenic ones, and were possibly responsible for the emergence of new pathogenic lineages [63]. Therefore, natural ecosystems are recognized to play a role as reservoirs of novel crop pathogens with a meaningful impact on disease management and biosecurity [64].

#### **5. Conclusions**

In this work, we reported a case-study investigating the taxonomy and SM production in the endophytic *Fusarium* strain A1021B. As a consequence of the ongoing updates in the phylogenetic relationships of *Fusarium* species, the analyses of mycotoxin profile and selected gene sequences lead us to identify this isolate as *F. babinda*. Our findings support previous observations that SM production in axenic cultures by *Fusarium* strains does not necessarily conform to genetically based analyses, and that this limitation could be overcome in vivo where interaction with the host plant or other endophytic microorganisms may result in the activation of silent genes.

Besides sequences deposited in GenBank, strain A1021B is available on request for inclusion in phylogenetic and metabolomic studies.

**Author Contributions:** Conceptualization, R.N. and F.V.; methodology, G.M., R.M., A.S., N.L. and F.V.; writing—review and editing, G.M., R.M., F.V. and R.N.

**Funding:** The research activity of F.V., A.S., R.M. and G.M. was funded by MIURPON [grant number Linfa 03PE\_00026\_1; grant number Marea 03PE\_00106]; MIUR-GPS [grant number Sicura DM29156]; POR FESR CAMPANIA 2014/2020- O.S. 1.1 [grant number Bioagro 559]; MISE [grant number Protection F/050421/01-03/X32].

**Conflicts of Interest:** The authors declare no conflict of interest.
