**1. Introduction**

Members of the cyanobacteria genera *Microcystis, Anabaena, Oscillatoria, Planktothrix* and *Nostoc* are able to synthesize cyanotoxic secondary metabolites such as microcystin, nodularin, cylindrospermopsin, saxitoxin or β-N-methylamino-L-alanine (BMAA), a non-proteinaceous amino acid form [1–5]. Thus, any food or ingredient originating from cyanobacterial biomass and destined for human consumption must be seriously scanned for the presence of these toxins [6]. This is particularly important in the case of Llayta. Llayta is a foodstuff consumed by rural Andean communities since pre-Columbian times [7,8]. The vernacular name Llayta identifies the dry biomass of macro colonies of a filamentous cyanobacterium that grows at Andean wetlands over 3000 m of altitude. The cyanobacterium has been isolated from Llayta colonies (denominated *Nostoc* sp. strain LLA-15) and its taxonomy and biochemical composition has been previously reported [9,10]. After harvesting and sun drying, Llayta can be purchased today at food markets in Tacna (southern Peru) and Arica and Iquique (northern Chile) and used for the preparation of local dishes [11]. Studies on the biochemical composition of Llayta showed that 60% of total amino acids were essential amino acids for humans and 32% of total fatty acids were polyunsaturated fatty acids [10]. To date, no data on the potential for toxin producing in Llayta have been reported, although there are no epidemiological reports that show that consumption of Llayta may cause human illnesses. Based on this body of evidence, we hypothesize

that Llayta available at southern Peruvian and northern Chilean food markets for human consumption is a biomass free of cyanotoxins. To support this hypothesis, we provide genetic and analytic evidence.

#### **2. Results and Discussion**

The target genes *mcyA*, *cyrJ*, *anaC*, *sxtA* and *sxtG* were not amplified in any sample. The *mcy*A gene can be used in the early warning of cyanotoxins with good certainty, but according to Jasser et al. (2017) [12], they preferentially amplify *Microcystis* and *Planktothrix* genera. On the other hand, the *mcy*E gene was only amplified from Llayta DNA. Since *mcy*E gene primer sequences are found in several microcystin producer genera, the detection of this gene in opposition to the absence of *mcy*A gene is reasonable. However, the gene *mcyE* was amplified from natural Llayta DNA only. Interestingly, DNA from strain LLA-15 did not amplify the *mcyE* gene, even though LLA-15 was isolated from the dry biomass of Llayta. The *mcyE* amplicon from Llayta DNA was recovered and sequenced, and the BLASTn at the NCBI data bank showed 96% identity with gen *mcyE* from *Nostoc* sp. 152, a known microcystin-producing strain [13]. Then, the amplified *mcy*E gene could belong to other microcystin-producing cyanobacteria present in Llayta colonies; in fact, the metagenomics of the closely associated microbiome of edible Llayta colonies show an abundance of bacteria from the phylum Proteobacteria and several cyanobacterial phyla dominated by *Nostoc* (38%), *Anabaena* (25%) and *Nodularia* (20%), with minor contributions of *Microcystis* and *Cylindrospermopsis* (unpublished data). In addition, it is important to highlight that molecular screening for cyanotoxins only gives us preliminary data for potential production, being crucial to confirm the data by analytical methodologies. For that reason, after molecular screening, the MC-LR evaluation by HPLC-PDA was performed in Llayta biomass extracts. MC-LR standard injection showed a peak at 8.6 min with an absorption maximum at 238 nm. The Llayta HPLC chromatogram showed a weak signal at 8.79 min with absorption maxima at 238 nm and 390 nm. Then, it was decided to conduct a MS for this weak signal, since this might be evidence for the presence of microcystin in the Llayta extract. As expected, the standard MC-LR microcystin LR showed an RT of 7.21 min and mass fragments of 995.37-977.36-866-599 m/z (Figure 1A). The Llayta sample in LC-MS did not show a signal at RT 7.21 min. The LC signal region with RT 6–7 min was analysed by MS, and did not show the fragment pattern expected for microcystin LR. Together, the evidence supports the notion that Llayta biomass does not contain a microcystin LR-like toxin (Figure 1B). Interestingly, the Llayta LC chromatogram showed a strong signal at approximately 14 min, which becomes a subject for future work, in order to discover which molecules are there.

**Figure 1.** Mass spectra for standards MC-LR and BMAA and Llayta extracts. (**A**): Microcystin LR standard (SIGMA). (**B**): Llayta extract in 50% methanol. (**C**): BMAA standard (SIGMA). (**D**): Llayta hydrolyzate in 50% methanol.

Dominant microcystins produced by *Nostoc* strains are DMadda or ADMadda variants, like the toxic [ADMAdda5] MC-LR (m/z 1009), the [DMAdda5] MC-LR (m/z 981) and the [D-Asp3, ADMAdda5] MC-LR (m/z 1023) [14]. The Llayta spectra (Supplementary Materials: Figures S1–S6) obtained did not contain any of the fragments reported for this MC-LR variants or for DMadda or ADMadda moieties due to the absence of the diagnostic ions at m/z 553 [Mdha-Ala-Leu-MeAspArg + H+] and m/z 627 [Arg-ADMAdda-Glu + H+]. DMAdda (m/z 121.06) and ADMAdda (m/z 163.08) specific fragment ions were also absent. The absence of all those reported [15,16] characteristic fragments confirms the safety of Llayta as a food ingredient with respect to cyanobacteria toxins.

MC-LR biosynthesis is dependent on the expression of a gene cluster that includes genes *mcy*A and *mcy*E and cyanobacteria are not able to produce the toxin if one or more of these genes are lost during evolutionary processes, as has been reported for *Microcystis aeruginosa*, *Microcystis viridis*, *Microcystis wesenbergii* and *Microcystis ichtyobable* [17–20]. A similar explanation can be inferred from our results, in which Llayta DNA amplified the gene *mcy*E but not the gene *mcy*A, and MC-LR was not present in Llayta extracts.

The LC chromatogram and MS spectrum for standard BMAA (Figure 1C) showed that BMAA migrated with an approximate RT of 5 min and yielded fragments of 119-102-88-76-73 m/z, in agreement with the literature [21–23]. In contrast, the Llayta hydrolysate showed a weak signal with an RT 5.5 min (Figure 1D), and its MS spectrum showed a different fragment pattern to that of standard BMAA.

As has been previously documented, cyanobacteria food supplements may be contaminated with cyanotoxins, either microcystins [24] or anatoxin-a [25], and the monitoring of these toxins is fundamental to prevent human intoxications. In our work, extracts of Llayta were analysed and did not show the presence of cyanotoxins. However, these results need further work to confirm the presence or absence of cyanotoxins in cyanobacterial samples used for human consumption, from different geographical regions in Peru and northern Chile, as well the seasonal effect on cyanotoxin production.
