**4. Discussion**

#### *4.1. Siderophore Production Revealed by CAS-Assay*

While effectively employed for the identification of numerous siderophore producing bacteria, the CAS assay has a notable limitation; microbial growth may be hindered due to a few factors [31]. Since the metal cation of interest weakly bound to the dye, it was less freely diffusible into the bacteria, resulting in lower availability. In addition, the medium needed to be at pH 6.8 for the indicative colour change to occur successfully. Finally, HDTMA has been known for its slight bacterial toxicity. If an organism had weak Fe transport, required a basic or acidic optimal pH, or was sensitive to HDTMA, it would of had reduced growth. In our experiments, we found that while a range of media compositions could be used, all AAP did have reduced growth on CAS plates, when compared to control. However, since growth did indeed occur, siderophore production could therefore be analyzed.

Other, more high-throughput, alternative methods were considered, but the chosen agarized CAS-assay was most ideal for determining siderophore and metallophore production. Recently, a bulk screening assay for siderophore detection was proposed [41]. However, both the "traditional" and "modified microplate" qualitative techniques described could not be used for our application, since there was an assumption that siderophores were always constitutively expressed, whereas cultures were grown in complex media without manipulating the concentration of any metal of interest. While this may be the case for some strains, the expression/production of most siderophores or other secondary metabolites usually requires induction from an external factor, which can be either the presence or absence of a stimuli [18]. For biologically significant metals, including Fe, Zn, and Cu, the related cation-specific metallophores are expected to be produced only under limiting conditions. In opposite, metallophores that act on V, Te, Se, or other more toxic metal(loid)s are presumably only synthesized when such toxins are present at higher concentrations. Therefore, growth on agarized plates that contained each metal of interest pre-bound to CAS-reagent as a stimulant for metallophore production was our chosen method. Future works are required to test varying concentrations of metal cations to determine which yields more/less production of specific metallophores.

Regarding the Fe-chelation, a range of phenotypes was observed, when detecting siderophore activity after the 5-day incubation (Figure A1). In particular, 4 phenotypes were distinguished among all isolates based on size of clearing/colour change zone. Negative results ( −) had to have bacterial growth, but without a change in medium opacity. The smallest zone of clearing (+), seen previously after prolonged growth [42], was likely not due to siderophore production [43]. Rather, this small aura could be due to high rates of metal uptake from the surrounding medium. Prolonged bacterial metabolism allowed for increased simple metal diffusion into cells, which reduced its amount in the near-by medium, rendering the dye in that narrow area void of metal, turning yellow. Hence, there is a very small <1 mm zone. In comparison, moderate or highly diffusible siderophore release was quite evident, and was segregated into two phenotypic groups. A zone <10 mm compared to a zone >10 mm, where each represented less or more diffusible secondary metabolite, respectfully, and where lower or higher concentrations were produced and captured additional metal cations. In this test, all 4 zonal varieties for Fe-chelation were discovered (Tables 1 and 2, and Figure 2).

Modifying the CAS assay to monitor chelation of elements other than Fe2+ included Mg2+, V3+, Mn2+, Co2+, Ni2+, Cu2+, Zn2+, Se4+ and Te2+ (Figure 4A). The method could be successfully adapted for all selected metal(loid)s, where only Mn2+, Co2+, Ni2+, Cu2+, Zn2+ had been proposed previously [44]. In addition, the assay could be used with variable nutrient and organic carbon concentrations that did not inhibit the activity. Furthermore, we discovered that a wide range of AAP produced metallophores, which bound a variety of metal(loid)s in addition to Fe (Tables 1 and 2). Some AAP had siderophores that specifically bound Fe only, including strains E1, E4(1), RB3, NM416, AM27, CK155, BC100, Z24, Z39, and J05. Most AAP with highly diffusible siderophores, and >10 mm zones on Fe-CAS plates, had activity towards a large variety of metal(loid)s. Few strains, including SS56 and

P4 preferentially bound metal(loid)s other than Fe. Since all 9 additional cations Mg2+, V3+, Mn2+, Co2+, Ni2+, Cu2+, Zn2+, Se4+ and Te2+ could be bound by metallophores, future work may consider a broader range of cations and the extent of metals that can be exogenously chelated. To determine if specific metal(loid)s were more readily chelated than others, all positive CAS assay results were tallied for each, Figure 4B. Here, multiple +++ represented strong production or significant interaction, and could be compared to weaker reactions such as +, or ++. In this way, Fe was the highest cumulative acquired cation, with the activity ranked as Fe > Zn > V > Te > Cu > Mn > Mg > Se > Ni > Co. It would appear slight variation in cation size, from Fe (55.85) to Co (58.93), had a strong impact on activity, where Fe was most frequently captured, and Co the rarest. Indeed, since strongest reactions existed for Fe and Zn, with less reactivity found for metals in between the sizes of these two within the periodic table, specific mechanisms likely existed to capture either metal, inferring cation specificity for each metallophore. Further analysis will be required to see if the broad range metal(loid) acquisition is due to the production of a single siderophore that reacts with a variety of metals, or the bacteria produced specific metallophores for each. AAP metallophore production could be explained by requirement of trace elements and the need of toxicity prevention. Both Fe and Zn are biologically necessary and commonly limited or unavailable in dissolved forms, therefore acquisition via specific siderophores would be an asset. The 3rd and 4th highest captured metal(loid)s were V and Te, which have known toxic properties, and could have been sequestered as a result of a protection mechanism alone. Bound to a metallophore, these cations would be too large to freely diffuse through outer membranes and be restricted from entering the cell, to prevent any toxic influence. The remaining 6 had reduced activity, likely because they are less toxic and easier available in microbial environment, and therefore less metallophores could be expected.

**Figure 4.** Variant metal(loid)s tested in CAS assay. ( **A**) Chosen elements highlighted in yellow. (**B**) Tallied number of positive results from Tables 1 and 2.

#### *4.2. Environmental Distribution of Siderophore Producers*

In relation to the origin of isolation a few patterns were observed. Collections of AAP strains that originated from hot springs, freshwater lakes, and biological soil crusts all had a high proportion of siderophore producers. In opposition, the isolates from marine, meromictic lake, and saline spring habitats produced less, or no siderophores. Rather, they seem to rely on sufficient metal uptake directly from the local microenvironment by diffusion. The main differentiating factor in this case is the requirement of NaCl. It appears that AAP capable of halotolerance or halophilic growth do not produce siderophores of equal activity or quantity as bacteria that do not depend on NaCl for growth. This

correlation may be due to each strains' reaction to osmotic pressure. Cells that are adapted to tolerate higher levels of NaCl will likely have additional cation pumps to survive naturally and resist the high levels of solutes in a saline environment. As seen elsewhere, cation pumps can be non-specific, where many are capable of removing several toxic cations from the cytoplasm of microorganisms [45]. In comparison, freshwater AAP are comparably less osmophilic, and would therefore have less need of copious cation pumps in membranes. Therefore, they are more likely to evolve defensive mechanisms that modify the local environment to suit their needs, including the production of external small molecules that would render cations less diffusible.

#### *4.3. Phylogenetic Diversity of Fe-Chelating AAP*

Comparing phylogenetic diversity and the production of siderophores on CAS plates by AAP delivered a few notable trends (Figure 2). Broadly, no studied representatives of the order *Rhodobacterales* or *Hyphomonadaceae* had siderophores, as no or <1 mm zone of colour change was present. The *Acetobacteraceae* that were in closest relation to known type species of AAP, strains RB-3T and CK155 had siderophores that only chelated Fe as discussed above. In comparison, strains P40 and J01 were more genetically distant from known type species of AAP, and also produced significant zones of clearing, ≥10 mm for all metal(loid)s tested, signifying their own group. The predominantly strong expression of siderophores by AAP among the *Methylobacteraceae* warrant further study as most type species in this clade have not been previously recognized as phototrophs. Of note, previous research had found that isolates relating to *Methylobacterium mesophilicum* and *M. extorquens* produced siderophores, but the activity had not been linked to aerobic anoxygenic phototrophy [46]. One clade of AAP which related by 98.7–99.1% 16S rRNA sequence similarity to *Bosea lupini*, including strains P13, SS335, and SS63 were all capable of strong production of metallophores, >10 mm zones for multiple metal(loid)s. In comparison, strains P4 and SS56, which have 99.6% relation to *M. phylloshaerae* and 99.6% to *M. branchiatum*, respectively, both had stronger reactions against metal(loid)s that were not Fe (Figure A2). This activity may be explained through the findings of related works, where methanotrophs produced methanobactin, a chalkophore, which is a Cu specific metallophore [15]. The isolates we have tested may indeed possess similar mechanisms as they sequester Cu strongly, as well as Mg, V, Mn, Zn, and Te more favorably than Fe.

*Sphingomonadaceae* could be separated into 3 groups, as those related to *Citromicrobium* did not produce siderophores, *Sphingomonas* relatives produced some siderophores that were predominantly Fe specific, while relatives of *Blastomonas* could produce siderophores that acted on all metal(loid)s tested. A few *Sphingomonas* relatives had been previously found to produce siderophores against Fe [47], but not other metal(loid)s. The *Blastomonas/ Erythromonas* grouping was of particular interest as most representatives revealed strong metallophore production against all 10 cations tested. Our results corresponded well with previous analysis of *E. ursincola*, strain KR99, which had very high resistance to V, Te, and Se oxides, internally reducing them to elemental states [48]. Since KR99 can both acquire Se, Te, and V via metallophore activity (Figure A2), and internally reduce metal(loid) oxides, it appears to require them in reduced form for some reason. Future study of such associations will determine if *E. ursincola* sequesters these cations as a protective measure, or uses them for some metabolic purpose. In comparison, the *Erythrobacteraceae* were not as concisely divided as other families, where those closest to *Erythromicrobium* had siderophores, but most *Porphyrobacter* had very small <1 mm or negligible zones. An exception was strain BE100, which branched distantly from its nearest relative *P. colymbi* (Figure 2), and showed a significant presence of metallophores, ~10 mm zones for all metal(loid)s except Te. Finally, strain EG19 that hailed from γ- rather than <sup>α</sup>-*Proteobacteria*, had moderate siderophore production for both Fe and Zn, <10 mm. While many siderophores have been discovered as products of bacteria within the γ -proteobacterial clade (Table S1), none have been documented as highly pigmented.

#### *4.4. Analysis of the Brown-Coloured Siderophore*

Gel purification of brown pigment produced by *C. halotolerans* (Figure 3), revealed that while CBB bound to proteins remained in the gel, both unbound CBB and brown pigment were released during destaining process. Comparing lanes 2–5, the sample prepared by resin concentration clearly contained brown pigment, but also accumulated proteins smaller than 26 kDa. The use of the 3 kDa cut off spin column did indeed remove these contaminants, as shown in lanes 6–8. Since the small brown pigment passed through the spin column, was purified on gel electrophoresis, and maintained activity on CAS plates, these tests confirmed that it acted as a siderophore. Furthermore, these procedures established the brown compound to clearly be under 800 Da, and approximated to be around ~341 Da when correlated to the ladder during TRIS-tricine gel electrophoresis (Figure S1). Both the estimated small size and the brown appearance of the siderophore synthesized by *C. halotolerans* were useful for its tentative identification. The most comparable small molecule described in literature was rhodotorulic acid (Table S1), a 344.4 Da siderophore that was pigmented red when bound to Fe [49]. However, this acid has only been naturally found in yeasts including *Rhodotorula pilimanae*, with no known bacterial producers [50]. With that in mind, hydroxamic acids are produced by both bacteria and fungi [51,52], and therefore similar secondary metabolites can be expected in other species. In addition, the colour disparity, red compared to orange-brown, may indicate an altered structure among siderophores produced by *R. pilimanae* and *C. halotolerans*, respectively.

Since *C. halotolerans* hails from the γ-*Proteobacteria*, comparisons must be drawn between its siderophore and those produced by other species in the γ-subclass. The most similar in size was acinetobactin, a 346.4 Da molecule from *Acinetobacter baumannii* expressed using the operon containing *basABCDEFGHIJ*, *bauABCDEF* and *barAB* genes [53]. *C. halotolerans* genome, published within the One Thousand Microbial Genomes Phase 4 (KMG IV) project by the DOE Joint Genome Institute, submitted online in 2019 with accession number PRJNA520330 [54], contained neither similar genes nor was the operon present, while using very low homology search. Due to the divergence between *C. halotolerans*' pigmented siderophore and *A. baumannii*'s lack of colour, and the absence of similar genes, we assume that acinetobactin was not the siderophore of *C. halotolerans*. Further structural analysis will be necessary to confirm the structural identity of this novel compound.
