From Laboratory towards Industrial Operation: Biomarkers for Acidophilic Metabolic Activity in Bioleaching Systems

In the actual mining scenario, copper bioleaching, mainly raw mined material known as run-of-mine (ROM) copper bioleaching, is the best alternative for the treatment of marginal resources that are not currently considered part of the profitable reserves because of the cost associated with leading technologies in copper extraction. It is foreseen that bioleaching will play a complementary role in either concentration—as it does in Minera Escondida Ltd. (MEL)—or chloride main leaching plants. In that way, it will be possible to maximize mines with installed solvent-extraction and electrowinning capacities that have not been operative since the depletion of their oxide ores. One of the main obstacles for widening bioleaching technology applications is the lack of knowledge about the key events and the attributes of the technology’s critical events at the industrial level and mainly in ROM copper bioleaching industrial operations. It is relevant to assess the bed environment where the bacteria–mineral interaction occurs to learn about the limiting factors determining the leaching rate. Thus, due to inability to accurately determine in-situ key variables, their indirect assessment was evaluated by quantifying microbial metabolic-associated responses. Several candidate marker genes were selected to represent the predominant components of the microbial community inhabiting the industrial heap and the metabolisms involved in microbial responses to changes in the heap environment that affect the process performance. The microbial community’s predominant components were Acidithiobacillus ferrooxidans, At. thiooxidans, Leptospirillum ferriphilum, and Sulfobacillus sp. Oxygen reduction, CO2 and N2 fixation/uptake, iron and sulfur oxidation, and response to osmotic stress were the metabolisms selected regarding research results previously reported in the system. After that, qPCR primers for each candidate gene were designed and validated. The expression profile of the selected genes vs. environmental key variables in pure cultures, column-leaching tests, and the industrial bioleaching heap was defined. We presented the results obtained from the industrial validation of the marker genes selected for assessing CO2 and N2 availability, osmotic stress response, as well as ferrous iron and sulfur oxidation activity in the bioleaching heap process of MEL. We demonstrated that molecular markers are useful for assessing limiting factors like nutrients and air supply, and the impact of the quality of recycled solutions. We also learned about the attributes of variables like CO2, ammonium, and sulfate levels that affect the industrial ROM-scale operation.


Introduction
The Chilean copper industry will be mostly based on the exploitation of sulfide ore reserves in the coming years [1] because of the exhaustion of oxide resources. Mineral concentration is the primary process currently used for copper extraction from sulfide ores.
Then, even when the copper production is expected to increase in Chile, a 50% decrease in the production of electrowon cathodes is estimated. That decrease means a concomitant loss of advantages like a possible reward from the quality of cathodes and additional revenues from the recovery and sale of valuable metals contained in the concentrates. On the other hand, concentrate exports and shipping have potential restrictions and additional costs, and variable charges are also associated with treatment and refining [1]. Nowadays, the development of chemical leaching processes for sulfide ores has gained broad attention in hydrometallurgy due to its faster kinetics and higher recovery when compared to bioleaching [2,3]. Copper leaching in concentrated chloride media under ambient conditions is already used in several Chilean operations to benefit secondary sulfide ore [4]. In contrast, for primary sulfide ores, pilot-scale tests can foresee acceptable copper recovery [3]. However, above all the advantages of the beneficiation processes for sulfide ores mentioned above, the high-energy consumption of crushing and grinding for concentrate production poses an impediment when working with marginal materials. On the other side, one of the main issues in the application of chloride leaching is related to costs and impact on infrastructure due to the necessity of changing metallic equipment in contact with Cl - [4]. In that scenario, copper bioleaching, mainly run-of-mine (ROM) copper bioleaching, is considered the best alternative for the extraction of marginal resources currently not considered part of the profitable reserves because of the cost associated with leading technologies in copper extraction [1]. It is foreseen that bioleaching will play a complementary role in concentration and chloride primary leaching operations, considering that it is not necessary to consider the capital cost for installing the solvent extraction and electrowinning (SX/EW) plants, which are already installed. This was the case at the ROM bioleaching plant at Minera Escondida Ltd. (MEL), where more than 250,000 t of copper as cathodes were produced in 2018, and half of them by bioleaching. Consequently, there is a "clearly crucial need to endow hydrometallurgy with technologies that allow it to increase its efficiency" [1]. The potential of bioleaching to enable recovery of metals from deposits to process available resources that are not amenable to metal recovery by any other means has been assessed, e.g., in Europe as well, aiming to develop environmentally sustainable mining operations [5].
One of the main issues of bioleaching at the mine site is the close control required of acidity, air and nutrients availability, temperature, water quality, and cell activity, among other variables, for better metallurgical performances [6][7][8][9][10][11]. However, it is not feasible to monitor these variables inside ROM industrial permanent heaps, as argued before [12,13]. Furthermore, in terms of mineralogy, the heterogeneous ore bed of sulfide exposure to leachate, solution permeability, and air distribution [14] challenges the control of these variables. ROM permanent heaps have had a remarkable engineering development during the last 20 years. The experience revealed the necessity to solve additional issues, like differential acid feeding, air distribution, and solution recirculation.
The microbial community inhabiting ROM bioleaching heaps at MEL has been monitored based on the relevant taxonomic groups and cell activity determination [15,16]. Following these strategies, we assessed the impact of the microbial community structure on the process's metallurgical performance [6,15,17]. The microbial diversity was also assessed in other bioleaching operations by using the same strategy [17].
We proposed that a state-of-the-art functional-based strategy to assess habitat-specific functional features should indirectly reveal a heap's environment conditions, such as bioavailable O 2 , nutrients, and stressful conditions [18,19]. A similar strategy was used to opportunely predict oil contamination in the marine environment [20]. However, up to our knowledge, the analysis of functional genes as biomining community traits was only reported by using cultured microorganisms and in laboratory-controlled conditions [21,22]. Our research work focuses on applying molecular microbial markers for a better control and bioleaching efficiency enhancement.
Based on the genome and metatranscriptomic analyses [9] and the predicted main functions involved in bacterial leaching, we determined the set of genes that better char-acterizes the microbial response to critical variables in the heap environment. Some of the selected genes were already recognized as the standard for microbial functional ecology [19]. We designed and validated the primers for qPCR, and we performed the experiments at different levels (cultures, column tests, and industrial heap) to validate those genes as biomarkers.
The information obtained from the validated real-time PCR array was incorporated in the development of one of the modules of a decision-support system (DSS) based on decision rules that transform the obtained knowledge into recommendations to the plant operator of the bioleaching process [23,24], assisting with the mine site. This kind of DSS is popularly used, e.g., in evidence-based medicine, to improve healthcare delivery [25]. A clinical decision-support system (CDSS) is a software designed to aid clinical decisionmaking. In a CDSS, the characteristics of an individual patient are matched to a computerized clinical knowledge base to present the clinician with patient-specific assessments or recommendations. The construction of the DSS for bioleaching processes considers a database for industrial data logging and storage, a knowledge base acquired by suitable statistical and computational tools, and finally, the translation of knowledge into action by applying recommendations that come to terms with operational limitations. The DSS is composed of five reasoning modules: (1) the impact of mineralogy/mineralization in the metallurgical performance, (2) microbial activity and copper recovery, (3) estimated temperature inside the heap, (4) marker genes for critical issues, and (5) inoculation/reinoculation requirements [23].
Leptospirillum ferriphilum DSM 14647 type-strain and IESL strain were grown in reactors (1 L) with 600 mL of ABS medium [28], using FeSO 4 7H 2 O (50 mM) as an energy source at 37 • C, pH 1.5, with agitation of 200 rpm, and three different concentrations of MgSO 4 7H 2 O: 25 mM (control, 2.5 g L −1 SO 4 2− ), 250 mM (25 g L −1 SO 4 2− ), and 500 mM (50 g L −1 SO 4 2 ) ( Table 1). Each condition was performed in triplicate. The strain DSM 14647 is the typestrain obtained from a culture collection, while IESL is a native strain isolated from the industrial process and maintained in active growth since that. Both L. ferriphilum strains have different life histories related to sulfate and probably have different tolerance levels.
Every 3 h during 3 to 4 days, a direct cell count was performed using an improved Neubauer chamber and phase contrast microscope for L. ferriphilum and At. thiooxidans cultures. Ferrous ion and tetrathionate oxidation activity were monitored by titration with K 2 Cr 2 O 7 and by potentiometric titration of produced H 2 SO 4, respectively [30]. All cultures were harvested in the late exponential growth phase when cells were in abundance and metabolically active. Cells were recovered by vacuum filtration through nitrocellulose membrane (0.2 µm) and were immediately preserved at −80 • C in RNAlater ® solution (Invitrogen™ AM7021, Carlsbad, California, USA) until nucleic acid isolation. We modeled growth (cells·L −1 ) and bacterial activity (Fe 2+ consumption and H 2 SO 4 generation) as a function of incubation time and nutrient or stressor concentrations (CO 2 ppm and SO 4 2− , respectively) in order to determine the effect of these variables. Modeling was carried out using a multiple regression model [31] using the Minitab 18 statistical software.

Bioleaching Column Test
Controlled bioleaching experiments run for 105 days at 21-25 • C. Columns (1 m high, 0.6 m diameter) were operated in an open-circuit system (without solution recirculation) and fed with raffinate from the MEL process [32]. Column (col.) tests were performed to assess the effect of air feeding, and NH 4 + and SO 4 (Table 1). Samples were collected at different operation times for each microorganism's quantification and transcriptomic analyses (Table 1). Notwithstanding the relevant metabolic activity previously observed on mineral-associated bacteria [33,34] and its recognized effect on mineral dissolution [35], mineral samples were neither obtained nor analyzed in this work. That decision was taken regarding the research work scope, which was to develop molecular markers for monitoring a mining operation where it is not possible to get representative mineral samples for periodic assessment [12].

Industrial Bioleaching Heap
The bioleaching heap process at MEL considers ROM material that has been characterized as low-grade sulfide ore and heap dimensions of 5000 m long by 2000 m wide [6]. Leaching horizontal sections denominated strips, and five lifts or levels are currently loaded (18 m height each). The strips are irrigated with raffinate from the top of the heap. Raffinate percolates until reaching the heap's base, enabling dissolved copper from the ore in a PLS.
In an attempt to feed the entire ore heap with the O 2 and CO 2 necessary for microbial growth and activity, the heap is force-aerated by mechanical blowers located at the base of the heap of each strip. Only some strips of the fourth lift are aerated from their bases as well ( Figure 1). The microbial community at the Escondida sulfide leach process was mainly composed of (i) mesophilic microorganisms, At. ferridurans D2, At. thiooxidans, and Ac. acidiphilum; (ii) thermotolerant microorganisms, L. ferriphilum and At. ferrooxidans IESL-32; (iii) moderate thermophiles, At. caldus, Firmicutes representatives from Sulfobacillus genus and Ferroplasma from the Archaea domain; (iv) extremely thermophilic genus, Sulfolobus [6]. Recently, pyrosequencing of 16S rRNA and metagenome sequencing of industrial PLS samples revealed the occurrence of Thermoplasmatales archaeon A or E-plasma [37].
Ten PLS samples were collected from five strips (S-132, S-327, S-410, S-413, and S-414) with different aeration conditions and NH 4 + and SO 4 2− levels ( Table 2). Briefly, PLS samples (10 L for RNA and 1 L for DNA isolation) from the aerated strip S-410 were obtained biweekly, both at the base of the first lift (drop-D) and the base of the fourth lift (M) of the heap, before (one sample) and after (two samples) an aeration interruption. Aeration suspension was carried out by turning off the aeration blowers (forced aeration) located at the base of the S-410, on the fourth lift (M), while the blowers in the base of the first lift (D) of the strip remained turned on. Besides, PLS samples (10 L for RNA and 1 L for DNA isolation) from the S-132, S-327, S-413 strips (permanently and fully aerated), and from the S-414 strip (permanently and partially aerated, with half of the blowers off) were collected. Operational and physicochemical (pH, Eh, NH 4 + , and SO 4 2− concentrations) characterization of samples were performed ( Table 2). Cells from the PLS of each strip were harvested, washed, and preserved as described above. Differences in the NH 4 + , SO 4 (Table 2). These samples were used to validate the biomarkers associated with nitrogen metabolism, osmotic stress, and ferrous/sulfur oxidation.

Candidate Genes Selection and Primer Design
To better understand the situation within the heap, it is necessary to know what the microorganisms are doing and how representative they are of the microbial community through gene expression studies and cell quantification. The candidate marker genes were selected to represent the predominant components of the microbial community inhabiting the industrial heap (At. ferrooxidans, At. thiooxidans, L. ferriphilum, and Sulfobacillus sp. CBAR13) and the metabolisms involved in microbial responses to changes in the heap environment that affects the process performance. Oxygen reduction, CO 2 and N 2 fixation, ammonium transport, iron, and sulfur oxidation, and response to osmotic stress were the metabolisms selected regarding the results of previous research in the system [7,8,10,11].
The metagenomic analysis of a DNA pool generated from industrial samples from the MEL bioleaching heap was previously performed [37]. At. thiooxidans, At. ferrooxidans, L. ferriphilum, and Sulfobacillus sp. related sequences were separated through mapping with CLC Genomic Workbench 8.1 (QIAGEN Bioinformatics, Redwood City, CA, USA.) against Acidithiobacillus, Leptospirillum, and Sulfobacillus available genomes. Four representative reconstructed genomes from industrial samples were chosen, respectively, for At. thiooxidans, At. ferrooxidans, L. ferriphilum, and Sulfobacillus sp. Reconstruction was performed based on the coverage and the sequencing depth. The sequences of the reconstructed genome were then annotated using the RAST server (https://rast.nmpdr.org/ (accessed on 28 December 2020)) [38].
Primers for RT-qPCR (Table S1) were designed for protein-coding genes related to CO 2 fixation, such as rcbL1, rcbL2, csoS2, csoS3, csoS4 from At. thiooxidans, and rcbL1, rcbL2 from At. ferrooxidans (Table 3). Primers were also designed for protein-coding genes related to synthetic pathways of compatible solutes, nitrogenase activity, ammonium transport and assimilation from L. ferriphilum, and protein-coding genes related to the ferrous ion and sulfur oxidation from Sulfobacillus sp. CBAR13 (Table 3). Additionally, primers for 16S rRNA, gyrA, and alaS housekeeping genes (Table 3) were designed for At. thiooxidans, At. ferrooxidans, L. ferriphilum, and Sulfobacillus sp. CBAR13 (Table S1). These housekeeping genes were used as endogenous controls for the relative gene-expression calculation. Table 2. Physicochemical characterization of industrial samples from the Minera Escondida Ltd. (MEL) bioleaching heap (strips S-405, S-410, S-132, Scheme 327. S-413, and S-414) sampled at different operation times for gene expression analysis associated with CO 2 availability, osmotic stress, nitrogen availability, and ferrous/sulfur oxidation. Gray areas mark the operational factors and their related metabolisms studied in the respective industrial samples. Industrial strips were sampled on: (1) 24 November 2015; (2) 29 December 2016; (3) 12 January 2017; (4) 26 January 2017; (5) 23 August 2017. The marks W0, W2, and W4 in S-410 correspond to three samplings performed biweekly before (0), after 2 weeks, and after 4 weeks since the blower was turned off. D represents samples collected from the heap base, M represents samples collected at the base of the fourth lift, and B represents a sample collected at the base of the second lift of the heap. L1 and L4 represent blowers present in the heap-base or the fourth lift base, respectively. ON and OFF are blowers turned on and turned off, respectively. Day refers to the strip operation day.

Sample
Day Heap Lift
All used primers were empirically validated in their specificity, repeatability, and sensibility, as described previously [11]. Pearson's (R) coefficients of 0.99 and optimal efficiencies (E) (100-110%) [41] were obtained for all selected primers (Table S1). The annealing temperature for all primers was 60 • C.

RNA Isolation and Transcriptional Analysis
RNA was isolated, and reverse transcription was performed as previously described [9,11]. Transcription dynamics of five At. thiooxidans (ATT) Calvin-Benson-Basham (CBB) genes (Table 3) were evaluated in ATT cultures, col.1, col.2 (Table 1), and ten industrial samples ( Table 2) by reverse transcription quantitative PCR (RT-qPCR) [6]. Transcription dynamics of two At. ferrooxidans (AFE) CBB genes (Table 3) were also evaluated on industrial samples. The AFE-CBB genes were not evaluated in the bioleaching column col.2 because At. ferrooxidans did not predominate in that microbial community (results not shown). Since ATT and AFE have the same CO 2 fixation mechanism (CBB pathway) but different gas requirements, we hypothesize that it is expectable that CBB genes from both species could have a similar response but different expression levels in the industrial environment. Table 3. Candidate genes analyzed for biomarkers identification and industrial validation. The CO 2 fixation genes are from the Calvin-Benson-Basham cycle. HKG is housekeeping genes. SRC is sulfur-reduced compounds. Sb. sp. is Sulfobacillus sp. Genes that were used as endogenous control are indicated with (*).
Transcription dynamics of four genes associated with ferrous oxidation and two genes associated with sulfur oxidation (Table 3) were evaluated in Sulfobacillus sp. CBAR13 (STO) pure cultures (Table 1) and industrial strips S-132, S-327, S-413, and S-414 ( Table 2).
The fold-change gene expression (ratio) was calculated by the Pfaffl model using REST software V2.0.13 [42]. The Pfaffl model has been successfully used in previous gene expression analyses for acidophilic microorganisms [6,9,11]. One sample of each sample-set was selected and used as a control sample. Relative gene expression was calculated as the transcription level in target samples relative to the control sample. All relative gene expression analyses were normalized against endogenous controls previously validated by the NormFinder algorithm [43]. Relative expression levels were expressed as Log 2 -ratio. Significant differences in relative expression were adjusted at −1 < Log 2 ratio < 1.

Cell Harvesting, DNA, and RNA Extraction
RNA quantity obtained from culture samples ranged between 20-200 ng/µL while the quality for 260/280 and 260/230 ratios ranged between 1.5-2.1 and 0.9-2.0, respectively, as expected for this kind of sample. The analysis performed on the RNA obtained from bioleaching columns and industrial-strip samples (Table S2) revealed that, in general, RNA presented good and acceptable quantity and quality ranges for transcriptomic RT-qPCR analyses.  (Figure 2). The growth rates of At. thiooxidans IESL-33 as assessed by cell counts was slower at lower CO 2 concentrations. Significant differences (p < 0.05) (Table S3) were observed between reactors operated at 50 and 500 ppm of CO 2 (Figure 2A). No significant differences were observed in growth between reactors operated at 50 and 100 ppm of CO 2 or reactors operated at 300 and 500 ppm of CO 2 . The microbial activity expressed as acid generation (H 2 SO 4 )-the product of bacterial sulfur oxidation activityshowed a lower rate of production in reactors with lower CO 2 availability ( Figure 2B).

Identification of Transcriptomic Markers Associated with CO 2 Availability
Culture batch reactors of At. thiooxidans (ATT) IESL-33 at 50 ppm, 100 ppm, and 500 ppm of CO 2 showed an over-expression of ATT-rbcL-1, ATT-csoS2, ATT-csoS3, and ATT-csoS4B genes, and an under-expression of ATT-rbcL-2 gene relative to the control condition (300 ppm CO 2 ) ( Figure 3A). The highest over-expression of these genes occurred in the culture growing with 50 ppm of CO 2 ( Figure 3A), suggesting an apparent transcriptomic response to low CO 2 availability. The ATT-csoS3 gene associated with carbonic anhydrase [44] was significantly over-expressed only at relatively low CO 2 levels ( Figure 3A), as was also observed in the column test ( Figure 3B). In the column test (col.2), non-significant differences were observed in the expression level of the ATT-rbcL-1 gene among samples with relatively low CO 2 levels (0-500 ppm CO 2 -out). Meanwhile, this gene's under-expression was evident in samples with relatively high CO 2 levels (1500-2000 ppm CO 2 -out) ( Figure 3B). Besides this, the over-expression of the ATT-rbcL-2 gene was evidenced at day 64 with 2000 ppm CO 2 -out.
The transcription dynamics of the ATT-csoS3 gene were similar to the ATT-rbcL-1 gene, except after the second decrease of CO 2 on day 76 (200 ppm CO 2 -out) when ATT-csoS3 was over-expressed ( Figure 3B).

Industrial Validation of Transcriptomic Markers Associated with CO 2 Fixation
Transcriptional analyses were performed using industrial strips sampled for aeration (Table 2) because the cDNA of strips S-410-D1, S-410-M2, S-413-D did not show PCR amplification for any of the candidate marker genes. This result was possibly due to the absence of the target microorganisms or the absence of the expression of those genes. Besides this, some genes like ATT-csoS2, ATT-csoS3, and rbcL2 from At. ferrooxidans (AFE) did not show amplification in any examined industrial samples (Figure 4). The absence of significant differences (Log 2 ratio < 1) in the relative expression level of the ATT-rbcL-1 gene among S-410-M0, S-410-D4, and S-410-M4 ( Figure 4A) suggested that similar levels of CO 2 were available in these samples. Only the PLS solution from S-410-D4, obtained from the drop four weeks after aeration interruption, showed a slight gene over-expression, which allows us to infer a relatively low CO 2 availability for At. thiooxidans. A different response was observed in At. ferrooxidans AFE-rbcL-1 gene, which was slightly over-expressed in S-410-D2 (obtained from the drop two weeks after aeration interruption), and no expression differences were observed in the rest of the samples ( Figure 4A). The ATT-rbcL-2 gene was over-expressed in samples S-410-D2 and S-410-M4 ( Figure 4A) and partially aerated strip S-414-D ( Figure 4B).  Lower expressions of nifH (involved in dinitrogen fixation, Table 3), glnA, and glnB (involved in the assimilation of ammonium, Table 3) were detected in most of the samples when the NH 4 + concentration was increased in the bioleaching column col.3 ( Figure 5). The transcriptional pattern observed for glnA and glnB genes was different in the first sample (operation day 76) after the increment of NH 4 + (day 62). On the contrary, the amt gene (involved in the transport of ammonium, Table 3) was only under-expressed when a high level of NH 4 + was present in col.3 ( Figure 5). All genes were over-expressed on day 104, 13 days after the second interruption of the NH 4 + supply in the column.

Industrial Validation of Transcriptomic Markers Associated with N 2 Metabolism
Transcriptional analyses for industrial validation of biomarkers associated with nitrogen fixation were performed using sampled strips ( Table 2). Some genes like glnA did not show amplification in any of the industrial samples. The relative gene expression was analyzed using two validated reference genes (16S rRNA and alaS).
An increase in the expression of nifH was observed when the NH 4 + level decreased to 3 mg L −1 in the industrial leaching solutions analyzed ( Figure 6). The pattern observed in the amt2 and glnA gene's transcription does not relate to the registered concentration of NH 4 + . It is crucial to consider that the level of NH 4 + reached in the columns was higher than the high level reported from the industrial solutions. The L. ferriphilum strain DSM 14647 growth rate was significantly lower (p < 0.0001) at 25 and 50 g L −1 SO 4 2− than at 2.5 g L −1 ( Figure 7A) (Table S3). However, there was not enough data to confirm the significance of the decrease in the oxidation activity measured by the Fe 2+ consumption at higher SO 4 2− concentrations ( Figure 7B) (Table S3). in the bioleaching column, col.3. Significant differences in relative expression were adjusted at −1 < Log 2 ratio < 1 (horizontal black lines). Gene expression was calculated relative to the control sample (log 0 values) that corresponds to operational day 61. The X-axis corresponds to operation days at the sampling time of col.3.  The native L. ferriphilum strain IESL-25 growth rate was significantly lower (p < 0.0001) at 25 and 50 g L −1 SO 4 2− than at 2.5 g L −1 SO 4 2− ( Figure 8A). The negative effect of high SO 4 2− levels was evidenced by the significant decrease (p < 0.05) in the rate of Fe 2+ consumption in cultures growing with 50 g L −1 SO 4 2− ( Figure 8B) (Table S3).

Identification of Transcriptomic Markers Associated with Osmotic Stress
Transcriptional analyses were performed with pure culture samples (three conditions on triplicate) of L. ferriphilum DSMZ 14647 growing under different SO 4 2− concentration and PLS samples from the column test (col.4) where SO 4 2− concentration was modified. The relative gene expression of treZ, treY, lamb, treX, galU, and gadB genes was analyzed using two validated reference genes (16S rRNA and alaS).
The transcriptional dynamic of the selected genes involved in the trehalose synthesis pathway in the reactor culture of L. ferriphilum type strain DSM 14647 showed an increase at high SO 4 2− levels ( Figure 9A) relative to the control sample (2 g L −1 SO 4 2− ). The transcription of the gadB gene only increased when culturing with 50 g L −1 SO 4 ( Figure 9A). Non-significant gene transcription differences were observed between the triplicates of L. ferriphilum DSMZ 14647 control cultures growing on 2.5 g L −1 SO 4 or between the different analyzed genes in the pure culture tests ( Figure 9A). and 103 in a bioleaching column, respectively, using as control the sample taken on operation day 66. Vertical bars represent the standard deviation (SD) calculated with biological replicates. Significant differences in relative expression were adjusted at −1 < Log 2 ratio < 1 (horizontal black lines). In (B), the X-axis corresponds to operation days at the sampling time of col.4.
Four of the six genes associated with compatible solutes in L. ferriphilum were quantified in the bioleaching column test (col.4) after the first and the second increment of sulfate concentration in the feeding solution ( Figure 9B) and relative to the non-intervened sample used as control (day 67). The over-expression of treYZ genes remained stable during both sulfate interventions performed on operation days 68 and 103, respectively ( Figure 9B). The lamB and galU genes were only over-expressed on day 91, after 23 days since the first sulfate increment from 50 to 80 g L −1 . The gene of the trehalose biosynthetic pathway OtsAB (I) represented by the gene galU maintains its expression level despite sulfate increase. A similar result was observed in the expression level of the lamB gene related to trehalose transport ( Figure 9B).

Industrial Validation of Transcriptomic Markers Associated with Osmotic Stress
Transcriptional analyses for industrial validation of biomarkers associated with osmotic stress were performed using the sampled strips (Table 2). Some genes like treX, gadB, and galU did not show amplification in the industrial samples. The relative gene expression was analyzed using the two validated reference genes (16S rRNA and alaS).
The treZ, treY, and lamB genes from L. ferriphilum were over-expressed in industrial PLS samples related to high SO 4 2− concentrations ( Figure 10A,B). The couple treYZ was over-expressed either in industrial strips obtained during the same sampling activity ( Figure 10A) as in strips obtained during independent samplings ( Figure 10B) with concentrations over 90 g L −1 SO 4 2 .

Biomarkers for Iron and Sulfur Oxidation in Sulfobacillus sp. CBAR13
3.5.1. Effect of the Energy Source on Growth of Sulfobacillus sp.
The Sulfobacillus sp. strain CBAR13 growth rate was significantly higher on SS medium with iron as the energy source than under DS medium with tetrathionate ( Figure 11).  Table 2). Significant differences in relative expression were adjusted at −1 < Log 2 ratio < 1 (horizontal black lines). Gene expression was calculated relative to the control samples (log 0 values) that correspond in (A,B), respectively, to S-132 and S-327. Figure 11. Cell growth of Sulfobacillus sp. CBAR13 strain growing with iron (Fe 2+ ) (black circles) and tetrathionate (white circles) as the energy source. Vertical bars represent the standard deviation (SD) calculated with biological replicates.

Identification of Transcriptomic Markers Associated with Ferrous and Sulfur Oxidation
Transcriptional analyses were performed with pure culture samples (two conditions on triplicate) of Sulfobacillus sp. CBAR13 growing with different electron donors (Fe 2+ and sulfur). The relative gene expression of cytbdl, sulf, chypII, carb, tehy, and fadp genes ( Table 2) were analyzed using the validated reference gene gyrA.
Transcriptional dynamics of the cytbdl, sulf, chypII, and carb genes from culture batch reactor samples of Sulfobacillus sp. CBAR13 growing on SS medium (50 mM FeSO 4 7H 2 O) showed a significant over-expression from one to six-folds relative to one of the replicates growing on DS medium (2 mM K 2 S 4 O 6 ) ( Figure 12A). Besides which, a significant underexpression from five to nine-folds for tehy and fadp genes associated with tetrathionate oxidation (Table 3) was observed in cultures growing on SS medium ( Figure 12A). Non-significant transcriptional differences were observed between the triplicates of Sulfobacillus sp. CBAR13 growing on DS cultures ( Figure 12A).

Figure 12.
Transcriptional dynamics of cytbdI, sulf, chypII, carb, tehy, and fadp genes from Sulfobacillus sp. CBAR13 (STO) growing in a pure culture test with different electron donors (SS with Fe 2+ and DS with tetrathionate) (A), and industrial bioleaching strips from the MEL-heap with different availabilities of Fe 2+ (0.24 and 0.04 g L −1 Fe 2+ in S-132-D and S-327-D, respectively) (B). Significant differences in relative expression were adjusted at −1 < Log 2 ratio < 1 (horizontal black lines). Gene expression was calculated relative to a control sample (log 0 values) that correspond in (A,B), respectively, to one biological triplicate from DS cultures and one technical triplicate from S-132. Vertical bars in each column correspond to the standard deviation (SD) calculated with biological and technical replicates.

Industrial Validation of Transcriptomic Markers Associated with Ferrous and Sulfur Oxidation
Transcriptional analyses for industrial validation of biomarkers associated with ferrous/sulfur oxidation (Table 2) showed a significant increment near six-folds on the gene tehy transcription in strip S-327 with 0.04 g L −1 Fe 2+ relative to strip S-132 with 0.24 g L −1 Fe 2+ ( Figure 12B and Table 2). Inversely, an under-expression was observed for sulf and chypll genes in the industrial sample with less Fe 2+ availability (S-327) ( Figure 12B). The relative gene expression was analyzed using the validated reference gyrA gene.
The analyses were performed using the sampled strips (Table 2). PCR amplification was not obtained from cDNA of strips S-413 and S-414 for any candidate marker genes. Genes cytbdl, carb, and fadp did not show amplification in the industrial samples ( Figure 12B).

Discussion
The results obtained for validating genetic markers associated with O 2 availability within the heap were previously reported [11].

Carbon Fixation
It is known that bioleaching processes rely on O 2 to promote the oxidation processes for sulfide mineral dissolution and on CO 2 as the carbon source for the growth of microorganisms, which catalyze the oxidation. An important issue in heap bioleaching technology is to install homogeneous aeration for providing enough O 2 and CO 2 in the environment to improve operational performance. The pH, temperature, ion concentration, carbonate content in the mineral, altitude, and activity of heterotrophic microorganisms also influence the CO 2 availability [28]. A significant decline of microbial oxidation rates was evidenced in CO 2 depleted zones in a controlled heap-like environment, sealed large columns op-erated isothermally at 40 • C and supplemented with additional CO 2 [45]. Besides, the effect of CO 2 was also determined in agitated reactors, and increased copper extraction was observed when CO 2 was supplied at 4% (4000 ppm) [46].
The results obtained from reactor experiments of At. thiooxidans growing at different CO 2 concentrations (Figure 2A,B) confirm previous studies regarding the necessity of elevated CO 2 concentrations for optimum growth in At. thiooxidans [47]. On the contrary, it has been reported that the closely related species At. ferrooxidans that use the same CBB cycle reach an optimum specific growth rate (µ) at 100 ppm of CO 2 [28]. However, At. thiooxidans maintains an increasing µ tendency until 3000 ppm of CO 2 [47] while a decline of µ was registered in At. ferrooxidans growing at CO 2 concentration above 100 ppm [28].
The characterization of the microbial community composition after air feeding interruption in col.2 evidenced a decrease of acidithiobacilli and L. ferriphilum and a similar abundance, or even an increase, of heterotrophic microorganisms like Thermoplasma sp. and Ferroplasma acidiphillum (Supplementary Figure S1). Then, the increased concentration of CO 2 on the output of col.2 ( Figure 3B) was probably due to heterotrophic microbial metabolism. In this experiment, the air was displaced with an equal flux of N 2 , so the gas-in mass input changes were discarded.
Previous works have shown that the growth of At. thiooxidans is favored by an elevated concentration of CO 2 (2500 ppm) [47]. Our results show that ATT rbcL-1 transcription was associated with relatively low CO 2 availability ( Figure 3A,B). Similar results in the transcription of rbcl-1 have been reported for At. ferrooxidans [9]. The over-expression of ATT rcbL-2 observed at the first airflow interruption in col.2 ( Figure 3B) suggests that the transcription of this gene could be associated with relatively high CO 2 levels, as it has been previously reported for other species [48]. It is generally concluded that RubisCo type II enzymes are adapted to function at low-O 2 and medium to high-CO 2 environmental levels [49], a feature supported by these results.
Apparently, relatively low CO 2 levels induced an increased transcription of ATT rbcL-1 and ATT csoS3 genes, as we have previously reported for acidithiobacilli [9]. The structural carboxysome csoS2 and csoS4B genes were over-expressed at both high and low CO 2 levels. A plausible explanation could be the differences in the turnover between the structural (e.g., CsoS2 and CsoS4B) and functional proteins (e.g., CsoS3), or it indicates a CO 2 -storage function in carboxysome. More evidence is required to confirm these hypotheses.
The transcriptional dynamics observed in the industrial samples ( Figure 4A,B) support previous findings that At. thiooxidans and At. ferrooxidans have different strategies to respond to changes in CO 2 levels, although both species share the same CO 2 uptake/fixation mechanism [50].
The over-expression of the ATT rbcL-2 gene ( Figure 4A,B) is probably suggesting a response of At. thiooxidans to high CO 2 and low O 2 levels as has been recognized for this gene [49] and observed in pure cultures and bioleaching column tests as well ( Figure 3A,B).
The consistent results of transcriptional dynamics obtained from the pure culture at the laboratory level, controlled column tests, and industrial samples allowed us to obtain real indicators for CO 2 availability in bioleaching systems. The industrial application of those indicators in monitoring forced aeration in a heap could influence the control and design of present and future bioleaching operations, respectively.
Finally, we show that the growth and the oxidation activity of At. thiooxidans IESL-33 was slower when the cells were exposed to a low concentration of CO 2 , also accompanied by a lower metabolic activity ( Figure 2 and Table S3), confirming the results of previous works [28]. Furthermore, we found that changes in the airflow input in bioleaching column tests and the MEL heap's industrial operation influence the microbial population ( Supplementary Figures S2 and S3). These changes can be related to metabolic shifts that microorganisms must overcome during periods of low oxygen concentration. Finally, the transcriptional analysis of the biomarkers for CO 2 metabolism of At. thiooxidans, rbcL-1 and rbcL-2 showed a similar response to that observed in At. ferrooxidans but at different CO 2 or O 2 concentrations.

Nitrogen Metabolism
The ore in a heap provides the micronutrients required for the microbial community's activity responsible for sulfide minerals dissolution. However, a supply of macronutrients, such as ammonium, potassium, and phosphate could be needed [51]. There is, as well, an input of ammonium in the system, caused by the ability of some microorganisms like At. ferrooxidans and some members of Leptospirillum genus to fix nitrogen from the atmosphere [8,51,52].
The ammonium concentration measured at the heap is on average 5.5 mg L −1 , and higher levels (up to 10 mg L −1 ) were observed after the start of irrigation of some new strips ( Figure S4). Similar levels were reported in leaching solutions on a heap bioleaching process from Talvivaara Mining Company [51]. Over-expression of genes involved in the nitrogen fixation (nifH) was observed in the column test when NH 4 + concentration decreased up to 3 mg L −1 , and a similar behavior was observed in the industrial leaching solutions analyzed ( Figure 6).
Previous results have shown a significant increase of the nifH expression level after 110 days of heap operation and an opposite profile for the amt gene of L. ferriphilum. Even though we have not measured the ammonium concentration in that opportunity, the results insinuated a higher availability of ammonium at the initial stages of the process and a metabolic shift to N 2 fixation when the operation time progressed. The results from ammonium analysis of randomly selected strip PLSs confirmed the occurrence of ammonium levels above the average during the initial irrigation of some strips from the first lift ( Figure S4).
A high diversity of L. ferriphillum was observed in the industrial heap with a predominance of sequences from Group IIc [6,37]. Group IIc is composed of strains with and without the capacity for N 2 fixation [8,[52][53][54]. The N 2 fixation capacity was confirmed in the IESL-25 strain [54] isolated from the industrial heap [55]. Different gene localization was observed in the nitrogen fixation operon of the pan-genome of Leptospirillum spp. retrieved from the heap [37].

Osmotic Stress
The primary response of bacteria to exposure to a high osmotic environment is the accumulation of specific solutes like K + , glutamate, γ-aminobutyrate, trehalose, proline, and glycine betaine at concentrations that are proportional to the osmolarity of the medium [56].
Trehalose is a sugar that has been described as the compatible solute mainly used by L. ferriphilum in response to osmotic stress. Increments of trehalose synthesis until 59 folds after exposition to high concentrations of MgSO 4 have been previously reported [57]. Genes involved in the three trehalose biosynthetic pathways: GalU-OtsAOtsB (I), TreY-TreZ (V), and TreS (IV) [58,59], were identified in the genome of L. ferriphilum Sp-Cl [54]. The transcription dynamics of genes involved in the TreYZ pathway observed in this study from pure cultures, controlled bioleaching columns, and industrial bioleaching heap (Figures 9 and 10) confirms that L. ferriphilum prefers the TreYZ pathway under osmotic stress triggered by high SO 4 2− concentrations. The over-expression of treYZ genes observed in all the mentioned scenarios (Figures 9 and 10) confirms the participation of the TreYZ pathway in the response to osmotic stress in all the analyzed sulfate concentrations in L. ferriphilum and the treYZ genes as industrially validated transcriptomic markers for the monitoring of the osmotic stress in bioleaching systems.
The UDP-glucose pyrophosphorylase enzyme coded by the galU1 gene [60] catalyzes a reaction between the glucose-1-phosphate and the UTP to form UDP-glucose. UDPglucose participates in glycogen synthesis, which, in turn, can be used as a precursor for the trehalose synthesis by the TreYZ pathway [61,62]. Previous studies report that the simultaneous over-expression of galU, otsAB, and treYZ genes triggers a six-fold increment in the trehalose production in Corynebacterium glutamicum [60,61]. The results of the present study allow us to infer that L. ferriphilum type-strain (DSMZ 14647) and L. ferriphilum "wild" strains from the industrial bioleaching heap use the TreYZ pathway to increase the trehalose synthesis under prolonged conditions of osmotic stress ( Figure 9A,B).
The lamB gene expression is associated with the trehalose transport from the cytoplasm to the cell periplasm. It can then be degraded to two glucose molecules by a trehalase (TreA) [63]. The over-expression of the lamB gene observed at all the analyzed sulfate concentrations ( Figures 9A,B and 10A) confirms that this gene is also transcribed by L. ferriphilum in response to osmotic stress (Figures 9 and 10). It is possible to infer that L. ferriphilum can regulate the intracellular concentration of solutes by lamB gene transcription when trehalose synthesis has been increased by the co-transcription of treYZ/galU genes.
The transcription of the gadAB genes is associated with the synthesis of the glutamate decarboxylase, a pyridoxal 5 -phosphate-dependent enzyme that catalyzes the glutamate conversion producing λ-aminobutyric acid (GABA) by an irreversible decarboxylation reaction [64]. From the gadB transcription-dynamic results observed only in the pure culture tests (Figure 9A), it is reasonable to assume that glutamate is probably not the compatible solute preferred by L. ferriphilum under osmotic stress conditions. This issue agrees with previously reported results confirming trehalose as the preferred compatible solute in L. ferriphilum [57]. The gad gene does not participate in trehalose synthesis, but it has been reported that the enzyme is required for trehalose protection against acid stress. The over-expression of both genes simultaneously could indicate the occurrence of acid stress [65].
The congruent transcription dynamic observed from treYZ and lamB genes on the pure cultures, bioleaching-column tests, and industrial bioleaching heap certainly allow us to validate these three genes as transcriptional markers associated with osmotic stress response in L. ferriphilum. The treYZ and lamB genes can be useful for eventual metabolic monitoring of industrial copper-bioleaching processes.

Ferrous and Sulfur Oxidation
The candidate genes associated with ferrous (cytbdI, sulf, chypII, carb) and sulfur (tehy, fadp) oxidation pathways (Table 3) were over-expressed and under-expressed, respectively, in pure cultures of Sulfobacillus sp. CBAR13 growing in the presence of iron as the only energy source (SS) relative to cultures growing with tetrathionate as electrons donor (DS) ( Figure 12A). The same transcriptional dynamics were observed for sulf, chypII, and tehygenes analyzed on the industrial samples ( Figure 12B), validating those genes as molecular markers for monitoring iron and sulfur oxidation in industrial heap-bioleaching systems. The cytbdI, carb, and fadp genes did not show amplification in the analyzed industrial samples, so they were proposed as putative molecular markers for the energy generation metabolism in Sulfobacillus until the industrial validation be reached. The absence of amplification could be due to the low sensibility of the primers. New primers will be designed and assayed for these three genes in future experiments.
The results observed agreement with the previously reported substrate behavior in tetrathionate-adapted and iron (II)-adapted cells of three species of Sulfobacillus genus. In that research, the authors concluded that a 'preference' for ferrous oxidation appears to be expected in this genus. The requirement of ferric ions for growth on tetrathionate would explain that feature [66].
While a 'preference' for ferrous oxidation appears to be shared in this genus, understanding how the cells convert the chemical energy to biomass will require significant further study.
Sulfobacillus genomes were predicted to possess tetrathionate hydrolase (tehy), which directs the disproportion of tetrathionate to generate sulfate, thiosulfate, and sulfur [67,68]. In this work, the tehy gene was under-expressed near ten-folds in pure cultures growing with ferrous ion as the electron donor (SS) ( Figure 12A) and over-expressed near six-folds in the industrial strip with a lower Fe 2+ availability (S-327) ( Figure 12B, Table 2). Similar transcriptional dynamics were observed in S. acidophilus, where the tehy gene was over-transcribed 1.94-folds and under-transcribed 0.47-folds in cultures grown in the presence of sulfur and FeSO 4 , respectively [69].
Blue-copper proteins are members of the cupredoxin superfamily and have been found related to aerobic ferrous iron oxidation in bioleaching systems (e.g., [70]). The sulf gene related to sulfocyanin blue-copper protein from Sulfobacillus was over-expressed in the CBAR13 strain growing on SS medium with Fe 2+ and under-expressed in the industrial strip with lower Fe 2+ availability ( Figure 12A,B). The transcription dynamics observed in the present study support the RNAseq [71], and the semi-quantitative RT-PCR [67] results showed previously. Both studies revealed an over-expression of putative genes associated with blue-copper-like proteins (sulfocyanin/rusticyanin) from S. thermosulfidooxidans on a chalcopyrite bioleaching column test when Sulfobacillus proportion in the column was >80% and the Fe 2+ oxidation rate measured at 45 • C reached 10 mg L −1 h −1 , and during growth on ferrous sulfate, respectively.
The sulf and chypII genes associated with the sulfocyanin protein and the cbb3-type cytochrome-c-oxidase-subunit-I, respectively, and both putatively related to iron oxidation in Sulfobacillus sp. CBAR13 were under-expressed in industrial strip S-327 (0.04 g L −1 Fe 2+ ) relative to S-132 (0.24 g L −1 Fe 2+ ) ( Figure 12B). However, the level of under-expression of those genes observed in the industrial strip with lower Fe 2+ availability was less evident than the over-expression of the ferrous ion oxidation genes evidenced in the Sulfobacillus sp. CBAR13 pure cultures growing with Fe 2+ as the energy source ( Figure 12). The stronger iron-oxidizer L. ferriphilum present could explain this result in the studied industrial MEL strips ( Figure S3). It has been reported that some electron chain components, linked to iron oxidation in Sulfobacillus sp. CBAR13, are down-regulated in the presence of a second iron oxidizer as L. ferriphilum [72]. The above is probably because the concentration of available iron is too low for utilization by Sulfobacillus sp. CBAR13. Meanwhile, L. ferriphilum can scavenge Fe 2+ at concentrations far below Sulfobacillus sp. CBAR13 capabilities but that L. ferriphilum is still able to use [72].
A set of 16 marker genes industrially validated throughout our work for O 2 , CO 2 , and N 2 availability, osmotic stress, and sulfur/ferrous oxidation study in Acidithiobacillus sp., Leptospirillum sp., and Sulfobacillus sp., conform the first qPCR-array for monitoring critical metabolic status in copper industrial bioleaching systems ( Figure 13). Furthermore, the rules constructed based on these results and already incorporated in the DSS, opportunely advise the operator about deviating parameters, that cannot be directly determined in the heap, from the optimal range for the microbial process.

Conclusions
The array of transcriptomic markers developed and validated in this study strongly represents the metabolic scenario involved in microbial responses to parameters that influence the industrial bioleaching heap process's productivity and efficiency. The transcriptomic dynamics resulting from the industrial strips analyses evidence of an active response to CO 2 and N 2 limitation, osmotic stress, and active iron and sulfur microbial oxidation.
Molecular markers were demonstrated to be useful for indirectly assessing critical environmental variables (events), which are known as limiting factors for the rate of copper leaching in bacterial-assisted plants. The results have allowed us to provide new data on parameters that should be improved to optimize bioleaching, such as CO 2 and NH 4 + concentrations. Moreover, this array, applied for periodically assessing the industrial bioleaching operations, will provide the amount of data needed to get the industry knowledge required to encourage innovation. The incorporation of those results in the DSS, by means of rules, transfers the knowledge acquired to the operator.
The possibility of transferring this approach to other bioleaching plants is high because the microbial community structure inhabiting the bioleaching processes is mostly a traceable one, where Acidithiobacillus, Leptospirillum, and Sulfobacillus are the predominant members.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/genes12040474/s1, Table S1: Primers designed for At. thiooxidans (ATT), At. ferrooxidans (AFE), L. ferriphilum (LII), and Sulfobacillus sp. CBAR13 (STO) and used for real-time quantitative PCR (RT-qPCR) analysis. Table S2: RNA quantity and quality obtained from the bioleaching column test and industrial bioleaching strip samples. Table S3: Estimated p-values for multiple regression models of the At. thiooxidans, L. ferriphilum IESL-25, and L. ferriphilum-type strain DSM 14647 growing at different CO 2 and SO 4 2− levels. Figure S1: Copy numbers of 16S rRNA by real-time PCR analysis of autotrophic (Atf and Att) and heterotrophic (Fac and Thsp) microorganisms in column experiment with on/off air feeding. Cut = air feeding interrupted. Atf: Acidithiobacillus ferrooxidans; Att: At. thiooxidans; Fac: Ferroplasma acidiphilum; Thsp: Thermoplasma sp. Figure S2: Characterization of the col.2 community in response to input-airflow interventions. Samples used in the transcriptional analysis are indicated with an asterisk and their O 2 levels (%) (*). Horizontal bars on the below the x-axis indicate the two periods when input-airflow was interrupted in the column (days 62 to 68 and days 91 to 104). The input (CO 2 -in) and output (CO 2 -out) CO 2 levels are shown. Figure