Construction of a silica-inducible expression vector and reporter plasmids
We previously reported that Thermus strains produced a specific Sip in the presence of supersaturated silica. This Sip exhibited high homology with a Fe3+-binding ABC transporter that plays a role in Fe3+ uptake. The expression of Sip is thought to be a response to iron starvation stress, which can be induced by the addition of supersaturated negatively charged colloidal silica that captures Fe3+ ions (Fujino et al 2016). The ferric uptake regulator (Fur), a repressor of iron uptake-related gene expression, responds to low environmental iron levels and is released form promoter region to enhance downstream gene transcription. In a previous study, the cultivation of Thermus cells in medium containing 10 mM silica led to an approximately 10-fold higher Sip expression level relative to that observed in the absence of silica (Doi et al 2009). Therefore, we speculated that the sip promoter, which controls protein expression via a mechanism induced by silicic acid, could be utilized to express heterologous proteins in Thermus.
The construction of the plasmid DNA construct used in this study is illustrated in Figure 1. We inserted a multi-cloning site (MCS) and 600-bps fragment of a putative silica-inducible promoter region from T. thermophilus HB8 into the pYK596 plasmid, which contains a thermostable hygromycin resistance gene. Next, we eliminated the Xho I site on the pYK596 backbone was eliminated by site-directed mutagenesis to yield the plasmid pSix1. To enable a reporter assay, a thermostable β-galactosidase gene was inserted into pSix1 at the Nde I and Xho I sites within the MCS. All the modified regions in this new βgal/pSix1 reporter plasmid were confirmed using an ABI 3130 genetic analyzer (Applied Biosciences, Foster City, CA, USA).
Effect of supersaturated silica on the growth of T. thermophilus
As described above, the addition of silicic acid to Thermus cells may induce stress. Therefore, the effects of silicic acid on the growth of Thermus cells was monitored. Growth inhibition was observed in T. thermophilus HB27 cells harboring βgal/pSix1 after cultivation in medium containing various concentrations of silica, particularly those corresponding to supersaturation (>5 mM at 70 °C; Figure S1). Only slight cell growth inhibition was observed at 6.7 mM silica, whereas significant inhibition was observed at 10 mM silica, as indicated by an optical density at 600 nm (OD660) value that was reduced to nearly half the value observed at 0 mM silica. Although growth appeared to be impeded by silica-induced iron deficiency, the lack of a severe effect of exposure to 10 mM silica on the growth of wild-type T. thermophilus HB27 (data not shown) suggests that the observed growth inhibition may be attributable to the allocation of various energy resources toward protein expression rather than cell growth.
Effect of the silicic acid concentration on β-galactosidase activity
β-Galactosidase is a convenient enzyme, as its activity can be quantitated precisely in liquid assays (Miller 1972). Consequently, it has been used widely as a reporter to monitor gene expression. Approximately 30 years ago, Koyama et al. first reported the use of a thermostable β-galactosidase gene as a reporter in Thermus strains (Koyama et al 1990). Since then, several researchers have attempted to use β-galactosidase assays in studies of Thermus (Park and Kilbane 2004, Renata Moreno 2003). However, wild-type Thermus cells exhibit a high level of background activity, and consequently it has been difficult to assess the precise level of promoter activity, particularly if the target promoter is relatively weak. To overcome this limitation, Fujita et al. recently developed a precise reporter assay system in which the background activity was reduced by disrupting all β-galactosidase genes in the host cells (Fujita et al 2015).
Despite those earlier findings, we selected wild-type HB27 as the host strain in our β-galactosidase assay because a previously reported qRT-PCR analysis revealed strong sip promoter activity (16-fold increase in response to 10 mM silica) (Fujino et al 2016). As expected, exposure to 10 mM silica-induced a sufficiently high level of β-galactosidase activity to enable a comparison between the induced and non-induced conditions. The β-galactosidase activity levels measured after a 24-h cultivation at various silica concentrations are indicated in Figure 2. Notably, the β-galactosidase activity level was negligibly lower in non-induced condition than in cells under inducing conditions. HB27 cells that harbored pSix1 empty vector exhibited an activity level <10 MU, similar to that measured in HB27 cells without plasmid (data not shown). Relative to cells harboring pSix1, cells harboring the βgal/pSix1 plasmid exhibited slightly higher levels of activity (17–26 MU) at when exposed to silica concentrations of 0–6.7 mM silica, suggesting leaky expression. However, when cultured in 10 mM silica, cells harboring βgal/pSix1 achieved β-galactosidase activity levels as high as 190 MU (11-fold higher than that observed at 0 mM silica), indicating strong sip promoter activity.
Exposure to supersaturated silica indirectly causes iron starvation, leading to induction of the sip promoter. Therefore, we also tested direct iron starvation caused by iron chelators such as 2,2’-dipyridyl (DP) and 1,10- phenanthroline (phen). However, the addition of DP and phen (0.5–5 mM) did not have a notable effect on β-galactosidase expression (data not shown). This result implies that Thermus can utilize chemically masked iron species, such as siderophores (Wandersman and Delepelaire 2004). Therefore, the addition of supersaturated silica is a superior method for initiating sip promoter-mediated transcription.
Effect of medium exchange on gene expression
As described above, β-galactosidase activities were measured after an uninterrupted culture in medium containing various silica concentrations. However, induction during the middle or late exponential growth phase is generally much more advantageous for protein expression. Unlike IPTG, highly concentrated silica is easily precipitated. Consequently, it is difficult to achieve an appropriate silicic acid concentration when using a stock solution. We used medium exchange to achieve silica-mediated induction during the late exponential growth phase. Briefly, cells harboring βgal/pSix1 were cultivated in normal TM medium without silica until the OD660 value reached 0.6. The cells were then collected by centrifugation and inoculated in fresh medium containing 10 mM silica. The β-galactosidase activity in the cultures increased gradually along with the duration of silica exposure, reaching a plateau at 12 h. As shown in Table S3, however, the activity levels after medium exchange were lower than those observed after uninterrupted cultivation in medium containing supersaturated silica. Although the maximum activity value achieved in a medium exchange culture was 3.6-fold higher than that observed before induction, this maximum activity was less than one-third of the value observed after an uninterrupted culture period in 10 mM silica. In a recent report, we similarly observed that medium exchange was ineffective in an E. coli-based protein expression system that utilizes silica as an inducer in advance (Fujino et al 2016b). As discussed in previous reports, this decrease in protein expression may be attributable to intracellular iron storage proteins. Because normal silica-free medium contains sufficient iron, cells grown in this medium can take up sufficient levels of iron to maintain growth and can store this element as bacterioferritin (Abdul-Tehrani et al 1999). As these bacterioferritin molecules are then passed to daughter cells, several generations may be required to provoke iron starvation. Therefore, exposure to silica throughout the cultivation period resulted in much higher protein expression levels.
Immunodetection of His-tagged β-galactosidase
We next examined whether our T. thermophilus-based heterologous expression system could be applied for practical use. In light of our previous analysis, β-galactosidase was induced by continuous exposure to 10 mM silica. After a 48-h cultivation, the cell extracts were subjected to SDS-PAGE and Western blotting (Figure 3). In the absence of silicic acid, no protein bands corresponding to β-galactosidase were observed in either the insoluble or soluble fraction after an SDS-PAGE analysis, and no corresponding bands were detected by Western blotting. By contrast, the extracts of cells exposed to 10 mM silica produced clearly visible bands on SDS-PAGE. Bands corresponding to His-tagged β-galactosidases were also detected on Western blots. Particularly strong bands were detected in the soluble fraction, suggesting that the recombinant form of β-galactosidase expressed in T. thermophilus was correctly folded.
As noted previously, inclusion body formation is sometimes observed when using bacterial cell factories, especially E. coli, are used to produce recombinant proteins for both research and industrial applications. When we expressed recombinant β-galactosidase in E. coli, we found that most of the protein was expressed in the insoluble fraction, indicating the formation of inclusion bodies (right panel in Figure 3). This discrepancy in protein production between the species may be attributable to the high temperature under which T. thermophilus is cultivated, as thermophilic enzymes occasionally require exposure to high temperatures to achieve correct folding into catalytically active forms (Diruggiero and Robb 1995, Goda et al 2005, Schultes and Jaenicke 1991, Siddiqui et al 1998). Our successful achievement of soluble protein expression in T. thermophilus suggests that our expression vector can serve as a useful genetic tool for the expression of thermostable gene products that would be insoluble when produced in E. coli.
Previously, Moreno et al. reported the achievement of heterologous gene expression in T. thermophilus at a practical level (Moreno et al 2005). The system developed by these authors adopted the nitrate reductase operon (nar operon) in T. thermophilus HB8 (Ramirez-Arcos et al 1998), which is strongly transcribed in response to nitrate under anoxic conditions. Therefore, their system requires the specific host strain T. thermophilus HB27::nar (transplanted with the nar cluster from HB8) to enable anaerobic growth in the presence of nitrate (Vazquez-Tello et al 2002). By contrast, our system, which is driven by a silica-inducible promoter, does not require a genetically modified host strain because Sip induction is common among Thermus species and is basically controlled by ferric uptake regulators that are inherent in most microorganisms. These features may enhance the convenience of our system and its applicability to hundreds of other strains in the genus Thermus. The results presented herein highlight the potential use of T. thermophilus as a host strain for the expression of thermostable proteins and other practical applications (e.g., microbial bioprocesses) that require relatively high temperatures.
Promoter deletion analysis
Next, deletion mutants of βgal/pSix1 were constructed to determine the minimum region of the active promoter. We found that 100-bp region adjacent to sip CDS exhibited significant promoter activity, whereas a 50-bp adjacent region exhibited no activity. The 5ʹ region corresponding to a span of -100 to -600 bp from the start codon of sip also did not exhibit promoter activity. Therefore, the essential sip promoter region appeared to be contained within -100 to -50 bp upstream of the start codon (Figure 4A). Interestingly, the 100-bp promoter yielded a β-galactosidase activity level of approximately 1,200 MU, which was 6 times higher than the level achieved with the 600-bp promoter region. This phenomenon was also confirmed by SDS-PAGE, which revealed that the shorter promoter region yielded a stronger thermostable β-galactosidase band (Figure 4B). Our results suggest that the 100-bp sip promoter region yielded much more efficient expression of the recombinant protein than the original expression vector pSix1. Consequently, we truncated the promoter region of pSix1 to 100 bp to yield pSix3 (Accession number: LC504201).
Mini-prep extraction revealed that pSix1 produced a lower yield than the original plasmid, pYK596, suggesting an issue of stability with the former construct. Because the sip promoter regions in the plasmid DNA (derived from T. thermophiles HB8) and in the host chromosomal DNA (T. thermophiles HB27) are so similar (99%), we suspected that homologous recombination might have occurred. Specifically, a single-crossover homologous recombination event would have led to the insertion of plasmid DNA into the chromosomal DNA. Figure S2_A presents a schematic of the gene arrangements of normal chromosomal DNA, pSix1, and chromosomal DNA in a single-crossover mutant. An analysis of total DNA against DNA from various plasmids confirmed the single-crossover homologous recombination event. As shown in Figure S2_B, βgal/pSix1 yielded a band that corresponded to the single-crossover region between chromosomal and plasmid DNA. No crossover associated bands were observed with any other plasmids, including pSix3, indicating the stable maintenance of these plasmids in the host cells.
In our experiments, the 100-bp promoter yielded the highest expression level. However, this promoter was also associated with a higher leaky expression level than the long promoter. We speculated that this might be attributable to a loss of Fur protein, which represses the sip gene. In a previous report, the estimated plasmid copy number of pTT8 (the precursor plasmid of pYK596) relative to chromosomal DNA was eight (Ohtani et al 2012, Takayama et al 2004). Given that 8-fold fur binding site located on plasmid DNA against single fur gene on chromosomal DNA, this leaky expression would be inevitable. Further modifications, such as an insertion of the fur gene into the expression vector, might be needed to achieve more strict expression control.
Homologous expression of β-galactosidase and heterologous expression of pisGDH via pSix3
We determined that pSix3 was most suitable for protein expression in T. thermophilus HB27. Subsequently, we successfully expressed β-galactosidase from Thermus sp. A4 as shown in Figure 4B. His-tagged β-galactosidase was purified using immobilized nickel-affinity chromatography, as confirmed by the appearance of a single band on SDS-PAGE (Figure 5A), and yielded 27 mg/L of culture. This expression level was the highest achieved using this T. thermophilus-based system. However, this result should be considered homologous expression because the target gene was β-galactosidase from a Thermus strain.
To confirm the ability of this system to induce heterologous expression, we attempted to express glutamate dehydrogenase from the hyperthermophilic archaeon Pyrobaculum islandicum (pisGDH), which requires heat treatment to reach a fully activated form. First, the original pisGDH gene was cloned into the pSix3 vector. However, this plasmid did not yield protein expression in Thermus cells, which we attributed to a codon bias between Thermus and Pyrobaculum. We then codon-optimized a pisGDH gene fragment for Thermus via artificial synthesis, which enabled the successful expression of pisGDH in Thermus cells (Figure 5B). After purification via affinity chromatography, this system yielded 9.5 mg pisGDH/L of culture. As expected, Thermus expressed a fully active form of pisGDH without any further treatment, thus emphasizing the superiority of this expression system for thermostable enzyme production. A specific activity analysis yielded a value of 4.90 µmol/min/mg for the enzymes expressed in Thermus, which was nearly identical to that of native pisGDH (3.51 µmol/min/mg).
Our findings support our speculation that fully activated thermostable enzymes could be expressed in Thermus cells cultured at high temperatures. However, we note that our original failure to express pisGDH from the original (i.e., not optimized for Thermus) sequence cannot be ignored. Thermus species have a high GC content (~70%) and therefore a very different codon usage pattern from that of other organisms. A new host strain containing tRNAs for codons that are rarely used in Thermus might be required to achieve universal translation.