Stay or Go: Sulfolobales Biofilm Dispersal Is Dependent on a Bifunctional VapB Antitoxin

ABSTRACT A type II VapB14 antitoxin regulates biofilm dispersal in the archaeal thermoacidophile Sulfolobus acidocaldarius through traditional toxin neutralization but also through noncanonical transcriptional regulation. Type II VapC toxins are ribonucleases that are neutralized by their proteinaceous cognate type II VapB antitoxin. VapB antitoxins have a flexible tail at their C terminus that covers the toxin’s active site, neutralizing its activity. VapB antitoxins also have a DNA-binding domain at their N terminus that allows them to autorepress not only their own promoters but also distal targets. VapB14 antitoxin gene deletion in S. acidocaldarius stunted biofilm and planktonic growth and increased motility structures (archaella). Conversely, planktonic cells were devoid of archaella in the ΔvapC14 cognate toxin mutant. VapB14 is highly conserved at both the nucleotide and amino acid levels across the Sulfolobales, extremely unusual for type II antitoxins, which are typically acquired through horizontal gene transfer. Furthermore, homologs of VapB14 are found across the Crenarchaeota, in some Euryarchaeota, and even bacteria. S. acidocaldarius vapB14 and its homolog in the thermoacidophile Metallosphaera sedula (Msed_0871) were both upregulated in biofilm cells, supporting the role of the antitoxin in biofilm regulation. In several Sulfolobales species, including M. sedula, homologs of vapB14 and vapC14 are not colocalized. Strikingly, Sulfuracidifex tepidarius has an unpaired VapB14 homolog and lacks a cognate VapC14, illustrating the toxin-independent conservation of the VapB14 antitoxin. The findings here suggest that a stand-alone VapB-type antitoxin was the product of selective evolutionary pressure to influence biofilm formation in these archaea, a vital microbial community behavior.

the Pseudomonas putida mqsRA toxin-antitoxin locus showed significant biofilm defects. Not only did TA systems promote P. putida biofilm formation, but the antitoxin MqsA repressed a sigma factor and a universal stress protein (30), illustrating the antitoxin's dual role of toxin neutralizer and transcriptional regulator.
Much less is known about the function of TA systems in the Archaea. Studies have shown that TA systems play a role in the heat shock response of Sa. solfataricus (31,32) and the response of Metallosphaera spp. to uranium exposure (33,34). TA systems were transcriptionally upregulated in the thermoacidophile Sa. solfataricus during heat shock, and deletion of the vapB6 antitoxin gene resulted in a heat-labile mutant (31). Furthermore, Metallosphaera prunae developed resistance to hexavalent uranium-more so than its close relative Metallosphaera sedula-by degrading its own cellular RNA via toxin ribonucleases, resulting in growth arrest and entry into a dormant state (33,34).
Currently, the toxin RNase component of type II TA systems has attracted the most attention. Rarely have separate antitoxin mutants been examined and in fewer cases has the regulatory ability of the antitoxin been investigated outside its own operon. Furthermore, toxin-antitoxin systems are often acquired through horizontal gene transfer, leading to a lack of nucleotide identity conservation, even within a species (35). Here, we focus on an archaeal type II antitoxin's regulatory function beyond its own promoter, in which a VapB antitoxin had a profound effect on biofilm formation in S. acidocaldarius. The nucleotide identity of this antitoxin, VapB14, was highly conserved across the Sulfolobales and responded similarly to biofilm growth in M. sedula, suggesting a significant role in regulating this phenomenon across the Sulfolobales.

RESULTS
Sulfolobus acidocaldarius biofilm transcriptome. Transcriptomics comparing biofilm to planktonic S. acidocaldarius cells identified genes associated with biofilm formation. Among the most biofilm-responsive genes was Saci_2184, expression of which increased 10.8-fold in the biofilm compared to planktonic culture, second only to the hypothetical protein gene Saci_0301 (Fig. 1). The high transcriptional response of Saci_0301 was previously reported, and its deletion causes a biofilm defect, which was found to be regulated by a noncoding RNA, RrrR (RNase R-resistant RNA) (36). Saci_2184 and Saci_2183 encode a putative VapB type II antitoxin and its cognate VapC toxin, respectively. Using TAfinder (37), 17 type II toxin-antitoxin pairs are predicted in the S. acidocaldarius genome (Table 1; see Fig. S1 in the supplemental material). The clustering of TA loci in the genome indicates a possible horizontal inheritance by mobile genetic elements (Fig. S1). Of these 17 potential TA pairs, two sets, Saci_1957/Saci_1956 and Saci_2111/Saci_2112, are not associated with any TA type and have predicted toxins of an abnormally long length, suggesting they are unlikely to be type II TAs. One identified TA pair was predicted as an MNT/HEPN-like system, which is common to thermophiles (38). MNT (minimal nucleotidyltransferase)-type antitoxins inactivate their cognate HEPN (higher eukaryotes and prokaryotes nucleotidebinding) ribonucleases by AMPylation (39). The remaining TAs were identified as the VapBC type, including Saci_2184 (referred to here as VapB14). Interestingly, the VapB14 predicted cognate VapC-type toxin Saci_2183 (referred to here as VapC14) was unresponsive in biofilm cells. In fact, except for vapB14, all other identified toxin or antitoxin genes were largely unresponsive in the biofilm (Fig. 1). All toxin and antitoxin genes, including vapB14, were between the 25th and 75th percentiles of the transcriptome profile for both the biofilm and planktonic conditions. Follow-up relative quantitative real-time PCR (qPCR) analysis of S. acidocaldarius MW001 biofilm and planktonic cultures revealed a significant 2.2-fold increase in vapB14 expression under the biofilm condition on day 3 ( Fig. 2A), confirming these results. Moreover, vapB14 is upregulated 2.9-fold in Saci_1223 (biofilm activator) mutant planktonic cells compared to the parent, suggesting that Saci_1223 is a repressor of vapB14 in planktonic growth ( Table 2).
Reduction of VapC14 toxin activity by its cognate VapB14 antitoxin. The VapC14 toxin and the VapB14 antitoxin were recombinantly expressed and tested for their associated activities. Purified fractions of the VapC14 toxin had a pinkish hue and correlated with the ;25-kDa VapC14 toxin band on an SDS-PAGE gel (Fig. S2). This coloration in VapC14-containing fractions may be due to the coelution of manganese ions, pinkish in aqueous solutions, that crystal structure studies indicate are present within VapC-type toxin active sites (17,18). Additionally, significant RNase activity was measured in 5 mg of the VapC14 compared to the no-protein control, confirming its function as an RNase-type toxin ( Fig. 2B; Fig. S3). No RNase activity was measured in the VapB14 antitoxin alone, and the VapC14 toxin's activity was completely abolished by the addition of the metal ion-chelating agent EDTA (Fig. S3). Addition of 10 mg of the VapB14 antitoxin led to a mild 15% reduction in detectable VapC14 toxin activity. However, VapC14 RNase activity was significantly reduced with the addition of 20 mg (34%) and 30 mg (30%) of VapB14, confirming the antitoxin function of VapB14 (Fig. 2B).
Regulation of the vapBC14 locus. Type II toxin-antitoxin systems are typically organized in an operon with the antitoxin upstream and overlapping the toxin or separated by a small intergenic region (35). This paradigm also applies to the vapBC14 locus as the vapB14 antitoxin gene is upstream of the vapC14 gene, with only a 25-bp intergenic region. Often type II antitoxins autorepress their own operon in conjunction with their cognate toxin through a process called conditional cooperativity; repression of the operon is dependent on the ratio of toxin to antitoxin in the regulating TA complex (27,28). If the ratio is skewed toward antitoxin binding, repression occurs; if the ratio is skewed toward toxin, repression is relieved. If autorepression was the only impact on the expression of the vapBC14 locus, then deletion of the vapB14 antitoxin gene would cause an increase in the transcription of vapC14. However, qPCR showed a significant 5.7-fold decrease in vapC14 expression in the DvapB14 mutant compared to the MW001 parent strain in planktonic cultures (Fig. 2C), indicating the VapB14 antitoxin may not be regulating the vapC14 promoter, as predicted. Role of the VapB14 antitoxin in planktonic and biofilm growth. Single and double deletion mutants were generated for the vapBC14 locus, and continuous monitoring of planktonic cultures was performed to determine culture fitness (Fig. 2D). The DvapBC14 toxin-antitoxin mutant and the DvapC14 toxin mutant grew similarly to the MW001 parent strain. However, planktonic growth of the DvapB14 antitoxin mutant exhibited a significant growth defect compared to any other strain at 48 and 72 h (Fig. 2D): 14% and 36% less than MW001, respectively. Neither the DvapC14 toxin nor the DvapBC14 toxin-antitoxin mutants were significantly different from the parent at any time point. Additionally, vapB14 expression increased a significant 2.6-fold in MW001 day 4 compared to day 1 planktonic cultures, indicating that this antitoxin may also play a role in late-stationary-phase growth ( Fig. 2A).
Using crystal violet staining, the vapBC14 mutant panel was evaluated for their ability to generate biofilms. Both the DvapC14 toxin and the DvapBC14 toxin-antitoxin mutant were biofilm overproducers compared to the MW001 parent, generating 47% and 65% more biofilm on day 3 and 124% and 119% more biofilm on day 4, respectively ( Fig. 3B and D). The DvapB14 antitoxin mutant was deficient in biofilm growth compared to the MW001 parent strain at every time point, with a significant difference on days 1 to 3 (Fig. 3B). Even accounting for the DvapB14 antitoxin mutant's growth defect by normalizing the biofilm data to the overall growth of the well (optical density at 600 nm [OD 600 ]), the DvapB14 antitoxin mutant retained a significant biofilm growth defect on day 2 (47% decrease) and day 3 (37% decrease) (Fig. 3C). The VapB14 antitoxin may act as an activator of biofilm formation by regulating genes such as abfr1, which encodes a S. acidocaldarius biofilm repressor, but more evidence is needed to support this possibility. The biofilm defect seen in the DvapB14 antitoxin mutant could be due to the unfettered VapC14 toxin targeting important biofilm RNAs, such as the transcript of the known Lrs14-like biofilm activator (Saci_1223).
Response of known biofilm genes to absence of the VapB14 antitoxin. The response of known biofilm genes in the DvapB14 antitoxin mutant was determined via qPCR on 3-day-old biofilm and planktonic samples. The biofilm repressor abfR1, Saci_1242  SEM. The DvapB14 mutant is significantly different from the MW001 parent at 48 h (l, P , 0.10) and 72 h (**, P , 0.05). Statistical differences were calculated using a two-way analysis of variance (ANOVA) followed by a Tukey honestly significant difference (HSD) post hoc test for panels A, B, and D and a Dunnett T3 test for panel C.
VapB Antitoxin Regulates Dispersal in the Sulfolobales mBio biofilm activator, and UV-inducible pilus genes (upsE and upsA) registered no response to deletion of the vapB14 antitoxin gene under any conditions tested (Fig. S4). Strikingly, a 3fold significant increase was observed in both arlB and arlX archaellum genes under the DvapB14 antitoxin mutant biofilm condition ( Fig. 4A and B). Upregulation of the archaellum in the DvapB14 biofilm, which triggers dispersal from the biofilm, is consistent with the crystal violet experiments showing the DvapB14 mutant produces significantly less biofilm. Furthermore, a slight increase was observed in both archaellum genes in the DvapBC14 double mutant, but no such increase was seen in the DvapC14 single mutant, indicating the importance of the VapB14 antitoxin alone in the regulation of S. acidocaldarius motility. The VapB14 antitoxin directly or indirectly represses the expression of key archaellum genes to minimize biofilm dispersal. Additionally, the deletion of the vapC14 toxin gene caused transcription of the aapA gene to be completely abolished in planktonic and biofilm cells, indicating that VapC14 toxin targets an aapA repressor such as AbfR1 (Fig. 4C). However, the binding of AbfR1 is dependent on its phosphorylation state, which differs across these conditions (15); relieving RNase degradation of abfR1 transcript would not result in complete abrogation of the aapA transcription in both planktonic and biofilm cells, as seen here. VapC14 is likely targeting an unknown repressor of aapA.
Deletion of the vapC14 toxin, with or without the presence of the VapB14 antitoxin, caused a significant increase in aapF in planktonic cells (3.4-fold in the DvapC14 mutant and 4.4-fold in the DvapBC14 mutant) compared to the MW001 parent (Fig. 4D). Antisense RNA transcripts are known to be within the Sa. solfataricus aapF homolog (Sso_2386) (40), and deletion of the aapF gene in S. acidocaldarius causes hyperarchaellation (41), which suggests that potential noncoding RNAs within the aapF gene may repress archaellum expression. The increase in aapF transcription in only DvapC14 toxin mutants indicates that the VapC14 toxin targets either aapF mRNA or these VapB Antitoxin Regulates Dispersal in the Sulfolobales mBio antisense transcripts, allowing them to accumulate in its absence. No significant change in aapF was measured in the DvapB14 antitoxin mutant under either condition, most likely due to the direct repression of aapF by AbfR1 in biofilm cells (14) and the low expression of vapB14 in planktonic cells (Fig. 1).
VapBC14 regulation of S. acidocaldarius surface structures. Transmission electron microscopy (TEM) of 4-day biofilms showed hyperarchaellation of the DvapB14 antitoxin mutant compared to the MW001 parent and DvapC14 toxin mutant (Fig. 5C). Moreover, an increase in archaella was also detected in the DvapB14 antitoxin mutant biofilm compared to any other strain via western blotting using anti-ArlB antibodies ( Fig. 5A and B). EM imaging and western blotting of the DvapBC14 toxin-antitoxin mutant biofilm displayed higher levels of archaella than the MW001 parent ( Fig. 5A to C), further confirming the toxin-independent role of the VapB14 antitoxin in the regulation of S. acidocaldarius dispersal. Additionally, both the DvapC14 toxin single mutant and the DvapBC14 double mutant lacked Aap pilus structures, confirming the qPCR results and indicating that the VapC14 toxin is a strong regulator of Aap pilus production. Furthermore, DvapC14 toxin mutant 4day planktonic cultures were devoid of most surface appendages, supporting the VapC14 toxin's regulation of Aap pili and archaella in planktonic cells. Thin structures referred to as "threads," which are structurally similar to type I pili (42), were unaffected by the VapBC14 replicates. Statistically significant difference compared to the MW001 parent under the same condition: *, P , 0.05. N.D., none detected as the cycle threshold was above the detection limit. All statistical differences were calculated using a two-way analysis of variance (ANOVA) followed by a Tukey honestly significant difference (HSD) post hoc test.
VapB Antitoxin Regulates Dispersal in the Sulfolobales mBio TA system as they were seen in every strain at similar levels. However, the DvapBC14 double mutant biofilm was hyperpiliated with Ups pili and hyperarchaellated in planktonic culture, which may be stress responses to the loss of other surface appendage structures. (Fig. 5C).
VapB14 antitoxin homologs across the Sulfolobales and beyond. It was surprising to find that removal of VapB14, comprised of 114 amino acids, had such a profound impact on growth physiology and biofilm formation processes in S. acidocaldarius. This raises the question of whether homologous antitoxins with similar roles exist in other Sulfolobales. In fact, homologs were identified in many Sulfolobales species (Fig. 6A). Interestingly, all surveyed species had at least one homolog of the VapB14 antitoxin, except for Acidianus brierleyi, Acidianus infernus, Stygiolobus azoricus, and Sulfuracidifex metallicus. Metallosphaera yellowstonensis was the only species with two vapB14 homologs. Conversely, weaker homologs of the VapC14 toxin were found in several Sulfolobales species with a much lower amino acid percentage of identity (Fig. 7). Additionally, synteny analysis using the SyntTax webtool (43) identified several species that contained homologous VapB14 proteins that were not colocalized with the VapC14 toxin gene. Sulfolobales species that have unpaired vapB14 homologs include several members of the Metallosphaera species (Metallosphaera hakonensis, Metallosphaera javensis, Metallosphaera prunae, Metalosphaera sedula, and Metallosphaera tengchongensis), Sulfodiicoccus acidiphilus, and Sulfuracidifex tepidarius. In fact, despite the conservation of a VapB14 homolog (Fig. 6A), a VapC14 homolog is absent from the genome of Sulfu. tepidarius (Fig. 7). This suggests that the role of the VapB14 VapB Antitoxin Regulates Dispersal in the Sulfolobales mBio antitoxin in biofilm development may be conserved among the Sulfolobales and less dependent on the conservation of its cognate toxin. To this point, the transcriptional response of the vapB14 homolog in Metallosphaera sedula, Msed_0871, in 3-day-old biofilms and planktonic cultures was consistent with the response of vapB14 in S. acidocaldrius (Fig. 6B). Since M. sedula Msed_0871 has among the lowest homologies found among the Sulfolobales, other vapB14 homologs with higher similarity likely also play a role in biofilm regulation. Moreover, M. sedula does possess both a vapB14 antitoxin (Fig. 6A) and a vapC14 toxin homolog (Fig. 7), but they are located at disparate locations in the genome, suggesting divergent functions. Toxin-antitoxin systems are often associated with mobile genetic elements and are highly susceptible to horizontal gene transfer between species (35). Because of the inherent mobility of these TA systems, it is common to see a large divergence in nucleotide identity even within the same species (35). However, VapB14 homologs across the Sulfolobales have conserved amino acid and nucleotide sequences (Fig. 6A). Furthermore, VapB14 homologs may be pivotal to regulating motility in Archaea as blastp results show VapB14 homologs in many members of the phylum Crenarchaeota and some examples in the Euryarchaeota. Homologs were even identified in motile mesophilic bacterial genera, such as the pathogenic mesophiles of Pseudomonas and marine bacteria of Nitrosococcus (see Data Set S1 in the supplemental material). While VapB14 clearly has an important role in the regulation of motility and its homologs are found in many motile prokaryotic species, this antitoxin is also present in many nonmotile genera, such as Acidianus of the Sulfolobales, thermophilic

DISCUSSION
The functional study of toxin-antitoxin systems remains controversial, with recent investigations suggesting no phenotypic response to stressors despite a measured transcriptional response (44). However, this study identifies a VapC14 toxin that significantly impacts RNA transcripts contained in the aapF gene and the production of archaella and Aap pilus structures. Additionally, VapC14 RNase activity, although not completely abolished, was significantly reduced by its cognate antitoxin, VapB14, confirming the canonical function of VapB14 as a VapC14-neutralizing antitoxin (Fig. 2B). However, deletion of vapB14 did not cause the expected upregulation of the vapC14 toxin gene (Fig. 2C) and may indicate a third-party regulator of the vapBC14 operon, which has some precedence (45,46). Saci_1223 is a potential candidate as its deletion results in upregulation of vapB14 in planktonic cells ( Table 2). Deletion of vapB14 could result in downstream polar effects; however, removal of vapB14 left no genetic scar, minimizing the potential of this deleterious effect. Furthermore, RNase assays were performed in vitro and may not have been representative of the native conditions inside the cell which could lead to stronger affinity of the VapB14 antitoxin for the VapC14 toxin. Also, antitoxins can be promiscuous, meaning that another antitoxin may aid in the reduction of VapC14 RNase activity and that VapB14 may bind a second toxin. The VapC14 toxin's activity was lower than previously seen for VapC-type toxins in M. sedula (34). This lower activity could be due to a higher specificity for a very narrow range of transcripts, such as aapF, that may not be well represented in the RNA probes available in the kit used for measuring RNase activity. Overall, RNase activity data demonstrated that VapC14 is an RNase-type toxin and that VapB14 does function as the antitoxin to this activity (Fig. 2B). The VapC14 toxin most likely activates attachment by targeting a strong repressor of the aapA archaeal adhesive pilus structural subunit. This is further supported by the complete lack of Aap pilus structures in the DvapC14 and DvapBC14 mutants (Fig. 5C). However, these mutants also are biofilm overproducers, which suggests they are employing an alternative attachment mechanism. Threads, the only non-type IV pilus surface filament on S. acidocaldarius, are present in EM images of all strains. Although the function of threads is still unknown, S. acidocaldarius Ups pilus, Aap pilus, and archaellum triple mutants are capable of making biofilm (6). Threads may be playing a compensatory attachment role in the DvapC14 and DvapBC14 mutants, allowing these strains to produce biofilm. Similarly, the Ups hyperpiliation of the DvapBC14 double mutant may also improve its biofilm production (Fig. 5C). The DvapBC14 double mutant biofilm also had an increase in archaella compared to the parent, which could contribute to biofilm formation as archaella can aid in initial attachment (6,47). Finally, S. acidocaldarius aapX and aapE mutants produce more extracellular matrix, which would contribute to biofilm formation (41). As there were no visible Aap pili in the DvapC14 single or DvapBC14 double mutant (Fig. 5C), it is reasonable to assume that AapX and AapE are absent in these strains yielding excess extracellular matrix. Overall, DvapC14 toxin and DvapBC14 double mutants may be biofilm overproducers through alternative attachment mechanisms and overexpression of extracellular matrix.
Additionally, in planktonic cells, aapF is significantly increased in the absence of the VapC14 toxin (Fig. 4D), which may contain antisense noncoding RNAs like those observed in the Sa. solfataricus aapF homolog (40). Furthermore, deletion of aapF in S. acidocaldarius causes an increase in archaella (41), suggesting that AapF or potential noncoding transcripts within the aapF gene repress archaella expression. An abundance of antisense aapF transcript may function as noncoding RNAs that posttranscriptionally downregulate archaellum gene expression. As is natural for a toxin-antitoxin system, VapC14 and VapB14 may apply opposing regulatory pressure on archaellum expression. VapC14 derepresses the archaella by degrading aapF mRNA or a noncoding RNA during planktonic growth. However, during biofilm growth VapB14 is highly expressed and transcriptionally represses the archaellum. VapB14 also behaves as a traditional antitoxin by neutralizing the VapC14 toxin, allowing the archaellum to be posttranscriptionally repressed (Fig. 8). While nutrient starvation is known to induce expression of the archaellum in S. acidocaldarius (48) through regulation by ArnR (12), the VapBC14 TA system's regulation of the archaellum is responsive to biofilm growth rather than nutrient availability.
Unlike Bacteria, noncoding RNAs are plentiful in the genomes of Archaea (40). Since Archaea lack sigma factors and have an abundance of type II toxin-antitoxin systems, this may point to RNase activity of type II toxins as an important regulatory mechanism within this domain. Specifically, homologs of both AapF and VapB14 are found in most surveyed Sulfolobales species (Fig. 6A; see Fig. S5, Data Set S3, and Text S1 in the supplemental material), suggesting that the mechanism of archaellum regulation described here may be a multispecies phenomenon that is present in all archaellated Sulfolobales. This is further supported by the similar upregulation of the M. sedula VapB14 homolog (Msed_0871) to biofilm growth (Fig. 6B). Additionally, conservation of vapB14 is independent of vapC14, as several Sulfolobales species either carry these genes at distinct locations or possess only a vapB14 antitoxin gene. Moreover, VapB14 homologs are prevalent in the archaeal phylum Crenarchaeota and present in some species of the phylum Euryarchaeota (see Data Set S1 in the supplemental material). While the importance of VapB14 in motile Archaea is evident, homologs are also found in some bacterial species and nonmotile organisms, indicating that VapB14 may also have other functions.
TA systems are prevalent in bacteria, archaea, and fungi (16), suggesting that the evolution of the current type II TA systems is not a recent occurrence. Fossil evidence has also shown that prokaryotic organisms of both bacterial and archaeal origins were forming multicellular biofilms more than 3 billion years ago (1). In fact, the earliest recorded occurrences of biofilms are in hydrothermal environments like those native to the species of the Sulfolobales (1). It is, therefore, possible that this VapBC14 TA system may have coevolved with the ability to form a biofilm within this thermophilic order. Furthermore, nucleotide sequence conservation of a type II TA system is atypical due to their association with mobile genetic elements and tendency toward horizontal gene transfer (35,49). However, VapB14 is highly conserved across the Sulfolobales (Fig. 6A), indicating an evolutionary selective pressure to maintain this small but important biofilm-regulating protein. Overall, the bifunctional VapB14 antitoxin has evolved as an important regulator of Sulfolobales, and perhaps archaeal, motility not only by inhibiting the activity of its cognate toxin but also through transcriptional repression of the archaellum (Fig. 8).

MATERIALS AND METHODS
RNase activity assay. RNase activity assays were performed as described previously using the RNaseAlert kit (Integrated DNA Technologies) (34). For each reaction mixture containing VapC14, 5 mg of VapC14 toxin was added, with or without the addition of 10, 20, or 30 mg of VapB14 antitoxin or 25 mM EDTA. A no-protein control and 10 mg of a VapB14 antitoxin-alone control were also performed. All reaction mixtures were prewarmed at 75°C for 5 min to activate the VapC14 toxin and VapB14 antitoxin.
Planktonic growth curves. Cultures were inoculated from frozen stocks in 50 mL 75°C prewarmed Brock's salts (pH 3) supplemented with 0.1% NZ-amine, 0.2% sucrose, and 0.01g/L uracil. Cultures of S. acidocaldarius MW001 and the DvapB14, DvapC14, and DvapBC14 mutants were grown aerobically in foam-stoppered flasks at a 1:5 volume/flask ratio at 75°C at 150 rpm. Cultures were monitored by back scatter at 520 nm every 15 min with the Cell Growth Quantifier (Scientific Bioprocessing, Inc.) for 72 h.
Crystal violet biofilm assay. The S. acidocaldarius MW001 parent and DvapB14 antitoxin, DvapC14 toxin, and DvapBC14 toxin-antitoxin mutant biofilms were grown in 1 mL of Brock's basal salts (pH 3) with 0.1% NZ-amine, 0.2% sucrose, and 0.01 g/L uracil on Sarstedt Cell 1 flat-bottom 24-well plates at 75°C for a period of 1 to 4 days. The outer wells of each 24-well plate were filled with 1 mL of water, and plates were incubated in a humidified box to reduce evaporation. Prior to staining, optical density at 600 nm (OD 600 ) was read as a measure of overall well growth. Supernatant was then removed, attached VapB Antitoxin Regulates Dispersal in the Sulfolobales mBio biofilm was stained with 500 mL 0.1% crystal violet, the biofilm was washed twice with 1 mL of water, crystal violet was solubilized with 500 mL 100% ethanol, and absorbance was read at 550 nm. RNA isolation and quantitative real-time PCR. S. acidocaldarius MW001 and mutant planktonic cultures were grown in 50 mL 75°C prewarmed Brock's salts (pH 3) supplemented with 0.1% NZ-amine, 0.2% sucrose, and 0.01g/L uracil. Cultures were inoculated at an OD 600 of 0.01 and grown aerobically in foam-stoppered flasks at a 1:5 volume/flask ratio at 75°C at 150 rpm for 4 days. On day 2, 20 mL of sterile 75°C deionized water was added to each flask. The entire culture of cells was centrifuged at 4,000 Â g for 5 min and resuspended in 1 mL of RNAlater (Invitrogen). S. acidocaldarius MW001 and mutant biofilm cultures were inoculated in the same prewarmed medium at an OD 600 of 0.01 in 150-by 20-mm Sarstedt Cell 1 tissue culture dishes (Sarstedt). Biofilms were incubated at 75°C in a sealed humidified box for 4 days. RNA was then extracted on days 1 to 4 for S. acidocaldarius MW001 and the DvapB14 antitoxin mutant and on day 3 for the DvapC14 toxin mutant and DvapBC14 double mutant using TRIzol reagent followed by the RNeasy kit (Qiagen) (50). Residual DNA was removed by a rigorous treatment with Turbo DNase (Invitrogen). RNA was determined to be relatively free of DNA contamination by qPCR, checking for amplification using secY primers (see Data Set S2 in the supplemental material) with RNA as the template. Relative qPCR was performed using SsoFast EvaGreen supermix (Bio-Rad) or SsoAdvanced Universal SYBR green supermix (Bio-Rad), and fold change values were calculated using the Livak method (51) with secY used as the normalizer.
M. sedula planktonic and biofilm cultures were inoculated as described above with the following exceptions. Cultures were inoculated in 50 mL 70°C prewarmed Brock's salts (pH 2) supplemented with 0.1% yeast extract and incubated at 70°C at 150 rpm for planktonic cultures and 70°C stationary for biofilm plates for 3 days. RNA was determined to be relatively free of DNA contamination by qPCR, checking for amplification using Msed_R0026 16s gene primers (Data Set S2) with RNA as the template. Relative qPCR was performed using SsoAdvanced Universal SYBR green supermix (Bio-Rad), and fold change values were calculated using the Livak method (51) with Msed_R0026 used as the normalizer.
Transmission electron microscopy of S. acidocaldarius surface appendage structures. Biofilms and planktonic S. acidocaldarius MW001 and mutant strains were grown as for RNA isolation. Biofilm was scraped off the petri dishes and resuspended in 1 mL growth medium. Five microliters of biofilm or planktonic cells were applied on freshly glow-discharged carbon/Formvar-coated copper grids (300 mesh; Plano GmbH) and incubated for 30 s. The excess liquid was blotted away, and cells were negatively stained with 2% uranyl acetate. Imaging was done with Hitachi HT7800 operated at 100 kV, equipped with an EMSIS Xarosa 20-megapixel CMOS camera.
Western blotting. To assay the production of ArlB in planktonic cells and biofilm, S. acidocaldarius MW001 and mutant strains were grown as for RNA isolation. Biofilm was scraped off the petri dishes and resuspended in 1 mL growth medium. The OD 600 s of cells from biofilm and planktonic cultures were determined. Cells were pelleted at 2,400 Â g in a tabletop centrifuge for 10 min and resuspended to a theoretical OD of 10 in 1Â SDS loading dye. The whole-cell samples were separated on SDS-PAGE and blotted on polyvinylidene difluoride (PVDF) membrane (Roche). The membrane was blocked with I-Block (Thermo Fisher Scientific) and incubated in primary antibody against ArlB (Eurogentec) overnight at 4°C. Afterwards, the membrane was incubated in secondary goat anti-rabbit antibody coupled to horseradish peroxidase (HRP) overnight at 4°C. Chemiluminescent signals were recorded with IBright 1500 (Invitrogen, Thermo Fisher Scientific) using Clarity Western ECL enhanced chemiluminescence blotting substrate (Bio-Rad).
Data availability. Microarray data are available at the Gene Expression Omnibus repository (NCBI) under accession no. GSE226483 for normalized data and accession no. GSM7077821, GSM7077822, GSM7077823, and GSM7077824 for raw data files. See Text S1 for additional methods.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.04 MB.