ABSTRACT
Halophilic archaea often inhabit environments with limited oxygen, and many produce ion-pumping rhodopsin complexes that allow them to maintain electrochemical gradients when aerobic respiration is inhibited. Rhodopsins require a protein, an opsin, and an organic cofactor, retinal. We previously demonstrated that in Halobacterium salinarum, bacterioopsin (BO), when not bound by retinal, inhibits the production of bacterioruberin, a biochemical pathway that shares intermediates with retinal biosynthesis. In this work, we used heterologous expression in a related halophilic archaeon, Haloferax volcanii, to demonstrate that BO is sufficient to inhibit bacterioruberin synthesis catalyzed by the H. salinarum lycopene elongase (Lye) enzyme. This inhibition was observed both in liquid culture and in a novel colorimetric assay to quantify bacterioruberin abundance based on the colony color. Addition of retinal to convert BO to the bacteriorhodopsin complex resulted in a partial rescue of bacterioruberin production. To explore if this regulatory mechanism occurs in other organisms, we expressed a Lye homolog and an opsin from Haloarcula vallismortis in H. volcanii. H. vallismortis cruxopsin-3 expression inhibited bacterioruberin synthesis catalyzed by H. vallismortis Lye but had no effect when bacterioruberin synthesis was catalyzed by H. salinarum or H. volcanii Lye. Conversely, H. salinarum BO did not inhibit H. vallismortis Lye activity. Together, our data suggest that opsin-mediated inhibition of Lye is potentially widespread and represents an elegant regulatory mechanism that allows organisms to efficiently utilize ion-pumping rhodopsins obtained through lateral gene transfer.
IMPORTANCE Many enzymes are complexes of proteins and nonprotein organic molecules called cofactors. To ensure efficient formation of functional complexes, organisms must regulate the production of proteins and cofactors. To study this regulation, we used bacteriorhodopsin from the archaeon Halobacterium salinarum. Bacteriorhodopsin consists of the bacterioopsin protein and a retinal cofactor. In this article, we further characterize a novel regulatory mechanism in which bacterioopsin promotes retinal production by inhibiting a reaction that consumes lycopene, a retinal precursor. By expressing H. salinarum genes in a different organism, Haloferax volcanii, we demonstrated that bacterioopsin alone is sufficient for this inhibition. We also found that an opsin from Haloarcula vallismortis has inhibitory activity, suggesting that this regulatory mechanism might be found in other organisms.
INTRODUCTION
Many integral membrane protein complexes, including those that carry out universal processes such as energy conversion and the response to environmental changes, require organic cofactors. There are compelling reasons for cells to balance the production of proteins and these corresponding cofactors, as both proteins and cofactors require energy for synthesis, and an imbalance of proteins and cofactors could potentially harm cells. This balance represents a challenging biological problem, since the mechanisms for synthesizing proteins and small molecules are not inherently linked.
To explore this fundamental question, we studied the relatively simple microbial rhodopsin complex that consists of a single opsin protein and a 20-carbon retinal cofactor. Microbial rhodopsins undergo a conformational change in response to light and function to transport ions or mediate phototaxis. After their initial discovery in halophilic archaea, microbial rhodopsins are now known to be widely distributed (reviewed in reference 1), and substantial evidence indicates that this distribution is partially due to frequent lateral gene transfer (2, 3). Although obtaining genes encoding opsins and retinal biosynthetic enzymes allows organisms to better exploit light in their environment, no benefit is gained if insufficient retinal is available to form functional complexes. Retinal biosynthesis may share intermediates with other cellular processes, and it is not known how organisms ensure that resources are devoted to produce the retinal needed for rhodopsins. In the work described here, we further characterized a regulatory mechanism in which opsins, when not bound by retinal, inhibit another process to preserve shared intermediates for retinal biosynthesis.
Bacteriorhodopsin (BR), a light-driven proton pump in the halophilic archaeon Halobacterium salinarum, was the first microbial rhodopsin described (4) and has been extensively studied as a model for membrane protein structure and activity (reviewed in references 5–7). Halophilic archaea thrive in high-salinity environments, such as salt crystallizer ponds, and often grow to densities of 107 to 108 cells ml−1 (8). With this high density, oxygen availability is limited, inhibiting aerobic respiration and necessitating the production of ion-pumping rhodopsins like BR. Under conditions of low oxygen, BR production is induced, and the BR apoprotein bacterioopsin (BO; UniProtKB accession no. P02945) and retinal are maintained in ∼1:1 stoichiometry (9). In H. salinarum, retinal is synthesized de novo with the final synthesis steps being the cyclization of lycopene to form β-carotene (10) and the subsequent cleavage of β-carotene to retinal (11). Lycopene is also used to synthesize bacterioruberins (12–14), 50-carbon carotenoids that may function to increase membrane rigidity (15) and provide protection from UV light (16). The conversion of lycopene to the first bacterioruberin intermediate is catalyzed by the lye-encoded enzyme lycopene elongase (Lye; UniProtKB accession no. Q9HPD9) (14, 17). Thus, lycopene is the last shared intermediate for bacterioruberin and retinal synthesis, and the Lye enzyme is a potential target for regulation of these pathways.
We previously described a novel regulatory mechanism in H. salinarum in which BO, when not bound by retinal, inhibits Lye activity so that bacterioruberin production is reduced and lycopene is available for synthesis of the retinal cofactor (Fig. 1). To determine the specificity of this inhibition, we then compared the activity of H. salinarum Lye to the Lye homolog from Haloferax volcanii, a related halophilic archaeon that does not express an opsin but does produce bacterioruberins. When expressed in H. salinarum, H. volcanii Lye also converted lycopene to the first bacterioruberin intermediate but was not inhibited by BO (14). Here, we build on this previous work by expressing opsin and Lye homologs in H. volcanii. We demonstrate that BO is sufficient to inhibit H. salinarum Lye in the absence of any other factors from H. salinarum. We also provide insight into potential features of Lye and BO that are required for BO-mediated inhibition of bacterioruberin synthesis. Lastly, we demonstrate that Haloarcula vallismortis cruxopsin-3 inhibits bacterioruberin synthesis catalyzed by the H. vallismortis lye homolog. This finding suggests that interactions between opsins and other proteins may be widespread and raises the possibility that this regulatory mechanism confers a selective advantage to organisms that express ion-pumping rhodopsins.
Lycopene is the last shared intermediate in retinal and bacterioruberin biosynthesis. CrtY, the lycopene cyclase that catalyzes the conversion of lycopene to β-carotene that is subsequently cleaved to form retinal. Lye, lycopene elongase that catalyzes the prenylation of lycopene to the first bacterioruberin intermediate. Broken line, the regulatory mechanism of BO inhibition of Lye activity.
RESULTS
H. volcanii lye homolog required for bacterioruberin synthesis.We previously determined that the H. volcanii lye homolog (HVO_RS16895; UniProtKB accession no. D4GTV9) rescued the Δlye phenotype in H. salinarum, and expression of H. volcaniilye in lycopene-producing Escherichia coli resulted in a bacterioruberin precursor (14). To test if the lye gene was required for bacterioruberin synthesis in H. volcanii, we constructed an in-frame deletion of residues 45 to 226 from the predicted 301-amino-acid (aa) protein. The H. volcanii Δlye strain lacked detectable bacterioruberin production and resulted in an accumulation of lycopene (Fig. 2). Bacterioruberin synthesis was restored with the introduction of either H. volcanii or H. salinarumlye on an expression vector (Fig. 2), demonstrating that Lye is required for the conversion of lycopene to bacterioruberins.
The H. volcanii lye gene is required for bacterioruberin synthesis. RP-UHPLC traces of carotenoid extracts from H. volcanii Δlye, H. volcanii H1209 (parental strain), and strains harboring plasmids expressing lye from H. volcanii, H. salinarum, and H. vallismortis as labeled. Positions of bacterioruberin and lycopene (lyc) standards are noted. Traces were normalized for total carotenoid concentration, corrected for slight differences in retention time using an internal standard, and offset along the vertical axis for clarity. Traces are representative of at least 3 replicates for each strain.
BO expression in H. volcanii sufficient to inhibit H. salinarum Lye activity. H. volcanii lacks genes for any opsins (18) and is highly unlikely to have any native factors that interact with opsins. H. volcanii therefore represents an ideal system to test the possibility that BO expression does not require other proteins from H. salinarum to inhibit lycopene elongase activity. We first used site-directed recombination (19) to place the H. salinarum lye gene at the H. volcanii lye locus. The H. salinarum Lye enzyme was functional, as bacterioruberin synthesis was restored in this strain (Fig. 3A). To test if the presence of BO was sufficient to inhibit the H. salinarum Lye enzyme, we transformed the strain expressing H. salinarum lye with an expression vector harboring bop, the gene that encodes BO. In H. volcanii with its native lye gene, bacterioruberin synthesis was unaffected by BO expression (Fig. 3A). In contrast, when H. salinarum lye replaced H. volcanii lye, BO expression reduced bacterioruberin to about 16% of the total carotenoids (Fig. 3A) with the remaining 84% as lycopene. To determine if this inhibition was a specific effect of the opsin, we added retinal during growth of the culture to convert the BO to BR. Retinal addition resulted in a partial rescue of bacterioruberin levels to 63% of the total carotenoid (bacterioruberin and lycopene) in the cells (Fig. 3A). These results indicated that BO is sufficient to inhibit the H. salinarum Lye enzyme.
Expression of BO in H. volcanii specifically inhibits H. salinarum Lye. (A) Box plot (Tukey's [43, 44]) indicating bacterioruberin levels as determined by RP-UHPLC analysis of H. volcanii cultures with H. volcanii or H. salinarumlye. Strains harbored a BO expression vector or empty vector (control) as indicated. Where noted, retinal was added during the growth of the culture. Carotenoid levels were quantified by RP-UHPLC as described in Materials and Methods. Heavy horizontal bars indicate the median values, and boxes demarcate the upper and lower quartiles. Whiskers extend to the smaller value of 1.5 times the interquartile range or the most extreme value, n ≥ 3. (B) Photographs of representative colonies of H. volcanii strains with native lye replaced with H. salinarum lye. Empty vector control (left), strain with plasmid allowing BO expression (middle), strain with plasmid allowing BO expression with retinal added to the plate to convert BO to BR (right). (C) Box plot indicating bacterioruberin levels (ruberin metric) as determined by analysis of colony color. H. volcanii cultures were plated on Hv-YPC medium supplemented with retinal as noted. Four-day-old colonies were photographed and digital images were analyzed as described in Materials and Methods. Heavy horizontal bars indicate the median values, and boxes demarcate the upper and lower quartiles. Whiskers extend to the smaller value of 1.5 times the interquartile range or the most extreme value, n ≥ 9. Asterisks indicate Bonferroni adjusted P of <0.05. NS, not significantly different.
Developing colony color analysis to assess bacterioruberin production.To more rapidly and efficiently quantify bacterioruberin production in H. volcanii strains, we developed an assay based on colony color. Bacterioruberin imparts a pink color, and this color is more intense in colonies with greater concentrations of bacterioruberin (Fig. 3B). Large colonies were grown on solid media, supplemented with retinal where noted, and pictures taken with a digital camera mounted to a dissection microscope under controlled lighting. Using image analysis software, a variety of color measures were recorded. These measures were modeled using beta regression for their ability to predict bacterioruberin as a percentage of total carotenoid in cultures as measured by reverse-phase ultra-high-performance liquid chromatography (RP-UHPLC) for multiple strains (see the supplemental material). Using the parameters of the model, we calculated a metric for bacterioruberin proportion. This ruberin metric facilitated high-throughput screening and comparisons of strains in the absence or presence of retinal (Fig. 3C). Although colony color analysis exhibited greater variability than RP-UHPLC, our mean values were strikingly similar to those determined by analysis of carotenoids from liquid culture (compare Fig. 3A and C).
H. salinarum Lye residues 69 to 120 necessary for BO-mediated inhibition.To identify the region of H. salinarum Lye required for BO-mediated inhibition, we constructed “hybrid” genes that contained portions of both H. salinarum and H. volcaniilye (Fig. 4A). These genes were integrated into the H. volcaniilye locus, and inhibition was tested by expression of BO. BO-mediated inhibition of Lye was not observed when the enzyme included only the N-terminal 68 amino acid residues from H. salinarum (Fig. 4B). However, hybrid Lye enzymes that included larger sections of H. salinarum Lye were inhibited by BO expression (Fig. 4B). Together, our data demonstrate that the H. salinarum Lye residues 69 to 120 are necessary to allow inhibition by BO. Unfortunately, hybrid genes that encoded an N terminus from H. volcanii did not confer substantial bacterioruberin production (data not shown), so we could not test if this region was sufficient to allow BO-mediated inhibition.
A specific region within H. salinarum Lye is necessary for BO-mediated regulation. (A) Protein alignment map of H. salinarum and H. volcanii Lye homologs and hybrids. The x axis represents the consensus alignment amino acid position, and the vertical lines within the protein sequences represent gaps in the alignment. Dashed lines delineate the H. salinarum Lye region required for inhibition by BO. (B) Box plot indicating bacterioruberin levels of strains expressing different versions of lye in the absence or presence of BO. Images were obtained and analyzed, and data were plotted as described for Fig. 3. n ≥ 9. Asterisks indicate Bonferroni adjusted P of <0.05. NS, not significantly different. (C) Template-based modeling of H. salinarum and H. volcanii Lye proteins. Structure predictions of H. salinarum (left) and H. volcanii (right) Lye. Structures are oriented so that the cytoplasm (in) is at the bottom and the extracellular environment (out) is at the top of the image. The regions highlighted in blue (H. salinarum) and cyan (H. volcanii) represent the identified 52-aa region required for inhibition by BO. The structural models for the two Lye homologs were generated using PHYRE2 (20) and superimposed using SuperPose (45). The structure files were highlighted and exported to image files using Geneious version 7 (46).
To assess potential structural differences in the region required for inhibition, we generated template-based secondary structure predictions (20) of H. salinarum and H. volcanii Lye proteins (Fig. 4C). Both homologs were predicted to be integral membrane proteins with nine transmembrane domains. For both predictions, the highest template match was to a structure of 4-hydroxyboenzoate octaprenyltransferase, a UbiA superfamily member from the thermophilic archaeon Aeropyrum pernix (21) with at least a 90% confidence score over more than 85% of the length of the Lye proteins. Similar to the predicted activity of Lye, this enzyme uses isopentenyl pyrophosphate as a substrate to attach prenyl groups to secondary metabolites. Based on this model, the 52-aa region necessary for inhibition has a similar overall structure in both H. salinarum and H. volcanii enzymes and consists of a portion of a cytoplasmic loop, a transmembrane helix, a very short extracellular loop, and another transmembrane helix (Fig. 4C).
Ablation of BO-mediated inhibition of bacterioruberin biosynthesis requires covalent binding of retinal.After observing that retinal addition to convert BO to BR largely reversed the inhibition of bacterioruberin biosynthesis (Fig. 3), we sought to determine if this reversal required covalent binding of retinal to BO. We expressed BO with the lysine residue at position 216 in the mature protein changed to alanine. This lysine (K216) is required for the Schiff-base linkage to retinal (22, 23). The degree to which K216A BO inhibited H. salinarum Lye activity was similar to that of wild-type BO (Fig. 5A). Retinal addition resulted in a slight, but not statistically significant (permutation test, Bonferroni adjusted, P = 0.11), increase in bacterioruberin, but the degree of the increase was substantially less than that resulting from wild-type BO (Fig. 5A).
BO lacking covalent retinal binding inhibits Lye; H. salinarum sensory opsin II does not inhibit Lye. (A) Box plot indicating bacterioruberin levels of H. volcanii strains with H. salinarum lye expressing BO, BO K216A, or SOII in the absence or presence of retinal. Images were obtained and analyzed, and data were plotted as described for Fig. 3. n ≥ 9. Asterisk indicates Bonferroni adjusted P of <0.05. NS, not significantly different. (B) Quantification of H. salinarum BO and SOII and H. vallismortis CO expression. H. volcanii strains harboring expression plasmids for indicated opsins with C-terminal His6 tags were grown in cultures with the same procedures used for carotenoid analysis. Cell lysates were normalized using total protein concentration and separated by polyacrylamide gel electrophoresis. Blots were probed with anti-His antibody and bands were quantified by densitometry. Median values compared to BO median set to 1. Error bars indicate 1 standard deviation, n = 3. Inset, representative immunoblot of His-tagged BO, SOII, and CO.
H. salinarum sensory opsin II does not inhibit Lye.In addition to BR, H. salinarum produces three other rhodopsins: the chloride ion pump halorhodopsin and two sensory rhodopsins. Since these opsins also require retinal, we tested whether they inhibited Lye activity. We constructed H. volcanii strains expressing each H. salinarum opsin. Immunoblotting of C-terminally His-tagged versions of these proteins revealed that only sensory opsin II (SOII; UniProtKB accession no. P71411), the protein component of sensory rhodopsin II (SRII), was expressed at levels comparable to, although less than, those of BO (Fig. 5B). Like BO, SOII was detected in the membrane fraction, indicating that it was appropriately targeted to the plasma membrane of H. volcanii (Fig. S3 in the supplemental material). Expression of SOII did not affect bacterioruberin synthesis, and the addition of retinal to this strain also did not change bacterioruberin production (Fig. 5A). This result indicates that BO has specific structural features that allow inhibition of Lye activity, and these features are not present in SOII.
Evidence of an opsin-Lye interaction in another organism.As a first step to determine if opsins in other organisms also inhibit the activity of lycopene elongase enzymes, we analyzed potential inhibition caused by cruxopsin-3 (CO; UniProtKB accession no. P94854), an opsin identified in H. vallismortis (24). Like BO, CO covalently binds retinal to form a rhodopsin that functions as a proton pump (25). An expression vector encoding CO was constructed and introduced into H. volcanii strains with H. volcanii or H. salinarum lye. CO expression was confirmed by immunoblotting of His-tagged CO, although at approximately half the abundance of BO (Fig. 5B), and the protein localized to the H. volcanii membrane (Fig. S3). CO expression in H. volcanii had no observable effect on bacterioruberin production catalyzed by H. volcanii or H. salinarum Lye enzymes (Fig. 6).
H. vallismortis Lye is specifically inhibited by H. vallismortis CO. Box plot indicating bacterioruberin levels of H. volcanii strains with H. volcanii, H. salinarum, or H. vallismortis lye and the indicated opsin in the absence or presence of retinal. Images were obtained and analyzed, and data were plotted as described for Fig. 3. n ≥ 9. Asterisks indicate Bonferroni adjusted P of <0.05. NS, not significantly different.
We then tested for CO-mediated inhibition of the H. vallismortis Lye enzyme. A similarity search of the H. vallismortis genome sequence (available from the Joint Genome Institute [http://genome.jgi.doe.gov ]) revealed a gene, BLV10_RS12820, predicted to encode a 294-aa protein (UniProtKB accession no. A0A1H2XMF1) homologous to H. salinarum Lye (60% identity over 276 amino acids) and H. volcanii Lye (60% identity over 273 amino acids). The H. vallismortis lye homolog rescued bacterioruberin synthesis when expressed from a plasmid (Fig. 2) or integrated into the H. volcaniilye locus (Fig. 6), confirming the functional identity of the gene. Expression of CO sharply reduced bacterioruberin production when catalyzed by H. vallismortis Lye, and this inhibition was partially relieved when retinal was added (Fig. 6). H. salinarum BO expression had no significant effect on bacterioruberin synthesis in the strain with H. vallismortis Lye (Fig. 6), indicating that the opsin-Lye interaction is specific. This result shows that opsin-mediated inhibition of Lye is not unique to H. salinarum and may be a common mechanism to regulate retinal production.
DISCUSSION
BO expression sufficient to inhibit H. salinarum Lye.By expression of H. salinarum genes in H. volcanii, we determined that BO is sufficient to inhibit the bacterioruberin biosynthesis catalyzed by H. salinarum Lye, and this inhibition is released by addition of retinal to convert BO to BR (Fig. 3). The most plausible explanation for these findings is that BO, in the absence of its retinal cofactor, directly binds Lye. When sufficient retinal is available to convert BO to BR, Lye is released and catalyzes the conversion of lycopene to bacterioruberins.
We also demonstrated that BO with the K216A point mutation, which does not covalently bind retinal, inhibits Lye activity even in the presence of retinal (Fig. 5A). Therefore, BO must undergo a structural change upon retinal binding that eliminates its inhibiting function. We are currently working toward identifying mutations in BO that allow simultaneous covalent binding of retinal and inhibition of Lye activity. These modified BR complexes may, as for wild-type BR, form the two-dimensional crystal lattice that allows relatively straightforward purification and structural determination of BR. These structural analyses would provide insight into the structure of the BO apoprotein and changes that occur upon retinal binding.
Inhibition of activity likely mediated by a specific region of Lye.By expression of hybrid genes that contained regions of both H. salinarum and H. volcanii lye, we determined that H. salinarum Lye residues 69 to 120 were necessary for inhibition (Fig. 4A and B). This region does not encompass either of the two conserved aspartate-rich motifs that likely mediate substrate binding or catalysis in UbiA prenyltransferases (21, 26), indicating that variability in this region could allow regulation while maintaining the catalytic core of the enzyme. For both Lye proteins, the transmembrane helices in the region required for inhibition are predicted to be accessible to possible interactions with other membrane proteins (Fig. 4C), suggesting that a potential BO-Lye interaction could occur between transmembrane regions. Interestingly, the signaling by sensory rhodopsins to their cognate transducers is known to be mediated through contacts within the membrane (27), raising the possibility of similar interactions between BO and Lye. Although the H. salinarum and H. volcanii Lye proteins are predicted to have high structural similarity in the portion of the protein necessary for regulation (Fig. 4C), the largest section of predicted disordered structure lies in the cytoplasmic loop within this region. This finding suggests that conformational changes in this loop may regulate enzyme activity.
Based on sequence similarity, Lye is a member of the UbiA superfamily, integral membrane proteins found throughout all domains of life that catalyze the addition of a prenyl group to a variety of substrates (reviewed in reference 28). Intriguingly, UbiA enzymes, including octaprenyltransferases encoded by CoQ, UbiA, and MenA genes, catalyze key steps in quinone biosynthesis and use polyprenyl pyrophosphate molecules that could also be used as precursors for retinal synthesis. Ion-pumping rhodopsins are often upregulated in low-oxygen conditions when the need for quinones for respiration and oxygen scavenging is reduced. Therefore, our findings raise the possibility that opsins inhibit these UbiA enzymes in a mechanism similar to that used by BO and CO to inhibit Lye. The UbiA superfamily is highly relevant to human health, as dysfunctional UbiA enzymes have been implicated in a variety of diseases, including Schnyder crystalline corneal dystrophy (29), infantile multisystem disease (30), and urologic cancer (31). In addition, selective inhibitors of these prenyltransferases have been shown to be effective against Mycobacterium tuberculosis (32). Our characterization of opsin-mediated Lye inhibition may allow insight into the regulation and activity of other members of the UbiA superfamily.
Opsin-mediated inhibition of Lye and similar enzymes represents an ideal regulatory mechanism for genes obtained through lateral transfer.In this work, we determined that H. vallismortis CO specifically inhibited the H. vallismortis Lye (Fig. 6), suggesting that H. vallismortis, like H. salinarum, inhibits bacterioruberin synthesis to conserve lycopene for retinal production. This raises the possibility that the described mechanism, in which an opsin inhibits a pathway that consumes retinal precursors, would be commonly found in organisms with ion-pumping rhodopsins. As further evidence that ion-pumping rhodopsins have specific features that mediate inhibition, we determined that expression of H. salinarum SOII had no effect on bacterioruberin accumulation (Fig. 5A). Although we cannot exclude the possibility that SOII folding is different when expressed in H. volcanii, our results are consistent with previous observations in H. salinarum that conversion of lycopene to bacterioruberin was nearly quantitative in the absence of BO (14). BR and SRII are structurally very similar (33, 34), but they have stark differences in expression patterns in H. salinarum; BR production is induced to high levels (∼10% of total cell protein) under microaerobic conditions, and SRII is produced in much smaller amounts (0.01 to 0.1% of total cell protein) under both aerobic and microaerobic conditions (35). Given these differences in expression, BO-mediated inhibition of Lye activity may have evolved to ensure that all available lycopene is directed toward retinal synthesis when cells required BR activity to survive microaerobic conditions.
Genes allowing rhodopsin production are commonly spread by lateral gene transfer, and many organisms may produce substantial amounts of ion-pumping rhodopsins to survive under conditions of limited resources (1). After genes are obtained, retinal production may be in competition with other processes in these organisms for shared intermediates like lycopene. Opsins that inhibit these processes to promote retinal synthesis may confer a selective advantage, especially when functional ion-pumping rhodopsins are needed for energy conversion in the absence of respiration. Therefore, this regulatory mechanism first described for H. salinarum could be found among organisms with ion-pumping rhodopsins in all three domains of life. More broadly, apoproteins, like BO, lacking their cofactors may have activities that have not been characterized.
MATERIALS AND METHODS
Growth conditions. H. volcanii was grown in complete (Hv-YPC) medium (36) supplemented with 40 μM thymidine. Liquid cultures were grown at 40°C with shaking at 250 rpm. Plates were incubated at 42°C. For H. volcanii cultures used for carotenoid quantification or opsin expression analysis, 10-ml cultures were started from individual colonies and grown for 24 h, diluted to an A660 of 0.05 in new medium, and grown for an additional 48 h to an A660 value of 1.7 to 2.0. There were no obvious growth differences in any H. volcanii strains. Four milliliters of these saturated cultures was used to inoculate 96 ml Hv-YPC in 250-ml flasks, and these 100-ml cultures were grown for 18 h. Where noted, 20 μl of 20 mM retinal (Sigma-Aldrich, St. Louis, MO) in isopropanol (or isopropanol alone for cultures without retinal) was added to the 100-ml cultures to a final concentration of 4 μM retinal.
H. volcanii strain construction. H. volcanii strains are listed in Table 1. Plasmids and primers used to construct strains are listed in Tables S6 and S7, respectively, in the supplemental material. Plasmids were propagated in E. coli DH5α (New England BioLabs, Ipswich, MA) grown in LB medium supplemented with ampicillin (100 μg/ml). H. volcanii transformation was conducted as previously described (37). Gene knockout and replacement mutations were made in H. volcanii strain H1209 (38) and derivative strains. For genomic recombinants, selection and counterselection of pyrE2 was conducted as described using plasmids derived from pTA131 (36). Plasmids for gene expression in H. volcanii were constructed using pTA963 (38). All plasmids introduced into H. volcanii were confirmed by sequencing the added or modified DNA regions. If appropriate, proper integration into the genome was confirmed by PCRs both upstream and downstream of the integrated region that included one primer not present on the integrative plasmid. Opsin- or Lye-encoding open reading frames introduced into H. volcanii were amplified by PCR and resequenced to confirm strain fidelity. T. Allers generously provided H1209, pTA131, and pTA963. H. volcanii DS70 (39) was graciously sent by K. Bidle.
H. volcanii strains used in this study
Carotenoid quantification and analysis from cultures.Carotenoids were extracted from H. volcanii similarly to the previously described method for H. salinarum (14). Cultures were centrifuged for 15 min at 11,000 × g. The pellet was washed in 100 ml medium salt solution and centrifuged for 15 min, and supernatant was removed. To facilitate removal of as much supernatant as possible, the pellet was centrifuged for an additional 5 min, and remaining liquid was removed by aspiration. Pellets were resuspended in 1 ml medium salt solution, transferred to screw-cap microcentrifuge tubes, and centrifuged for 5 min at 16,100 × g. The supernatant was removed, and tubes were purged with N2. Cell pellets were stored at −80°C until carotenoid extraction.
For extraction, the pellet was lysed in 1 ml lysis solution (0.016 mg/ml DNase [Sigma-Aldrich]; 0.05% NaN3) for 1 h with shaking at 1,500 rpm in the dark. The cell lysate, ∼1.5 ml, was then vigorously mixed in an amber vial with 9 ml of acetone for 20 min. Hexane (5 ml) and water (1 ml) were added and briefly mixed vigorously. The organic (upper) layer was collected, and the remaining sample was extracted again with 3 ml of hexane. The organic layers were combined and evaporated to dryness under an N2 stream. The pigments were resuspended in 250 μl ethyl acetate and filtered through a 0.45-μm polytetrafluoroethylene filter before analysis with RP-UHPLC.
For RP-UHPLC, samples were fractionated on a Shimadzu (Kyoto, Japan) UHPLC system using a C18 column (2.1 by 50 mm, 1.9-μm particle size), and absorbance at 490 nm was monitored with a photodiode array detector. The mobile phase was a gradient of solvent A (0.05% ammonium acetate) and solvent B (acetonitrile-methanol-chloroform, 75:15:10) eluting at 0.4 ml min−1. The solvent change over an 11.5-min sample run was programmed as follows: elution with 70% B for 1 min, gradient to 100% B for 2.5 min, isocratic elution with 100% B for 4.5 min, gradient to 70% B for 0.1 min, and reequilibration with 70% B for 3.4 min. The standard curve for lycopene was generated with commercial lycopene (Sigma-Aldrich). The bacterioruberin standard curve was generated from pooled H. volcanii (strain H1209) cell extracts. Bacterioruberin concentration was calculated spectroscopically using an extinction coefficient in acetone (E) of 141 mM−1 cm−1 at 490 nm (40). The lycopene present in the samples used for the bacterioruberin standards was calculated to contribute less than 0.5% to the absorbance at 490 nm. All peaks eluting between 1.2 and 5.2 min that displayed spectra characteristic of bacterioruberins were included in the bacterioruberin concentration. Lycopene precursors and other carotenoids potentially present in minor concentrations were not included in the analysis.
Colony color analysis to assess bacterioruberin production.Cultures were grown from colonies for 48 h, and the A660 value was adjusted to 1.5 prior to plating. Five-microliter droplets of culture were placed on Hv-YPC plates supplemented with 100 μl isopropanol (control) or 100 μl 20 mM retinal (dissolved in isopropanol) for a final concentration of approximately 50 μM retinal. The droplets were allowed to dry, and the plates were incubated at 42°C. Each colony was photographed at approximately 24-hour increments with a Moticam 5.0 MP camera (Motic, Hong Kong, China) mounted on a VistaVision dissection microscope (VWR, Radnor, PA). Consistent lighting was achieved with a ring light connected to a light source with a 150-watt halogen bulb with a 380- to 780-nm emission spectrum. Images were obtained with Motic Image Plus, version 2.0 software. The field was “white balanced” prior to each session to ensure photographic consistency. Wild-type (H1209) H. volcanii was included on every plate to allow normalization of the data.
For color quantification of colonies, each image was sampled using Photoshop CS6 software (Adobe, San Jose, CA). The eyedropper sampling tool was set to the maximum area (101 by 101 pixels), and the sampled area was chosen for color consistency, avoiding any visible shadows or highlighting artifacts. After sampling, the color window was used to display the measurements of that color: hue, saturation, brightness, red, green, blue, cyan, magenta, yellow, and black. Values were normalized by comparison to wild-type (H1209) colonies on the same plate. These measurements were recorded for each genotype and used in beta regression modeling (41) to predict bacterioruberin levels as determined by RP-UHPLC analysis. The regression analysis indicated that a combination of hue and saturation after 4 days of colony growth was the best predictor of bacterioruberin content. This model was used to estimate parameters for a function incorporating hue and saturation for the calculation of a ruberin metric, which predicts the proportion of bacterioruberin for genotypes grown in the absence or presence of retinal. Comparisons of colonies based on their ruberin metric were made using a permutation test as implemented by the R package perm (42). For comparisons involving group sample sizes of 9 or less, the permutation test was made with exact computation. Asymptotic approximation was used for larger sample sizes. Bonferroni adjustment was applied when multiple comparisons were conducted using the same data. Details of model development and calculation of the ruberin metric are included in the supplemental materials.
Quantification of opsin expression in H. volcanii. H. volcanii strains harboring plasmids for expression of opsins with C-terminal His6 tags were constructed. Cultures were grown and cell pellets were obtained as described for carotenoid analysis. Pellets were lysed in 1.5 ml of lysis solution (1.3 mM Benzonase nuclease [Sigma-Aldrich], 1× protease inhibitor cocktail [Sigma-Aldrich P2714], 10 mM Tris-HCl, pH 8.5) and incubated with rotation at 37°C for approximately 1.5 h to allow degradation of DNA. Total protein was quantified by bicinchoninic acid protein assay (Pierce, Rockford, IL). Cell lysates were diluted to protein concentrations of 12 mg/ml using Laemmli sample buffer (Bio-Rad, Hercules, CA), with 2-mercaptoethanol (355 mM) as a reducing agent. Protein samples (30 μl) were separated by polyacrylamide gel electrophoresis, and opsin levels were determined by immunoblotting with α-his antibody (Genscript, Piscataway, NJ). The blots were subsequently probed with anti-mouse IgG secondary antibody conjugated to horseradish peroxidase (Genscript) and developed with a Picosignal chemiluminescence kit (Pierce) by following the manufacturer's instructions. Densitometry quantification was performed using Quantity One software (Bio-Rad).
ACKNOWLEDGMENTS
R.F.P. designed the research, directed the biological interpretation of the data, and wrote the manuscript. A.M.P. assisted in the experimental design and data analysis and performed much of the strain construction. S.M.G. developed the colony colorimetric method, conducted data analysis, and constructed some of the strains used in the study. D.R.A. generated the statistical model for colony colorimetry and conducted data analysis. E.L.S. conducted experiments to quantify carotenoids and proteins. All authors read and approved of the paper.
We thank Kang Mo Ku and Moo Jung Kim for conducting the RP-UHPLC analysis. Sara Lagerholm, Vanesa Silvestri, and Marie Estock provided valuable technical assistance. Many undergraduate students contributed their time to this project, including Adam Lavertu, Antoinette Dummer, Kyle Hughes, Ray Nakada, Katherine Metayer, Erika Smith, Gracey McGrory, Christine Wamsley, Abby Gregory, Sarah Lane-Reticker, Emily Hoylman, Mary Schulz, and Nick Waldner. We are indebted to Frank Fekete, David Hall, and James Scott for insightful scientific advice during our discussions.
Research reported in this project was supported by an Institutional Development Award from the National Institute of General Medical Sciences (NIGMS) of the National Institutes of Health under grant P20GM103423. Additional funding was provided by the Colby College Natural Science Division and award R15GM094735 from NIGMS to R.F.P.
FOOTNOTES
- Received 3 May 2017.
- Accepted 31 July 2017.
- Accepted manuscript posted online 7 August 2017.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00303-17 .
- Copyright © 2017 American Society for Microbiology.
