The putative lyase genes cpeS, cpeZ, and mpeV from the RS9916 genome (Supplementary Figure 1) were amplified via polymerase chain reaction (PCR) using a standard Pfu DNA polymerase system (ThermoFisher Scientific, Waltham, MA) and synthetic forward and reverse oligonucleotide primers with engineered restriction endonuclease sites (Eurofins MWG Operon, Huntsville, AL). Primers used to amplify genes by PCR for the construction of these expression vectors are previously published (Carrigee, 2020; Carrigee et al., 2020a,b). Amplified fragments were separately cloned into compatible Novagen Duet vectors using corresponding restriction enzymes as listed in Supplementary Tables 1, 2 or were previously published (Carrigee, 2020; Carrigee et al., 2020a,b). Expression vectors used in this study (Supplementary Table 2) include two previously described (Shukla et al., 2012; Kronfel et al., 2019b).

The mpeV gene from WH8020 was amplified via PCR using Platinum SuperFi II DNA Polymerase as a master mix in lieu of the Pfu system as previously described (Carrigee et al., 2020b). PCR reactions were performed using the standard High Fidelity PCR protocol from ThermoFisher Scientific (Waltham, MA). WH8020 substrate genes were cloned as an operon for PEI (cpeBA) with the beta subunit in frame with a hexahistidine tag (HT) and inserted into pET-DUET independently. WH8020 mpeV was cloned in-frame with the coding region of the HT and inserted into pCDF-DUET1. For co-expression of multiple genes required to characterize MpeV, some genes were subcloned into both multiple cloning sites (MCSs) for compatibility as follows. RS9916 cpeZ was cloned using Platinum SuperFi PCR protocol and inserted in frame after the sequence encoding a HT into MCSI of pCOLA-Duet containing non-tagged (NT) cpeS in MCSII (Carrigee, 2020; Carrigee et al., 2020b). The RS9916 cpeA gene sequence was inserted into multiple cloning site I (MCSI) pET-Duet (Novagen, Madison, WI) in frame with the sequence encoding a hexahistidine tag (HT). cpeB was subsequently subcloned into MCSII to achieve (MCSI/MCSII/vector) RS9916 HTcpeA/HTcpeB/pET-DUET and RS9916 HTmpeA/HTmpeB/pET-DUET as previously described (Carrigee, 2020; Carrigee et al., 2020b).

A single-site variant of WH8020 CpeB was created by mutating the H141 residue to L (H141L) using combined overlap extension PCR method adapted from (Hussain and Chong, 2016) using the Platinum SuperFi protocol (Supplementary Table 1). A PCR fragment containing the entire mutant WH8020 cpeBA operon was cloned into MCSI of pET-DUET vector containing resulting plasmids listed in Supplementary Table 2. All plasmids were sequenced for verification of clones and mutations (Eurofins Genomics LLC, Louisville, KY).

Initial experiments for heterologous protein expression were performed using E. coli grown in Luria Bertani (LB) medium. However, we used modified, auto-induced medium for maximal protein yield (Studier, 2005). This involved a 100-ml LB starter culture of E. coli cells grown at 37°C overnight; then this was added to a liter of auto-induced medium composed of LB containing 1 mM MgSO₄, 25 mM (NH₄)₂SO₄, 50 mM KH₂PO_4, 50 mM Na₂HPO₄, 0.5% glycerol, and 0.05% glucose at 18°C with appropriate combinations of antibiotics ampicillin (Ap: 100 μg·ml⁻¹), chloramphenicol (Cm: 34 μg·ml⁻¹), kanamycin (Km: 50 μg·ml⁻¹), and spectinomycin (Sp: 100 μg·ml⁻¹). With either medium, once the OD_600nm reached 0.6, cultures were induced with 1 mM isopropyl 1-β-D-thiogalactopyranoside, after which the cells were allowed to grow at 18°C for an additional 24 h before being harvested by centrifugation at 11,000 ×g for 8 min in a Sorvall RC 5C Plus centrifuge (Kendro Laboratory Products, Newtown, CT). Cell pellets were stored at −20°C until ready for purification and analysis. The wet weight of all cell pellets was measured and recorded prior to storage at −20°C.

The histidine-tagged (HT) proteins were purified as previously described (Carrigee et al., 2020b). Briefly, cell pellets were resuspended at 3.0 ml·g⁻¹ complete with mini protease cocktail (Thermo Scientific, Waltham, MA), 0.01 mg·ml⁻¹ lysozyme (Fisher Scientific, Hampton, NH), and passed through a French Pressure Cell Press at 18,000 psi three times. All samples purified by cobalt affinity chromatography, dialyzed to remove imidazole, and then concentrated by ultrafiltration through an Amicon Ultra centrifugal filter unit (10 kDa cutoff; Novagen/EMD Millipore Corp., Darmstadt, Germany) and stored at −20°C.

Purified protein was quantified using Bradford colorimetric assay (BioRad, Hercules, CA) and diluted to obtain equal concentrations across co-expressions for direct comparison when possible. Absorbance spectroscopy was performed using Perkin Elmer Lambda 35 UV/VIS or Shimadzu UV-2600 UV–Vis spectrophotometers followed by fluorescence spectroscopy using a Perkin Elmer LS55 (Waltham, MA) with excitation at 490 nm (PEB) or 440 nm (PUB; slit widths were set at 10 nm). Proteins were subsequently resolved by 15% (w/v) polyacrylamide gel electrophoresis (PAGE) in the presence of sodium dodecyl sulfate (SDS) and ultimately visualized by Coomassie blue staining (Saunée et al., 2008). To visualize proteins with bound bilin, gels were subjected to zinc-enhanced fluorescence using ChemiDoc MP imaging system (Bio-Rad, Hercules, CA) with excitation at 460–490 nm (PUB) and 520–545 nm (PEB).

WH8020 cells were obtained from the Roscoff Culture collection.¹ Cultures of WH8020 were grown at 22°C in PCR-S11 media and acclimated for at least 7 days in either blue light (BL) or green light (GL) and PBS were collected as previously described (Sanfilippo et al., 2016; Mahmoud et al., 2017; Sanfilippo et al., 2019b). Fluorescence excitation spectra were recorded at an emission of 575 nm, using a Perkin Elmer LS-50B spectrofluorometer. The fluorescence excitation 495–545 nm ratio was calculated and used as a proxy for the molar PUB to PEB ratio, as described (Humily et al., 2013).

PBS were purified using methods previously described (Shukla et al., 2012; Sanfilippo et al., 2016, 2019a). Samples were dialyzed overnight against 5 mM Na phosphate buffer (pH 7.0) and subsequently purified via high performance liquid chromatography (HPLC) using methods outlined in (Shukla et al., 2012; Sanfilippo et al., 2016). Phycobiliproteins were monitored from 210 to 700 nm with specific channels monitoring for total protein (280 nm), PUB (490 nm), and PEB (550 nm). Relevant fractions were vacuum-dried and kept at −20°C prior to digestion for mass spectrometric (MS) analysis as previously described (Biswas et al., 2011; Shukla et al., 2012; Sanfilippo et al., 2016). Purified proteins were dialyzed against 2 mM sodium phosphate buffer (pH 7.0) and 1 mM β-mercaptoethanol. One aliquot of trypsin (dimethylated trypsin from porcine pancreas; Sigma, St. Louis, MO) was added to 2% (w/w) from a 20 μg ml⁻¹ stock to the denatured protein mixtures and incubated at 30°C for 3 h in the dark (Shukla et al., 2012). The reaction was quenched by adding 30% (v/v) glacial acetic acid. Digested peptides were passed through a pre-equilibrated C8 Sep-Pak cartridge (Waters Corporation, Milford, MA), and thereafter the eluted sample was vacuum dried and stored at −80°C before LC–MS² on a Thermo Orbitrap Fusion Lumos instrument. MS1 scans were obtained with a resolution of 120,000 and mass range of 300–2,000 m/z. Data dependent HCD were acquired with a 3 s cycle time, quadrupole precursor isolation window of two 2 m/z and resolution of 30,000 with 30% relative collision energy. Samples were separated by an Easy NanoLC1200 HPLC (ThermoFisher) equipped with a 75 μm × 15 cm Acclaim PepMap100 separating column (Thermo Scientific) downstream of a 2 cm guard column (Thermo Scientific). Buffer A was 0.1% formic acid in water. Buffer B was 0.1% formic acid in 80% acetonitrile. Peptides were separated on a 30 min gradient from 4% B to 33% B. All data processing was performed with Thermo XCalibur 4.0, Proteome Discover 2.1 (Thermo Scientific), and a local copy of ProteinPropector 5.22.1 (prospector.ucsf.edu).

Gene sequences from Synechococcus strains RS9916 and WH8020 were retrieved from Cyanorak (Garczarek et al., 2021). Amino acid sequences were analyzed using the ClustalW alignment tool from MacVector software V. 12.7.5 (MacVector Inc., Apex, NC) and Phyre² prediction system (Kelley et al., 2015). The model of the MpeV-CpeB complex was obtained using the AlphaFold2 multimer implemented in the ColabFold server based on the protein sequences of MpeV and CpeB from RS9916 (Jumper et al., 2021; Mirdita et al., 2022). The PEB chromophore was then manually docked into the MpeV/MpeB complex structure using Coot (Emsley and Cowtan, 2004).

Initially, we wanted to confirm that the bilin composition of PEI and PEII from wild type (WT) WH8020 PBS was as previously described (Ong and Glazer, 1991). Cells used in this previous study were isolated from cultures grown in white light, a light condition known to elicit a pigment phenotype equivalent to GL in CA4 strains (Ong and Glazer, 1991; Everroad and Wood, 2006; Humily et al., 2013). Here we isolated PBS from native WH8020 cells grown in two separate light conditions: GL for maximum PEB production and BL for maximum PUB production (Supplementary Figure 2). PBPs were then separated by HPLC with relative absorbance and chromatograms obtained using 550 nm (PEB) and 495 nm (PUB) for bilin content and 280 nm for total protein (Figures 3A–C; Shukla et al., 2012; Carrigee et al., 2020b). The numbered peaks in Figures 3D–G were collected, digested with trypsin, and analyzed by LC/MS/MS to identify the proteins present in each peak. As shown in Table 1, peaks labeled as MpeA, CpeA, CpeB, and MpeB were verified by MS coverage (Table 1; Figure 3). As expected for this CA4-A strain, variable bilin content was detected among the α-subunits depending on light quality (GL or BL), and the changes in PUB:PEB ratio observed in the spectra for CpeA and MpeA matched those documented for RS9916 during the CA4 process (Figures 3H,I; Kehoe, 2010; Shukla et al., 2012; Nguyen, 2018; Carrigee et al., 2020b). Unfortunately, we were not able to identify all bilins attached at each light condition for each protein by LC tandem MS, but we can infer the likely bilin content from our spectra and using the previous characterizations of WH8020 in white light (equivalent to GL; Ong and Glazer, 1991) and RS9916 in BL and GL (Shukla et al., 2012) as shown in Figure 3I.

As expected from previous work showing that the CA4 process affects only MpeA and CpeA (Shukla et al., 2012), no changes were observed in the MpeB and CpeB proteins in GL vs. BL (Figures 3E,G). As previously reported by Ong and Glazer for WH8020 grown in white light (Ong and Glazer, 1991) and for RS9916 in both GL and BL (Shukla et al., 2012), WH8020 MpeB binds a doubly-linked PUB at C50, 61 and two PEB (C159 and C82) in a 1:2 ratio, or a PUB:PEB of ~0.58 independent of light condition (Figures 3E,G; Ong and Glazer, 1991). However, while RS9916 CpeB maintains bilin composition of one PUB (doubly-linked to C50, 61) and 2 PEB (C82 and C165) for CpeB regardless of growth conditions, HPLC analysis showed that WH8020 CpeB has PEB bound to all sites, including the doubly-linked C50, 61 position, consistent with what Ong and Glazer reported in 1991 (Ong and Glazer, 1991). The fraction collected for CpeB (Figure 3E, peak 2) only contains PEB (no PUB is detected) and no bilin content changes were detected in GL or BL.

MpeV was first suggested as a putative lyase by Wilbanks and Glazer, after sequencing of a large fraction of the PBS rod genomic region from WH8020 (Supplementary Figure 1; Wilbanks and Glazer, 1993). ClustalW analysis of protein sequences from RS9916 and WH8020 revealed 97.3% similarity between the CpeB substrates, 97.8% similarity between MpeB substrates, and 72.2% similarity between the MpeV homologs (Supplementary Figure 3). This led us to hypothesize that MpeV could be a lyase acting at C50, 61 on CpeB and perhaps also the lyase-isomerase acting on C50, 61 on MpeB in WH8020.

A comparison of Synechococcus CpeB sequences sorted by pigment type (PT) is shown in Supplementary Figure 4A [for review on PTs, see (Six et al., 2007; Humily et al., 2013; Grébert et al., 2022)]. All strains of Synechococcus belonging to PT2 (green light specialists with PEI and PEB only), PT3a (green light specialists with PEI, PEII, and a constitutively low PUB:PEB), about half of PT3dA (CA4-A) including WH8020, and the only representative of PT3eA (RCC307 that is genetically undistinguishable from typical PT3dA but exhibits only faint variations in the PUB:PEB ratio), all possess a H at position 141 in CpeB. In contrast, all strains that are either PT3c (typical BL specialists), PT3f (a rarer type of BL specialists), PT3dB (CA4-B), the other half of PT3dA including RS9916, and a natural mutant strain (BIOS-E4-1 that also display a BL specialist phenotype; Humily et al., 2013; Grébert et al., 2018) possess a L at position 141 in CpeB (Supplementary Figure 4A, blue highlights). By comparison for MpeB, all PT3a (or 3eA) strains exhibited a H whereas most other strains have a L, consistent with the fact that the former have a PEB and all others a PUB at C50, 61 (Supplementary Figure 4B). The only exception to this rule is PT3f strains which have a M at this position.

For simplicity, all recombinant proteins from RS9916 are hereafter prefixed with an “RS” and all WH8020 proteins are hereafter prefixed with a “WH” while generic referral to protein(s) from both strains will remain unprefixed (e.g., CpeA vs. RSCpeA or WHCpeA). We sought to determine the activity of WHMpeV compared to that of RSMpeV using our heterologous E. coli expression system with various substrates. All protein co-expressions analyzing β-subunits as substrates were designed to also express α-subunits in an effort to increase solubility of β-subunits as previously shown (Anderson and Toole, 1998; Kronfel et al., 2019b; Carrigee et al., 2020b). Our earlier work with RS9916 showed that the enzymes RSCpeS and RSCpeZ were required to obtain enough chromophorylated (attaching PEB at C82 of CpeB and MpeB), soluble RSCpeB and RSMpeB substrate to allow RSMpeV to function (Carrigee et al., 2020b). Therefore, these genes from RS9916 were expressed concomitantly in trials testing RSMpeV and WHMpeV activity (Figure 4). Even though detectable amounts of α-subunits (CpeA) co-purified with their respective β-subunits (CpeB; Figures 4E–G), CpeA was expectedly not chromophorylated by the available lyases (see section LC–MS–MS analyses of recombinant co-expressions below). CpeB was present in all co-expressions as determined by total protein staining using Coomassie blue (Figure 4G). Control co-expressions to assess bilin addition and site specificity included CpeB/CpeA expressed without a lyase, or expressed in the presence of RSCpeZ, RSCpeS, and/or MpeV as outlined in Supplementary Table 3. Absorbance spectra from recombinant co-expressions of RSMpeV or WHMpeV enzymes with RSCpeB, WHCpeB, or WHMpeB demonstrate that WHMpeV acts as a PEB lyase-isomerase, doubly ligating PUB on RSCpeB (Figure 4A, co-expression 1, red line; absorbance peak 491.5 nm) and WHMpeB (Supplementary Figure 5; absorbance peak at 493 nm), but it does not isomerize PEB when attaching it to WHCpeB (Figures 4B,C, co-expression 4, black dashed line). The addition of a doubly-ligated PEB on WHCpeB subunit is detected as a prominent shoulder at 534.0 nm in addition to a peak at 556.5 nm from the C82 PEB ligated by RSCpeS (Figures 4B,C, co-expression 4, black dashed line). This is consistent with observations by Kronfel et al. for a paralogous E/F type lyase called CpeF which doubly ligates a PEB on CpeB at an equivalent position in the freshwater cyanobacterium Fremyella diplosiphon (Kronfel et al., 2019b). We conclude that WHMpeV is capable of isomerizing activity on both WHMpeB and RSCpeB but that the WHCpeB substrate prevents the isomerization activity of WHMpeV.

As mentioned above, WHCpeB contains H at position 141 whereas RSCpeB contains L at this position (Supplementary Figures 3, 4). Molecular modeling of the structure of WHCpeB suggested that H141 should be positioned close to the C50, 61 bilin position and at ~7.4 Å from C50 (Figure 5). When PEB is docked with its ring A close to C50 and ring D close to C61, aspartic acid (D)54 of WHCpeB is poised toward the bilin, potentially interacting with the pyrrole nitrogen atoms of rings B and C (Figure 5). The arginine (R)57 residue on WHMpeV points toward the propionate chain of the C ring of PEB whereas MpeV-D116/R89 line the binding pocket near D ring (Figure 5B). Linkage at the A ring of PEB to C50 is a critical step in chromophore attachment, isomerization and stability for RSCpeB (Carrigee et al., 2020b). A site-directed mutation was introduced into WHCpeB converting H141 to leucine (H141L; Figure 5; Supplementary Table 1) to test whether this residue affected isomerization activity by WHMpeV. Absorbance data from the co-expressions including RSCpeZ, RSCpeS, and WHMpeV as outlined in Supplementary Table 3 revealed that this single substitution provided sufficient change(s) within the binding pocket to allow isomerization to occur (Figures 4, 5). Indeed, when native WHCpeB is expressed in the presence of RSCpeZ, RSCpeS, and WHMpeV, we see evidence of a doubly-ligated PEB at the C50, 61 position (Figures 4B–D, co-expression 4, black dashed line, and Table 2). However, when we perform this same co-expression using pHTCpeB(H141L)A (Supplementary Table 2), we see a doubly-linked PUB at the C50, 61 position (Figures 4C,D, co-expression 6 purple line, co-expression 5 green line, and Table 2). These findings indicate that the microenvironment of the WHCpeB binding pocket surrounding C50 plays a major role in WHMpeV isomerization activity at C50, 61.

Table 2 shows peak areas obtained for extracted ion chromatograms (M + 4H)⁴⁺ ions of the tryptic peptide for CpeB containing residues 38–78. The peptide sequence is the same in both the RSCpeB and WHCpeB and is shown in Supplementary Tables 4–7. The negative control sample (WHCpeBA, no MpeV) shows no bilin attachment and interestingly, MS–MS data suggest C61 and C73 are linked by a disulfide bridge. Complete ligation of bilin to C50 and C61 was observed for the positive control sample (WHCpeB + WHMpeV). RSMpeV appears to modify WHCpeB and WHCpeB(H141L) at a slower rate than WHMpeV modifies WHCpeB as we observed unmodified, disulfide linked 38–78 peptide in the mixture. Intriguingly, the co-expressions where WHMpeV acted on RSCpeB or WHCpeB(H141L) showed evidence of bilins singly attached to the 38–78 peptide as well as normal double attachment and some unmodified substrate with no bilins attached at all. It is likely that the disulfide bond formed during purification and processing, as the cytoplasm of E. coli is generally a reducing environment (Gąciarz et al., 2017). However, its formation may indicate a lack of sufficient activity by MpeV or insufficient folding by CpeB to achieve the doubly ligated chromophore in some of these combinations (Table 2). Representative mass spectra supporting these observations comprise Supplementary Figures 5–7; Supplementary Tables 4–7, contain lists of tandem mass spectral fragment matches. In all samples containing RSCpeS, a bilin attached at C82 was observed, as previously described (Carrigee et al., 2020b). We did not observe any bilin modifications on CpeA in these co-expressions.

To explore the possible role of CpeB H141 in conferring the lyase activity of MpeV, we built a structure model of MpeV in complex with two substrates: CpeB and PEB, using manual docking aided by AlphaFold2 implemented in the ColabFold server (Jumper et al., 2021; Mirdita et al., 2022; Figure 5). In this model, the bilin pigment adopts an anti, syn, anti-configuration in which the ring A and ring D are in close proximity of their respective C anchors (i.e., C50 and C61 of CpeB) and the pyrrole nitrogen atoms of PEB rings B/C are in hydrogen-bonding distances with CpeB-D54 (Figure 5). Interestingly, we note that the side chain at the 141 position of CpeB would directly interact with the ring B propionate (Figure 5). For example, in WH8020 CpeB, H141 potentially forms a hydrogen bond with the ring B propionate while L141 of RS9916 CpeB cannot. This difference may explain the absence of the isomerase activity of MpeV on the CpeB substrate of WH8020 because the hydrogen bond between H141 and ring B propionate group may confer steric hinderance for the bilin transformation required by the PEB to PUB isomerization during a ligation reaction (Yang et al., 2007).