It is since many years textbook knowledge that the concentration of the second messenger cGMP is regulated in animal rod and cone cells by type II rhodopsins via a G-protein signaling cascade. Microbial rhodopsins with enzymatic activity for regulation of cGMP concentration were only recently discovered: in 2014 light-activated guanylyl-cyclase opsins in fungi and in 2017 a novel rhodopsin phosphodiesterase (RhoPDE) in the protist Salpingoeca rosetta (SrRhoPDE). The light regulation of SrRhoPDE, however, seemed very weak or absent. Here, we present strong evidence for light regulation by studying SrRhoPDE, expressed in Xenopus laevis oocytes, at different substrate concentrations. Hydrolysis of cGMP shows an ∼100-fold higher turnover than that of cAMP. Light causes a strong decrease in the Km value for cGMP from 80 to 13 µM but increases the maximum turnover only by ∼30%. The PDE activity for cAMP is similarly enhanced by light at low substrate concentrations. Illumination does not affect the cGMP degradation of Lys296 mutants that are not able to form a covalent bond of Schiff base type to the chromophore retinal. We demonstrate that SrRhoPDE shows cytosolic N- and C-termini, most likely via an eight-transmembrane helix structure. SrRhoPDE is a new optogenetic tool for light-regulated cGMP manipulation which might be further improved by genetic engineering.
Rhodopsins are retinylidene proteins which consist of the protein part ‘opsin’ and a covalently bound retinal as a chromophore. Photoreceptors in the animal eyes are classified as type II rhodopsins. Their activation regulates the electrical potential via G proteins and a signaling chain . In the photo transduction system of the mammalian eyes, the cGMP level is decreased by light through rhodopsin via a G-protein, activating a phosphodiesterase (PDE). As the PDE therefore controls the activity of a cyclic nucleotide-gated cation channel, illumination leads to a hyperpolarization of the photoreceptor cell and ultimately to neuronal signaling to the brain.
Microbial rhodopsins are known as type I rhodopsins. The first microbial rhodopsin was identified from Halobacterium halobium  (later correctly identified as H. salinarum), which was then characterized as a light-activated proton pump [3,4]. New microbial rhodopsins were afterwards discovered, like the chloride pump halorhodopsin [5,6], sensory rhodopsins  and the direct light-gated ion channels, the channelrhodopsins [8–10].
The first microbial rhodopsin with proven enzyme activity was found by Avelar et al.  in the fungus Blastocladiella emersonii  and named BeGC1. We and others proved its light-activated guanylyl-cyclase activity by heterologous expression and named it cyclase opsin or Cyclop  (also known as RhGC  and RhoGC ). In different systems, it was shown to be a tightly light-controlled optogenetic tool, very specific for light-activated cGMP production. Cyclop from B. emersonii (BeCyclop) is the first rhodopsin with a proven localization of the N- and C-termini in the cytosol and most likely eight-transmembrane helices (8-TM) [12,14], contrary to all the previous rhodopsins with a characteristic 7-TM structure and an extracellular N-terminus.
Another recently discovered microbial rhodopsin is rhodopsin PDE (abbreviated RhPDE  or RhoPDE ), an opsin fused to a PDE domain (Figure 1A). SrRhoPDE is a unique enzyme rhodopsin, so far found only in the genome of Salpingoeca rosetta. It was expressed in HEK293 cells and shown to be a slightly light-activated PDE which is 10-fold more active with cGMP as a substrate than with cAMP . Illumination increased its hydrolytic activity for cGMP 1.4-fold . Lamarche et al.  also expressed this protein heterologously in HEK293 cells, but observed no influence of light on activity of purified RhoPDE.
RhoPDE is an 8TM rhodopsin.
Here, we show that SrRhoPDE is a PDE with a light-regulated Michaelis–Menten constant and high hydrolysis activity for cGMP, by studies with a different expression system, the Xenopus laevis oocyte. Illumination leads at low substrate concentrations to a significant activation of cGMP and cAMP hydrolysis. Additionally, in our system, its hydrolysis activity is ∼100 times higher for cGMP than for cAMP.
Materials and methods
The DNA of SrRhoPDE was synthesized by GeneArt Strings DNA Fragments (Life Technologies, Thermo Fisher Scientific) according to the protein sequence in the data base (XP_004998010.1), codon-optimized to Mus musculus, with a BamHI restriction site before the start codon and a HindIII restriction site after the stop codon.
The synthesized fragment was then inserted into the pGEMHE within N-terminal BamHI and C-terminal HindIII restriction sites. The YFP::RhoPDE was made by inserting 5′-BamHI-YFP (no stop codon)-XhoI-3′ and 5′-XhoI-RhoPDE-HindIII-3′ to above-mentioned pGEMHE vector. RhoPDE::YFP was made using 5′-BamHI-RhoPDE (no stop codon)-XhoI-3′ and 5′-XhoI-YFP-HindIII-3′. The BIFC-RhoPDE construct was made by inserting RhoPDE sequence without stop codon to the BiFC (Bimolecular Fluorescence Complementation) vector , with KpnI and XhoI restriction sites.
Point mutations were made using the Quickchange Method with the following primer pairs: K296M-qcF (5′-GCCGCCAtgGTTGGCCTGGCCA-3′) and K296M-qcR (5′-GCCAACcaTGGCGGCGTAATC-3′) for making K296M, K296A-qcF (5′-GCCGCCgctGTTGGCCTGGCCA-3′) and K296A-qcR (5′-GCCAACagcGGCGGCGTAATC-3′) for making K296A.
The sequence was confirmed by complete DNA sequencing. Plasmids were linearized by NheI digestion and used for in vitro generation of cRNA with the AmpliCap-MaxT7 High Yield Message Maker Kit (Epicentre Biotechnologies).
Twenty nanograms or 30 ng of cRNA (indicated in every individual figure) were injected into Xenopus oocytes. Injected oocytes were then incubated for 3 days with ND96 (96 mM NaCl, 5 mM KCl, 0.1 mM CaCl2, 1 mM MgCl2, 5 mM HEPES and pH 7.6) in an 18°C incubator, supplemented with 1 µM all-trans retinal.
Membrane extraction and PDE activity assay
Xenopus oocyte membrane extraction protocol was done according to ref. . Solution A for extracting and resuspending membranes contained 100 mM NaCl, 75 mM Tris, 5 mM DTT, 5% glycerol, pH 7.3 with HCl. Membranes were resuspended by mixing the pellet with 8 µl of solution A per oocyte for cGMP and 1 µl of solution A per oocyte for cAMP measurements. This membrane suspension was added in a ratio of 1 : 10 to the reaction buffer with final concentration as follows: 0.1 mM cGMP-Na/cAMP-Na or as indicated in individual figure, 5 mM MgCl2 or as indicated, 100 mM NaCl, 75 mM Tris/Tris–Cl, 5 mM DTT, pH 7.3. Aliquots of 10 µl were taken at indicated times and the reaction was immediately stopped by the addition of 190 µl stop solution. The amount of cGMP (or cAMP) was determined using the DetectX High Sensitivity Direct Cyclic GMP (or cAMP) Chemiluminescent Immunoassay Kit (Arbor Assays). PDE activity was then calculated by the cGMP (or cAMP) consumption over time.
Fluorescence emission was used to quantitate the YFP-tagged wild-type (wt) or mutated RhoPDE protein amount in different samples, similar as in ref. . The fluorescence emission values of membranes containing YFP-tagged RhoPDE or of known amounts of YFP were measured by a Fluoroskan Ascent microplate fluorometer. The fluorescence of the known amounts of YFP yielded a linear relationship with YFP amount, enabling us to obtain a calibration curve for YFP protein measurement. The protein amount was calculated by the YFP fluorescence with y = (x + 0.015)/(27.7 * 0.02) pmol, where x = fluorescence emission value, made in the same buffer. To calculate the RhoPDE turnover, the PDE activity of the same sample was measured by the activity assay.
Illumination was mostly done with a 473 nm laser (Changchun New Industries Optoelectronics Tech). The light intensities of different experiments were all adjusted to ∼0.1 mW/mm2 and measured with a Laser Check intensity meter (Coherent, Dieburg, Germany).
Fluorescence pictures were taken 3 days after injection with a confocal microscope (Leica DM6000).
SrRhoPDE is an 8-TM microbial opsin
To our knowledge, the guanylyl BeCyclop is the first opsin with experimentally proven cytosolic N- and C-termini , most likely by a structure containing 8-TMs. Our conclusion was based on sequence prediction and a BiFC experiment  and was recently confirmed independently by another group using a different method .
Sequence analysis suggests that also SrRhoPDE is an 8-TM protein with an extra N-terminal transmembrane helix, TM0 (Figure 1B). A BiFC construct was made by fusing the N-terminal half of YFP (YN) to the C-terminus and the C-terminal half of YFP (YC) to the N-terminus of the opsin part of SrRhoPDE, i.e. amino acid 1–327 (Figure 1B). A clear fluorescence suggested that the N- and C-termini of the opsin domain are on the same side of the plasma membrane (Figure 1C). Considering the cytosolic localization of PDE activity and the sequence alignment with other rhodopsins, both termini have to be cytosolic and SrRhoPDE should have 8-TMs.
Light-regulated PDE activity of
The activity of SrRhoPDE was then studied by an in vitro reaction with membrane fragments from oocytes expressing SrRhoPDE. As shown in Figure 2, light could increase the SrRhoPDE activity for cAMP and cGMP, whereas no PDE activity was observed with membranes from control oocytes, not expressing RhoPDE (Supplementary Figure S1). In accordance with the recent publications [15,16], we found a relatively strong PDE activity (especially for cGMP) already in the dark. However, in contrast with a previously published 1.4-fold light activation , we observed ∼2-fold activation of cGMP hydrolysis at 100 µM cGMP (Figure 2A and Supplementary Figure S1C) and ∼3-fold light activation at 25 µM cGMP (Figure 2C). We also observed >100 times (instead of 10 times; ) higher activity in light for cGMP than for cAMP (Figure 2A,B and Supplementary Figure S1). The much weaker PDE activity at 100 µM cAMP was increased by light ∼5-fold (Figure 2B and Supplementary Figure S1D).
RhoPDE is a light-regulated PDE.
The SrRhoPDE activity and light regulation depend on Mg2+ concentration (Figure 2C), when measured at 25 µM cGMP. The SrRhoPDE activity was dramatically reduced and no obvious light regulation could be seen at a Mg2+ concentration of 0 mM. With 1 mM Mg2+, the PDE activity increases and light activation became ∼2-fold. Mg2+ (2 mM) could further increase the light activity 1.5-fold without influencing the dark activity, thus increasing the light activation to ∼3-fold. Mg2+ (5 mM) then showed no obvious difference to 2 mM Mg2+.
YFP fusion to
To study the mechanism of SrRhoPDE activity in greater detail, mutation of critical amino acids can be helpful. However, the quantification of expressed protein amount is needed to exclude that altered activity of a mutant is not simply caused by a different expression level. For this purpose, fusing a fluorescence tag-like YFP proved to be helpful .
N- and C-terminal YFP fusion constructs were tested, as in contrast with 7-TM rhodopsins with an extracellular N-terminus, this 8-TM protein allows YFP fusion to a cytosolic N-terminus. The two YFP fusion constructs turned out to be quite different. As determined from YFP fluorescence of membranes, the N-terminal YFP fusion construct showed a much better expression level than the C-terminal YFP fusion (Figure 3A). For YFP::RhoPDE with 100 µM cGMP, we determined a good light activation of cGMP hydrolysis, similar to unlabeled RhoPDE (Figure 3B). The C-terminal YFP fusion, however, exhibited weaker expression and a strong reduction in light activation, due to increased activity in the dark (Figure 3A,B). When measuring cAMP hydrolysis at 100 µM cAMP, we observed for the N-terminal YFP fusion construct ∼2.6-fold light activation (Figure 3C), instead of L/D (activity in light to activity in dark) = 5 without YFP fusion (Figure 2B). The N-terminal YFP fusion construct was then used for further mutation experiments.
Expression and activity after YFP fusion to N- or C-terminus of RhoPDE.
Mutation of the retinal-binding lysine abolishes light regulation of
Effects on expression and activity of Lys296-mutated YFP::RhoPDE.
We therefore tested YFP::RhoPDE mutants where the K296 was replaced by an alanine (K296A) or methionine (K296M) and observed good expression in oocyte membranes, similar to wt YFP::RhoPDE (Figure 4B). As expected, no light regulation of PDE activity was seen with K296A and K296M (Figure 4C). The wt YFP::RhoPDE showed ∼2.3-fold light activation from the same batch of oocyte and the same reaction condition (100 µM cGMP, Figure 4C).
The amount of expressed protein was calculated by measuring the fluorescence intensity of wt or mutated YFP::RhoPDE-expressing membranes. For this purpose, we generated a calibration curve with fluorescence intensities of known YFP amounts, see Materials and methods and ref. . A typical YFP::RhoPDE expression (wt or K296 mutant) amounted to ∼0.5 pmol per oocyte (Table 1). The turnover of wt YFP::RhoPDE at 100 µM cGMP was calculated to be 28 ± 5 s−1 in light, which is very similar to what Lamarche et al.  determined at 5 mM cGMP, ranging from 19 to 28 s−1. For the dark turnover at 100 µM cGMP, we obtained 12 ± 2 s−1 (Table 1). Surprisingly, however, K296A showed a higher turnover (∼23 s−1), which is more similar to wt in the light, and K296M showed a lower turnover (∼13 s−1), more like wt in the dark. This suggested two different conformations of these two lysine mutants, important for the RhoPDE activity.
|.||Protein amount (pmol/oocyte) .||Dark turnover (s−1) .||Light turnover (s−1) .|
|YFP::RhoPDE||0.46 ± 0.04||12 ± 2||28 ± 5|
|K296A||0.50 ± 0.04||22 ± 4||23 ± 0.8|
|K296M||0.41 ± 0.05||14 ± 2||13 ± 3|
|.||Protein amount (pmol/oocyte) .||Dark turnover (s−1) .||Light turnover (s−1) .|
|YFP::RhoPDE||0.46 ± 0.04||12 ± 2||28 ± 5|
|K296A||0.50 ± 0.04||22 ± 4||23 ± 0.8|
|K296M||0.41 ± 0.05||14 ± 2||13 ± 3|
Turnover refers to reaction at 20°C, starting with 100 µM cGMP. Fluorescence emission values of YFP::RhoPDE and mutants were used to calculate the protein amount with a standard YFP fluorescence curve, made in the same buffer. n = 3, errors = SD.
Light-regulated substrate affinity of
The observation that the ratio of L/D is larger at lower cGMP concentrations (Figure 2A, L/D = ∼2 at 100 µM cGMP; Figure 2C, L/D = 3 at 25 µM cGMP) led to the hypothesis that light might influence the Km (Michaelis–Menten constant) value for cGMP. We then tested the activity with different initial cGMP concentrations, ranging from 1 to 250 µM. Plotted data are fitted with a Michaelis–Menten equation. As shown in Figure 5A,B, maximum hydrolysis speed was not changing so much by light with 1.2 nmol/min in dark and 1.6 nmol/min in light, both calculated with the protein from one oocyte. But the Km value for cGMP changed significantly with illumination: Km = ∼80 µM in dark and ∼13 µM in light. When we then measured the RhoPDE activity with 7.5 µM starting cGMP, the L/D activity ratio increased to ∼4 (Figure 5C). We also determined the Km values for cGMP of the two light-insensitive mutants K296A and K296M (see Supplementary Figure S3). The activity of both mutants was tested in darkness, as they were completely light-insensitive (Figure 4). As expected from the measured turnover at 100 µM cGMP, we found a low Km value of 13 µM cGMP for K296A in the dark (Supplementary Figure S3A), just as for wt RhoPDE in light. For K296M, however, a Km value of 63 µM was determined (Supplementary Figure S3B), more similar to the Km value of 80 µM for wt RhoPDE in the dark.
Light-regulated substrate affinity of RhoPDE.
The present study on a new microbial rhodopsin, SrRhoPDE with an 8-TM topological structure, clearly demonstrates light-activated PDE activity. Illumination causes 4- to 5-fold light activation for cGMP and cAMP hydrolysis at low substrate concentrations, while illumination does not affect the activity of Lys296 mutants (K296M and K296A). As these mutants are not able to covalently bind retinal via a Schiff base, they lack the ability to transmit the signal of retinal isomerization to the protein. The experiments with the two mutants hint at two different conformations (K296A more like RhoPDE in light and K296M more like RhoPDE in dark) and exclude any artifactual light activation of SrRhoPDE, caused by our illumination conditions. Yoshida et al.  found that illumination increased the RhPDE (RhoPDE) hydrolytic activity with cGMP 1.4-fold, when measuring at initial 100 µM cGMP, where we determined a ratio (L/D) of ∼2. The observed lower L/D ratio might be because of a different expression system and/or different reaction conditions. Previously, we observed with the fungal light-activated guanylyl cyclase from B. emersonii, BeCyclop, that crude membranes from HEK293 cells, expressing BeCyclop, showed a lower L/D ratio than BeCyclop-expressing membranes from Xenopus oocytes . Compared with the HEK293 cells, Xenopus oocytes can be homogenized faster and more gently by simply pipetting the oocytes with a small-opening pipette for membrane extraction which might cause less damage to membrane proteins. We also found that pH influences the RhoPDE activity. A lower light-activation effect was detected at pH 6.5, used by Yoshida et al., than at pH 7.3, as used by us (Supplementary Figure S2). Another possible factor is the binding of retinal to the protein which is absolutely necessary for light-sensitivity of rhodopsins. As shown by the K296A/M mutants, a lacking Schiff base bond to retinal yields a completely light-insensitive protein which, however, still shows enzymatic activity. In fact, we can only assume that, in our study, all wt RhoPDE molecules contain retinal, as we supplement the medium with 1 µM all-trans retinal immediately after cRNA injection. This retinal supplementation has led to better expression  and higher photocurrents for most rhodopsins tested by us [8,9,18]. But if retinal saturation of RhoPDE is not 100%, then the retinal-free ‘opsin-PDE’ will skew the measured Km for cGMP in light. The kcat determined by Lamarche et al.  is ranging from 19 to 28 s−1, similar to the turnover in light, as determined by us (28 ± 5 s−1 at 100 µM cGMP which is equivalent to a maximal turnover of 32 ± 6 s−1). More interestingly, we found that light decreases the Km value for the substrate cGMP by a factor of ∼6 (from 80 to 13 µM), but influences the maximum hydrolysis turnover only slightly (increases by 30%). This explains why Lamarche et al.  cannot see significant light activation when using 5 mM cGMP as a substrate. In fact, our own initial experiments, using 1 mM cAMP or cGMP as a substrate, also yielded no convincing conclusion about light activation. Illumination did not affect the activity of the Lys296 mutants (K296M and K296A), which are unable to form a covalent Schiff base bond to retinal. Surprisingly, the light-insensitive K296M mutant shows a Km value for the substrate cGMP of 63 µM, similar to the Km of wt RhoPDE in the dark; whereas the light-insensitive K296A mutation decreased the Km value to the same value as light did for wt RhoPDE. In the Michaelis–Menten formalism, our results mean that product formation (k2) is nearly unaffected by illumination, but the affinity of the enzyme for cGMP increases (k1+ increases or/and k1− decreases). We speculate that the light-induced conformational change of rhodopsin is transduced to the PDE domain where it leads to a higher affinity for cGMP, probably by a decrease in k1−, i.e. slower cGMP dissociation. Hopefully this question will be answered by structure determinations in the future. For this purpose, the two mutants with quite different Km values, K296A with similarity to a light-activated RhoPDE and K296M with similarity to a dark-adapted RhoPDE, might help to elucidate the mechanism.
Our bimolecular fluorescence complementation (BiFC) experiment, together with the immunofluorescence result of Lamarche et al. , is strong evidence that RhoPDE has 8-TMs with cytosolic N- and C-terminal ends but not the ‘classical’ 7-TMs, as suggested by Yoshida et al. in text and Figure 6 . This 8-TM architecture is for opsins unusual, but was already demonstrated for another microbial opsin, the BeCyclop, first by us  and then corroborated by Trieu et al. . We also showed that N-terminal YFP fusion to SrRhoPDE is preferable to C-terminal YFP fusion, because C-terminal YFP fusion reduced the expression level and strongly reduced the light regulation (by increasing dark activity).
So far, two families of well-characterized enzyme rhodopsins, Cyclop and RhoPDE, were discovered which both show 8-TM protein structure. In the zoospore of B. emersonii, BeCyclop (BeGC1) is necessary for its reproduction , possibly through a cGMP-dependent K+ channel . S. rosetta belongs to choanoflagellates, which are commonly considered as the closest living relatives of the animals. As S. rosetta is until now not studied in depth, we currently lack hints about the function of SrRhoPDE in its natural host. SrRhoPDE offers, however, a new possibility for optogenetic cGMP/cAMP manipulation by decreasing cGMP/cAMP with light. The light-activated guanylyl-cyclase BeCyclop has already exhibited its power as an optogenetic tool [12,13] for increasing cGMP by illumination. Currently, we do not know if a possibly tighter light regulation of SrRhoPDE in vivo relies on some unknown factors in S. rosetta. In the future, it might be possible to improve the L/D activity ratio of RhoPDE, either by further studies in S. rosetta or by genetic engineering.
Blastocladiella emersonii guanylyl cyclase 1 (GC1)
Bimolecular Fluorescence Complementation
cyclase opsin; dpi, days post injection
- K, Lys
light to dark ratio (of activity)
Y.T., S.G., S.Y. and G.N. designed and performed the experiments and analyzed the data. S.G. and G.N. wrote the ﬁrst draft of the paper and all authors revised the paper and approved the ﬁnal version to be published.
This work was supported by grants from the Deutsche Forschungsgemeinschaft to Georg Nagel [TRR 166/A03]. G.N. acknowledges the support provided by the Prix-Louis-Jeantet. The PhD study of Y.T. is supported by China Scholarship Council (CSC[AQ5] ).
We thank all members of the laboratory for mutual help in routine lab work and for discussion.
The Authors declare that there are no competing interests[AQ6] associated with the manuscript.
These authors contributed equally to this work.