Functional Comparison of RGS9 Splice Isoforms in a Living Cell

Functional comparison of RGS9 splice isoforms in a living cell

Kirill A. Martemyanova,b, Claudia M. Krispelc,d, Polina V. Lishkoa, Marie E. Burnsc,d,1, and Vadim Y. Arshavskya,e,1

aDepartment of Ophthalmology, Harvard Medical School, Boston, MA 02114; bDepartment of Pharmacology, University of Minnesota, Minneapolis, MN 55455; cCenter for Neuroscience and dDepartment of Ophthalmology & Vision Science, University of California, Davis, CA 95618; and eDepartments of Ophthalmology and Pharmacology, Duke University, Durham, NC 27710

Edited by Jeremy Nathans, Johns Hopkins University School of Medicine, Baltimore, MD, and approved November 10, 2008 (received for review September 9, 2008) Two isoforms of the GTPase-activating protein, regulator of G be lost before the effector is activated (17, 18). In contrast, protein signaling 9 (RGS9), control such fundamental functions as RGS9–2 does not require PDE␥ for high-affinity interaction vision and behavior. RGS9–1 regulates phototransduction in rods with transducin but instead uses the PDE␥-like domain on its and cones, and RGS9–2 regulates dopamine and opioid signaling unique C terminus to increase the affinity between RGS9–2 and in the basal ganglia. To determine their functional differences in its target G proteins (19). To investigate whether the unique the same intact cell, we replaced RGS9–1 with RGS9–2 in mouse properties of RGS9–1 are essential for attaining normal ampli- rods. Surprisingly, RGS9–2 not only supported normal photore- tude and time course of the photoresponse, we generated a sponse recovery under moderate light conditions but also outper- transgenic mouse in which rods expressed RGS9–2 instead of formed RGS9–1 in bright light. This versatility of RGS9–2 results RGS9–1 and studied the physiological consequences of this from its ability to inactivate the G protein, transducin, regardless replacement. of its effector interactions, whereas RGS9–1 prefers the G protein- effector complex. Such versatility makes RGS9–2 an isoform ad- Results vantageous for timely signal inactivation across a wide range of Characterization of the RGS9–2 Transgenic Mouse. Transgenic mice stimulus strengths and may explain its predominant representa- expressing RGS9–2 instead of RGS9–1 in rods were generated by tion throughout the nervous system. expressing the cDNA containing the coding sequence of the mouse RGS9–2 gene under the control of the rhodopsin promoter de- ͉ ͉ G proteins phototransduction RGS proteins scribed in (20) (Fig. 1B; see Experimental Procedures for details). Animals carrying the RGS9–2 transgene were crossbred to RGS9 roteins of the regulators of G protein signaling (RGS) family knockout mice previously shown to have normal photoreceptor Pare ubiquitous regulators of signal duration in many G morphology and protein composition except for the lack of protein pathways. Their RGS homology domain is directly RGS9–1 and virtually complete loss of G␤5 (9). Western blot responsible for accelerating the GTPase activity of G protein analysis of retinal extracts indicated that two independent trans- ␣-subunits, but most RGS proteins also contain additional genic lines displayed similar robust expression of RGS9–2 (Fig. 1C). domains, which vary greatly among the family members. Little is The retinal morphology in these RGS9–2-expressing mice was known about the functional roles of these non-catalytic domains, normal (Fig. 1D), with no signs of retinal degeneration. Consis- although they are thought to contribute to the specificity of RGS tently, the total amount of rhodopsin in their retinas measured at interactions (reviewed in 1, 2). RGS9 is one of the better-studied 3 months of age also was normal (466 Ϯ 57 pmol per retina vs. 470 Ϯ multidomain RGS proteins and exists in two splice isoforms 110 pmol per retina in wild-type mice; SEM, n ϭ 4). The transgenic (3–5), both of which form constitutive complexes with the type expression of RGS9–2 was accompanied by ϳ 70% restoration of ␤ ␤ 5 G protein subunit, G 5 (6–8). The difference between these the expression of its constitutive subunit, G␤5 (photoreceptor- isoforms resides in the structure of their C-termini: a short 18-aa specific long-splice variant; Fig. 2A), indicating that both RGS9 sequence in RGS9–1 is replaced with 209 residues in RGS9–2 isoforms can stabilize the G␤5 expression level in rods. Immuno- (Fig. 1A). staining of retina cross-sections with anti-RGS9–2 antibodies re- RGS9–1 is expressed exclusively in rod and cone photorecep- vealed that virtually all immunoreactivity was confined to the tors where it sets the duration of electrical responses to light by light-sensitive compartment of the rod cell, the outer segment (Fig. accelerating the GTPase activity of transducin (3). In mice, the 1E), replicating the pattern of RGS9–1 localization in wild-type lack of RGS9 causes a drastic delay in photoresponse recovery rods (3). (9), and its mutation in humans leads to difficulties in adjusting to bright light and seeing moving objects (10). RGS9–2 is Quantification of RGS9–2⅐G␤5 Expression Level in the Retinas of expressed predominantly in the striatum, where it controls Transgenic Mice. Before assessing the expression level of RGS9–2 reward behavior and movement coordination by regulating D2 in transgenic rods, we re-examined the quantification of RGS9–1 dopamine and ␮-opioid receptor signaling (11–15). RGS9 in wild-type rods. Accurate determination of RGS9–1 levels in knockout mice display augmented sensitivity to rewarding prop- rods is very important because the time course of the rod’s light erties of morphine and cocaine (12, 13) and rapidly develop dyskinesias following administration of dopamine receptor an-

tagonists (15). Author contributions: K.A.M., M.E.B., and V.Y.A. designed research; K.A.M., C.M.K., and The physiological significance of having two RGS9 isoforms is P.V.L. performed research; K.A.M., C.M.K., P.V.L., M.E.B., and V.Y.A. analyzed data; and unknown. High-affinity interaction of RGS9–1 with transducin K.A.M., M.E.B., and V.Y.A. wrote the paper. requires that transducin first binds its effector, the ␥-subunit of The authors declare no conﬂict of interest. cGMP phosphodiesterase (PDE␥) (16). As a result, both This article is a PNAS Direct Submission. ␥ RGS9–1 (9) and PDE (17) are needed for timely GTP hydro- 1To whom correspondence should be addressed. E-mail: [email protected] or lysis by transducin and normal recovery of the rod from light [email protected]. excitation. We hypothesized that the biological role for such a This article contains supporting information online at www.pnas.org/cgi/content/full/ dual requirement is to ensure that the G protein would not 0808941106/DCSupplemental. inactivate before effector binding and therefore no signal would © 2008 by The National Academy of Sciences of the USA

20988–20993 ͉ PNAS ͉ December 30, 2008 ͉ vol. 105 ͉ no. 52 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808941106 Downloaded by guest on September 24, 2021 Fig. 2. Transgenic expression of RGS9–2 in rods rescues the expression of the long-splice isoform of G␤5. (A) Quantitative Western blot analysis of G␤5in retinal lysates. Varying amounts of lysates from RGS9–2 and wild-type mice normalized by the amounts of rhodopsin were loaded on the same gel, and the G␤5 band was visualized using an antibody raised to the N-terminal epitope of the G␤5 long-splice isoform (6) and the infrared imaging detection (Odyssey Imager, LI-Cor Biosciences). (B) Quantitative analysis of G␤5 levels from the experiment in A.G␤5 band intensities were measured by Odyssey v.1.2 software (LI-Cor Biosciences). The data are taken from one of three similar experiments.

with even small amounts of rod outer-segment material obtained from the RGS9 knockout mouse significantly reduced its im- munodetection efficiency, a phenomenon noted before for quantification of other proteins (23). Therefore, to account for the microenvironment of the blotted proteins, the RGS9–1⅐G␤5 standards were mixed with rod outer-segment material from RGS9 knockout mice and analyzed alongside wild-type rod outer-segment samples matched by the amount of rhodopsin. These experiments explain the variation in previous estimates using different amounts of the outer-segment material and reveal that wild-type rods contain RGS9–1 in a 1:269 Ϯ 18 molar ratio with rhodopsin (SEM; n ϭ 8), which is about twofold more than the highest previous estimates and is comparable with the PDE amount in rods (22, 24). We next assessed the expression level of RGS9–2⅐G␤5in transgenic rods. In this case, we concentrated on quantifying the long-splice isoform of G␤5, which exists exclusively as a constitutive equimolar complex with RGS9 in photoreceptors (6, 9, 21, 25). We chose G␤5 rather than RGS9–2 to avoid the inherent differences in antibody recognition by individual RGS9 iso- Fig. 1. Transgenic expression of RGS9–2 in rods of RGS9 knockout mice. (A) ␤ Domain composition of RGS9 isoforms. Both isoforms share four domains: the forms. Furthermore, unlike G 5, the two RGS9 isoforms run at disheveled-EGL10-pleckstrin (DEP) domain, which mediates their attachment different positions of the SDS gel, which makes their quantifi- to membrane anchors R9AP and R7BP; the R7 family homology (R7H) domain; cations differentially sensitive to the protein microenvironment the G protein ␥-subunit-like (GGL) domain, which binds to G␤5; and the RGS effects shown in Fig. S1. The data in Fig. 2 demonstrate that homology (RGS) domain. RGS9–1 has an 18-aa C-terminal tail (CT), whereas transgenic retinas expressed 70% Ϯ 4% (n ϭ 3) of the G␤5, and RGS9–2 has a 209-aa PDE type 6 ␥-subunit-like (PGL) domain. (B) Genetic thus ϳ 70% of the entire RGS9–2⅐G␤5 complex, of wild-type construct for the transgenic expression of RGS9–2. (C) Western blot detection retinas. However, the data presented in the next section indicate ␮ of RGS9–2 expression in mouse retina lysates containing 20 g rhodopsin; #5 that the expression of this RGS9–2⅐G␤5 varied significantly and #10 designate two independent founder lines examined in this study. (D) Retinal morphology of 2-month-old mice containing RGS9–2 transgene ex- among individual rods. pressed on the RGS9 knockout background (RGS9–2) and their wild-type littermates (RGS9–1). Plastic-embedded 1-␮m-thick retina cross-sections were Single-Cell Recordings Reveal Heterogeneous Expression of RGS9–2 in stained with toluidine blue. (E) Immunolocalization of RGS9–2 in transgenic Transgenic Rods. To assess the ability of RGS9–2 to inactivate retinas. Frozen sections obtained from RGS9–2 transgenic animals or their transducin in intact rods, we used suction electrodes to record wild-type littermates were stained as described in Materials and Methods. their responses to brief flashes of varying strength (Fig. 3). The DIC ϭ differential interference contrast image from the transgenic retina; light sensitivity of RGS9–2 rods was the same as that of the ϭ ϭ ϭ ϭ GC ganglion cells; INL inner nuclear layer; IPL inner plexiform layer; IS wild-type rods (Table 1). Specifically, their responses to single photoreceptor inner segments; ONL ϭ outer nuclear layer; OPL ϭ outer plexiform layer; OS ϭ photoreceptor outer segments. photons reached the same peak amplitude at the same time, and the flash strength eliciting half-saturating response also was indistinguishable from that of wild-type rods. This finding indi- response depends strongly on the RGS9 expression level (21), cates that the basic parameters of phototransduction activation but previous quantifications have yielded results varying from in RGS9–2 rods were normal. However, responses from 1:1640 to 1:610 molar ratio with rhodopsin (3, 22). The data from RGS9–2 rods showed unusual variability in photoresponse re-

supporting information (SI) Fig. S1 offer a potential explanation covery. Based on the exponential time constant of recovery, NEUROSCIENCE for such variation. Mixing recombinant RGS9–1⅐G␤5 standards ␶REC, the 45 individual rods analyzed in this study were separated

Martemyanov et al. PNAS ͉ December 30, 2008 ͉ vol. 105 ͉ no. 52 ͉ 20989 Downloaded by guest on September 24, 2021 Fig. 3. Representative ﬂash-response families of RGS9–2 transgenic rods. Responses of transgenic rods were categorized based on their recovery kinetics, which were (A) normal (‘‘wild-type-like’’), (B) somewhat slower than normal (‘‘intermediate’’), or (C) resembled responses of RGS9 knockout rods (‘‘knockout-like’’). Flash strengths were roughly 8, 50, 180, 600, and 1950 photons/␮m2.

into three distinct groups (Fig. 3 and Table 1). One population RGS9–2 Rods Have Faster Bright-Flash Inactivation Kinetics than (‘‘wild-type-like’’; ␶REC between 0.1 and 0.3 s; n ϭ 28) exhibited Wild-Type Rods. At dim and moderate flash strengths, photore- dim flash responses that were indistinguishable from responses sponses from wild-type-like transgenic rods were identical to of wild-type rods. The second group of only two cells (‘‘RGS9 those from wild-type rods (Fig. 4 and Table 1). A close inspection knockout-like’’; ␶REC Ͼ 2 s) exhibited responses very similar to of the average single-photon responses showed perfect agree- those of RGS9 knockout rods (9). The third population (‘‘in- ment throughout the entire time course (Fig. 4 A, B). The same termediate’’; n ϭ 15) exhibited intermediate dim flash response was true for all responses to flashes producing up to 4,000 kinetics (␶REC between 0.3 and 1.00 s). Because these three rod activated rhodopsin molecules (R*) (Fig. 4C). This agreement populations differed only in their response recovery, it is likely demonstrates that under physiological conditions, RGS9–2 com- that they expressed variable amounts of RGS9–2, as occasionally pletely substitutes for RGS9–1 in transducin inactivation and observed in transgenic lines using this promoter (20). does not interfere with PDE activation in this light regimen. Cell-to-cell variability in RGS9–2 expression allows us to Interestingly, these recordings did not show the approximately ␥ explain why the Western blot determination of G␤5 in Fig. 2B twofold inhibition of the RGS9–2 by PDE identified in previ- indicated that the amount of RGS9–2⅐G␤5 in whole retinas from ous biochemical experiments (19). Such inhibition would be the RGS9–2 mice was ϳ 70% of normal. Assuming that our expected to produce slower transducin inactivation and therefore slower response recovery. Potential explanations include sample size in the electrophysiological experiments reasonably the much higher concentrations of interacting proteins in intact reflected the total rod population of transgenic retinas and that rods and their persistent membrane localizations, neither of wild-type-like rods contained normal RGS9 levels, knockout- which can be achieved in vitro. like rods contained none, and intermediate rods roughly half of However, the most surprising result was that rods expressing the wild-type levels, the ‘‘average’’ RGS9–2⅐G␤5 expression ϳ RGS9–2 recovered significantly faster than wild-type rods in level would fall at 79%, which is in good agreement with the response to bright flashes (Fig. 5A). The most common way of 70% value determined by quantitative Western blotting. assessing recovery from bright saturating flashes is to measure ⅐ ␤ The reason that the upper limit of the RGS9–2 G 5 expres- the dominant time constant of recovery, ␶ (Fig. 5B), which is ⅐ ␤ D sion in transgenic rods matches the amount of RGS9–1 G 5in determined experimentally from the slope of the relation be- ⅐ ␤ wild-type rods is because the amount of RGS9–1 G 5 expression tween response saturation time and the natural log of flash in rods is set ultimately by the amount of its membrane anchor, strength (the Pepperberg plot; see refs. 28, 29). In normal mouse RGS9 anchor protein (R9AP) (21, 26). The same rule should rods, ␶D is ϳ 200 ms for responses to flashes up to a natural log apply to both RGS9 isoforms because they share an identical of the flash strength of ϳ ln i ϭ 9(ϳ4,000 R*/flash) and reflects R9AP-binding domain and both bind to R9AP with very high the rate of transducin GTPase (21). For flashes activating over affinity (8, 27). Therefore, in the rest of this study we concen- 4,000 R*, the relation rises more steeply (or ‘‘breaks’’), reflecting trated on comparing the wild-type-like group of transgenic rods a significant slowing in the response recovery rate. Such slowing with wild-type rods. This comparison provided a unique oppor- likely results from a depletion of one or more proteins respon- tunity to evaluate the functional properties of RGS9–1 and sible for normal deactivation of the phototransduction cascade RGS9–2 in cells expressing them in equal amounts. (29). Strikingly, RGS9–2 rods did not display a break in the

Table 1. Electrophysiological characteristics of RGS9–2 transgenic rods Elementary Time to

Io (photons/ SF (pA/photons Amplitude Dim Flash Integration peak ␶D1 tD2 2 Ϫ2 Rod Type Id (pA) ␮m ) ␮m ) (pA) ␶rec (sec) Time (sec) (msec) (sec) (sec)

Wild-type-like 13.9 Ϯ 0.5 56 Ϯ 3 0.18 Ϯ 0.02 0.56 Ϯ 0.10 0.19 Ϯ 0.01 0.22 Ϯ 0.01 105 Ϯ 7 0.23 Ϯ 0.01 0.32 Ϯ 0.03 (28) (26) (24) (21) (24) (24) (24) (25) (24) Intermediate 12.6 Ϯ 0.3 49 Ϯ 6 0.20 Ϯ 0.02 0.47 Ϯ 0.13 0.48 Ϯ 0.06 0.55 Ϯ 0.06 120 Ϯ 17 0.9 Ϯ 0.2 0.8 Ϯ 0.1 (15) (14) (14) (14) (14) (14) (14) (13) (13) KO-like 13.4, 14.6 70.9, 46.6 0.18, 0.24 0.61, 0.45 2.12, 2.66 1.48, 2.50 203, 203 8.98, 9.04 11.3, 8.11 (2) (2) (2) (2) (2) (2) (2) (2) (2) Grand average 13.6 Ϯ 0.4 54 Ϯ 3 0.19 Ϯ 0.01 0.53 Ϯ 0.07 0.40 Ϯ 0.08 0.43 Ϯ 0.07 115 Ϯ 8 0.9 Ϯ 0.3 1.1 Ϯ 0.4 (45) (42) (40) (37) (40) (40) (40) (40) (39)

Number of cells (n) in each group are given in parentheses. For n Ͼ 10, values are mean Ϯ SEM. Id, Dark current as determined by the saturating response amplitude; Io - Flash strength that elicited a half-maximal response; ␶D1, Dominant time constant of recovery for saturating ﬂashes producing up to 4000 R*; ␶D2, Dominant time constant of recovery for saturating ﬂashes producing more than 4000 R*.

20990 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808941106 Martemyanov et al. Downloaded by guest on September 24, 2021 Fig. 5. RGS9–2 speeds recovery from bright flashes. (A) Population average bright-flash responses from wild-type-like RGS9–2 (red, n ϭ 30) and C57BL/6 rods (black, n ϭ 23). Dotted line indicates the 10% recovery level that was used to calculate the time in saturation. (B) Pepperberg plot of the time spent in saturation (Tsat) as a function of the natural log of the flash strength (ln i) in photons/␮m2 for 25 RGS9–2 rods and 22 C57BL/6 rods. The slopes of the relations (␶D) for C57BL/6 (wild-type) rods were 0.238 s up to ln i ϭ 9(solid line) and 0.624 s beyond ln i ϭ 9(dashed line). The slope for wild-type-like RGS9–2 rods was 0.234 s. Error bars represent SEM.

cient photoresponse recovery over a broader range of light Fig. 4. Responses of wild type-like RGS9–2 rods to dim-to-moderate flashes intensities than normal rods expressing the photoreceptor- are normal. (A) Population average single-photon responses of wild-type-like specific RGS9–1. This result is not intuitive and inevitably raises RGS9–2 (red, n ϭ 21) and C57BL/6 (black, n ϭ 31) rods. (B) Same traces as in A the question why rods employ an apparently ‘‘inferior’’ isoform shown on an expanded time scale. The similarity in rising phases indicates no of this protein. Although we realize that in biology the question changes in phototransduction amplification. Error bars represent SEM. (C) ‘‘why’’ is not nearly as legitimate as the question ‘‘how,’’ search- Population averages of entire families of flash responses of RGS9–2 wild-type- ing for the answer provides an opportunity to revisit the func- ϭ ϭ like (red, n 30) and C57BL/6 9 (black, n 23) rods. Dots represent SEM. For tional and evolutionary specializations of the visual transduction C57BL/6 rods, the weighted flash strengths ranged from 5 to 2099 photons pathway. ␮mϪ2 roughly by factors of 2. Flash strengths for the population of transgenic rods were 3.6% dimmer. The average dark currents (in pA) were 13.3 (C57BL/6) and 14.3 (RGS9–2). Why do Photoreceptors Use RGS9–1? Effective single-photon detection by rods requires that the signal from the activated ␶ rhodopsin to PDE be relayed with high efficiency. In our Pepperberg relation at any flash strength. The D of these rods ␥ was the same as that of wild-type rods before the breaking point previous study (17), we demonstrated that PDE is required for and remained unchanged up to the brightest flashes that could timely GTP hydrolysis by RGS9–1 in vivo and proposed that this be delivered by our source (producing over 50,000 R*). Thus requirement contributes to this coupling efficiency by preventing RGS9–2 mediates timely transducin inactivation over a broader transducin inactivation before binding PDE (see also 18). How- range of flash strength than RGS9–1. ever, we now demonstrate that there is no detectable difference The ability of transgenic rods to recover more rapidly from in the kinetics of the photoresponse rising phases in rods brighter flashes is likely to arise from the major difference expressing RGS9–1 or RGS9–2, despite the ability of RGS9–2 between RGS9 isoforms revealed in biochemical experiments: to inactivate free transducin efficiently. Although this result may RGS9–1 requires PDE␥ to inactivate transducin rapidly, and seem surprising initially, it simply reveals that in the intact cell, RGS9–2 does not. Binding to PDE␥ increases transducin’s transducin interacts with PDE much more rapidly than with affinity for RGS9–1 by more than 20-fold (16), whereas the either RGS9 isoform, even though PDE and RGS9 isoforms are PDE␥-like domain of RGS9–2 relieves this requirement (19). expressed in comparable amounts. Indeed, transducin is known Because rods contain ϳ 10-fold more transducin than PDE to bind and activate PDE within a few milliseconds (31), whereas (reviewed in 30), bright light can cause transducin activation in the average time for either RGS9 isoform to inactivate trans- excess of PDE. In wild-type rods, this excess transducin is ducin is ϳ 100 times longer (the value of the ␶D; Fig. 5 and Table expected to be inactivated by RGS9–1 more slowly than the 1). Therefore, the coupling efficiency between transducin and PDE-bound transducin, and this difference presumably causes PDE is so high that specific binding properties of a given RGS9 the break in the Pepperberg relation. In contrast, RGS9–2 inactivates this excess transducin very efficiently, making trans- isoform are insignificant for preserving signal amplification. genic rods able to recover more rapidly from brighter flashes and On the other hand, although we typically consider ‘‘faster’’ to preventing the break in the Pepperberg relation (Fig. 5B). be better, there are some practical considerations that suggest that the slower transducin inactivation in bright light may be Discussion desirable for rod function. In rods, massive light-driven trans-

The central observation of this study is that rods expressing the location of transducin from the outer segment to other subcel- NEUROSCIENCE brain-specific splice isoform of RGS9, RGS9–2, achieved effi- lular compartments (reviewed in 32, 33) can contribute to light

Martemyanov et al. PNAS ͉ December 30, 2008 ͉ vol. 105 ͉ no. 52 ͉ 20991 Downloaded by guest on September 24, 2021 adaptation (34) and may be neuroprotective by reducing energy follows. The typical length of a rod outer segment in our metabolism through the mechanism of decreasing the amount of recordings was ϳ 20 ␮m and contained ϳ 5 ϫ107 rhodopsin transducin available for activation (35, 36). Because the extent of molecules (44). Based on the rhodopsin to PDE ratio of ϳ 300:1 translocation is proportional to the time that transducin is active (24), there are ϳ 170,000 PDE molecules per rod, a number (37, 38), the inability of RGS9–1 to inactivate efficiently the reasonably close to the estimated ϳ 130,000 activated trans- transducin produced in excess of PDE␥ may benefit this process. ducins, particularly given the major differences in experimental Another idea is that the presence of RGS9–1 in photorecep- conditions under which individual numbers used in these calcu- tors reflects the evolutionary origin of the phototransduction lations were obtained. cascade. RGS proteins exist in all eukaryotic organisms, whereas In summary, comparison of RGS9–1 and RGS9–2 in con- no PDE␥ genes have yet been discovered in species more ancient trolling the light responses of rods has revealed that RGS9–2 ␥ than lamprey (39). Thus, PDE emerged in evolution during the not only completely substituted for RGS9–1 under dim to time when high-affinity transducin interactions with RGS pro- moderate light intensities but also outperformed RGS9–1 in ␥ teins already existed. Accordingly, we speculate that the PDE bright light. Although we argue that the presence of RGS9–1 prototype may not have had such a high affinity for transducin in photoreceptors may reflect the evolutionary history of its as the modern-day PDE␥ and that the formation of the entire ␥ cellular function, RGS9–2 seems to be a more versatile transducin-PDE -RGS9 molecular module was possible because isoform of this protein that can function efficiently anywhere of the selection for an RGS protein isoform that would not rapid inactivation of its G protein target(s) is needed. Such inactivate transducin before it binds and activates PDE. The versatility may explain its wider expression pattern throughout similarities in sequence and function between the C terminus of the nervous system. RGS9–2 and PDE␥ (19) also raise the question of whether one ␥ evolved from the other. Consistently, the RGS9 and PDE genes Materials and Methods are located on the same region of chromosome 17 (within loci DNA Constructs, Rod Outer Segments, Recombinant Proteins, and Antibodies. 17q24–17q25 in the human genome). However, because of the Generation of expression constructs for RGS9–1 and G␤5 (long-splice variant) current gaps in the genome sequences of the lamprey and its was described in (19). The recombinant RGS9⅐G␤5 complexes were expressed immediate predecessor, the hagfish (see 40 for a review on in the insect Sf9/baculovirus system, purified by Ni-NTA chromatography, and evolution of the eye), we cannot determine which of these their concentrations were measured by UV spectroscopy as in (45). Osmotically alternatives is more probable. intact mouse rod outer segments were prepared as in (17). Rhodopsin concentration in all photoreceptor membrane and retina lysate preparations was ␧ ϭ PDE Rate-Limits Rod Response Recovery to Bright Flashes. Our determined spectrophotometrically using 500 40,000. Transducin was pu- discovery that RGS9–1 and RGS9–2 differ in their ability to rified from frozen bovine retinas (46), and its concentration was determined mediate photoresponse recovery in bright light also has revealed based on the maximum amount of rhodopsin-catalyzed GTP␥S binding (47). the mechanism limiting rod temporal resolution under those Western blot and immunohistochemical detection of RGS9–2 was performed ␤ conditions. The temporal resolution of vision depends on timely using sheep antibodies against its C-terminal epitope (8). G 5 was detected recovery of the light-sensitive current in photoreceptors. In using the sheep antibody against the N-terminal epitope of its long-splice isoform (6). RGS9–1 was detected using the sheep antibody against the normal mammalian rods, flashes that activate from 1 to 4,000 R* RGS9–1 fragment called ‘‘RGS9c’’ (6). produce responses that recover at an invariant rate. On the molecular level, this rate is determined by the inactivation of the Generation of RGS9–2 Transgenic Mouse. The DNA region encoding mouse complex between transducin and PDE caused by the RGS9–1- RGS9–2 was subcloned into pBamH4.4, a rod-specific mammalian expres- stimulated transducin GTPase activity (21). However, for flashes sion vector. In the resulting plasmid the ORF of RGS9–2 was located under activating more than 4,000 R* the time constant of recovery the control of a 4.4-kb mouse opsin promoter region and supplied with the slows, indicating that the biochemical reaction that rate-limits polyadenylation signal of the mouse protamine gene (20). The construct recovery has changed. This slowing (at the break of the Pep- was injected into the pronuclei of oocytes from superovulated females of perberg relation) was proposed to arise from either slowed BDF1 strain (F1 of C57/Bl6 ϫ DBA/2 from Charles River Laboratories). The rhodopsin inactivation or slowed GTP hydrolysis (29), but transgene integration was determined by PCR analysis of tail DNA. To neither of these putative explanations had been tested experi- establish the transgenic line, where RGS9–2 is expressed on a RGS9 knock- mentally. Here, we demonstrate that transgenic replacement of out background, founders were crossed with RGS9 knockout mice for two RGS9–1 with RGS9–2 abolishes this slowing, so that photore- generations (9). sponse recovery rate remains invariant from 1 to at least 50,000 Immunohistochemistry. For immunohistochemical detection of RGS9–2 in ret- R*. Therefore response recovery to flashes exciting 4,000– ina sections, eyes were fixed for 4 h with paraformaldehyde (4% in PBS) at 4 °C, 50,000 R* normally is limited by the inactivation of transducin ␥ cryoprotected with 30% sucrose in PBS at 4 °C, and mounted in embedding produced in excess of PDE , which is not a good substrate for medium (Tissue-Tek OCT Compound, Sakura Finetek). Frozen sections were RGS9–1. Thus, normally, it is the amount of PDE␥, not the obtained, rehydrated, blocked with PBT1 (PBS, 0.1% Triton-100, 1% BSA slowing of rhodopsin inactivation, that sets the dominant time (wt/vol), 5% heat-inactivated goat serum) for 1 h, incubated with the same constant for recovery of responses to very bright flashes. primary antibody as used for Western blotting in PBT1 overnight at 4 °C, The conclusion that in normal rods the amount of activated washed four times with PBT2 (PBS, 0.1% Triton-100, 1% BSA), and incubated transducin exceeds the amount of available PDE after a flash with Alexa 488-conjugated secondary antibodies in PBT2 for 2 h. After being producing more than 4,000 R* is generally consistent with washed twice with PBT2 for 5 min and with PBS for 5 min, sections were available biochemical estimates of phototransduction activation mounted in anti-fading media (Pierce) and analyzed using a laser-scanning (see ref. 41 for a detailed review). Each of the 4,000 R* is confocal microscope. predicted to activate up to ϳ 400 transducin molecules per second (the highest rate of ϳ 150/s measured in rod outer- Single-Cell Electrophysiology. Suction electrode recordings from dark-adapted segment suspensions at 22 °C (42) projected to the rate at 37 °C rods were conducted as described in (25). The average response to a large number (Ͼ30) of flashes was considered to be in the linear range if its mean using the relation from (43)). The average lifetime of each R* amplitude was less than 20% of the maximal response amplitude. These can be as long as the value of the nondominant time constant of dim-flash responses were used to estimate the form of the single-photon ϳ the phototransduction cascade inactivation, 0.08 s (21). Thus response using the ‘‘variance to mean’’ method described in (48). Integration the total number of transducin molecules activated by 4,000 R* time was used as a measure of the duration of the incremental flash response is ϳ 4,000 ϫ 400 ϫ 0.08 Ϸ130,000. This number is comparable and is defined as the time integral of the average linear response divided by to the total number of PDE molecules in a rod, calculated as its peak amplitude (49). The time that a bright-flash response remained in

20992 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0808941106 Martemyanov et al. Downloaded by guest on September 24, 2021 saturation was calculated as the time interval between the midpoint of the Kessler and Peter Yoo for technical assistance. This work was supported by the ﬂash and the time at which the current recovered by 10%. National Eye Institute grants EY018139 (K.A.M.), EY14047 (M.E.B.), EY012859 (V.Y.A.), EY012576 (Core Grant for Vision Research to UC-Davis), and EY5722 ACKNOWLEDGMENTS. We thank Edward N. Pugh for helpful comments on (Core Grant for Vision Research to Duke University) and by Research to Prevent the manuscript, Norman A. Michaud for plastic sectioning, and Christopher Blindness (M.E.B. and V.Y.A.).

1. Burchett SA (2000) Regulators of G protein signaling: a bestiary of modular protein 25. Krispel CM, Chen CK, Simon MI, Burns ME (2003) Prolonged photoresponses and binding domains. J Neurochem 75:1335–1351. defective adaptation in rods of G␤5Ϫ/Ϫ mice. J Neurosci 23:6965–6971. 2. Hollinger S, Hepler JR (2002) Cellular regulation of RGS proteins: modulators and 26. Keresztes G, et al. (2004) Absence of the RGS9⅐G␤5 GTPase-activating complex in integrators of G protein signaling. Pharmacol Rev 54:527–559. photoreceptors of the R9AP knockout mouse. J Biol Chem 279:1581–1584. 3. He W, Cowan CW, Wensel TG (1998) RGS9, a GTPase accelerator for phototransduction 27. Anderson GR, et al. (2007) Expression and localization of RGS9–2/G␤5/R7BP complex in Neuron 20:95–102. vivo is set by dynamic control of its constitutive degradation by cellular cysteine 4. Granneman JG, et al. (1998) Molecular characterization of human and rat RGS 9L, a proteases. J Neurosci 27:14117–14127. novel splice variant enriched in dopamine target regions, and chromosomal localiza- 28. Pepperberg DR, et al. (1992) Light-dependent delay in the falling phase of the retinal tion of the RGS 9 gene. Mol Pharmacol 54:687–694. rod photoresponse. Visual Neuroscience 8:9–18. 5. Rahman Z, et al. (1999) Cloning and characterization of RGS9–2: a striatal-enriched 29. Lyubarsky AL, Pugh EN, Jr (1996) Recovery phase of the murine rod photoresponse alternatively spliced product of the RGS9 gene. J Neurosci 19:2016–2026. reconstructed from electroretinographic recordings. J Neurosci 16:563–571. 6. Makino ER, Handy JW, Li T, Arshavsky VY (1999) The GTPase activating factor for 30. Pugh EN, Jr, Lamb TD (2000) in Molecular Mechanisms in Visual Transduction, eds. transducin in rod photoreceptors is the complex between RGS9 and type 5 G protein Stavenga DG, DeGrip WJ, Pugh, EN, Jr. (Elsevier, Amsterdam), Vol. 3, pp. 183–255. beta subunit. Proc Natl Acad Sci USA 96:1947–1952. 31. Pugh EN, Jr, Lamb TD (1993) Amplification and kinetics of the activation steps in ␤ 7. Kovoor A, et al. (2000) Co-expression of G 5 enhances the function of two Ggamma phototransduction. Biochim Biophys Acta 1141:111–149. subunit-like domain-containing regulators of G protein signaling proteins. J Biol Chem 32. Calvert PD, Strissel KJ, Schiesser WE, Pugh EN, Jr, Arshavsky VY (2006) Light-driven 275:3397–3402. translocation of signaling proteins in vertebrate photoreceptors. Trends Cell Biol 8. Martemyanov KA, Yoo PJ, Skiba NP, Arshavsky VY (2005) R7BP, a novel neuronal 16:560–568. protein interacting with RGS proteins of the R7 family. J Biol Chem 280:5133–5136. 33. Artemyev NO (2008) Light-dependent compartmentalization of transducin in rod 9. Chen CK, et al. (2000) Slowed recovery of rod photoresponse in mice lacking the GTPase photoreceptors. Mol Neurobiol 37:44–51. accelerating protein RGS9–1 Nature 403:557–560. 34. Sokolov M, et al. (2002) Massive light-driven translocation of transducin between the 10. Nishiguchi KM, et al. (2004) Defects in RGS9 or its anchor protein R9AP in patients with two major compartments of rod cells: a novel mechanism of light adaptation. Neuron slow photoreceptor deactivation. Nature 427:75–78. 34:95–106. 11. Garzon J, Rodriguez-Diaz M, Lopez-Fando A, Sanchez-Blazquez P (2001) RGS9 proteins 35. Burns ME, Arshavsky VY (2005) Beyond counting photons: trials and trends in verte- facilitate acute tolerance to ␮-opioid effects. Eur J Neurosci 13:801–811. brate visual transduction. Neuron 48:387–401. 12. Rahman Z, et al. (2003) RGS9 modulates dopamine signaling in the basal ganglia 36. Fain GL (2006) Why photoreceptors die (and why they don’t) BioEssays 28:344–354. Neuron 38:941–952. 37. Kerov V, et al. (2005) Transducin activation state controls its light-dependent translo- 13. Zachariou V, et al. (2003) Essential role for RGS9 in opiate action Proc Natl Acad Sci USA cation in rod photoreceptors. J Biol Chem 280:41069–41076. 100:13656–13661. 38. Lobanova ES, et al. (2007) Transducin translocation in rods is triggered by saturation of 14. Cabrera-Vera TM, et al. (2004) RGS9–2 modulates D2 dopamine receptor-mediated the GTPase-activating complex. J Neurosci 27:1151–1160. Ca2ϩ channel inhibition in rat striatal cholinergic interneurons. Proc Natl Acad Sci USA 39. Muradov H, Boyd KK, Kerov V, Artemyev NO (2007) PDE6 in lamprey Petromyzon 101:16339–16344. marinus: implications for the evolution of the visual effector in vertebrates. Biochem- 15. Kovoor A, et al. (2005) D2 dopamine receptors colocalize regulator of G-protein istry 46:9992–10000. signaling 9–2 (RGS9–2) via the RGS9 DEP domain, and RGS9 knock-out mice develop 40. Lamb TD, Collin SP, Pugh EN, Jr (2007) Evolution of the vertebrate eye: opsins, dyskinesias associated with dopamine pathways. J Neurosci 25:2157–2165. 16. Skiba NP, Hopp JA, Arshavsky VY (2000) The effector enzyme regulates the duration of photoreceptors, retina and eye cup. Nat Rev Neurosci 8:960–976. G protein signaling in vertebrate photoreceptors by increasing the affinity between 41. Arshavsky VY, Lamb TD, Pugh EN, Jr (2002 G proteins and phototransduction Annu Rev transducin and RGS protein. J Biol Chem 275:32716–32720. Physiol 64:153–187. 17. Tsang SH, et al. (1998) Role for the target enzyme in deactivation of photoreceptor G 42. Leskov IB, et al. (2000) The gain of rod phototransduction: reconciliation of biochem- protein in vivo. Science 282:117–121. ical and electrophysiological measurements. Neuron 27:525–537. 18. Arshavsky VY, Pugh EN, Jr (1998) Lifetime regulation of G protein-effector complex: 43. Heck M, Hofmann KP (2001) Maximal rate and nucleotide dependence of rhodopsin- emerging importance of RGS proteins. Neuron 20:11–14. catalyzed transducin activation: initial rate analysis based on a double displacement 19. Martemyanov KA, Hopp JA, Arshavsky VY (2003) Specificity of G protein-RGS protein mechanism. J Biol Chem 276:10000–10009. recognition is regulated by affinity adapters. Neuron 38:857–862. 44. Lyubarsky AL, Daniele LL, Pugh EN, Jr (2004) From candelas to photoisomerizations in 20. Lem J, Applebury ML, Falk JD, Flannery JG, Simon MI (1991) Tissue-specific and the mouse eye by rhodopsin bleaching in situ and the light-rearing dependence of the developmental regulation of rod opsin chimeric genes in transgenic mice. Neuron major components of the mouse ERG. Vision Res 44:3235–3251. 6:201–210. 45. Skiba NP, et al. (2001) RGS9-G␤5 substrate selectivity in photoreceptors. Opposing 21. Krispel CM, et al. (2006) RGS expression rate-limits recovery of rod photoresponses. effects of constituent domains yield high affinity of RGS interaction with the G Neuron 51:409–416. protein-effector complex J Biol Chem 276:37365–37372. 22. Zhang X, Wensel TG, Kraft TW (2003) GTPase regulators and photoresponses in cones 46. Ting TD, Goldin SB, Ho Y-K (1993) Purification and characterization of bovine trans- of the eastern chipmunk. J Neurosci 23:1287–1297. ducin and its subunits. Methods Neurosci 15:180–195. 23. Lyubarsky AL, et al. (2005) Mole quantity of RPE65 and its productivity in the gener- 47. Fung BK, Hurley JB, Stryer L (1981) Flow of information in the light-triggered cyclic ation of 11-cis-retinal from retinyl esters in the living mouse eye. Biochemistry 44:9880– nucleotide cascade of vision. Proc Natl Acad Sci USA 78:152–156. 9888. 48. Mendez A, et al. (2000) Rapid and reproducible deactivation of rhodopsin requires 24. Pentia DC, Hosier S, Cote RH (2006) The glutamic acid-rich protein-2 (GARP2) is a high multiple phosphorylation sites. Neuron 28:153–164. affinity rod photoreceptor phosphodiesterase (PDE6)-binding protein that modulates 49. Baylor DA, Hodgkin AL (1973) Detection and resolution of visual stimuli by turtle its catalytic properties. J Biol Chem 281:5500–5505. photoreceptors. J Physiol (Camb) 234:163–198. NEUROSCIENCE

Martemyanov et al. PNAS ͉ December 30, 2008 ͉ vol. 105 ͉ no. 52 ͉ 20993 Downloaded by guest on September 24, 2021