Advertisement

Circadian Activation of Bullfrog Retinal Mitogen-activated Protein Kinase Associates with Oscillator Function*

  • Yuko Harada
    Affiliations
    Department of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo and CREST, Japan Science and Technology Corporation, Tokyo 113-0033, Japan
    Search for articles by this author
  • Kamon Sanada
    Affiliations
    Department of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo and CREST, Japan Science and Technology Corporation, Tokyo 113-0033, Japan
    Search for articles by this author
  • Yoshitaka Fukada
    Correspondence
    To whom correspondence should be addressed: Dept. of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo, Hongo 7-3-1, Bunkyo-Ku, Tokyo 113-0033, Japan. Tel./Fax: 81 3 5802 8871
    Affiliations
    Department of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo and CREST, Japan Science and Technology Corporation, Tokyo 113-0033, Japan
    Search for articles by this author
  • Author Footnotes
    * This work was supported in part by grants-in-aid from the Japanese Ministry of Education, Science, Sports and Culture.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
      The vertebrate retina retains a circadian oscillator, and its oscillation is self-sustained with a period close to 24 h under constant environmental conditions. Here we show that bullfrog retinal mitogen-activated protein kinase (MAPK) exhibits anin vivo circadian rhythm in phosphorylation with a peak at night in a light/dark cycle. The phosphorylation rhythm of MAPK persists in constant darkness with a peak at subjective night, and this self-sustained rhythm is also observed in cultured retinas, indicating its close interaction with the retinal oscillator. The rhythmically phosphorylated MAPK is detected only in a discrete subset of amacrine cells despite ubiquitous distribution of MAPK throughout the retinal layers. Treatment of the cultured retinas with MAPK kinase (MEK) inhibitor PD98059 suppresses MAPK phosphorylation during the subjective night, and this pulse perturbation of MEK activity induces a significant phase delay (4–8 h) of the retinal circadian rhythm in MAPK and MEK phosphorylation. These observations strongly suggest that the site-specific and time-of-day-specific activation of MAPK contributes to the circadian time-keeping mechanism of the retinal clock system.
      SCN
      suprachiasmatic nucleus
      MAPK
      mitogen-activated protein kinase
      MEK
      MAPK kinase
      TBS
      Tris-buffered saline
      ZT
      zeitgeber time
      CT
      circadian time
      LD
      12 h light/12 h dark
      DD
      constant darkness
      GABA
      γ-aminobutyric acid
      TH
      tyrosine hydroxylase
      The physiology and behavior of living organisms from bacteria to humans show daily fluctuations with a period of approximately 24 h, and the oscillations controlled by autonomous circadian oscillators are termed circadian rhythms (
      • Takahashi J.S.
      , ). These rhythms can be synchronized (entrained) to environmental time cues such as light or temperature but are sustained under constant environmental conditions in the absence of time cues (
      • Takahashi J.S.
      ). A lot of studies on the Drosophila melanogaster clock system have demonstrated that a transcription/translation-based feedback loop is involved in generating the circadian rhythmicity (,
      • Young M.W.
      ). In Drosophila, a complex of two transcription factors, dCLOCK and CYCLE (dBMAL1), positively regulates both period (per) andtimeless (tim) transcription through CACGTG E-box elements found in their promoters (
      • Allada R.
      • White N.E.
      • So W.V.
      • Hall J.C.
      • Rosbash M.
      ,
      • Darlington T.K.
      • Wager-Smith K.
      • Ceriani M.F.
      • Staknis D.
      • Gekakis N.
      • Steeves T.D.L.
      • Weitz C.J.
      • Takahashi J.S.
      • Kay S.A.
      ,
      • Rutila J.E.
      • Suri V.
      • Le M.
      • So W.V.
      • Rosbash M.
      • Hall J.C.
      ), leading to an increase in protein levels of both PER and TIM. PER and TIM proteins form a heterodimer and translocate to the nucleus where PER/TIM negatively regulate dCLOCK/CYCLE-induced transcription of their own promoters (
      • Darlington T.K.
      • Wager-Smith K.
      • Ceriani M.F.
      • Staknis D.
      • Gekakis N.
      • Steeves T.D.L.
      • Weitz C.J.
      • Takahashi J.S.
      • Kay S.A.
      ,
      • Sehgal A.
      • Price J.L.
      • Man B.
      • Young M.W.
      ,
      • Vosshall L.B.
      • Price J.L.
      • Sehgal A.
      • Saez L.
      • Young M.W.
      ,
      • Gekakis N.
      • Saez L.
      • Delahaye-Brown A.-M.
      • Myers M.P.
      • Sehgal A.
      • Young M.W.
      • Weitz C.J.
      ,
      • Myers M.P.
      • Wager-Smith K.
      • Rothenfluh-Hilfiker A.
      • Young M.W.
      ,
      • Saez L.
      • Young M.W.
      ). Thus, these positive and negative elements constitute an autoregulatory feedback loop.
      In vertebrates, autonomous circadian oscillators are located in several neuronal tissues such as the retina, pineal gland, and suprachiasmatic nucleus (SCN)1 (
      • Takahashi J.S.
      ,
      • Kasal C.A.
      • Menaker M.
      ,
      • Deguchi T.
      ,
      • Cahill G.M.
      • Besharse J.C.
      ,
      • Welsh D.K.
      • Logothetis D.E.
      • Meister M.
      • Reppert S.M.
      ,
      • Tosini G.
      • Menaker M.
      ,
      • Whitmore D.
      • Foulkes N.S.
      • Strähle U.
      • Sassone-Corsi P.
      ). Homologs of per, clock, and bmal1 have been identified in vertebrates, and they are expressed in SCN of mammals (reviewed in Refs. and ). It is now postulated that the molecular framework of the circadian oscillator is conserved from flies to mammals, although each clock component appears to function in a slightly different manner (reviewed in Refs.
      • Sassone-Corsi P.
      and
      • Hardin P.E.
      • Glossop N.R.J.
      ).
      In addition to the transcriptional and translational regulation, post-translational modifications of clock gene products play important roles in the clock system. Drosophila PER, TIM, and CLOCK proteins are phosphorylated in a daily/circadian manner (
      • Edery I.
      • Zwiebel L.J.
      • Dembinska M.E.
      • Rosbash M.
      ,
      • Zeng H.
      • Qian Z.
      • Myers M.P.
      • Rosbash M.
      ,
      • Lee C.
      • Bae K.
      • Edery I.
      ), and another clock gene product, DOUBLETIME, closely related to human casein kinase Iε, has been shown to reduce the stability of PER by phosphorylation (
      • Price J.L.
      • Blau J.
      • Rothenfluh A.
      • Abodeely M.
      • Kloss B.
      • Young M.W.
      ,
      • Kloss B.
      • Price J.L.
      • Saez L.
      • Blau J.
      • Rothenfluh A.
      • Wesley C.S.
      • Young M.W.
      ). Also, Drosophila TIM is degraded upon receiving light, probably in a tyrosine kinase- and proteasome-dependent manner (
      • Myers M.P.
      • Wager-Smith K.
      • Rothenfluh-Hilfiker A.
      • Young M.W.
      ,
      • Zeng H.
      • Qian Z.
      • Myers M.P.
      • Rosbash M.
      ,
      • Naidoo N.
      • Song W.
      • Hunter-Ensor M.
      • Sehgal A.
      ). Although these posttranslational events significantly contribute to the circadian timing mechanism in Drosophila, little is known about the molecular nature of protein kinases, phosphatases, and proteases in vertebrate clock systems.
      Recently, we have demonstrated circadian phosphorylation of mitogen-activated protein kinase (MAPK) in the chick pineal gland (
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      ). Importantly, circadian-phosphorylated MAPK was localized in follicular pinealocytes (
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      ), where the pineal circadian oscillator is located (
      • Nakahara K.
      • Murakami N.
      • Nasu T.
      • Kuroda H.
      • Murakami T.
      ), suggesting an intracellular linkage between the circadian activation of MAPK and the oscillator. This is also true of the mouse SCN (
      • Obrietan K.
      • Impey S.
      • Storm D.R.
      ), although the phase of the circadian activation of MAPK in the mouse SCN is opposite to that of pineal MAPK (i.e. daytime-versus nighttime-specific activation; Refs.
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      and
      • Obrietan K.
      • Impey S.
      • Storm D.R.
      ). Here we investigated the MAPK activation cycle and its contribution to the oscillator in the retina, which is another well defined clock-containing tissue of vertebrates. It was clearly demonstrated that bullfrog (Rana catesbeiana) retinal MAPK exhibitsin vivo and in vitro circadian rhythm in activation peaked at nighttime, and this rhythmicity was found in a discrete subset of amacrine cells, which could be a candidate for clock location in the retina. Noticeably, the retinal circadian rhythms in MAPK and MAPK kinase (MEK) phosphorylation were phase-delayed by a pulse perturbation of MEK activity, suggesting that the MEK-MAPK pathway plays an important role in maintenance of the circadian time-keeping mechanism basically common to vertebrate clock systems.

      EXPERIMENTAL PROCEDURES

       Animals and Preparation of Tissue Homogenate

      Adult bullfrogs (R. catesbeiana) were entrained to 12-h light/12-h dark (LD) cycles for 7 days before experiments. The bullfrogs were sacrificed by decapitation, and their retinas were immediately homogenized with 500 μl of ice-cold lysis buffer (20 mmTris-HCl, 1% (v/v) Triton X-100, 10% (v/v) glycerol, 137 mm NaCl, 2 mm EDTA, 50 mm NaF, 1 mm sodium orthovanadate, 4 μg/ml aprotinin, and 4 μg/ml leupeptin, pH 8.0 at 4 °C). Then the homogenate was centrifuged (10,000 × g, 10 min), and the supernatant (80 μl) was mixed with 20 μl of 5× SDS-polyacrylamide gel electrophoresis sample solution followed by boiling for 8 min. All of the manipulations during the dark period were performed under infrared light (>800 nm) with the aid of darkroom goggles (NEC, Tokyo, Japan).

       Tissue Culture

      Adult bullfrogs were entrained to LD cycles for 7 days, and the retinas were isolated 2–4 h before the end of the light period. Each retina was placed on a Millicell CM membrane (Millipore Corp.) with its photoreceptor layer upward and then cultured under 90% O2/5% CO2 at 21 °C in 2 ml of defined culture medium (
      • Besharse J.C.
      • Dunis D.A.
      ) supplemented with 5% fetal bovine serum (JRH Biosciences), 100 μm hydroxytryptophan, 100 μg/ml streptomycin sulfate, 100 units/ml penicillin, 100 μml-ascorbic acid, and amino acid mixture (0.61 mm Arg, 0.17 mm Cys, 1.00 mm Glu, 0.27 mm Gly, 0.17 mm His, 0.33 mmIle, 0.33 mm Leu, 0.41 mm Lys, 0.99 mm Met, 0.17 mm Phe, 0.33 mm Thr, 0.04 mm Trp, 0.17 mm Tyr, and 0.33 mm Val, each at a final concentration). At the end of the light period, the tissues were transferred to constant darkness (DD) and kept in culture. The retinal homogenate was prepared from the cultured retinas as described above.

       Immunoblotting

      Proteins separated by SDS-polyacrylamide gel electrophoresis (13% acrylamide for Fig. 1 A or 8.5% acrylamide for the others) were transferred to polyvinylidene difluoride membranes (Millipore Corp.). The blots were blocked in 1% (w/v) skim milk in TBS (pH 7.4) for 1 h at 37 °C and then incubated at 4 °C overnight with a primary antibody in the blocking solution. MAPK was detected by anti-MAPK antibody (sc-154, 1:5000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and the dually phosphorylated form of MAPK was detected by anti-active MAPK antibody (1:1000; Promega) or rabbit polyclonal anti-phospho-MAPK antibody (1:1000; New England Biolabs). MEK was detected by anti-MEK1/2 antibody (1:1000; New England Biolabs), and the dually phosphorylated form of MEK was detected by anti-phospho-MEK1/2 antibody (1:1000; New England Biolabs). The primary antibodies were detected by horseradish peroxidase-conjugated anti-rabbit IgG antibody (1:5000; Kirkegaard & Perry Laboratories, Inc.), and the positive signals were visualized by an enhanced chemiluminescence detection system (Renaissance; PerkinElmer Life Sciences). When reprobed, the blots were stripped according to the manufacturer's protocol.
      Figure thumbnail gr4
      Figure 4Dose-dependent inhibition of retinal MAPK phosphorylation by PD98059. The retinas were cultured from ZT10, treated with 0.1% Me2SO (A, control) or with PD98059 (8 μm in B, 80 μm in C) from ZT12 of day 1, and transferred to the DD condition. On day 2, two retinas were collected at each indicated time point from culture for preparing the homogenate in the dark. Equal amounts of proteins (40 μg) were immunoblotted with anti-active MAPK antibody, and the same blot was reprobed with anti-MAPK antibody. The levels of phosphorylated MAPK were quantified by a densitometric scanning of the immunostained membrane, and the values were shown in D as percentages relative to the highest value (a point at CT12 in control). The data are a representative set of two independent experiments with similar results.
      Figure thumbnail gr1
      Figure 1Circadian rhythm in MAPK phosphorylation in the bullfrog retina. A, the bullfrog retinal homogenate (40 μg of proteins) was immunoblotted with anti-MAPK antibody (lane 1), anti-active MAPK antibody (lane 2), or anti-phospho-MAPK antibody (lane 3). B, the bullfrogs were maintained under LD cycles for 7 days and then transferred to the DD condition. The retinas were isolated at the indicated time of day 8 or 9 in the dark, and equal amounts of proteins (40 μg) of the retinal homogenate were immunoblotted with anti-active MAPK antibody (upper panel) or with anti-MAPK antibody (lower panel). The shaded and black areas at the bottom represent the subjective day and subjective night, respectively.

       Preparation of Tissue Sections and Immunofluorescence Analyses

      Cultured retinas were fixed in 4% paraformaldehyde in 0.1 m sodium phosphate buffer (pH 7.4) for 4 h at 4 °C. Then the tissues were cryoprotected with 20% sucrose in sodium phosphate buffer, embedded in the OCT mounting medium (Sakura, Japan), frozen by using liquid nitrogen, and stored at −80 °C until use. Finally, 10-μm-thick sections were cut out from the embedded tissues, mounted on gelatin-coated glass slides, and air-dried. The sections on glass slides were pretreated with a blocking solution (10% bovine serum albumin, 1% horse or goat normal serum, 0.5% Triton X-100 in TBS, pH 7.4) for 30 min at room temperature and then incubated with a primary antibody diluted in the blocking solution at 4 °C for 72 h. After rinsing with TBS, the sections were treated with a secondary antibody for 24 h at 4 °C and again washed with TBS. Then the sections were mounted with Vectashield mounting medium (Vector Laboratories). The primary antibodies used were mouse monoclonal anti-phospho-MAPK antibody (1:50; New England Biolabs), rabbit anti-MAPK antibody (1:400; Santa Cruz Biotechnology), rabbit anti-tyrosine hydroxylase antiserum (1:1000; Protos Biotech Corp.), and rabbit anti-glutamic acid decarboxylase antiserum (1:1000; Chemicon). The secondary antibodies used were fluorescein-conjugated horse anti-mouse IgG (1:1000; Vector Laboratories) and Alexa-568-conjugated goat anti-rabbit IgG (1:2000; Molecular Probes, Inc., Eugene, OR). TO-PRO-3 (Molecular Probes) was used for staining of cell nuclei. Images of the sections were viewed with a confocal laser-scanning microscope (Leica, TCS-NT).

      RESULTS

       Circadian Rhythm of MAPK Phosphorylation

      MAPKs are highly conserved serine/threonine kinases. When a bullfrog retinal homogenate (a mixture of those prepared in the midday and midnight) was subjected to an immunoblot analysis, a single band of 41-kDa protein was recognized by anti-MAPK antibody (Fig.1 A, lane 1). This protein was assigned as bullfrog MAPK, because (i) the molecular mass was close to the reported value of Xenopus MAPK (42 kDa; Ref.
      • Gotoh Y.
      • Nishida E.
      • Matsuda S.
      • Shiina N.
      • Kosako H.
      • Shiokawa K.
      • Akiyama T.
      • Ohta K.
      • Sakai H.
      ) and (ii) this protein band was also recognized by two kinds of antibodies, anti-active MAPK antibody (lane 2) and anti-phospho-MAPK antibody (lane 3), which are specific to MAPK phosphorylated on both threonine and tyrosine residues (
      • Gotoh Y.
      • Nishida E.
      • Matsuda S.
      • Shiina N.
      • Kosako H.
      • Shiokawa K.
      • Akiyama T.
      • Ohta K.
      • Sakai H.
      ,
      • Anderson N.G.
      • Maller J.L.
      • Tonks N.K.
      • Sturgill T.W.
      ,
      • Payne D.M.
      • Rossomando A.J.
      • Martino P.
      • Erickson A.K.
      • Her J.-H.
      • Shabanowitz J.
      • Hunt D.F.
      • Weber M.J.
      • Sturgill T.W.
      ). Since MAPK is activated by the dual phosphorylation (
      • Anderson N.G.
      • Maller J.L.
      • Tonks N.K.
      • Sturgill T.W.
      ), these antibodies are capable of detecting MAPK activation in the bullfrog retina.
      To investigate a possible involvement of MAPK in the bullfrog retinal clock system, the phosphorylation state of retinal MAPK was estimated in the retinal homogenate prepared at various time points from bullfrogs maintained in DD (Fig. 1 B) and LD (Fig.2 B) conditions. Immunoblot analysis of the retinal homogenate revealed that under the DD condition, MAPK phosphorylation peaked in the mid- to late-subjective night (CT18–24; under the DD condition, CT0 and CT12 refer to the time of lights on and off in the previous LD cycle, respectively), and it became dephosphorylated in the late subjective day (Fig. 1 B,upper panel). The MAPK phosphorylation rhythm is also observed under the LD cycle (Fig. 2 B, upper panel) with a phase similar to that in the DD condition. The levels of MAPK protein were nearly constant throughout the period both in DD and LD conditions (Figs. 1 B and 2 B,lower panels). Clearly, the daily change in MAPK phosphorylation is regulated by a self-sustaining mechanism characteristic of a circadian oscillator.
      Figure thumbnail gr2
      Figure 2Entrainment of MAPK phosphorylation rhythm to the shifted LD cycle. The bullfrogs maintained in LD cycles for a week were transferred to the inverted LD cycle at the day 1 and maintained for 8 days in the new LD cycle (A). The retinal homogenates were prepared at the indicated time points (shown byasterisks in A), and the samples in day 0 (B), day 3 (C), and day 8 (D) were immunoblotted with anti-active MAPK antibody (B–D,upper panels) or with anti-MAPK antibody (B–D,lower panels). The white and black areas at the bottom represent the periods of light and dark, respectively.

       Phase Shift of MAPK Phosphorylation Rhythm by Light

      Another feature of the circadian rhythm is its adaptive property (entrainment) of the phase to a shifted daily change in environmental conditions such as LD cycle. To investigate the relationship between the retinal MAPK phosphorylation and the oscillator, we examined how the phosphorylation rhythm responds to the shifted LD cycle. The bullfrogs were maintained for 7 days under LD cycles and then transferred to a reversed LD cycle. The retinal homogenates were prepared at various time points in the day (day 0) before the transfer to the new LD cycle and in days 3 and 8 thereafter (Fig. 2 A). In the original LD cycle (in day 0), MAPK phosphorylation peaked in the mid- to late-night (ZT20–24; ZT0 is lights on, and ZT12 is lights off) and declined in the daytime (Fig.2 B, upper panel). At the day 3 in the new LD cycle, the phosphorylation level had a peak at the LD transition (ZT12–18; Fig. 2 C, upper panel), and at day 8, the phosphorylation became evident only in the nighttime (Fig.2 D, upper panel). This fluctuation profile at day 8 is very similar to that observed at day 0, indicating a complete phase shift of phosphorylation rhythm within a week. The gradual entrainment of the phosphorylation rhythm to the shifted LD cycle resembles that of mouse per1 expression rhythm observed in the mouse SCN (
      • Sun Z.S.
      • Albrecht U.
      • Zhuchenko O.
      • Bailey J.
      • Eichele G.
      • Lee C.C.
      ), and this suggests a close linkage between the circadian oscillator and the phosphorylation rhythm of retinal MAPK.

       In Vitro Circadian Phosphorylation of MAPK in the Retina

      In amphibians, circadian oscillators are located not only in the retina but also in the hypothalamus. For example, the circadian locomotor activity rhythm of Xenopus laevis is lost by lesion of the hypothalamus including the SCN (
      • Harada Y.
      • Goto M.
      • Ebihara S.
      • Fujisawa H.
      • Kegasawa K.
      • Oishi T.
      ). Also, the neural connections of the hypothalamus to the retina have been reported neuroanatomically inRana esculenta (
      • Kemali M.
      • De Santis A.
      ). In order to evaluate the contribution of the retinal circadian system to the phosphorylation rhythm of retinal MAPK, the state of MAPK phosphorylation was characterized in the isolated bullfrog retinas. As shown in Fig.3 A (upper panel), MAPK phosphorylation in the cultured retina showed a daily rhythm even under the DD condition with a peak in the early subjective night (CT12–16), whereas MAPK protein levels were nearly constant throughout the period (Fig. 3 A, lower panel). The rhythmicity of MAPK phosphorylation persisted for at least 2 cycles (Fig. 3 B), indicating that the molecular cycle is linked to the endogenous circadian clock system in the retina. The peak time of the phosphorylation level is advanced by 4–6 h relative to that observed in in vivo experiments under the DD condition (Fig.1 B). Such a difference in phase between in vivoand in vitro experiments has often been observed in studies on various clock systems (
      • Kasal C.A.
      • Menaker M.
      ,
      • Deguchi T.
      ,
      • Whitmore D.
      • Foulkes N.S.
      • Strähle U.
      • Sassone-Corsi P.
      ,
      • Greve P.
      • Voisin P.
      • Grechez-Cassiau A.
      • Bernard M.
      • Collin J.-P.
      • Guerlotte J.
      ), and it has been attributed to the effects of tissue dissection and/or culture.
      Figure thumbnail gr3
      Figure 3Circadian rhythm of MAPK phosphorylation in the cultured retina. The retinas were isolated from the bullfrogs entrained to LD cycles for 7 days, cultured from ZT10, and transferred to the DD condition at ZT12 of day 1. Two retinas were collected from culture for preparing the homogenate in the dark every 4 h (A) or 6 h (B). Equal amounts of proteins (40 μg) were immunoblotted with anti-active MAPK antibody (A, upper panel) or anti-phospho-MAPK antibody (B, upper panel), and the same blots were reprobed with anti-MAPK antibody (A and B,lower panels).

       Role of MAPK Phosphorylation Rhythm in the Circadian System

      The autonomous rhythm of MAPK phosphorylation in the cultured retina suggests a role of MAPK acting as an output of the clock system or alternatively as a part of the oscillation mechanism itself. To investigate the latter possibility, we examined whether a transient inhibition of retinal MAPK activity affects the circadian clock system. For this purpose, we used PD98059 (
      • Dudley D.T.
      • Pang L.
      • Decker S.J.
      • Bridges A.J.
      • Saltiel A.R.
      ,
      • Alessi D.R.
      • Cuenda A.
      • Cohen P.
      • Dudley D.T.
      • Saltiel A.R.
      ), a specific and reversible inhibitor of MEK responsible for dual phosphorylation and activation of MAPK (
      • Zheng C.-F.
      • Guan K.-L.
      ,
      • Seger R.
      • Krebs E.G.
      ). The bullfrog retinas were treated with 8 or 80 μm PD98059 from ZT12 (on day 1) and transferred to the DD condition, and the level of MAPK phosphorylation was investigated during the drug treatment in the dark (on day 2). The time-of-day-specific phosphorylation of MAPK (Fig.4 A and open circles in Fig. 4 D) was inhibited in a dose-dependent manner by the drug, and 80 μm PD98059 almost completely suppressed the phosphorylation peak (Fig. 4 C and closed squares in Fig. 4 D). During the drug treatment, the levels of MAPK protein showed no significant change (Fig. 4,A–C, lower panels). These results demonstrate that MEK is an upstream kinase responsible for circadian phosphorylation of retinal MAPK and coincide nicely with the observation of the circadian oscillation in the level of MEK phosphorylation (see below and Fig.5 B). We then investigated the temporal change in MAPK phosphorylation under the DD condition after the treatment of the cultured retinas with 80 μm PD98059 for 12 h (at ZT10–22). The activities of MAPK and MEK seemed to be fully recovered after the removal of PD98059, because the maximal phosphorylation levels of MAPK and MEK in the drug-treated experiment were similar to those in a control one (Fig. 5). However, the 12-h treatment with PD98059 in the first day of culture induced a delay (4–8 h) of the peak time in the phosphorylation rhythm in the following days (Fig. 5 A, third panel) as compared with that in a control experiment (Fig. 5 A, first panel). In the parallel experiment, we also examined the MEK phosphorylation level representing the MEK kinase activity. As shown in Fig. 5 B, we observed a similar phase delay of MEK phosphorylation rhythm by the pretreatment of PD98059. These results indicate that the retinal circadian clock still remains oscillating even after the transient loss of MEK activation, which vice versa gave a strong effect on the phase of the rhythm. The MEK-MAPK pathway seems to play a pivotal role in circadian time keeping mechanism in the retinal clock.
      Figure thumbnail gr5
      Figure 5Effect of PD98059 on the circadian rhythm of MAPK phosphorylation. The retinas were cultured from ZT8, treated with 80 μm PD98059 or Me2SO for 12 h (ZT10–22) on day 1, and transferred to the DD condition at ZT12 on day 1. At each indicated time point in day 2–3, two retinas were collected, and the homogenates (40 μg proteins) were subjected to immunoblotting with anti-phospho-MAPK antibody (A,first and third panels), with anti-MAPK antibody (A, second and forth panels), with anti-phospho-MEK antibody (B, first andthird panels), or with anti-MEK antibody (B,second and fourth panels). The doublet bands (43- and 45-kDa proteins in B) were assigned as bullfrog MEKs, since they were recognized by both anti-MEK antibody and anti-phospho-MEK antibody.

       Localization of Circadian-phosphorylated MAPK in the Retina

      To determine the site of circadian-phosphorylated MAPK in the retina, thin sections were prepared from the cultured retinas at various time points under the DD condition, and they were subjected to immunostaining. Immunoreactivity to MAPK was widespread throughout the retinal layers at any time within a day (Fig.6, B and F). On the other hand, no significant immunoreactivity to phosphorylated MAPK was observed in the retinal section prepared at CT24 (Fig. 6 E), while at CT12 a discrete subset of amacrine cells showed strong positive signals at their cell bodies and moderate signals at their processes in the inner plexiform layer (Fig. 6 A). In a more magnified view (Fig. 6, I–K), phosphorylated MAPK is located mainly in the cytosol (green) and diffusely in the nucleus (yellow) at CT 12. The peak time in signal intensity at the cell bodies was consistent with that observed in the immunoblot analysis (Fig. 3), indicating that, despite ubiquitous distribution of MAPK in the retina, the circadian phosphorylation and activation of retinal MAPK occur only in a subset of amacrine cells.
      Figure thumbnail gr6
      Figure 6Localization of circadian-phosphorylated MAPK in the bullfrog retina. A–H, the bullfrog retinas were cultured as described in the legend to Fig. , and 10-μm-thick sections were prepared from the retina fixed at CT12 (A–D) or at CT24 (E–H) on the second day of culture, followed by immunostaining. Shown are the immunofluorescence images labeled with anti-phospho-MAPK antibody (A and E,green) and with anti-MAPK antibody (B andF, red). C and G are the merged images (A + B and E +F, respectively), and D and H are their differential interference contrast images, respectively. The positions of time-of-day-specific positive signals at the inner nuclear layer (INL) and the inner plexiform layer (IPL) are indicated by a white arrowhead andwhite arrows, respectively. I–S, the cultured bullfrog retina was fixed at CT12, and 10-μm-thick sections were prepared. The retinal sections were immunostained by anti-phospho-MAPK antibody (I, green), and cell nuclei were stained by TO-PRO-3 (J, red). The retinal sections were immunostained with a combination of anti-phospho-MAPK antibody (L, green) and anti-TH antiserum (M, red) or with a combination of anti-phospho-MAPK antibody (P, green) and anti-glutamic acid decarboxylase antiserum (Q,red). K, N, and R show the merged images (I + J, L +M, and P + Q, respectively), andO and S show their differential interference contrast images, respectively. PRL, photoreceptor layer;ONL, outer nuclear layer; OPL, outer plexiform layer, GCL, ganglion cell layer. Scale bars, 50 μm (D, H, O, andS) and 20 μm (K).
      Retinal amacrine cells have been classified into a variety of subtypes, which are characterized by diverged neurotransmitters such as γ-aminobutyric acid (GABA), glycine, dopamine, and acetylcholine (
      • Kolb H.
      ). To investigate the type of amacrine cells showing circadian phosphorylation of MAPK, the retinal sections were immunostained by antibodies to two markers. One is tyrosine hydroxylase (TH), an enzyme responsible for dopamine synthesis (
      • Osborne N.N.
      • Patel S.
      • Vigny A.
      ), and the other is glutamic acid decarboxylase for synthesis of GABA (
      • Lam D.M.K.
      • Su Y.Y.T.
      • Swain L.
      • Marc R.E.
      • Brandon C.
      • Wu J.-Y.
      ). In the double staining analysis, the TH antiserum immunostained some amacrine cells (Fig.6 M), but the phospho-MAPK- and TH-immunoreactive cells were mutually exclusive (Fig. 6 N). Similarly, the glutamic acid decarboxylase antiserum labeled a subset of amacrine cells and horizontal cells (Fig. 6 Q), but they are not identical to the phospho-MAPK-immunoreactive amacrine cells (Fig. 6 R). These results indicate that the phospho-MAPK-immunoreactive amacrine cells represent nondopaminergic and non-GABAergic neurons.

      DISCUSSION

      We have previously demonstrated circadian activation of MAPK in the chick pineal gland (
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      ), and others showed an overt rhythm of MAPK phosphorylation in the mouse SCN (
      • Obrietan K.
      • Impey S.
      • Storm D.R.
      ). The present study demonstrating rhythmic phosphorylation of MAPK in the bullfrog retina, together with those in the two defined pacemaker structures, avian pineal gland and mammalian SCN, strongly suggests that phosphorylation rhythm of MAPK is conserved among various animal species and among clock-containing tissues. The phosphorylation rhythms of MEK and MAPK persisted in vitro under the DD condition (Figs. 3 and 5), and hence they seem to represent outputs of the retinal circadian oscillator. On the other hand, the pulse perturbation of the MEK activity induced the phase shift of the rhythm (Fig. 5), implying that the MEK-MAPK pathway affects the oscillator. Thus, the MEK-MAPK pathway acts as if it were both an input and output of the retinal oscillator. One of the possible models postulates these separate roles of the MEK-MAPK pathway, but this requires careful consideration of the subcellular distribution of circadian-activated MAPK (see below). A more straightforward model is that circadian-activated MAPK feeds back the circadian signal to the oscillator, forming a secondary loop. This would explain why the MEK-MAPK pathway and the circadian oscillator mutually affect each other. We previously demonstrated a marked phase shift of the circadian rhythm induced by a transient suppression of MEK activity in the chick pineal gland (
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      ). Taken together, the MEK-MAPK pathway is likely to play an important role in the circadian time-keeping mechanism common to the vertebrate clock systems, although a physiological significance of circadian activation of MAPK in the mammalian SCN remains to be elucidated (
      • Obrietan K.
      • Impey S.
      • Storm D.R.
      ). The phosphorylation rhythm of MAPK seems to reflect an overall oscillation of the Raf-MEK-MAPK,
      Y. Hayashi, K. Sanada, and Y. Fukada, manuscript in preparation.
      but we cannot exclude a formal possibility that another MEK target distinct from MAPK is responsible for the phase shift induced by the PD98059 treatment (Fig. 5).
      An alternative role of MAPK in the vertebrate clock system could be that in synchronization among nearby clock cells via extracellular signals. For instance, melatonin and dopamine are released from retinal cells in a circadian manner (
      • Cahill G.M.
      • Besharse J.C.
      ,
      • Wirz-Justice A.
      • Da Prada M.
      • Reme C.E.
      ,
      • Adachi A.
      • Nogi T.
      • Ebihara S.
      ), and these seem to contribute to coupling of individual oscillators distributed among clock cells (
      • Adachi A.
      • Nogi T.
      • Ebihara S.
      ,
      • Liu C.
      • Weaver D.R.
      • Jin X.
      • Shearman L.P.
      • Pieschl R.L.
      • Gribkoff V.K.
      • Reppert S.M.
      ). MAPK might be regulated by these transmitters mediating synchronization of closely located cellular oscillators. In this case, MAPK is expected to respond quickly to the change in environmental conditions. In the present study, we observed a rather slow response of MAPK phosphorylation to the change in LD cycle condition (Fig. 2), suggesting a tighter association of MAPK with the central oscillator in the retina.
      In the chick pineal gland, MAPK phosphorylation rhythm is observed in follicular pinealocytes (
      • Sanada K.
      • Hayashi Y.
      • Harada Y.
      • Okano T.
      • Fukada Y.
      ), where the circadian oscillator is located (
      • Nakahara K.
      • Murakami N.
      • Nasu T.
      • Kuroda H.
      • Murakami T.
      ). Given its critical role in the circadian time-keeping mechanism (Fig. 5), circadian-activated MAPK could be a potential marker for detecting clock cells. In the Xenopus and mouse retinas, it has been demonstrated that the photoreceptor cells retain or constitute the circadian clock system regulating the melatonin rhythm in culture (
      • Cahill G.M.
      • Besharse J.C.
      ,
      • Tosini G.
      • Menaker M.
      ). In contrast, mouse per1, clock, andbmal1 are most strongly expressed in the inner nuclear layer of the mouse retina (
      • Gekakis N.
      • Staknis D.
      • Nguyen H.B.
      • Davis F.C.
      • Wilsbacher L.D.
      • King D.P.
      • Takahashi J.S.
      • Weitz C.J.
      ), and newly identified clock genes mousecry1 and cry2 (
      • van der Horst G.T.J.
      • Muijtjens M.
      • Kobayashi K.
      • Takano R.
      • Kanno S.
      • Takao M.
      • de Wit J.
      • Verkerk A.
      • Eker A.P.M.
      • van Leenen D.
      • Buijs R.
      • Bootsma D.
      • Hoeijmakers J.H.J.
      • Yasui A.
      ,
      • Kume K.
      • Zylka M.J.
      • Sriram S.
      • Shearman L.P.
      • Weaver D.R.
      • Jin X.
      • Maywood E.S.
      • Hastings M.H.
      • Reppert S.M.
      ) are expressed in the inner nuclear layer but not in the photoreceptor layer (
      • Miyamoto Y.
      • Sancar A.
      ). Thus, the identity of clock cells in the vertebrate retina is still controversial. The present study demonstrating the site-specific and time-of-day-specific activation of retinal MAPK (Fig. 6, Aand E) is consistent with the notion that the amacrine cells in the inner nuclear layer are the potential clock cells. This does not exclude a possibility that the bullfrog photoreceptor cells also retain the circadian oscillator with a lower amplitude in MAPK phosphorylation cycle or with a slightly different mechanism regarding the contribution of MAPK. In the circadian clock system within the amacrine cells, MAPK may be linked with the endogenous oscillator at the cell bodies in one hand, whereas MAPK at their plexus (see Fig. 6 A) may respond to the extracellular signals possibly for the entrainment of the clock or for synchronization among clock cells. It is possible that rhythmically activated MAPK could constitute a secondary loop that is interconnected with the core feedback loop at intracellular and/or intercellular levels. These loops could regulate and stabilize each other, and they may provide a basis for general properties of the circadian clock system such as period length (
      • Dunlap J.C.
      • Loros J.J.
      • Liu Y.
      • Crosthwaite S.K.
      ).
      A subset of amacrine cells synthesize dopamine (
      • Witkovsky P.
      • Schütte M.
      ,
      • Djamgoz M.B.A.
      • Wagner H.-J.
      ) and release it rhythmically (
      • Wirz-Justice A.
      • Da Prada M.
      • Reme C.E.
      ,
      • Adachi A.
      • Nogi T.
      • Ebihara S.
      ). It is unlikely, however, that the circadian activation of MAPK is linked intracellularly to the rhythmic release of dopamine, because phospho-MAPK- and TH-immunoreactive cells are not identical (Fig. 6 N). On the other hand, about half of the amacrine cells in the human retina contain GABA (
      • Crooks J.
      • Kolb H.
      ), and GABA seems to be involved in the regulation of dopamine synthesis/release, which indirectly affects retinal melatonin synthesis (
      • Kazula A.
      • Nowak J.Z.
      • Iuvone P.M.
      ,
      • Boatright J.H.
      • Rubim N.M.
      • Iuvone P.M.
      ,
      • Boatright J.H.
      • Rubim N.M.
      • Iuvone P.M.
      ). This led to speculation that circadian-activated MAPK is located at GABAergic amacrine cells, but this was not the case (Fig. 6 R). It is intriguing as well to see a possible colocalization of known clock gene products within the amacrine cells, but our ongoing efforts have been unsuccessful due to no cross-reactivity of commercially available antibodies to clock components of this animal species.
      The immunoreactivity to circadian-phosphorylated MAPK is localized both in the nucleus and cytosol of the immunopositive amacrine cells (Fig.6 K), suggesting that circadian-activated MAPK phosphorylates nuclear and/or cytosolic proteins as targets. DrosophilaPER, TIM, and CLOCK proteins are phosphorylated in a daily/circadian manner (
      • Edery I.
      • Zwiebel L.J.
      • Dembinska M.E.
      • Rosbash M.
      ,
      • Zeng H.
      • Qian Z.
      • Myers M.P.
      • Rosbash M.
      ,
      • Lee C.
      • Bae K.
      • Edery I.
      ), and in vertebrates, putative clock genes are expressed in the retinal inner nuclear layer (
      • Gekakis N.
      • Staknis D.
      • Nguyen H.B.
      • Davis F.C.
      • Wilsbacher L.D.
      • King D.P.
      • Takahashi J.S.
      • Weitz C.J.
      ,
      • Miyamoto Y.
      • Sancar A.
      ), where circadian-regulated MAPK resides. MAPK might regulate biochemical activities and/or stabilities of these clock proteins, through phosphorylation by itself or via MAPK-activated protein kinases such as p90 ribosomal S6 kinase. Another possible target downstream of MAPK is Ca2+/cAMP-response element-binding protein. Recently, it has been demonstrated that mammalian SCN and Drosophilaexhibit circadian rhythms in Ca2+/cAMP-response element-mediated gene expression (
      • Belvin M.P.
      • Zhou H.
      • Yin J.C.P.
      ,
      • Obrietan K.
      • Impey S.
      • Smith D.
      • Athos J.
      • Storm D.R.
      ). Since Ca2+/cAMP-response element-binding protein is phosphorylated by p90 ribosomal S6 kinase (
      • Xing J.
      • Ginty D.D.
      • Greenberg M.L.
      ), it seems likely that the p90 ribosomal S6 kinase-Ca2+/cAMP-response element-binding protein signaling pathway is regulated by MAPK for circadian time keeping of the oscillator.

      REFERENCES

        • Takahashi J.S.
        Annu. Rev. Neurosci. 1995; 18: 531-553
        • Dunlap J.C.
        Cell. 1999; 96: 271-290
        • Young M.W.
        Annu. Rev. Biochem. 1998; 67: 135-152
        • Allada R.
        • White N.E.
        • So W.V.
        • Hall J.C.
        • Rosbash M.
        Cell. 1998; 93: 791-804
        • Darlington T.K.
        • Wager-Smith K.
        • Ceriani M.F.
        • Staknis D.
        • Gekakis N.
        • Steeves T.D.L.
        • Weitz C.J.
        • Takahashi J.S.
        • Kay S.A.
        Science. 1998; 280: 1599-1603
        • Rutila J.E.
        • Suri V.
        • Le M.
        • So W.V.
        • Rosbash M.
        • Hall J.C.
        Cell. 1998; 93: 805-814
        • Sehgal A.
        • Price J.L.
        • Man B.
        • Young M.W.
        Science. 1994; 263: 1603-1606
        • Vosshall L.B.
        • Price J.L.
        • Sehgal A.
        • Saez L.
        • Young M.W.
        Science. 1994; 263: 1606-1609
        • Gekakis N.
        • Saez L.
        • Delahaye-Brown A.-M.
        • Myers M.P.
        • Sehgal A.
        • Young M.W.
        • Weitz C.J.
        Science. 1995; 270: 811-815
        • Myers M.P.
        • Wager-Smith K.
        • Rothenfluh-Hilfiker A.
        • Young M.W.
        Science. 1996; 271: 1736-1740
        • Saez L.
        • Young M.W.
        Neuron. 1996; 17: 911-920
        • Kasal C.A.
        • Menaker M.
        Science. 1979; 203: 656-658
        • Deguchi T.
        Science. 1979; 203: 1245-1247
        • Cahill G.M.
        • Besharse J.C.
        J. Neurochem. 1990; 54: 716-719
        • Welsh D.K.
        • Logothetis D.E.
        • Meister M.
        • Reppert S.M.
        Neuron. 1995; 14: 697-706
        • Tosini G.
        • Menaker M.
        Science. 1996; 272: 419-421
        • Whitmore D.
        • Foulkes N.S.
        • Strähle U.
        • Sassone-Corsi P.
        Nat. Neurosci. 1998; 1: 701-707
        • Reppert S.M.
        Neuron. 1998; 21: 1-4
        • Sassone-Corsi P.
        Nature. 1998; 392: 871-874
        • Hardin P.E.
        • Glossop N.R.J.
        Science. 1999; 286: 2460-2461
        • Edery I.
        • Zwiebel L.J.
        • Dembinska M.E.
        • Rosbash M.
        Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 2260-2264
        • Zeng H.
        • Qian Z.
        • Myers M.P.
        • Rosbash M.
        Nature. 1996; 380: 129-135
        • Lee C.
        • Bae K.
        • Edery I.
        Neuron. 1998; 21: 857-867
        • Price J.L.
        • Blau J.
        • Rothenfluh A.
        • Abodeely M.
        • Kloss B.
        • Young M.W.
        Cell. 1998; 94: 83-95
        • Kloss B.
        • Price J.L.
        • Saez L.
        • Blau J.
        • Rothenfluh A.
        • Wesley C.S.
        • Young M.W.
        Cell. 1998; 94: 97-107
        • Naidoo N.
        • Song W.
        • Hunter-Ensor M.
        • Sehgal A.
        Science. 1999; 285: 1737-1741
        • Sanada K.
        • Hayashi Y.
        • Harada Y.
        • Okano T.
        • Fukada Y.
        J. Neurosci. 2000; 20: 986-991
        • Nakahara K.
        • Murakami N.
        • Nasu T.
        • Kuroda H.
        • Murakami T.
        Brain Res. 1997; 774: 242-245
        • Obrietan K.
        • Impey S.
        • Storm D.R.
        Nat. Neurosci. 1998; 1: 693-700
        • Besharse J.C.
        • Dunis D.A.
        Exp. Eye Res. 1983; 36: 567-580
        • Gotoh Y.
        • Nishida E.
        • Matsuda S.
        • Shiina N.
        • Kosako H.
        • Shiokawa K.
        • Akiyama T.
        • Ohta K.
        • Sakai H.
        Nature. 1991; 349: 251-254
        • Anderson N.G.
        • Maller J.L.
        • Tonks N.K.
        • Sturgill T.W.
        Nature. 1990; 343: 651-653
        • Payne D.M.
        • Rossomando A.J.
        • Martino P.
        • Erickson A.K.
        • Her J.-H.
        • Shabanowitz J.
        • Hunt D.F.
        • Weber M.J.
        • Sturgill T.W.
        EMBO J. 1991; 10: 885-892
        • Sun Z.S.
        • Albrecht U.
        • Zhuchenko O.
        • Bailey J.
        • Eichele G.
        • Lee C.C.
        Cell. 1997; 90: 1003-1011
        • Harada Y.
        • Goto M.
        • Ebihara S.
        • Fujisawa H.
        • Kegasawa K.
        • Oishi T.
        Biol. Rhythm Res. 1998; 29: 30-48
        • Kemali M.
        • De Santis A.
        Exp. Brain Res. 1983; 53: 193-196
        • Greve P.
        • Voisin P.
        • Grechez-Cassiau A.
        • Bernard M.
        • Collin J.-P.
        • Guerlotte J.
        Biochem. J. 1996; 319: 761-766
        • Dudley D.T.
        • Pang L.
        • Decker S.J.
        • Bridges A.J.
        • Saltiel A.R.
        Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 7686-7689
        • Alessi D.R.
        • Cuenda A.
        • Cohen P.
        • Dudley D.T.
        • Saltiel A.R.
        J. Biol. Chem. 1995; 270: 27489-27494
        • Zheng C.-F.
        • Guan K.-L.
        J. Biol. Chem. 1993; 268: 11435-11439
        • Seger R.
        • Krebs E.G.
        FASEB J. 1995; 9: 726-735
        • Kolb H.
        Eye. 1997; 11: 904-923
        • Osborne N.N.
        • Patel S.
        • Vigny A.
        Histochemistry. 1984; 80: 389-393
        • Lam D.M.K.
        • Su Y.Y.T.
        • Swain L.
        • Marc R.E.
        • Brandon C.
        • Wu J.-Y.
        Nature. 1979; 278: 565-567
        • Wirz-Justice A.
        • Da Prada M.
        • Reme C.E.
        Neurosci. Lett. 1984; 45: 21-25
        • Adachi A.
        • Nogi T.
        • Ebihara S.
        Brain Res. 1998; 792: 361-369
        • Liu C.
        • Weaver D.R.
        • Jin X.
        • Shearman L.P.
        • Pieschl R.L.
        • Gribkoff V.K.
        • Reppert S.M.
        Neuron. 1997; 19: 91-102
        • Cahill G.M.
        • Besharse J.C.
        Neuron. 1993; 10: 573-577
        • Tosini G.
        • Menaker M.
        Brain Res. 1998; 789: 221-228
        • Gekakis N.
        • Staknis D.
        • Nguyen H.B.
        • Davis F.C.
        • Wilsbacher L.D.
        • King D.P.
        • Takahashi J.S.
        • Weitz C.J.
        Science. 1998; 280: 1564-1569
        • van der Horst G.T.J.
        • Muijtjens M.
        • Kobayashi K.
        • Takano R.
        • Kanno S.
        • Takao M.
        • de Wit J.
        • Verkerk A.
        • Eker A.P.M.
        • van Leenen D.
        • Buijs R.
        • Bootsma D.
        • Hoeijmakers J.H.J.
        • Yasui A.
        Nature. 1999; 398: 627-630
        • Kume K.
        • Zylka M.J.
        • Sriram S.
        • Shearman L.P.
        • Weaver D.R.
        • Jin X.
        • Maywood E.S.
        • Hastings M.H.
        • Reppert S.M.
        Cell. 1999; 98: 193-205
        • Miyamoto Y.
        • Sancar A.
        Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 6097-6102
        • Dunlap J.C.
        • Loros J.J.
        • Liu Y.
        • Crosthwaite S.K.
        Genes Cells. 1999; 4: 1-10
        • Witkovsky P.
        • Schütte M.
        Vis. Neurosci. 1991; 7: 113-124
        • Djamgoz M.B.A.
        • Wagner H.-J.
        Neurochem. Int. 1992; 20: 139-191
        • Crooks J.
        • Kolb H.
        J. Comp. Neurol. 1992; 315: 287-302
        • Kazula A.
        • Nowak J.Z.
        • Iuvone P.M.
        Vis. Neurosci. 1993; 10: 621-629
        • Boatright J.H.
        • Rubim N.M.
        • Iuvone P.M.
        Vis. Neurosci. 1994; 11: 1003-1012
        • Boatright J.H.
        • Rubim N.M.
        • Iuvone P.M.
        Vis. Neurosci. 1994; 11: 1013-1018
        • Belvin M.P.
        • Zhou H.
        • Yin J.C.P.
        Neuron. 1999; 22: 777-787
        • Obrietan K.
        • Impey S.
        • Smith D.
        • Athos J.
        • Storm D.R.
        J. Biol. Chem. 1999; 274: 17748-17756
        • Xing J.
        • Ginty D.D.
        • Greenberg M.L.
        Science. 1996; 273: 959-963