Acyl coenzyme A-binding protein (ACBP) is phosphorylated and secreted by retinal Müller astrocytes following protein kinase C activation


Address correspondence and reprint requests to Zuyuan Qian, Neurological Sciences Institute, Oregon Health Sciences University West Campus, 505 NW 185th Avenue, Beaverton, OR 97006, USA. E-mail:


Horizontal optokinetic stimulation of rabbit retina in vivo evokes increased expression of acyl coenzyme A-binding protein (ACBP), also known as ‘diazepam binding inhibitor,’ from retinal Müller cells. If the expressed ACBP were also secreted by Müller cells, then stimulus-evoked secretion of ACBP could influence the activity of GABAA receptor-expressing retinal neurons. In this study, we examine in vitro whether ACBP is secreted by Müller glial cells and Müller-like QNR/K2 cells following stimulation with elevated levels of KCl and phorbol myristic acetate (PMA). KCl and PMA stimulation evoked secretion of threonine-phosphorylated ACBP. A sequence analysis of ACBP shows that it has five potential phosphorylation sites: Two threonine sites fit a protein kinase C phosphorylation pattern. Two threonine sites fit a casein kinase II (CK2) pattern. One serine site fits a CK2 pattern. As CK2 is not expressed in QNR/K2 cells, it is probable that protein kinase C accounts for the phosphorylation of ACBP in these cells and for the PMA-evoked secretion of ACBP. Serine phosphorylation was constitutive. Horizontal optokinetic stimulation increased threonine-phosphorylated ACBP in rabbit retina. Phosphorylation of ACBP may influence its target affinity. We used a proteolytic fragment of ACBP, octadecaneuropeptide (ODN), to investigate how threonine phosphorylation influences its affinity for GABAA receptors. Threonine-phosphorylated ODN had a stronger affinity for GABAA receptors than did unphosphorylated ODN or unphosphorylated ACBP. We conclude that stimulus-induced Müller cell secretion of phosphorylated ACBP could influence the GABAergic transmission in neighboring retinal neurons.

Abbreviations used

acyl coenzyme A-binding protein


1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′tetraacetic acid


bovine serum albumin


casein kinase II


Dulbecco’s modified Eagle’s medium




fetal bovine serum


histidine-Acyl coenzyme A-binding protein


horizontal optokinetic stimulation


Krebs-ringer buffer




phosphate-buffered saline


protein kinase C


phorbol myristic acetate


phosphorylated octadecaneuropeptide


serum-free media


Tris-buffered saline 0.1% Tween 20

Acyl coenzyme A-binding protein (ACBP), an 87 amino acid protein, also known as ‘diazepam binding inhibitor,’ displaces diazepam from the GABAA receptor (Guidotti et al. 1983; Knudsen 1991; Knudsen et al. 1993). ACBP is present in many different tissues, with high levels present in the liver, brain, and retina (Gray et al. 1986; Mocchetti et al. 1986; Shoyab et al. 1986; Owens et al. 1989; Bovolin et al. 1990; Alho et al. 1991). Within the brain, the highest level of ACBP is found in the cerebellum (43 μM) (Costa and Guidotti 1991) where it is expressed by Bergmann glial astrocytes (Alho et al. 1985, 1988). In other brain regions, ACBP is also expressed by astrocytes (Tonon et al. 1990), tanycytes (Tonon et al. 1990; Malagon et al. 1993), ependymocytes lining the third ventricle (Tonon et al. 1990; Malagon et al. 1993; Do-Rego et al. 2001), and the subgranular layer of the dentate gyrus (Yanase et al. 2002).

In the retina, ACBP is expressed by radially oriented Müller glial cells and by stellate astrocytes (Holländer et al. 1991). The concentration of ACBP Müller glial cells is higher than the 10–50 μM range found in the brain as a whole. High glial concentrations of ACBP and the close apposition of astrocytic glial processes to neurons suggest that glial astrocytes could secrete biologically significant levels of ACBP.

When administered intraventricularly, ACBP antagonizes the anxiolytic actions of diazepam (Guidotti et al. 1983). In mice intraventricular injection of ACBP alters the preference to food (Manabe et al. 2001) and also decreases the pentobarbital-induced sleeping time (Dong et al. 1999). Intra-cisternal injections of ACBP increases low voltage fast EEG activity (Barmack et al. 2004). In solutions prepared from brain lysates, an antibody to the GABAAα1 receptor co-immunoprecipitated ACBP (Barmack et al. 2004). While diazepam enhances the chloride current at the GABAA receptor, ACBP may reduce it (Bormann 1991).

Secretion of ACBP by glial cells could provide an important mechanism by which GABAergic activity is regulated in the brain and in the retina. In the retina, GABAergic synaptic transmission is critical for proper functioning of direction-selective retinal circuitry. When GABAA receptors in retinal eye cups are blocked by the application of GABAA receptor specific antagonists, directional selectivity of ganglion cells is reduced (Kittila and Massey 1995, 1997).

Previously, we have shown that long-term horizontal binocular optokinetic stimulation (HOKS) in the posterior→anterior direction increases the expression of ACBP in Müller glial cells (Barmack et al. 2004). If ACBP, released from astrocytes, modulates retinal GABAergic transmission, then increased ganglion cell activity should also increase secretion of ACBP. Here, we investigate whether ACBP is secreted by rabbit Müller glial cells and by Müller-like QNR/K2 cells derived from quail neural retina in vitro (Pessac et al. 1983). We show that 3-day-old primary cultures of Müller glial cells as well as QNR/K2 cells secrete ACBP. We also show that HOKS in vivo increases the phosphorylation of ACBP in the retina. We further show that phosphorylation of ACBP increases its affinity for the GABAA receptor.

Materials and methods

Anesthesia and surgery

Pigmented rabbits were anesthetized with intramuscular injections of ketamine hydrochloride (50 mg/kg), xylazine (6 mg/kg), and acepromazine maleate (1.2 mg/kg). Two inverted bolts were fixed to the cranium (Barmack and Nelson 1987). The two screws mated with a device to restrain horizontal head movement during HOKS. Rabbits were housed and handled according to the guidelines of the National Institutes of Health on the use of experimental animals. All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Oregon Health & Science University.

Long-term binocular HOKS

Rabbits were placed in a restrainer at the center of a cylindrical optokinetic drum with a contour-rich pattern on the interior wall (diameter, 110 cm and height, 115 cm). The head of the rabbit was fixed to the restrainer by a spring-loaded coupling that mated with implanted head screws. The coupling maintained a head pitch angle of 12° with respect to earth horizontal. This angle corresponds to that maintained by unrestrained rabbits (Hughes 1971; Barmack and Nelson 1987; Soodak and Simpson 1988). The head coupling permitted small movements in the sagittal plane but prevented head movements in the horizontal plane. This method of restraint caused no pressure on any part of the body. The rabbit maintained its normal posture. All four paws remained in contact with the support surface.

The optokinetic drum rotated at a constant angular velocity (5°/s), stimulating the left eye in the P→A direction and the right eye in the anterior→posterior direction (A→P). HOKS lasted 48 h. Every 8 h, the rabbit was removed from the drum and allowed access to food and water. After 48 h of HOKS, the rabbit was anesthetized. The left and right eyes were removed and the retinas were prepared for analysis as described below.

Tissue culture for QNR/K2 cells

QNR/K2 (CRL-2533) cells were obtained from American Type Culture Collection (Manassas, VA, USA). These cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS). Cells were maintained in 10% CO2 at 37°C. The cells were removed by trypsinization and split 1 : 3 every 3 days. Cells were grown for three passages after being brought out of freeze-down in liquid N2 and then used through passage 12.

Müller glial cell preparation and cell culture

Adult rabbit eyes were removed from rabbits deeply anesthetized by a killing dose of ketamine hydrochloride (100 mg/kg), xylazine (12 mg/kg), and acepromazine maleate (2.4 mg/kg). The eyes were placed on ice in CO2-independent medium (Hanks’ balanced salt solution without calcium, without magnesium; Hyclone, Logan, UT, USA) for 2–5 min before the cornea was removed and the vitreous was drained. The neural retina was separated from the pigment epithelium and washed twice in Ringer’s solution without Ca2+ supplemented with 2.5 mM EGTA. Cell dissociation was obtained by treating each retina with 2.2 U of activated papain (Worthington Biochemical, Lakewood, NJ, USA) for 40 min at 37°C. Papain enzyme activity was stopped by the addition of DMEM with 10% FBS. Then, 0.005% Dnase (Worthington Biochemical) was added. The tissue was further dissociated by gentle trituration using a 10 mL pipette (Hauck et al. 2003). Dissociated cells were collected by centrifugation (400 g, 5 min), re-suspended in DMEM containing 10% FBS and plated directly onto 60 mm cell culture plates (Corning, Acton, MA, USA) to achieve an 80% density the following day. Plated cells were allowed to attach for 16 h at 37°C in an incubator. Non-attached cells were removed by gentle agitation (panning). Cultures were assessed immunohistochemically and used in secretion experiments within 3 days after harvest, as immunohistochemical changes in Müller glial cells are not observed within this period (Hauck et al. 2003).

Antibodies and immunochemistry

Histidine-tagged ACBP (his-ACBP, AF407578) was generated by a bacterial expression system (TOPO TA; Invitrogen). The correct expression sequence of his-ACBP was confirmed by N-terminal protein sequence analysis. His-ACBP was used to immunize goats (Alpha Diagnostics International, San Antonio, TX, USA). The crude antibody to his-ACBP was affinity-purified using a calmodulin-binding peptide (pCal-ACBP), with the tag being placed on the C-terminus, using a pCal vector system (Stratagene, La Jolla, CA, USA). Ten milligrams of pCal-ACBP was bound to a 1 mL HiTrap™ NHS-activated column (Amersham Pharmacia AB, Uppsala, Sweden). Approximately 50 mL of crude antibody was run through the column. The column was washed with 10 mL of phosphate-buffered saline (PBS). The bound antibody was eluted with 5 mL of 100 mM glycine, pH 2.6, and subsequently dialyzed against PBS with three changes of buffer at 4 h each. To test specificity, the antibody was pre-absorbed with a 10-fold molar excess of purified pCal-ACBP for 2 h at 37°C. This pre-absorbed antibody did not bind to his-ACBP in western blots (data not shown).

Other immunological reagents included affinity-purified polyclonal antibody to rabbit phospho-threonine (Zymed, San Francisco, CA, USA), mouse monoclonal antibody to phospho-serine (Millipore, Billerica, MA, USA), and mouse anti-rabbit and mouse anti-goat horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotech, Santa Cruz, CA, USA).

Monoclonal antibodies to vimentin (Chemicon, Temecula, CA, USA) and glutamine synthetase (BD Transduction laboratories, Palo Alto, CA, USA), as well as polyclonal antibodies to cd11b (Advanced targeting systems, San Diego, CA, USA) and his-ACBP antibody were used to immunolabel primary cultures of rabbit Müller glial cells and QNR/K2 cells on slides overnight at 4°C after fixation in 4%p-formaldehyde. Fluorescent secondary antibodies were labeled with Alexa Fluor 350 and 568 (Molecular probes, Eugene, OR, USA). All primary and secondary antibody incubations used for labeling cultures were prepared at concentrations of 1 : 100.

Immunoblotting ACBP from rabbit retina lysates

Rabbit retina lysates were prepared from dissected pieces of the central retina. These pieces were sonicated 3× for 15 s in lysis buffer: 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM EGTA, 0.25% Igepal, 0.4% deoxycholate, a protease inhibitor mixture tablet containing 3.0 mg antipain, 0.5 mg bestatin, 1.0 mg chymostatin, 0.5 mg leupeptin, 0.5 mg pepstatin, 3.0 mg phosphoramidon, 20 mg Pefabloc SC, 10 mg EDTA, and 0.5 mg aprotinin (Roche, Branchburg, NJ, USA) for each 15 mL of buffer. Retinal lysates were rocked for 1 h at 4°C and then centrifuged at 12 000 g for 10 min at 4°C to remove insoluble debris. The supernatant was resolved on a 4–12% gradient Bis–Tris gel, and analyzed for ACBP content by western blot. The amount of ACBP present in each retina was determined by densiotometric measurement of the ACBP band on western blots using enhanced chemiluminescence (Amersham, Piscataway, NJ, USA). Samples of lysates isolated from optokinetically stimulated retinas were adjusted to contain near equivalent amounts of ACBP. These samples were run on western blots. The blots were probed first with the affinity-purified his-ACBP antibody to verify that equivalent amounts of ACBP were present. The blots were stripped of his-ACBP antibody, as described above. The stripped blots were washed with Tris-buffered saline 0.1% Tween 20 (TBST) 2× for 5 min, blocked with 1.5% bovine serum albumin (BSA) and then probed with the antibody to phospho-threonine to determine the degree of phosphorylation.

Gel electrophoresis and immunoblotting

Proteins were separated on 4–12% gradient bis–Tris gels (Invitrogen). After electrophoresis, proteins were transferred onto Immobilon-P-membranes, blocked with TBST containing 1.5% BSA, incubated with primary antibody in TBST for 2 h at 25°C, washed 3× with TBST, and probed with horseradish peroxidase-conjugated mouse anti-goat or mouse anti-rabbit secondary antibodies. In some experiments, blots were stripped of antibodies by incubating 2× for 15 min with 0.1 M glycine, pH 2.5. The blots were washed with TBST 2× for 5 min, blocked with 1.5% BSA, and then probed with a different primary antibody.

Secretion of his-ACBP in response to high KCl or phorbol myristic acetate

QNR/K2 cells were transfected with the PCDNA3.1/V5-his-TOPO TA expression vector containing the ACBP gene with a C-terminal V5 epitope followed by a C-terminal his-tag (six consecutive histidines) (Invitrogen). The addition of the V5 epitope and his-tag resulted in a construct of ∼13 kDa. Transfection was performed using the Transfast transfection reagent (Promega, Madison, WI, USA) according to the manufacturer’s protocol. Briefly, QNR/K2 cells in 10 cm dishes were transfected at 70–80% confluence after being split the previous day; 15 μg of plasmid and 45 μL of Transfast reagent, at a lipid to DNA ratio of 1 : 1, were added to these cells. After 1 h, the transfection reagents were removed and the cells were maintained in DMEM 10% FBS. These cells were used for secretion experiments 36 h after transfection.

Transfected cells were maintained overnight in DMEM 10% FBS, washed 2× with modified Krebs-ringer buffer (KRB) (120 mM NaCl, 1.0 mM KCl, 15 mM HEPES, pH 7.4, 2.0 mM CaCl2, 24 mM NaHCO3, 1.0 mM MgCl2, and 1 mg/mL BSA) and then maintained in the same buffer for 1 h. The buffer was removed and replaced by KRB in which the KCl concentration was raised to 50 mM. NaCl was simultaneously lowered by 50 mM. QNR/K2 cells were incubated in this buffer for the indicated times. After treatment, the buffer was collected and spun at 400 g for 5 min. to remove any cells floating in the media. His-ACBP secreted into the buffer was purified using his-binding beads, nickel-impregnated agarose beads (Ni-NTA; Qiagen, Hilden, Germany), incubated, with rocking, in the buffer for 1 h at 25°C. The beads were poured into a column and washed with three column volumes of low imidazole-containing buffer (20 mM imidazole in 50 mM Na2HPO4, pH 8.0, and 300 mM NaCl). His-ACBP was eluted using 250 mM imidazole. Samples were analyzed by western blot.

QNR/K2 cells, transfected as above, were placed in DMEM without serum for 18 h and then treated with 0.002% dimethylsulfoxide (DMSO) vehicle or 100 nM phorbol myristic acetate (PMA) in 0.002% DMSO for the indicated times. His-ACBP was isolated as above. His-ACBP was measured by western blot.

Immunoprecipitation of ACBP from QNR/K2 and Müller glial cells, media, and cell lysates after PMA stimulation

Equal numbers of QNR/K2cells, not containing the his-ACBP construct, were plated in order to achieve 80% confluence. The next day, these cells were washed 3× with serum-free DMEM (SFM), and then left in SFM 18 h. Müller glial cells were treated as above with the exception that they were placed in SFM 2 days after plating. Cells were treated with PMA for 1–40 min. Extracellular media were removed and centrifuged at 400 g for 5 min at 4°C to remove any remaining cells. A complete mini-protease inhibitor pill and 1% Igepal were added to the supernatant. The supernatant was incubated with 20 μL of protein A/G agarose for 1 h to remove proteins that bind non-specifically to the beads. The beads were removed by centrifugation at 500 g for 1 min. ACBP was immunoprecipitated by adding 6 μg of the affinity-purified his-ACBP antibody to the sample. Samples were incubated at 4°C and rocked overnight. The next morning 50 μL of protein A/G agarose were added and the incubation was continued for 2 h at 4°C. The sample was washed 3× with PBS, re-suspended in 20 μL of sodium dodecyl sulfate–polyacrylamide gel electrophoresis loading buffer at a 2x concentration, and boiled for 10 min. Samples were analyzed by western blot and probed with the affinity-purified his-ACBP antibody.

Cells from the above treatments were placed in lysis buffer A and sonicated 3×. Insoluble debris was removed by centrifugation at 20 000 g for 10 min at 4°C. Equal amounts of protein from the cell lysates were cleared of proteins that non-specifically bind to protein A/G agarose as above. ACBP was immunoprecipitated and analyzed by western blot as above.

Identification of proteins that interact with PKC-δ

We identified proteins from QNR/K2 cells, retinal, and brain lysates that specifically interacted with his-protein kinase C (PKC)-δ. Lysates were incubated with histidine-binding beads at 4°C overnight to eliminate non-specific histidine-binding. Next, we mixed his-PKC-δ (expressed in QNR/K2 and immobilized on histidine-binding beads) with the lysate supernatant. The lysate and beads were co-incubated at 4°C for 1 h. Residual non-specific binding proteins were removed by five washes of low imidazole-containing buffer. Remaining proteins were eluted with high imidazole-containing buffer. The eluted proteins were separated with a 4–12% gradient Bis–Tris gel. They were transferred on to polyvinylidene difluoride membrane and detected by immunoblotting with the antibodies to his-ACBP and PKC-δ isoforms.

PMA stimulation evokes Ca2+ independent secretion of his-ACBP

QNR/K2 cells were incubated for 1 h in modified KRB ± 1 mM CaCl2, ± 75 μM 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′tetraacetic acid (BAPTA). The QNR/K2 cells were stimulated for the indicated times in modified KRB containing 100 nM PMA. Control QNR/K2 cells were treated with BAPTA alone or with DMSO alone. The media were collected, his-ACBP was immunoprecipitated and visualized by western blot as above.

Immunoprecipitation of ACBP from QNR/K2 and Müller glial cells after KCl stimulation

Equal numbers of cells were seeded onto 60 mm (Müller cells) or 100 mm (QNR/K2 cells) tissue culture dishes to achieve ∼80% confluence. Cells were maintained overnight in DMEM 10% FBS, washed 2× with modified KRB (1 mM KCl, 120 mM NaCl, 15 mM HEPES, pH 7.4, 24 mM NaHCO3, 1 mM MgCl2, 2 mM CaCl2, and 1 mg/mL BSA) and then fresh buffer for 1 h. The buffer was removed and replaced by KRB in which the KCl concentration was raised 50 mM and the NaCl concentration was lowered simultaneously 50 mM. The cells were incubated in this buffer for 1–30 min. After treatment, the buffer was collected and spun at 400 g for 5 min to remove any cells floating in the media. The supernatant was incubated with 20 μL protein A/G agarose beads for 1 h to remove proteins that bind non-specifically to the beads. The beads were removed by centrifugation at 500 g for 1 min. His-ACBP was immunoprecipitated by adding 6 μg of affinity-purified his-ACBP antibody to the sample. Samples were incubated and rocked at 4°C overnight. The next morning 50 μL of protein A/G agarose beads were added and the incubation was continued for 2 h. The sample was washed 3× with PBS, re-suspended in 20 μL of sodium dodecyl sulfate–polyacrylamide gel electrophoresis loading buffer at a 2x concentration, and boiled for 10 min. The samples were analyzed by western blot and probed with the affinity-purified his-ACBP antibody.

Cell viability assay

Untransfected QNR/K2 cells and Müller glial cells were plated at 80% confluence and maintained in SFM overnight. The cells were treated for 30–60 min with 100 nM PMA in DMSO, 50 mM KCl + DMSO, or DMSO alone. After treatment, CellTitre-Blue reagent (Promega) was added at 1/5 reagent/media ratio. Plates containing the cells were returned to 37°C for 1 h. The media were removed and placed on ice until the absorbance was measured at 570 and 600 nm. The average absorbance at 600 nm was subtracted from the 570 nm absorbance for each treatment. All absorbance measurements were below 0.5 at 600 nm or 570 nm. The absorbance of the reagents at 570 nm increases with the number of viable cells present that that can convert resazurin into a resorufin. Ratios of experimental/control absorbance were calculated.

ODN competes with ACBP for binding with GABAA receptors

His-ACBP (5 μg) was isolated from a bacterial expression system by attachment to histidine-binding beads. His-ACBP was incubated overnight at 20–23°C with 1 mL of a supernatant from a rabbit cerebellar lysate (1 mg/mL protein) that had been solubilized in lysis buffer A (containing a protease inhibitor tablet as above). The cerebellar lysate was pelleted at 20 000 g for 30 min, prior to use of the supernatant as a source of GABAA receptors. A peptide competing for binding to GABAA receptor was added to the mixture. Competing peptides (AnaSpec, San Jose, CA, USA) included phosphorylated and unphosphorylated octadecaneuropeptide (ODN) at the indicated concentrations. ODN was phosphorylated (p-ODN) at threonine 3 and 9 sites. BSA was used as a control protein in the competition assay to determine if the blocking action of the ODN peptides was specific.

Samples were washed with lysis buffer A, and pelleted at 400 g 3×, and subsequently eluted with 250 mM imidazole, pH 8.0, resolved on a 4–12% gradient Bis–Tris gel, and analyzed by western blot. The competition assay measured the residual amount of GABAAα1 that co-purified with his-ACBP bound to his-binding beads after co-incubation by densitometer analysis of western blots. The blot was cut at ∼25 kDa. The upper half was probed with an antibody to the GABAAα1 receptor [GABAA Rα1 (N-19): sc-7348; Santa Cruz Biotech] diluted 1 : 500. The lower half was probed with the affinity-purified his-ACBP antibody diluted 1 : 10 000. The blots were analyzed as described above.

Statistical analysis

Measurements were made from digital images of western blots using densitometry and analyzed using the Student’s t-test with two tails and assuming unequal variance.


Optokinetic stimulation evokes threonine-phosphorylation of retinal ACBP

Long-term HOKS of the rabbit in vivo evokes increased expression and phosphorylation of ACBP in retinal lysates taken from the eye stimulated in P→A direction relative to lysates taken from the eye stimulated in the A→P direction (Fig. 1). Western blots prepared from these retinal lysates were immunolabeled with affinity-purified his-ACBP antibody (Fig. 1b).

Figure 1.

 ACBP expression and ACBP threonine-phosphorylation were increased during horizontal optokinetic stimulation (HOKS). (a) During HOKS at 5°/s one eye is stimulated in the posterior→anterior direction (P→A). The other was stimulated in the anterior–posterior direction (A→P). (b) HOKS increased expression of ACBP in the eye receiving P→A stimulation. Western blots show an increased expression of ACBP, identified by the affinity-purified his-ACBP antibody, in both cytosol and membrane fractions. The lower bands indicate the use of an antibody to actin as an equal loading control. (c) Three rabbits received 48 h of HOKS, after which the fractions from each eye were equilibrated for ACBP (upper bands). This blot was stripped and re-probed with an antibody to phospho-threonine. HOKS increased threonine phosphorylation in the retinae stimulated in the P→A direction.

Based on the densitometric ratio of total ACBP in blots from eyes stimulated in opposite directions, the concentration of ACBP in samples of lysates was equilibrated and a second western blot was run. The second western blot was immunolabeled with affinity-purified his-ACBP antibody. The blots were subsequently stripped of antibodies and then immunolabeled with an antibody to phospho-threonine. Lysates from eyes stimulated in the P→A direction were more densely threonine-phosphorylated than were lysates from eyes stimulated in A→P direction (Fig. 1c).

In vitro models of cell-specific secretion of ACBP

Currently, it is not possible to test directly whether ACBP is secreted by Müller glial cells in vivo. Consequently, we used primary cultures of rabbit retinal Müller glial cells and immortalized Müller-like QNR/K2 cells to investigate cell-specific secretion of ACBP in vitro. QNR/K2 cells were derived from 7-day quail embryos transformed with Rous sarcoma virus (Pessac et al. 1983). Plated onto a slide, QNR/K2 cells retained a Müller-like morphology. They had a bipolar appearance with characteristic brush endings at either end. Both Müller glial and QNR/K2 cells were immunolabeled by antisera for vimentin, glutamine synthetase, and his-ACBP (Fig. 2). Vimentin is a non-specific glial cell marker. Glutamine synthetase is a protein critical for the recycling of GABA and glutamate. In the retina, glutamine synthetase is expressed exclusively by Müller glial cells (Germer et al. 1997). All cells in the Müller glial cell and QNR/K2 cell preparations stained positively for the specific Müller glial cell marker glutamine synthetase and had the morphology characteristic of Müller glial cells. An antibody to cd11b, a marker for microglia (Matsubara et al. 1999), failed to label either QNR/K2 or Müller glial cells, providing a negative control. These findings suggest that QNR/K2 cells share several characteristics of Müller cells and provide a model system with which to test secretory activity. Use of the QNR/K2 cell line provided adequate amounts of cellular and secreted his-ACBP more rapidly than could be performed using only primary Müller glial cells.

Figure 2.

 Müller glial and QNR/K2 cells express ACBP, vimentin, and glutamine synthetase. Primary cultures of rabbit Müller glial cells, 3 days after harvest, and QNR/K2 cells were immunolabeled with an affinity-purified antibody to his-ACBP (a and b), an antibody to vimentin (c and d) and an antibody to glutamine synthetase (e and f). All three proteins were expressed in both cell types.

QNR/K2 and Müller glial cells secrete ACBP after incubation in elevated KCl

QNR/K2 cells, transfected with his-ACBP, increased secretion of his-ACBP after a brief exposure to elevated extracellular KCl (Fig. 3a). The concentration of KCl was increased to 50 mM for 0–30 min, while the concentration of NaCl was reduced simultaneously by 50 mM to avoid osmotic changes. Samples of ∼5 × 106 QNR/K2 cells were maintained in 1 mM KCl in Krebs-modified ringer buffer for 1 h. His-ACBP was isolated from cell culture medium with his-binding beads. Western blots of the isolated proteins revealed increased secretion of his-ACBP from QNR/K2 cells treated with elevated concentrations of KCl (Fig. 3a). Replacing the modified KRB with fresh buffer of the same composition did not evoke increased secretion of his-ACBP. QNR/K2 cells incubated for an additional 30 min in 1 mM KCl in modified KRB without a buffer change did not increase secretion of his-ACBP. These initial secretion experiments used his-tags for isolation of ACBP because the affinity-purified his-ACBP antibody had not yet been prepared.

Figure 3.

 KCl and PMA treatments induce secretion of ACBP by Müller and QNR/K2 cells. Müller and QNR/K2 cells were maintained in Kreb’s ringer buffer overnight. After this conditioning phase, both cell types were further incubated with modified Kreb’s ringer buffer as described in Materials and methods. (a) KCl-induced his-ACBP secretion from QNR/K2 cells. His-ACBP was isolated with his-tagged beads. Lane 1: QNR/K2 cells after overnight incubation in buffer; Lane 2: The buffer was replaced with a fresh test buffer for 30 min; Lane 3: Cells were treated for 30 min with the test buffer, containing 50 mM KCl; and Lane 4: Bacterially expressed his-ACBP was run as a standard. (b) KCl-induced ACBP secretion from Müller glial cells. Secreted ACBP was immunoprecipitated from the extracellular media. Lanes 1–3: Müller cells were stimulated with 50 mM KCl for the indicated times; Lane 4: ACBP secretion was measured after further 30 min incubation in fresh buffer; and Lane 5: Bacterially expressed his-ACBP standard. (c) QNR/K2 cells were incubated with 100 nM phorbol myristic acetate (PMA) in DMSO for 0–30 min. PMA treatment evoked his-ACBP secretion. His-ACBP secretion was not evoked when either cell type was treated alone with DMSO. (d) Müller cells were incubated with 100 nM PMA in DMSO for 0–30 min. PMA treatment evoked ACBP secretion. (e) Treatment of QNR/K2 cells (black bars) and Müller glial cells (gray bars) with 0.002% DMSO, 100 nM PMA, or 50 mM KCl did not reduce cell viability as measured by CellTitre-Blue reagent.

Müller glial cells also secreted ACBP after incubation in elevated KCl. Samples of ∼2 × 106 Müller glial cells, maintained in 1 mM KCl in Krebs modified ringer buffer for 1 h were incubated in elevated levels of KCl for 0–30 min (Fig. 3b). Each sample was immunoprecipitated with the affinity-purified his-ACBP antibody, analyzed by western blot, and probed with the affinity-purified his-ACBP antibody.

PMA treatment increases PKC activity and induces secretion of ACBP from QNR/K2 and Müller glial cells

Phorbol myristic acetate, a PKC activator, increases protein secretion in both calcium-dependent and -independent modes (Billiard et al. 1997). We examined the possibility that ACBP secretion depends on PKC activation by stimulating transfected QNR/K2 cells and Müller glial cells with PMA for variable durations. PMA stimulation evoked secretion of ACBP by both QNR/K2 cells (Fig. 3c) and Müller glial cells (Fig. 3d). His-ACBP, secreted from QNR/K2 cells, was concentrated using his-binding beads. ACBP secreted from Müller cells was concentrated by immunoprecipitation using the affinity-purified his-ACBP antibody. DMSO, the solvent for PMA, had no effect by itself.

Treatment of QNR/K2 cells or Müller glial cells with KCl or PMA does not cause cell death

If incubation of QNR/K2 or Müller glial cells with KCl or PMA caused cell death, then release of ACBP into the media could be misinterpreted as ACBP secretion. We examined this possibility using a cell viability assay. We measured the survivability of treated QNR/K2 and Müller glial cell cultures and compared it to the survivability of equal volumes of untreated control cultures (Fig. 3e). There was no significant difference between treatment and control (p > 0.5). These data indicate that increased concentration of his-ACBP in the media of treated QNR/K2 cultures was due to cell secretion, not cell death.

PMA phosphorylates his-ACBP in QNR/K2 cells and induces its secretion

Phorbol myristic acetate treatment of QNR/K2 cells increased his-ACBP phosphorylation both inside the cells and secreted into the extracellular media. But was PMA-evoked phosphorylation necessary for his-ACBP secretion? We incubated QNR/K2 cells with 100 nM PMA for 0–80 min and then separately analyzed the phosphorylated ACBP recovered from the cells and extracellular media. We used phospho-threonine- and phospho-serine-specific antibodies to characterize PMA-evoked phosphorylation.

The concentration of threonine-phosphorylated cytoplasmic his-ACBP increased during the first 40 min of PMA treatment and decreased thereafter (Fig. 4a). The concentration of threonine-phosphorylated his-ACBP recovered from the extracellular media peaked 20 min after the PMA treatment was initiated (Fig. 4b). PMA treatment caused no change in the time course of serine phosphorylation of his-ACBP (Fig. 4c). These data suggest that his-ACBP secretion is influenced by threonine phosphorylation.

Figure 4.

 PMA treatment phosphorylates intracellular and secreted his-ACBP in QNR/K2 cells. QNR/K2 cells were incubated with 100 nM PMA in DMSO for 0–80 min as described in Materials and methods. Phosphorylated his-ACBP from both cell lysate and extracellular media was analyzed by western blot and probed with antibodies to phosphorylated threonine and serine. (a) PMA treatment of QNR/K2 cells threonine-phosphorylates intracellular his-ACBP. (b) PMA treatment also threonine-phosphorylates his-ACBP secreted by QNR/K2 cells. (c) His-ACBP is serine-phosphorylated constitutively in QNR/K2 cells. This constitutive phosphorylation is not enhanced by PMA treatment.

Identification of phosphorylation sites on ACBP

We scanned the amino acid sequence of ACBP, using the ScanProsite tool from Expasy to identify patterns, profiles, and motifs stored in the PROSITE database (Swiss Institute of Bioinformatics, Geneva, Switzerland). ACBP has five potential phosphorylation sites: Two threonine sites fit a PKC phosphorylation pattern (42–44 TeR and 65–67 TsK). Two fit a casein kinase II (CK2) pattern (36–39 TvgD and 65–68 TskE). One serine site fits a CK2 pattern (2–5 SqaE) (Fig. 5a). No other sites were found.

Figure 5.

 PKC and casein kinase II phosphorylation sites on ACBP. Potential phosphorylation sites are defined by known amino acid sequences. (a) Phosphorylation sites on ACBP occur at the numerically designated positions. PKC sites occur at threonine 42 and 65. Casein kinase II sites occur at serine 2, threonine 36, and threonine 65. (b and c) We screened QNR/K2 cells, brain cells, and retinal cells with antibodies to PKC-δ and casein kinase II. PKC-δ was expressed in all three tissues. Casein kinase II was expressed only in the rabbit brain and retina. (d) His-PKC-δ, expressed in transfected QNR/K2 cells, ‘pulled down’ proteins from rabbit retinal lysates and QNR/K2 cells. Western blots identified proteins that co-precipitated with PKC-δ. Lane 1: his-PKC-δ was separated from QNR/K2 cells using his-tag-binding nickel beads; Lane 2: his-binding beads were incubated with QNR/K2 cells untransfected with his-PKC-δ as a control; Lane 3: his-PKC-δ was incubated with rabbit retina lysate; Lane 4: his-binding beads were incubated with rabbit retina lysate without the inclusion of his-PKC-δ; Lane 5: Bacterially expressed his-PKC-δ was run alone as a control; and Lane 6: Unphosphorylated, bacterially expressed his-ACBP was also run as a control. The first row of western blots was probed with an antibody to PKC-δ. The second row was probed with the affinity-purified his-ACBP antibody. PKC-δ co-eluted with his-ACBP expressed in QNR/K2 cells. ACBP expressed in rabbit retinal lysates (lower molecular weight form).

We used antibodies to PKC-δ and CK2 to screen QNR/K2 cells, rabbit brain, and retina for kinases associated with PMA-induced secretion. PKC-δ was expressed in all three tissues (Fig. 5c). CK2 was expressed in rabbit brain and retina, but not in QNR/K2 cells (Fig. 5b). Consequently, CK2 could not be responsible for ACBP phosphorylation in QNR/K2 cells. Nor could it be responsible for secretion of ACBP from these cells.

We used his-PKC-δ, expressed in transfected QNR/K2 cells, to ‘pull down’ proteins that co-precipitated with PKC-δ. ACBP was identified as one of the proteins. It co-eluted with PKC-δ in rabbit retina and QNR/K2 cells (Fig. 5d). These data indicate that PKC-δ interacts with ACBP and may be responsible for its phosphorylation.

PMA induces secretion of his-ACBP independently of calcium concentration in QNR/K2 cells

Secretion of proteins can be evoked by PMA stimulation in the presence or absence of calcium. We examined whether calcium was required for the secretion of ACBP by placing transfected QNR/K2 cells in calcium-free media with BAPTA for 30 min. Subsequently, the cells were stimulated with 100 nM PMA. His-ACBP secreted into the medium was isolated by immunoprecipitation with the affinity-purified his-ACBP antibody. The removal of calcium had no effect on his-ACBP secretion. The secretion of his-ACBP by cells treated with BAPTA alone was not changed relative to the secretion of his-ACBP from untreated cells (Fig. 6). These observations suggest that calcium independent forms of PKC stimulate ACBP secretion.

Figure 6.

 PMA stimulation of QNR/K2 cells evokes Ca2+ independent secretion of his-ACBP. QNR/K2 cells were incubated in modified Kreb’s ringer buffer (KRB) for 30 min under the described conditions. Mean his-ACBP secretion was analyzed by western blot and measured by densitometry. Standard errors are indicated. The histograms show that PMA evoked increased secretion of his-ACBP. Increased secretion occurred independently of calcium concentration. Treatment with DMSO and BAPTA alone did not influence his-ACBP secretion.

Untransfected QNR/K2 cells secrete ACBP

It could be argued that secretion of his-ACBP by transfected QNR/K2 cells is artificial because its higher concentration in transfected cells. Consequently, we examined PMA-evoked secretion of ACBP by untransfected QNR/K2 cells. Cytoplasmic and secreted ACBP were immunoprecipitated and probed with the affinity-purified his-ACBP antibody (Fig. 7a and b).

Figure 7.

 PMA stimulation induces secretion of ACBP from untransfected QNR/K2 cells. After incubation in SFM for 18 h, QNR/K2 cells were treated for the indicated times with 100 nM PMA. ACBP was immunoprecipitated from treated cells (a), and from media (b). The isolated ACBP was analyzed by western blots probed with the affinity-purified his-ACBP antibody. (a and b) The western blots are representative of five experiments. (c) Mean ACBP released following PMA stimulation was measured by densitometry. Standard errors are indicated. PMA induced secretion of ACBP 20–30 min after it was added to the media (p < 0.01, indicated by asterisks).

The highest cytoplasmic concentration of ACBP was observed prior to PMA incubation. The highest concentration of secreted ACBP occurred at 20 min after the onset of PMA incubation. Secreted ACBP was increased by a factor of 2.8 over the zero time control (Fig. 7b and c). After PMA incubation for 40 min, ACBP secretion returned to untreated control levels, suggesting that internal stores of ACBP may have been depleted by stimulation and that secreted ACBP may have been degraded by proteases present in the media prior to the addition of protease inhibitors. Bacterially expressed his-ACBP, used as a positive control, had a higher molecular weight than ACBP expressed by QNR/K2 cells. The molecular weight of the bacterially expressed his-ACBP was increased by the attachment of a V5 epitope combined with a 6xhis tag on the C-terminus.

Threonine-p-ODN interacts with GABAA receptors

Threonine phosphorylation of ACBP in rabbit retinas optokinetically stimulated in the P→A direction may change the affinity of ACBP for GABAA receptors. We tested this possibility using a peptide derivative of ACBP, ODN. ODN consists of amino acids 34–51 of ACBP. It is generated by tryptic digestion of ACBP. ODN is found in rat cerebellum and cortex (Mocchetti and Costa 1987) and has a 3.5-fold higher affinity for the GABAA receptor than does ACBP (Ferrero et al. 1986). We synthesized both unphosphorylated and phosphorylated forms of ODN. The phosphorylated form was threonine-phosphorylated at positions 3 and 9. These positions corresponded to threonine residues 36 and 42 on intact ACBP.

We used a competition assay to evaluate whether the phosphorylated state of ODN changed its affinity for GABAAα1. His-ACBP was incubated with a lysate of the rabbit cerebellum as a source of GABAA receptors. We used the cerebellum as the source of GABAA receptors because of the greater abundance of cerebellar tissue. Unphosphorylated or phosphorylated ODN at two concentrations (10 and 20 μM) was added to the lysate to compete with his-ACBP for binding to GABAAα1. His-ACBP and bound GABAAα1 were collected and run on western blots. GABAAα1 and his-ACBP were identified with affinity-purified his-ACBP antibody and GABAAα1 antibody (Fig. 8a and b). Treatment of cerebellar lysates with 10 and 20 μM unphosphorylated ODN caused 22% and 31% reductions, respectively, in GABAAα1 bound to his-ACBP. Treatment with 10 and 20 μM p-ODN caused 57% and 82% reductions of GABAAα1 bound to his-ACBP. The addition of 1.5 mM BSA had no effect on the binding of GABAAα1 to his-ACBP.

Figure 8.

 Phosphorylated ODN displaces his-ACBP from GABAAα1 receptors with high affinity. A competition assay was used to test the affinity of phosphorylated and unphosphorylated ODN for GABAAα1 receptors. His-ACBP (5 μg), attached to histidine binding beads, was incubated overnight with a lysate from rabbit cerebellum in the presence of peptides that might compete with ACBP for binding with GABAAα1 receptors. (a) Samples were washed 3× with lysis buffer A, eluted with 250 mM imidazole, pH 8.0, and analyzed by western blot. The blot was cut at ∼25 kDa. The upper half was probed with an antibody to GABAAα1. The lower half was probed with the affinity-purified his-ACBP antibody. Co-incubated proteins included: Lane 1: None; Lane 2: ODN, 10 μM; Lane 3: ODN, 20 μM; Lane 4: ODN, phosphorylated at threonine 3 and threonine 9 (p-ODN) 10 μM; Lane 5: p-ODN, 20 μM; Lane 6: BSA, 1.50 mM; Lane 7: Incubation without cerebellar lysates; and Lane 8: GABAAα1 positive control (rat brain microsomes). A representative western blot illustrates six separate experiments. (b) Densitometric measurements of western blots showed that ODN and p-ODN displaced his-ACBP from GABAAα1 receptors following co-incubation. Phosphorylated-ODN displaced more his-ACBP than did unphosphorylated ODN. Significant decreases in GABAAα1 receptors bound to his-ACBP following treatment with ODN and p-ODN are indicated at p < 0.025 (*) and p < 0.001 (**).


Secretion of ACBP by QNR/K2 and Müller glial cells

Acyl coenzyme A-binding protein is transcribed, expressed, and secreted by retinal Müller glial cells and QNR/K2 cells following incubation in PMA, a PKC activator. The secretion of ACBP by Müller glial cells enlarges the possible roles of astrocytes in the regulation of neuronal activity. Müller glial cells are not the only astrocytes that express ACBP. Cerebellar Bergmann glial astrocytes (Barmack et al. 2004). Leydig cells and Sertoli cells also express ACBP (Garnier et al. 1993). Stimulation of cultured astrocytes by either β-amyloid peptides or pituitary adenylate cyclase activating polypeptide evokes expression and secretion of ACBP (Tokay et al. 2005). Exposure to the protein kinase A inhibiter, H89, or somatostatin has the opposite effect (Masmoudi et al. 2003, 2005).

Analysis of secreted ACBP

Acyl coenzyme A-binding protein has five potential phosphorylation sites: Two threonine sites fit a PKC phosphorylation pattern (42–44 TeR and 65–67 TsK). Two fit a CK2 pattern (36–39 TvgD and 65–68 TskE). One serine site fits a CK2 pattern (2–5 SqaE).

The crystal structure of ACBP isolated from bovine liver provides insight into the location of phosphorylation sites (Andersen and Poulsen 1992). Only 6/87 amino acids of rabbit retinal ACBP differ from those of the bovine liver diazepam-binding inhibitor. In 5/6 of these, the discrepant amino acids share homologies based upon charge. Three threonine residues and a serine residue are located close to the protein exterior, making these sites available for phosphorylation.

We have not identified which of the threonine residues is phosphorylated by PMA stimulation. In mammals, CK2 sites are constitutively phosphorylated (Meggio and Pinna 2003). In the present experiment, increased secretion occurred only in response to PKC activation. In QNR/K2 cells phosphorylation by CK2 is not possible as it is not expressed. We observed threonine phosphorylation of ACBP in vitro following PMA stimulation and in vivo following optokinetic stimulation. Threonine phosphorylation may be important for secretion of the active form of ACBP.

In ODN, threonine residues 3–6 and 9–12 correspond to residues 36–39 (TvgD) and 42–44 (TeR) on ACBP. Phosphorylation of these residues enhanced the affinity of ODN for GABAA receptors isolated from the cerebellum. This suggests that threonine phosphorylation of ACBP may be required for optimal binding of ACBP to GABAA receptors.

While our findings demonstrate that secreted ACBP is phosphorylated, we have not identified a specific site(s) necessary for secretion. We also have not shown how secretion of ACBP would be affected by blocking phosphorylation of one or more of these sites. Nor have we shown which phosphorylation site(s) on ACBP influence its interaction with GABAAα1. These questions can be best addressed with in vitro secretion experiments using ACBP plasmids that lack specific phosphorylation site(s).

ACBP and Müller glial cell function

Müller glial cells are the major glial cell type in the vertebrate retina (Uckermann et al. 2004). Known secretory roles for Müller glial cells include secretion of apolipoprotein E (Amaratunga et al. 1996) and nerve growth factor (Dicou et al. 1994). A similar secretory pattern has been observed in primary astrocytic cultures prepared from rat hippocampus (Calegari et al. 1999). Treatment of cultured astrocytes from mouse brain with depolarizing K+ also increases secretion of ACBP (Lamacz et al. 1996). Secretion of ACBP is not because of exocytosis as ACBP lacks a signal peptide, and Brefeldin A, which blocks the vesicular exocytosis, does not affect secretion of ACBP (Lafon-Cazal et al. 2003). Export of peptides without a signal sequence can occur through ATP-driven membrane translocations (Marty et al. 2005). In the present study, we found that secretion of ACBP occurs in the presence of BAPTA which should deplete intracellular stores of calcium. However, treatment of astrocytes with phorbol esters and ionomycin in the presence of extracellular calcium causes significantly greater secretion of secretogranin II than occurs in the absence of extracellular calcium (Calegari et al. 1999). Treatment of astrocytes with β-amyloid peptides not only induces secretion of ACBP (Tokay et al. 2005) it also increases intracellular calcium (Jalonen et al. 1997). Intracellular recordings from pyramidal cells in hippocampal slices indicate that slow transient currents can be evoked by glutamate release from astrocytes (Angulo et al. 2004). These combined observations suggest that astrocytes participate in regulated secretory pathways that include secretion of ACBP as well as several other proteins (Haydon and Carmignoto 2006).

Optokinetic stimulation of the retina activates cells in both the inner and outer nuclear layers. Increased neuronal activity increases the extracellular concentration of K+ (Karowski and Proenza 1977). Müller glial cells are highly permeable to K+ through four different voltage-gated K+ channels (Newman 1985; Brew et al. 1986; Newman and Reichenbach 1996; Chao et al. 1997). Consequently, Müller glial cells may secrete ACBP in response to stimulus-evoked increases in extracellular K+.

Once secreted, ACBP could associate with GABAA receptors expressed by rods, cones bipolar cells, horizontal cells, amacrine cells, and ganglion cells (Dreher et al. 1992; Euler and Wassle 1998; Johnson and Vardi 1998; Kim et al. 1998; Zucker and Ehringer 1998; Grunert 2000; Zhang et al. 2003). ACBP binding could block the chloride channel in the GABAA receptors and reduce inhibitory post-synaptic potentials (Fig. 9).

Figure 9.

 Secretion of phosphorylated ACBP and its interaction with GABAA receptor. 1. PMA activates PKC. Activated PKC binds to ACPB. 2. ACBP is phosphorylated at threonine 42 and/or threonine 65. 3. Threonine-phosphorylated ACBP is secreted from the cell. 4. Phosphorylated ACBP binds to GABAA receptor, inducing a conformational change (X) that reduces a chloride current.

The GABAA receptor is usually composed of α, β, and γ subunits. The benzodiazepine binding site requires specific α and γ subunits (Pritchett et al. 1989; Pritchett and Seeburg 1991). ACBP associates with the α subunit of the GABAA receptor complex. His-ACBP pulls down the GABAAα1 from brain and retinal lysates. Antibodies to GABAAα1 co-immunoprecipitate ACBP from cerebellar lysates (Barmack et al. 2004). However, we have not excluded the possibility that ACBP may also associate with other GABAA receptor subunits and only indirectly with the GABAAα1 subunit. Further testing with addition GABAA receptor subunit specific antibodies will be required to answer these questions.


We thank Dr Vadim Yakhnitsa for his thoughtful comments on an earlier draft of this manuscript. We also thank Ms Mary Westcott for her careful preparation of tissue for immunohistochemistry. This work was supported by grants from the National Eye Institute to NHB (EY04778 and EY018561).