Neuropeptide FF (FLFQPQRF-NH 2 ) and its Fragments Bind to α 2 δ Subunit of Voltage-Gated Calcium Channels

—Purpose: Gabapentin, a drug for neuropathic pain, exerts its therapeutic effect via inhibition of the  2  subunit of N-type Ca 2+ channels. Thus, finding peptides that specifically displace gabapentin from its binding site may lead to the development of new drugs. Methods: Displacement of bound [ 3 H]-gabapentin in membrane preparations of rat cerebral cortex and of human Cav2.2/β 3 /α 2 δ 1 expressed in CHO cell line. Results: Neuropeptide FLFQPQRF-NH2 specifically displaced bound [ 3 H]-gabapentin in membrane preparations from rats and CHO cells. Truncation of the C-terminus of FLFQPQRF-NH2 by three amino acid residues to produce FLFQP-NH 2 improved the displacement of gabapentin. FLFQP-NH 2 displaced bound [ 3 H]-gabapentin with IC 50 and K i values of 2.7 µM and 1.7 µM, respectively. Deletion of two amino acid residues (FQ) in the middle of the FLFQP-NH 2 sequence yielded FLP-NH 2 that displaced bound [ 3 H]-gabapentin with a lower affinity. IC 50 and K i values were 11.9 µM and 7.8 µM, respectively. Neutral binding cooperativity existed when of FLFQP-NH 2 , FLP-NH 2 and gabapentin when incubated together. FLFQPQRF-NH 2 but not FLFQP-NH 2 displaced bound [ 3 H]-gabapentin to membrane preparations of human Cav2.2/  3 /  2  1 expressed in CHO cells. Conclusion: FLFQPQRF-NH 2 , FLFQP-NH 2 and FLP-NH 2 displace bound gabapentin in membrane preparations of rat cerebral cortex. Binding cooperativity was detected when GBP/FLFQP-NH 2 /FLP-NH 2 were incubated together. These novel binding sites may provide new approaches to human Cav2.2/β3/α2δ1 were incubated for 30 min at 25ºC in 10 mM HEPES, pH 7.4 with 20 nM [ 3 H]-GBP and 10 µM of each peptide. The counts per minute value for 100% binding for [ 3 H]-GBP alone were 19,335±3,600 cpm and for non-specific binding were 428±100 cpm (mean  S.D. of 6 determinations). Results are an

Voltage-gated calcium channels (CaV1.x, CaV2.x, and CaV3.x) are an important route of calcium entry into the cell. The channels consist of a pore protein and four types of auxiliary subunits: 1, 2, , , and in some tissues an additional γ subunit (8,9). α2δ has been implicated in the development of neuropathic pain caused by damage or disease of the somatosensory nervous system. The α2 protein is an extracellular glycoprotein connected through disulfide bonds to the δ protein (10,11). α2δ1 of the N-type channel (CaV2.2) is a target for treatment of neuropathic pain (12,13). Three Cav2.2 antagonists, gabapentin (GBP, Neurontin), pregabalin (Lyrica), and ziconotide are approved drugs for chronic pain. α1 of L-type channel (CaV1.2) also seems to be involved in pain signaling and chronic pain (14).
Gabapentin with a chemical structure of 1-(aminomethyl)-cyclohexaneacetic acid and its analog pregabalin (3-(aminomethyl)-5methylhexanoic acid; 3-isobutyl; GABA) bind to α2δ1 and α2δ2 subunits but not to α2δ3 or α2δ4 (15)(16)(17)(18)(19). Their binding decreases the level of CaV2.2 in the cell membrane but it does not block the channel pore. Both compounds are common drugs for neuropathic pain (20,21). GBP binds to an RRR (arginine residue) motif located at the N-terminal of a VWA (von Willebrand factor) domain in the α2 protein (21). Replacing a single arginine residue to form RRA motif abrogates GBP binding to the α2 ______________________________________ Corresponding Author: Hanna Skubatz; E-mail address: skubatch@neopropain.com protein and consequently, GBP does not alleviate neuropathic pain in a pain model in mice (22).
The present study demonstrates that FLFQPQRF-NH2 and its fragments (FLFQP-NH2 and FLP-NH2) at µM concentrations displace bound [ 3 H]-GBP in membrane preparations of rat cerebral cortex and human Cav2.2 expressed in CHO cells. However, no displacement of bound [ 3 H]-GBP) was observed when GBP/FLFQP-NH2/FLP-NH2 were incubated together suggesting a novel binding site(s) that cross interacts with GBP receptor α2δ1.

Peptide synthesis
Four peptides (FLFQP-NH2, FLP-NH2, FLHyp-NH2 and FQP-NH2) were custom synthesized with purity greater than 95% using a standard solid-phase method (AnaSpec Inc., Fremont, CA, USA). The peptides were synthesized from all L-amino acids. Cis-4-hydroxy-L-proline residue was used in the synthesis of FLHyp. The peptides were purified by reversed phase HPLC using a C18 column (Acclaim™ 120 C18 Column; Thermo Fisher Scientific Inc. Waltham, MA, USA) coupled to a linear ion trap mass spectrometer (LTQ XL, Thermo Fisher Scientific Inc.). Mass spectra at the range of m/z values of 200 to 2000 were used to calculate the molecular mass of the peptides (Fig. 1). A singly charged peptide ion was the most intense peak in the mass spectra of FLFQP-NH2 at m/z value of 651 (A). FLP-NH2 was present as a singly charged ion at m/z value of 376.2 (B). FLHyp-NH2 (C) and FQP-NH2 (D) were also present as singly charged ions at m/z values 392 and 391, respectively. Dimeric ions and other charged ions commonly formed in the gas phase environment of the mass Spectrometer were also detected. Sodium adduct peaks were observed at 22 m/z units above the peptide ions. The deviation of the calculated molecular masses from the theoretical masses was less than 1 Da. After purification and spectrum analysis, the peptides were lyophilized and kept dry at 4ºC.
Membranes preparation of rat brain cortex Wistar male rats (175 ±25 g body weight) were sacrificed by CO2 overexposure and the cerebral cortex was dissected and used for membrane preparations (BioLASCO Taiwan Co., Taipei, Taiwan). All procedures were done at 4°C as described by Glee et al (16).
Brain tissue was removed from the skull into buffer A (5 mM Tris-HCl, pH 7.4, containing 0.32 M sucrose). Cerebral cortex tissue was removed from the brain tissue and homogenized in 10 volumes (v/w) of buffer A in a glass-Teflon homogenizer using ~10-15 up-and-down strokes. The homogenate was centrifuged at 1,000g for 10 min. The resulting pellet was washed twice with buffer A and the supernatants were collected and centrifuged at 21,500g for 20 min. The supernatant was removed and the resulting crude membrane pellet was resuspended in 5 mM Tris-HCl, pH 8.0 and stirred for 1 h at 4ºC. The suspension was centrifuged at 40,000g for 30 min and the pellet was resuspended in buffer B (5 mM Tris-HCl, pH 7.4, containing 1.2 M sucrose). The suspension was layered onto the top of a discontinuous sucrose density gradient. The gradient was centrifuged at 100,000g for 90 min. The fraction containing the cerebral cortex membranes was recovered from the 0.9 M -1.2 M sucrose interface and resuspended in 5 mM Tris-HCl, pH 7.4. The suspension was centrifuged at 40,000g for 30 min and the membrane fraction was kept at -80°C until use. Membrane fractions were prepared by Eurofins Scientific (Taiwan).

Membranes preparation of CHO cells expressing
Cav2.2/β3/α2δ1 and Cav1.2/β2/α2δ1 Cell lines Stable cell lines expressing (under tetracycline induction) human N-and L-type Cavs in CHO (Chinese hamster ovary) cells were constructed as described (23). One stable cell line expressed Cav2.2/β3/α2δ1. Neuronal CACNA1B gene that encodes the poreforming subunit of N-type channel was co-expressed with its β3 and α2δ1. The other cell line expressed Cav1.2/β2/α2δ1. Neuronal CACNA1C gene that encodes the pore-forming subunit of L-type channel was co-expressed with its β2 and α2δ1. Cell lines were produced by ChanTest corp (OH, USA).

Preparation of Membranes
Confluent cells (75-85% confluency) were washed and harvested in cold DPBS (Dulbecco's phosphate buffered saline without Ca 2+ or Mg 2+ ) by scrapping from the plates. Cell pellets were prepared by centrifugation at 100g for 10 min. The cells were resuspended in 10 mM Tris-HCl pH 7.5, 2 mM EDTA and a protease inhibitor cocktail (1 tablet per 250 ml buffer, cOmplete, Roche Diagnostics Corp., IN, USA). Subsequently, the cells were homogenized in a Polytron homogenizer. The homogenate was centrifuged at 38,000g for 20 min. This procedure was repeated twice and the membrane pellets were combined and transferred into a glass Dounce homogenizer. The membrane pellet was homogenized in a minimal volume of 10 mM Tris-HCl (pH 7.5) containing 10% sucrose using 10 up-and-down strokes. The resulting membrane suspension at a protein concentration of 2.5 mg/ml was aliquot and kept at -80ºC until use. Membrane suspensions were prepared by ChanTest corp (OH, USA).

Displacement of bound [ 3 H]-gabapentin by peptides in membrane preparations
Binding of [ 3 H]-GBP (90-120 Ci/mmol) to membrane preparations was performed as previously described (16,24). Briefly, a 20 µg aliquot of membrane protein was incubated with 20 nM [3H]-GBP in 96-well microplates (U96 PP-0.5 ml, Thermo Fischer Scientific Inc.) in the absence or presence of increasing concentrations of a specific peptide in 10 mM HEPES, pH 7.4. The peptides were either dissolved in the incubation buffer or in the incubation buffer containing 1% DMSO. The membrane preparations were filtered and washed with 100 µM unlabeled GBP in 10 mM HEPES, pH 7.4 to remove non-specific binding. The filters were then counted using a microplate scintillation counter (TopCount NXT, Packard Instrument Company, CT, USA) to determine bound [ 3 H]-GBP. Math-IQ software was used for curve fitting and calculation of various binding parameters (IDBS, MA, USA).

Displacement of bound [ 3 H]-gabapentin by
FLFQPQRF-NH2 and its fragments in rat cerebral cortex membranes.
[ 3 H]-Gabapentin bound to membrane preparations of rat cerebral cortex with high affinity and produced a steep monophasic curve (Fig. 2). The Kd and Bmax values were 38 nM and 6.8 pmol/mg protein, respectively and in agreement with published data. The saturation binding data were compatible with a single high-affinity binding site (Hill coefficient = 0.9) with IC50 and Ki values of 40 nM and 26 nM, respectively. Thus, the displacement of GBP from its high affinity-binding site was used to screen for peptide ligands.
FLFQPQRF-NH2 was truncated at the Cterminus to form FLFQP-NH2 and at the N-and Cterminal to form FQP-NH2. Ten µM of FLFQP-NH2 displaced 88% of bound [ 3 H]-GBP, a slight improvement in the displacement capacity relative to FLFQPQRF-NH2. Ten µM of FQP-NH2 displaced 71% of bound [ 3 H]-GBP. These results revealed that the two truncated peptides retain the capability to displace bound [ 3 H]-GBP from α2δ. However, FLFQP-NH2 seemed to be the most efficient (Fig. 2). The displacement of bound [ 3 H]-GBP resulted in a shallow, rightward shift of the curve in a concentration-dependent manner. The displacement curve was monophasic suggesting an interaction with a single binding site. The IC50 and Ki values for the displacement of bound GBP by FLFQP-NH2 were 2.7 µM and 1.7 µM, respectively. FLFQP-NH2 is a weaker ligand than GBP (~67 fold differences in IC50 and Ki). It may suggest that FLFQP-NH2 (molecular weight of 651) interacts with a large surface of α2δ whereas GBP (molecular weight of 171) interacts with only one arginine residue (R 217 ). In another sets of experiments FLP-NH2 displaced bound [ 3 H]-GBP in a concentration dependent manner (Fig. 3). Two amino acid residues (FQ) from the middle of FLFQP-NH2 sequence were omitted to form FLP-NH2. The IC50 and Ki values were 11.9 µM and 7.8 µM, respectively. These values were  4-fold lower than the values obtained for FLFQP-NH2. Removal of the sequence in the middle (FQ) reduced the binding affinity to the to the GBP binding site (Table 2). It suggests that the two amino acid residues make a significant contribution to the binding of FLFQP-NH2 to α2δ1. Replacing of proline residue in FLP-NH2 with hydroxyproline (cis-4-hydroxy-L-proline) residue decreased the displacement of bound [ 3 H]-GBP suggesting that the hydroxyl group of hydroxyproline interferes with the binding of FLP-NH2 (Table 2).
Bound [ 3 H]-GBP was not displaced when FLFQP-NH2 + FLP-NH2 were added concomitantly to the incubation medium ( Table 2). At 10 µM concentration each of the peptide displaced  40-50% bound [ 3 H]-GBP (Fig. 2). About 100% displacement was expected but instead the displacement efficiency was zero. One plausible explanation is that the peptides bind to a site distinct from the GBP binding site and cooperativity exists between these binding sites. As a consequence of this interaction bound [ 3 H]-GBP was not displaced.  (Table 3). On the other hand, FLFQP-NH2 exhibited a low displacement capability of bound [ 3 H]-GBP compared to that observed in membrane preparations of rat cerebral cortex (Table 1). One reason for changes in the displacement ability of FLFQP-NH2 is the differences in membrane constituents and environments. Similar results were obtained with membrane preparations of human L-type channel, Cav1.2/β2/α2δ1 expressed in CHO cells (data not shown). The IC50 value for [ 3 H]-GBP in membrane preparations of Cav1.2/β2/α2δ1 was 29 nM.

Interpretation of binding data
Gabapentin exists in two distinct conformations in solution with three elements for interaction. Alkylamino and carboxylate charged groups and a hydrophobic hydrocarbon ring (25). 2/ interacts with GBP and with the amino acid leucine that closely resembles gabapentinoid (26). FLFQPQRF-NH2, FLFQP-NH2 and FLP-NH2, on the other hand, can adopt many conformations in solution and they probably interact with a larger surface of 2/ than GBP (27). The data may also suggest that FLFQPQRF-NH2 and its fragments interact with α2δ at distinctly different sites than GBP. The peptides may interact with another binding site, an allosteric site, and in such a case the peptides may have identified a novel allosteric site sites in 2/. However, it has been shown that a single binding site interacts with structurally different ligands (30). Further studies will determine the peptides binding site(s). Study of the cross-interactions between GBP/FLFQP-NH2/FLP-NH2 binding sites in 2/ can provide new information for design a better drug. The muscarinic acetylcholine receptors, for example, have two distinct allosteric sites that modulate sensitivity to ligands (31). Cooperativity with an orthosteric site can switch between negative, positive, and neutral cooperativity depending on the presence or absence of a specific ligand. Allosteric modulation of the GBP binding site in 2/ may affect its efficacy and thus its analgesic and antiepileptic properties. It has been shown that spermine (28) and Ruthenium red (29) bind to allosteric sites in 2/ that is distinguished from GBP binding site. Thus, it is possible that FLFQPQRF-NH2, FLFQP-NH2 and FLP-NH2 bind to additional distinguished binding Potential ligands of 2/ FLFQPQRF-NH2, FLFQP-NH2 and FLP-NH2 may be putative endogenous ligands of the 2. One cleavage product of NPFF, QRF-NH2, was detected in mouse brain (32). It is possible that more cleavage sites exist in NPFF to allow the production of various peptides. Whether the FLFQP-NH2 and FLP-NH2 exist under physiological conditions is an unanswered question. Further structure and binding relationship studies have to be performed and the peptides' biological activity has to be determined. Whether these shorter peptides are functionally relevant in human brain biology remains unknown.
GBP binding affinity to the 2/ subunit expressed in CHO cells was similar to its binding to membrane preparations of rat cerebral cortex. FLFQPQRF-NH2 displaced bound [ 3 H]-GBP from the two different preparations suggesting the presence of a common binding site. However, the membrane environment of CHO cells affected FLFQP-NH2 binding and it did not displace bound [ 3 H]-GBP. The reason for this is unknown at the moment. The peptides can provide preliminary information on novel binding sites within α2δ and they may serve as a template for the development of new drugs for neuropathic pain (33).
NPFF (FLFQPQRF-amide) and its fragments and pain homeostasis NPFF at a cellular level exerts antinociceptive, antiopioid activity via its receptors, NPFF1R and NPFF2R (34,35). NPFF can modulate nociception depending on the pharmacological dosage and locations of administration (36)(37)(38). NPFF suppressed pain in a mouse model of neuropathic pain depending on the site of administration. The mRNA level of NPFF remained unchanged in a model of carrageenan induced inflammatory pain (2). The analgesic effects of FLFQP-NH2 and FLP-NH2 in a range of animal models of pain have to be studied to determine their involvement in the elevation of pain.
Currently, the GBP binding site in α2δ has no endogenous ligand. GBP affects the 2/ recycling and thus prevents neuropathic pain (39,40). It is possible that FLFQPQRF-NH2, FLFQP-NH2 and FLP-NH2 are endogenous ligands that modulate acute and neuropathic pain sensitivity and therefore involved in keeping homeostasis of pain sensation (41). These preliminary results provide a starting point for the identification of the binding sites of FLFQPQRF-NH2, FLFQP-NH2 and FLP-NH2 and their biological involvement in pain and other diseases such as epilepsy.  (%)  FLP-NH2  100  99  FLHyp-NH2  100  87  FLFQP-NH2 + FLP-NH2 10 -3 Membrane preparations (20 µg protein) were incubated with a mixture of 20 nM [3H]-GBP and one or two peptides at a concentration of 10 µM for 30 min at 25ºC. Specific binding of [ 3 H]-GBP without any peptide was 5,650 cpm and nonspecific binding was 710 cpm. The data are expressed as percentage of specific binding. Displacement greater than 50% was considered significant. Results are an average of duplicates. Hyp is the abbreviation for cis-4-hydroxy-L-proline.