Characterization of the novel cardiolipin binding regions identified on the protease and lipid activated PKC‐related kinase 1

Abstract

Protein kinase C‐related kinase 1 (PRK1) or PKN is a protease and lipid activated protein kinase that acted downstream of the RhoA or Rac1 pathway. PRK1 comprises a unique regulatory domain and a PKC homologous kinase domain. The regulatory domain of PRK1 consists of homologous region −1 (HR1) and −2 (HR2). PRK1‐(HR1) features a pseudosubstrate motif that overlapped with the putative cardiolipin and known RhoA binding sites. In fact, cardiolipin is the most potent lipid activator for PRK1 in respect of its either auto‐ or substrate phosphorylation activity. This study was thus aimed to characterize the binding region(s) of cardiolipin that was previously suggested for the regulatory domain of PRK1. The principal findings of this work established (i) PRK1‐(HR1) folded into an active conformation where high affinity binding sites (mainly located in HR1a subdomain) were accessible for cardiolipin binding to protect against limited Lys‐C digestion, (ii) the binding nature between acidic phospholipids and PRK1 (HR1) involved both polar and nonpolar components consistent with the amphipathic nature of the known cardiolipin‐binding motifs, (iii) identification of the molecule masses of the Lys‐C fragments of PRK1‐(HR1) complexed with cardiolipin molecule, and (iv) appreciable reductions in the secondary structural contents at 222 nm measured by circular dichroism analyses demonstrated the binding of cardiolipin elicited the disruptive effect that was most evident among all phospholipids tested, suggestive of a functional correlation between the extents of helical disruption and PRK1 activation.

Abbreviations

HR

heptad repeat

Lys‐C or endo Lys‐C

endoproteinase Lys‐C

MALDI‐TOF MS

matrix‐assisted laser desorption/ionization time of flight mass spectrometry

PKCs

protein kinase C

PKN

protein kinase novel

PRK1

protein kinase C‐related kinase 1

Introduction

Protein kinase C‐related kinase 1 (PRK1; also known as PKN or PAK1) is a trypsin or lipid activated kinase that acts as an effector downstream of RhoA or Rac1, the small GTP‐binding proteins that regulate the assembly of focal adhesion and actin stress fibers, in vivo.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 Other studies demonstrated that PRK1 was required for the survival of cardiac myocytes12 and guanosine‐mediated protection of neurons under stressed conditions.13 PRK1 was also indicated to promote the growth and development of prostate cancer cell.14 The structural features of PRK1 include an N‐terminal regulatory domain, hinge region, catalytic domain, and C‐terminal variable region (Fig. 1).3, 9 Overall PRK1 regulatory domain contains two internal domains, HR1 and HR2. The PRK1‐(1‐313) of the HR1 domain contains three heptad repeats (HR1a‐c).3 The HR2 domain corresponds to the nPKC C2‐like (or Vo) domain (PRK1‐398‐512).2, 3, 15 The HR1a subdomain contains pseudosubstrate (I) in PRK1‐(39‐53) and RhoA binding site in PRK1‐(7‐155) (Fig. 1).7, 8, 16, 17, 18 More recently, second pseudosubstrate (II) in PRK1‐(176‐186) and Rac1 binding site was also identified within the HR1b subdomain (Fig. 1).19 In addition to HR1a‐b subdomains, another pseudosubstrate (III) site was identified within the HR2 domain of PRK1 (Fig. 1).20, 21 Indeed, the maximal activity of PRK1 was observed only by removal of all pseudosubstrate motifs of its regulatory domain.21 The X‐ray crystallographic analysis revealed that the helical structures of PRK1‐(HR1a) were arranged in anti‐parallel direction and confirmed the structural details of the interactions between RhoA and PRK1‐(HR1a) subdomain.4 It was thus envisaged that the activation of PRK1 involved the steps of binding to RhoA or other biological activator(s) via its regulatory domain to release the pseudosubstrate motifs from catalytic domain. Phospholipids are the main constituents of cell membrane where PRK1 was recruited and subsequently activated by RhoA GTPase.22, 23 In addition, the effects of phospholipids on PRK1 activity were previously investigated and demonstrated that cardiolipin was the most effective lipid activator for hepatic PRK1.2, 15 Like RhoA, phospholipids could function in a comparable mode in promoting the biological activity of PRK1. Interestingly, a putative cardiolipin‐binding site (PRK1‐50‐77) identified within the N‐terminal region of the HR1a domain was shown to overlap with the pseudosubstrate (I) and RhoA binding sequences of PRK1 (Fig. 1).3, 7, 16 Thus, it seemed that cardiolipin binding might at least cause a disruption of the interaction between HR1a and pseudosubstrate (I) site, as has been suggested for the RhoA binding interaction. Similar to HR1a, HR1b subdomain contains a putative pseudosubstrate (II) motif (Fig. 1) overlapping with the Rac1 binding region.19 Hence, it was of interest to investigate the additional phospholipid (e.g., cardiolipin) binding site(s) located within the HR1b subdomain. The close proximity of the pseudosubstrate, cardiolipin, and RhoA or Rac1 binding sites within the N‐terminal region of PRK1 suggests that HR1 domain is of functional importance in the regulation of its biological activities.

Figure 1.

The domain structures of PRK1. The PRK1 structure consists of an N‐terminal regulatory domain that is linked by a hinge region to a C‐terminal catalytic domain. The PRK1‐(HR1) DNA plasmid encodes for PRK1‐(1‐313) containing an N‐terminally 6× histidine tag and three heptad repeats (HR1a‐c). The locations of the three pseudosubstrate motifs (I–III), cardiolipin, and RhoA binding regions of the HR1 and HR2 domains are also indicated in this diagram.

The aim of this study was to devise an integrated strategy with the following biochemical and biophysical approaches to determine and characterize the cardiolipin‐binding site of PRK1 by (i) limited endo Lys‐C digestions of the recombinant PRK1‐HR1‐(1‐313) in the absence and presence of cardiolipin, phosphatidic acid, phosphatidylcholine, or phosphatidylserine followed by Coomassie blue and immunostaining analyses (ii) biophysical analyses to compare the binding affinity for PRK1‐(HR1) and phospholipids interaction under low and high salt concentrations, and (iii) matrix‐assisted laser desorption/ionization time of flight mass spectrometry (MALDI‐TOF MS) to identify and compare the Lys‐C fragments generated in the absence and presence of cardiolipin.

Results

Structural analyses for the PRK1‐(HR1) protein

To investigate the structural and biochemical properties of the N‐terminal HR1 regions of the regulatory domain of PRK1, the recombinant 6× histidine‐tagged PRK1 (residues 1–313) (referred to as PRK1‐(HR1)) was constructed and expressed in bacterial cells (Fig. 1). The C‐terminal residues (314–946) were not included for expression (Fig. 1). PRK1‐(HR1) was refolded and monitored by the PRK1‐Trp13 intrinsic fluorescence [Fig. 2(a)]. The spectra obtained from the refolded, denatured, and free tryptophan were compared and demonstrated that the PRK1‐Trp13 of the refolded protein samples was in a folded conformation with a λ _max peaked at 347 nm [Fig. 2(a)]. The molecular mass of the purified recombinant protein was further analyzed by MALDI‐TOF MS. Two major peaks detected with average mass to charge (m/z) ratios of 38,982 Da (main peak) and 19,469 Da (minor peak). A third relatively small peak with m/z of 78,587 Da was also evident. The values for the two larger species were consistent with the calculated molecular masses of 38,885.7 and 77,771.4 Da for monomer and dimer [the inset in Fig. 2(b)] forms, respectively. The m/z 19,469 Da species corresponded to the MH²⁺ molecular ion of the PRK1‐(HR1) protein. The detection of only these three peaks confirmed the high level of purity of the sample.

Figure 2.

Two forms of PRK1‐(HR1) were identified in the refolded preparations. (a) The spectra obtained from refolded PRK1 (HR1), PRK1‐(HR1) polypeptide, and free tryptophan by intrinsic fluorescence spectrometry were compared. The concentration used for all samples was 1.5 μM. The spectra were scanned at the excitation wavelength, 296 nm and emission wavelength ranged from 310 to 410 nm. All spectra were buffer background subtracted. (b) The MALDI‐TOF mass spectrum shown is the average mass spectra obtained from 100 laser shots for PRK1‐(HR1). Ribonuclease A (M _w = 13,862.3) was used for external calibration. The m/z peaks of 38,982 and 19,469 represent [M + H]⁺¹ and [M + H]⁺², respectively. The inset shows the peak with m/z ratio 78,587 Da in the dimeric form. (c) Peaks “a” (dimer) and “b” (monomer) fractions were eluted by the S‐200 size exclusion chromatography. Lane 1: Protein markers; lane 2: 2 μg of PRK1‐(HR1); lane 3: Peak “a” 100 kDa; lane 4: Peak “b” 58 kDa. The S‐200 column was calibrated by the protein markers (inset) as described in Materials and Methods. The integrity and identity of the eluted “a” and “b” peak fractions were confirmed by Coomassie blue and immunostaining by anti‐PRK1/PKN‐(1‐150) analysis, respectively.

To determine the conformation of the PRK1‐(HR1), the refolded protein was applied to Sephacryl (S‐200) size exclusion chromatography [upper panel in Fig. 2(c)]. The molecular mass of two individual peak fractions was estimated to be 100,000 and 58,000 Da, corresponding to dimer (“a”) and monomer (“b”) peak, respectively. Calculations of the areas under the two peaks estimated the proportions of the monomers and dimers were 54% and 46%, respectively (Table S1). The identity of the protein samples was confirmed to be the PRK1‐(HR1) protein by anti‐PRK1/PKN‐(1‐150) [lower panel in Fig. 2(c)]. To investigate whether the dimer and monomer are in a chemical equilibrium, the eluted dimer preparation (peak “a” fraction) was re‐chromatographed on the S‐200 size exclusion column under the same buffer conditions. The S‐200 profile indicated that the re‐chromatographed peak “a” fraction eluted as a single peak at the identical elution volume as before (data not shown). This observation suggested that a component of PRK1‐(HR1) preparation (50%) was refolded into a form, which favored dimerization. The remaining fraction of protein was refolded into a monomer form, which appeared more likely to reflect the natural form of the HR1 domain in the native PRK1 monomer.2, 5

Monomer is the active cardiolipin binding form of PRK1‐(HR1)

The interactions between PRK1‐(HR1) proteins and phospholipids were first investigated by circular dichroism (CD) spectroscopy. The CD spectra for PRK1‐(HR1) in the presence and absence of cardiolipin, phosphatidylcholine, and phosphatidylserine were compared [Fig. 3(a)]. The cardiolipin concentration used in this experiment was 3.3 μM, in excess of its concentration (EC₅₀ = 1.7 μM) required to effect half‐maximal activation of purified rat liver PRK1.15 These results confirmed that the phospholipid‐induced decrease in ellipticity at 222 nm was associated with reduction in α helical secondary structure [Fig. 3(a)]. Notably, the binding of cardiolipin affected the most significant decrease in the helical contents of PRK1‐(HR1).

Figure 3.

The interactions between PRK1‐(HR1) and phospholipids. (a) Samples containing 3.3 μM PRK1‐(HR1) were preincubated with or without 3.3 μM cardiolipin, phosphatidylcholine, and phosphatidylserine in 20 mM sodium phosphate pH 7.4, 25 mM NaCl and analyzed by CD spectroscopy. The arrow indicated the CD signal variations detected at 222 nm in the absence and presence of phospholipids. The spectra are the average of five scans after subtracting the buffer background. (b) Samples of 1.3 μM PRK1‐(HR1) were incubated with cardiolipin and phosphatidic acid (upper) or phosphatidylcholine and phosphatidylserine (lower). The molar ratio of protein: Lipid was 1:10 in all cases. All spectra were measured at the excitation wavelength of 296 nm and emission wavelength range of 310–410 nm using 1 nm intervals; the spectra shown represent the average of eight scans. The buffer background was subtracted in each of the fluorescence emission spectra shown above.

The protein versus phospholipid interaction was further analyzed by the intrinsic fluorescence spectrometry in the absence and presence of phospholipids [Fig. 3(b)]. The quenching % elicited by the binding of cardiolipin at 347 nm was 74%, a marked reduction of the PRK1‐Trp13 emission intensity, and greater than 46%–48% affected by other phospholipids.

To further investigate which of conformation of PRK1‐(HR1) was selective for the binding of cardiolipin, the protein eluted from either peak “a” or “b” was preincubated with cardiolipin followed by the intrinsic fluorescence spectrometry. It was demonstrated that the presence of cardiolipin only quenched the monomeric fraction with a 70% of spectral intensity reduction at 347 nm [Fig. S1(A), S1(B)]. That was an observation similar to the quenching effect (74%) determined for the post S‐200 protein preparations containing both monomer and dimer [Fig. 3(b)]. It was thus indicated that the monomeric PRK1‐(HR1) represents the active cardiolipin (or phospholipids) binding fraction in the refolded protein samples.

Phospholipid‐protected regions of the PRK1‐(HR1) domain

The appearances of those highly conserved cardiolipin‐binding LR(K)X motifs identified throughout the entire HR1 domain (Fig. S3) suggested that the N‐terminal regions of PRK1 regulatory domain contain multiple cardiolipin‐binding sites.24 Indeed, the experiments of the phospholipid protection on the HR1 domain against Lys‐C digestion demonstrated that only the presence of either cardiolipin or phosphatidic acid protected the PRK‐(HR1) domain to yield a prominent 16‐kDa Lys‐C resistant fragment [Fig. 4(a), lanes 3 and 6]. In addition, this 16 kDa fragment was immunoreactive to the anti‐PRK1/PKN‐(1‐150), suggesting that the regions of protection were mainly located N‐terminally of the regulatory domain of PRK1. These observations suggested that both cardiolipin and phosphatidic acid protected those specific lysine residues on PRK1‐(HR1). Quantification by densitometry showed that a twofold higher level of the Lys‐C resistant polypeptide was generated in the presence of cardiolipin, compared with that of phosphatidic acid [Fig. 4(a)].

Figure 4.

Lys‐C digestion of the PRK1‐(HR1) in the presence and absence of phospholipids.

(a) Upper panel: PRK1‐(HR1) (4 μg) was preincubated with various phospholipids at 36°C, using a protein: lipid of 1:1.5 in molar ratio. The digestion was carried out at the Lys‐C: PRK1‐(HR1) ratio of 1:50 (w/w) in buffer 25 mM Tris–HCl pH 7.4, 1 mM EDTA at 36°C for 10 min. The Lys‐C digests were analyzed by SDS‐PAGE, followed by immunostaining analysis using anti‐PRK1/PKN‐(1‐150) (Upstate). Lower panel: the optical density of the 16 kDa Lys‐C peptide bands was quantitated by densitometry. (b) Upper panel: In the absence of cardiolipin, PRK1‐(HR1) (5.5 μg) was digested at Lys‐C: protein ratio of 1:100 (w/w) in the same buffer used in (a) at 36°C for 40, 80, and 120 minutes (lanes 3–5). In presence of cardiolipin, protein sample was digested under the protein: cardiolipin molar ratio of 1:150 in (a) at 36°C for 40–120 minutes (lanes 6–8). Lane 1 was the undigested full length protein and lane 9 was the lysozyme as a size reference. The individual digests were analyzed by SDS‐PAGE and Coomassie blue staining. Lower panel: the optical density of the 40, 29, 19–21, and16 kDa Lys‐C peptide bands was quantitated by densitometry. The solid and dashed line represent with or without cardiolipin, respectively. Due to the size of the 6–13 kDa fragments band, it was not included in this analysis.

In order to further compare the banding patterns of the RPK1‐(HR1) Lys‐C digests, a time course study was carried out in the absence and presence of cardiolipin [Fig. 4(b)]. The digests were analyzed by SDS‐PAGE and Coomassie blue staining to enable detection of any C‐terminal PRK1‐(HR1) fragments that were deficient in epitopes for anti‐PRK/PKN‐(1‐150). In the absence of cardiolipin, full length PRK1‐(HR1) (40 kDa) was converted rapidly into small peptide fragments [6– 13 kDa, Fig. 4(b) lanes 3–5]. The addition of cardiolipin at the high protein: lipid ratio of 1:150 protected most of the PRK1‐(HR1) protein against Lys‐C digestion during the 40  minutes incubation [Fig. 4(b), lane 6]. In longer Lys‐C digestions (60–120 minutes), there was a progressive breakdown of the 40‐kDa protein into the 29 kDa, 19–21 kDa with selective accumulation of the 16‐kDa Lys‐C resistant fragments [Fig. 4(b), lane 8].

Quantitative binding analyses for PRK1‐(HR1) versus phospholipids interactions under different conditions of ionic strength

To investigate the chemical nature of the PRK1‐(HR1)—phospholipid interaction, the dissociation constants (Kd) were determined by PRK1‐Trp13 fluorescence emission quenching spectrometry under low and high salt concentrations. The impact of the salt concentrations on the conformations of PRK1‐(HR1) was first analyzed (Fig. S2). A similar spectral characteristic was observed when the protein was under 25 or 300 mM of NaCl (Fig. S2). The phospholipid‐quenched signals on the PRK1‐Trp13 were measured and the fractional saturation values were calculated as described previously.25 The stoichiometry of the protein versus phospholipid binding was estimated to be one molecule lipid bound per free monomer as confirmed by the Scatchard analyses [Fig. 5(a–d)].

Figure 5.

Phospholipids binding curves for PRK1‐(HR1) in the presence of either 25 or 300 mM NaCl. The protein‐phospholipid binding quenching signals were converted into fractional saturation curves for PRK1‐(HR1) versus (a) cardiolipin, (b) phosphatidic acid, (c) phosphatidylserine, and (d) phosphatidylcholine in 20 mM sodium phosphate pH 7.4, containing either 25 mM NaCl (rectangle) or 300 mM NaCl (triangle). The Scatchard analyses estimated that the value of binding function was 1 (1:1 molar ratio) for protein and phospholipid interaction. The Cs* value represents the unbound concentration of the phospholipids. The Kd values were calculated and summarized in Table 1.

At 25 mM NaCl, the Kd values determined for cardiolipin, phosphatidic acid, phosphatidylserine, and phosphatidylcholine were 28.1, 39.7, 38.4, and 10 nM, respectively (Table 1). Of all the phospholipids, the binding affinity of phosphatidylcholine to PRK1‐(HR1) was the highest by up to four‐fold. The elevation of the salt concentration to 300 mM increased the Kd values for all phospholipids investigated (Table 1), suggesting the binding of phospholipid to the HR1 domain is sensitive to ionic strength. The most significant increase in Kd value was between PRK1‐(HR1) and cardiolipin (Table 1). The negative effect of increased ionic strength on the affinity of PRK1‐(HR1) for cardiolipin and other phospholipids (reflected in the increase in their Kds; Table 1) indicated that the high affinity binding events observed at low salt were dependent on electrostatic interactions. It was suggested that these interactions involved negatively charged phospholipid phosphoryl groups and positively charged amino acid residues in addition to any nonpolar interactions between the fatty acyl side chains of the phospholipid and hydrophobic amino acid side chains of the protein. Thus, the substantial negative effect of high salt on the Kd value for cardiolipin (5.2 fold increase), and to a lesser extent for phosphatidylcholine, phosphatidic acid, and phosphatidylserine, suggested that electrostatic interactions were of importance in the binding of phospholipids to PRK1‐(HR1) (Table 1).

Table 1.

PRK1‐(HR1)‐phospholipids binding parameters

Phospholipid names	K d 25 a , b (nM)	K d 300 a , b (nM)	K d 300 /K d 25
Cardiolipinc	28.1 ± 2.0	146.1 ± 5.1	5.2
Phosphatidic acidd	39.7 ± 2.3	83.9 ± 3.0	2.1
Phosphatidylserinee	38.4 ± 2.3	77.8 ± 2.5	2.0
Phosphatidylcholinef	10.0 ± 1.0	24.7 ± 2.3	2.5

^a K _ds were calculated as described in Materials and Methods.

^bThe ±SE of K _d (s) were generated from the linear regression analyses of Scatchard plot analyses.

^cData obtained from Figure 5(a).

^dData obtained From Figure 5(b).

^eData obtained from Figure 5(c).

^fData obtained from Figure 5(d).

MALDI‐TOF mass spectrometric analyses of the PRK1‐(HR1)‐Lys‐C fragments protected by cardiolipin

The phospholipid protection experiments established that the N‐terminal region of the PRK1 contains highly specific cardiolipin or phosphatidic acid binding sites (Fig. 4). Further study was carried out by MALDI‐TOF MS analyses of the Lys‐C digests in the absence and presence of cardiolipin to confirm the sequence identity of those HR1 regions of protection (Fig. S3). In the absence of cardiolipin, the relatively more prominent species were in the mass range 2401–10,410 Da [Fig. 6(a)]. The molecular masses of the 8553 and 6817 Da fragments best matched within experimental error for the PRK1‐(HR1)‐Glu⁵⁴‐Lys¹³³ (8490.5 Da) and PRK1‐(HR1)‐Gly⁻¹⁹‐Lys⁴³ (6838.4 Da) fragments, respectively [upper panel in Fig. 6(a), Table 2]. These results indicated that Lys‐C cleavage occurred in Region 1 at residues Lys⁴³ and Lys⁵³ and in Region 2 at Lys¹³³ (Fig. S3; Table 2). The major 6668 and 6896 Da species corresponded to the PRK1‐(HR1)‐Ile¹⁷⁷‐Lys²³⁶ and ‐Thr¹⁷⁵‐Lys²³⁶ Lys‐C fragments of PRK1‐(HR1), respectively [upper panel in Fig. 6(a), Table 2]. These results showed that Lys cleavage occurred within Regions 3–4 (Fig. S3). The other prominent species detected were the 2401 and 3593 Da species corresponding to the Leu²⁶⁷‐Lys²⁸⁷ and Gly²⁸⁸‐Lys³²² species, respectively [upper panel in Fig. 6(a)], indicating that cleavage occurred in Regions 5–6 (Fig. S3).

Figure 6.

MALDI‐TOF mass spectrometry analyses of the Lys‐C digestion of PRK1‐(HR1) in the absence and presence of cardiolipin. PRK1‐(HR1) (5.5 μg) was preincubated without (a) or with (b) cardiolipin and digested with Lys‐C. Upper panel: The peaks detected in the m/z range from 2100 to 10,000. Lower panel: The peaks detected in the m/z range from 10,000 to 40,000. The peaks in figure originating from fragmentation of PRK1‐(HR1) are summarized in Table 2 (minus cardiolipin) and Table 3 (plus cardiolipin). In the upper panel of (a), # denotes the peak with m/z of 15,433 Da that did not correspond to any species predicted as a Lys‐C digestion produced of PRK1‐(HR1) and was considered to be a contaminant. In the upper panel of (a), * denotes the peaks represent matrix adduct. In the lower panel of (b), * denotes the molecule ions corresponding to dimeric and trimeric forms of cardiolipin molecules.

Table 2.

MALDI‐TOF mass spectrometric peaks of PRK1‐(HR1)‐Lys‐C digests in the absence of cardiolipin

Observed mass (MH+)	PRK1‐(HR1)‐Lys‐C fragment	Predicted mass
13,566	Leu52‐Lys176	13,557.3
12,147	Gln143‐Lys251	12,114.4
10,410	Asp159‐Lys251	10,409.7
8767	Leu52‐Lys133	8726.6
a8553	Glu54‐Lys133	8490.5
a6896	Thr175‐Lys236	6903.6
a6817	Gly‐19‐Lys43	6838.4
a6668	Ile177‐Lys236	6673.4
a3593	bGly288‐Lys322	3595.1
a2401	Leu267‐Lys287	2399.8

^aProminent peptide mass spectrometric signals.

^bBoth Lys³²² and Lys³³⁶ are located within the non‐PRK1 extension region.

The experiments with the addition of cardiolipin was first performed by direct analyses of cardiolipin that identified the molecular ions of monomeric, dimeric, and trimeric cardiolipin with respective m/z value of 1506, 3003, and 4518 Da (Fig. S4). The molecular masses of cardiolipin molecule(s) were taken into account for the calculations of those Lys‐C fragment masses identified in MALDI‐TOF MS. Two of the prominent species, with estimated MH⁺ m/z values of 16,618 and 18,093 Da, were detected in the digests containing cardiolipin [lower panel in Fig. 6(b)]. The size of 16,683 Da fragment matched the Gly⁻¹⁹‐Lys¹³³ sequence (16,618 Da). The mass difference of 1475 Da between the 16,618 and 18,093 Da species was likely attributed by the mass of the cardiolipin monomer (1513 Da) (Fig. S4). This finding suggested that the 18,093 Da peak corresponded to the MH⁺ ion for the complex between a molecule of cardiolipin and the PRK1‐(HR1)‐Gly⁻¹⁹‐Lys¹³³ fragment. It should be noted that the high lipid to protein molar ratio used in the Lys‐C digestions, and the acidic (trifluoro acetic acid [TFA]) conditions used in sample preparations for MALDI‐TOF MS may have favored the survival of this complex under the mass spectrometric conditions [Fig. 6(b)]. Thus, the results of the mass analyses of the species in the 16–18 kDa range confirmed that cardiolipin protected the PRK1‐(HR1) lysine residues within Region 1 (Lys⁴¹, Lys⁴³, Lys⁴⁸, Lys⁵¹, and Lys⁵³; Tables 3 and 4, Fig. S3). Additional evidence for protection of Region 1 lysine residues was the observation of the 8131 Da species, corresponding to the Gly⁻¹⁹‐Lys⁵³ fragment detectable only in plus cardiolipin digests [upper panel in Fig. 6(b), Table 3]. The 7610 Da species [upper panel in Fig. 6(b)] was detected only in the presence of cardiolipin and matched the mass of the Gln¹³⁴‐Lys¹⁷⁴ (4619.4 Da) (Fig. 7) complexed with two molecules of cardiolipin (3000 Da). It provided an evidence of cardiolipin protection on the lysine clusters located within Regions 2–3 (Fig. S3). The 19,679 Da peak matched the mass of the Gln¹⁴³‐Lys³²² fragment (19,712.5 Da) that contains Regions 3–6 [lower panel in Fig. 6(b), Fig. S3]. Interestingly, this Gln¹⁴³‐Lys³²² fragment deficient of the epitope of anti‐PRK1/PKN‐(1‐150) was consistent with the 19–21 kDa species detectable only by Coomassie blue staining [Fig. 4(b)]. The other possible match for the 19,679 peak was that a complex between the Gly⁻¹⁹‐Lys¹³³ fragment and two molecules of cardiolipin (19,631 Da) (Table 3).

Table 3.

MALDI‐TOF mass spectrometric analyses of PRK1‐(HR1)‐Lys‐C digests in the presence of cardiolipin

Observed mass	PRK1‐(HR1)‐Lys‐C fragments (+/‐CAa)	Predicted mass	Lipid‐peptide complex
b19,679	Gln143‐Lys322 (or Gly‐19‐Lys133 + CA (2))	19,712.5 (or 19,630.7)	No Yes
b18,093	Gly‐19‐Lys133 + CA (1)	18,117.7	Yes
b16,618	Gly−19‐Lys133	16,617.7	No
13,623	bGln143‐Lys251 + CA (1) (or Leu52‐Lys176)	13,614.4 (13,557.3)	Yes No
13,126	Glu54‐Lys174	13,137.1	No
b12,125	Gln143‐Lys251	12,114.4	No
10,407	Asp159‐Lys251	10,409.6	No
b8131	Gly‐19‐Lys53	8115.3	No
b7610	aGln134‐Lys174 + CA (2)	7619.5	Yes
6830	Gly−19‐Lys43	6838.4	No
b6769	Leu260‐Lys322	6777.8	No
b5979	Leu267‐Lys322	5994.9	No
b5432	cLeu267‐Lys287 + CA (2)	5399.8	Yes
4500	CA (3)	4500.0	–
b3595	Gly288‐Lys322	3595.1	No
3009	CA (2)	3000.0	–
b2398	Leu267‐Lys287	2399.8	No

Predicted mass values for Lys‐C fragment‐cardiolipin complexes: assumes masses for individual components: a: Gln¹⁴³‐Lys²⁵¹ (12,114.4); b: Gln¹³⁴‐ Lys¹⁷⁴ (4619.5); c: Leu²⁶⁷‐Lys²⁸⁷ (2399.8).

^aCA (1–3): cardiolipin (numbers of monomer cardiolipin molecule).

^bProminent peptide mass spectrometric signals.

Table 4.

Cardiolipin protection sites on PRK1 (HR1) protein against Lys‐C digestion

Lys‐C fragments	Lysine residues	aDegree of protection
Gly‐19‐Lys133 and Gly‐19‐Lys53	Lys43, Lys48, Lys51 and Lys53	++++
Glu54‐Lys174	Lys133	+
Glu54‐Lys174 and Gln134‐Lys174	Lys140 and Lys142	++
Gln143‐Lys322, Gln134‐Lys174, Val141‐Lys174 and Gln143‐Lys251	Lys158 and Lys161	+++
Gln143‐Lys322 and Gln143‐Lys251	Lys174 and Lys176	++
Gln143‐Lys322 and Gln143‐Lys251	Lys236 and Lys246	++
Gln143‐Lys322	Lys251, Lys259 and Lys266	+
Leu260‐Lys322 and Leu267‐Lys322	Lys287	++

^a++++: very strong; +++: strong; ++: good; +: poor.

Figure 7.

Comparison of the MALDI‐TOF mass spectrometric PRK1‐(HR1) Lys‐C‐fragment patterns in the presence and absence of cardiolipin. There are six lysine cluster regions (1–6) defined in the diagrammatic representation of the PRK1‐(HR1) polypeptide (Mr 38856.7). Region 1 is located in HR1a; Regions 2 and 3 are located in HR1b and Regions 4 and 5 are located in HR1c. Region 6 is adjacent to the C‐terminal end of HR1c. Upper panel: Lys‐C fragments generated in the absence of cardiolipin as summarized in Table 2. Lower panel: Lys‐C fragments generated in the presence of cardiolipin as summarized in Table 3. The relative thickness of the bands representing individual Lys‐C fragments illustrated in the diagram represents the relative abundance of individual peaks observed in Figure 5.

Discussion

The results of this study established that the high affinity interactions between phospholipids and PRK1‐(HR1) are associated with substantial conformational changes, detected by both CD and tryptophan fluorescence. In CD analyses, a significant decrease of the secondary structural contents of PRK1‐(HR1) elicited by cardiolipin binding was observed [Fig. 3(a)]. This disruptive property seems related to the levels of activation of the PRK1 by phospholipids,2, 15 reflecting a possible correlation between its effectiveness in secondary structural disruption and enzyme activation. However, past study indicated that merely by the removal of PRK1‐(HR1) domain was insufficient to effect the maximal activation of PRK1.21 Because both PRK1‐(HR1) and ‐(HR2) contain pseudosubstrate motifs (Fig. 1), it was suggested that additional sites are present in HR2 domain for phospholipid (i.e., cardiolipin) interaction. Interestingly, it was demonstrated previously that the pseudosubstrate III motif of the PRK1‐(HR2) is a region for the binding of arachidonic acid.21 The substrate phosphorylation activity of rat liver PRK1 was increased by the presence of cardiolipin with two orders of magnitude greater than the addition of arachidonic acid.15 It is thus possible that cardiolipin could also target the region(s) within the HR2 domain to affect a similar disruptive effect as that was observed with the PRK1‐(HR1) domain.

The binding analyses by PRK1‐Trp13 fluorescence quenching identified the high affinity (electrostatic) and low affinity (hydrophobic) binding event under low and high salt condition, respectively (Table 1, Fig. 5). Overall, it demonstrated an amphipathic type of binding interactions took place between phospholipid and HR1 domain. Indeed, an inspection of the 3D structure of the PRK1‐(HR1a) subdomain revealed the high‐affinity cardiolipin‐binding sites mainly located on the Region 1 (Figs. 7 and 8, Table 4). The preferred cardiolipin‐binding site(s) identified by the Lys‐C protection assay could be the target site(s) for physiological regulation by either cardiolipin or a yet‐to‐be identified cardiolipin‐like lipid activator. The PRK1‐(HR1) domain could have also contained additional phospholipid‐binding sites (arginine‐based binding sites) that escaped detection in the Lys‐C‐based proteolytic protection assay. Notwithstanding this qualification, it is interesting that the cardiolipin‐protected Region 1 sequence implicated in high‐affinity cardiolipin‐binding overlaps the PRK1‐(50–77) sequence, which displays appreciable similarity with the defined cardiolipin‐binding sites in the mitochondrial cytochrome c oxidase subunit IV.3 In addition to Region 1, the cardiolipin‐binding sites were identified within the Regions 2–3 of HR1b and Regions 4–5 of HR1c subdomains. The primary sequence alignment analysis indicated that PRK1‐(HR1a) and ‐(HR1b) share a 30% of identity (59% similarity), suggesting that both HR1 subdomains could share comparable biological functions. However, a 3D structural superimposition analysis for HR1a and HR1b estimated that the r.m.s.d. value was 6.704 (>1), suggesting their respective structural arrangement in three dimension was quite distinct. In fact, the structural analyses for complexed HR1a_RhoA and HR1b_Rac1 revealed that the interfacial contacts of RhoA identified on HR1a are markedly varied with those of Rac1 on HR1b.19

Figure 8.

PRK1‐(HR1a) subdomain high affinity cardiolipin‐binding sites. The X‐ray crystallographic structure of PRK1‐(13‐98) recombinant protein (PDB code: 1CXZ)4 illustrates the proposed Region 1 PRK1‐(HR1a) cardiolipin‐binding sites, as indicated by the outlined residues. The two potential sites shown contain structural features of cardiolipin‐binding sites: (a) and (b), site 1, consisting of PRK1‐Lys43 and ‐Arg44 adjacent to the PRK1‐Leu70, ‐Leu77,‐Leu84 and Leu91; or (c) and (d), site 2, consisting of PRK1‐Arg47 and ‐Lys48 adjacent to PRK1‐Leu52, ‐Leu59 and ‐Leu66. The PRK1‐Trp13 residue is shown in yellow, arginine residues are in cyan, lysine residues are in purple, and hydrophobic leucine residues in green.

The highest binding affinity measured between PRK1‐(HR1) and phosphatidylcholine was unexpected as the extent to which the PRK1 activation by the presence phosphatidylcholine was not as potent as those of other lipids investigated (e.g., cardiolipin and phosphatidic acid).2, 15 Unlike cardiolipin and phosphatidic acid, phosphatidylcholine did not provide PRK1 with any protection on HR1a‐(1‐133) against endo Lys‐C digestion [Fig. 4(a)]. The location of the phosphatidylcholine binding site was thus suggested to be distinctive to those of acidic phospholipid with distances away from those pseudosubstrate motifs (Fig. 1). A primary structural comparison indicated that PRK1‐(HR1b)‐(127‐154) shared a 32% of identity to the sequence of a known phosphatidylcholine binding motif within the phosphatidylcholine transfer protein26 (Fig. S5). Since the binding affinities of PRK1‐(HR1) to phospholipid were at least two‐fold of magnitude greater than those to RhoA or Rac1,5, 19, 27 the activation of PRK1 is likely initiated mainly by the initial contacts established between HR1a‐b and the phospholipids of cell membrane. It was thus envisaged that the interactions between PRK1‐(HR1) and phosphatidylcholine or cardiolipin could take place in advance of the RhoA or Rac1 binding events. RhoA has been established as a GTPase protein that was required for the regulation of actin cytoskeleton organization.28 Interestingly, a homologous actin‐binding motif was identified within the HR1 domain of PRK1‐(50‐55).3 A related observation by Mukai et al. was that α‐actinin was phosphorylated upon its binding to PRK1 in a phosphatidylinositol 4,5‐bisphosphate (PIP₂)‐dependent manner.29 The PIP₂‐dependence of the reaction was suggested to be a consequence of the activation of PRK1 by the phospholipid.30 Furthermore, the expression of the kinase‐inactive PRK1 mutant disrupted the actin assembly in Rat1‐IR cells,31 indicating that RhoA GTPases regulated cytoskeleton organization through PRK1 to induce actin assembly and formation of stress fibers.28, 31 Further evidence for a role of PRK1 in cytoskeletal regulation is that the over‐expression of an inactivated mutant of the PRK1 kinase, phosphoinositide‐dependent protein kinase‐1 (PDK1) also disrupted actin cytoskeletal reorganization in cultured mammalian, indicating that PRK1 could function as a mediator of both RhoA‐GTPase and PI‐3 kinase signaling pathways.31

Cardiolipin is a potent activator of PRK1 in vitro 2, 15 and was found distributed predominantly in mitochondria.23 PRK1 has been found to phosphorylate a Mitochondrial Phospho Protein, MIPP65.32 However, the subcellular PRK1 was found mainly located in the cytosolic‐ or membrane‐bound fractions,17 and, there is no evidence for a role of mitochondrial cardiolipin in the regulation of PRK 1 activity, in vivo. Phosphatidic acid, which is related to cardiolipin, was regarded as a candidate regulator for the PRK1 given its effectiveness as an, in vitro, activator of the enzyme.2, 15 There have been several reports for the regulation of phosphatidic acid in response to stimulation of cells by anabolic regulations. For example, phospholipase D (PLD) was demonstrated to mediate hydrolysis of phosphatidylcholine to release phosphatidic acid in response to phorbol ester‐stimulation of HeLa cells.33 Rizzo et al. demonstrated a five‐fold increase in phosphatidic acid in Rat‐1 IR cells in response to insulin stimulation.34 It was thus suggested phosphatidic acid could act as a potential mediator of PLD‐dependent cell signaling pathway as well as physiological regulator of PRK1 via an activation mechanism similar to that of the cardiolipin. Further experiments seem warranted to elucidate the interplay between phospholipids and RhoA (or Rac1) in regulating the biological activity of PRK1.

Materials and Methods

Reagents

All reagents used were of analytical grade. The oligonucleotides were obtained commercially. Restriction enzymes, T4 DNA ligase, and DNA molecular mass markers (λDNA ladder from PstI digestion) were from Promega. endo Lys‐C (Lys‐C) were obtained from Boehringer Mannheim (Germany). Vent® DNA polymerase was from New England Biolab. Immunochemical and secondary antibody, an affinity‐purified horseradish peroxidase‐conjugated goat anti‐rabbit immunoglobulin, was purchased from Silenus Laboratory (Australia). The chemiluminescence reagents were from NEN Life Science. Pre‐stained protein markers were from Life Technology. Ni‐NTA metal affinity chromatography matrix was from QIAGEN (Germany). The anti‐PRK1/PKN‐(1‐150) antibody was obtained from Upstate. All lipids preparations were purchased from the Sigma and Aldrich. Cardiolipin (diphosphatidyl‐glycerol) from bovine heart; phosphatidic acid (1,2‐diacyl‐sn‐glycerol‐3‐phosphate) from egg yolk; phosphatidylserine (1,2‐diacyl‐sn‐glycerol‐3‐phospho‐l‐serine) from bovine brain; and phosphatidylcholine (1,2‐diacyl‐sn‐glycerol‐3‐phosphocholine) from egg yolk.

Recombinant constructs and bacterial expression

The cDNA sequence corresponding to rat liver PRK1‐(1‐313) was cloned into pET15b expression vector (Novagen). In‐frame DNA sequences were confirmed (ABI Prism™ Big Dye, Australian Genome Research Facility, Melbourne). Single colonies of pET15b‐PRK1‐HR1‐(1‐313) transformed E. coli strain BL21 (DE3) λ was used to express the recombinant proteins.

Ni‐NTA affinity purification of the PRK1‐(HR1)

The purification of PRK1‐(HR1) was carried out according to the protocol described previously.35 The Ni‐NTA‐affinity‐purified recombinant proteins were dialyzed against the buffers: 30 mM HEPES pH 7.4, 0.3M NaCl, 3 mM reduced glutathione, 5% (v/v) glycerol, 5 mM glycine, 0.005% (v/v) Tween 20, and 0.5 mM PMSF at 4°C for 16 h. The dialysates were desalted by further dialysis against buffers: 20 mM (Na₂HPO₄ + NaH₂PO₄), pH 7.4, 25 mM NaCl, and 0.5 mM PMSF at 4°C for 12 h. Protein concentration of the dialysate fraction was determined and the proteins were stored in –70°C until use.

Analytical Sephacryl‐200 size exclusion chromatography of the PRK1‐(HR1) protein

The PRK1‐(HR1) preparations were separately applied to a 25 mL Sephacryl 200 (S‐200) size exclusion column pre‐equilibrated with the column buffer: 20 mM (Na₂HPO₄ + NaH₂PO₄) pH 7.4, 150 mM NaCl. The column was calibrated with the following protein standards: cytochrome C (12,500 Da), soybean trypsin inhibitor (31,000 Da), ovalbumin (43,000 Da), and β‐galactosidase (116,000 Da).The protein elution profiles were monitored at A_280nm at 4°C. Protein concentrations for peak fractions collected were determined and proteins were stored in –70°C until use.

CD spectroscopy

The UV CD spectra of the PRK1‐(HR1) proteins were recorded at 25°C on an AVIV 62DS CD spectrophotometer (Lakewood, NJ). The CD spectra of the PRK1‐(HR1) protein were measured in the range, 195–250 nm, using a quartz cell with 1 mm light path. The spectra were recorded as the average of five scans for each recombinant protein sample after correction for the baseline contribution of buffer and phospholipid. The data were reported as the mean residue molar ellipticities ([θ]) or degree centimeter squared per mole values.

PRK1‐Trp¹³ emission fluorescence spectrometry

Intrinsic tryptophan fluorescence was measured using the LS‐5 luminescence (Perkin Elmer). The intrinsic fluorescence of the single tryptophan residue (Trp¹³) of PRK‐(HR1) was used to monitor the solution structures of the proteins. Spectra were obtained using an excitation wavelength of 296 nm and carried out emission scans at 0.5 nm intervals between 310 and 410 nm. The 2 mL quartz cuvette used in this experiment was pretreated with dichlorodimethylsilane to minimize the loss of protein due to adsorption to glass.

Preparation of the phospholipid vesicles

The phospholipids were dissolved in analytical grade methanol‐chloroform solvent and the phospholipid stock solutions stored at –20°C. Prior to use in binding studies, the aliquots of the phospholipid stock solutions were dried under a stream of N₂ gas and the phospholipids resuspended in buffer: 20 mM (NaH₂PO₄ + Na₂HPO₄) pH 7.4, 25 mM NaCl. The phospholipid vesicles were then prepared by sonication as described previously.36

Calculation of the lipid‐protein fractional saturation f_a and Kd value

The lipid‐induced protein tryptophan fluorescence quenching F/F _o value can be defined as:

$$\text{Quenching}\%=F/F_{\mathrm{o}}\times 100\left(\%\right)$$

where F is the intrinsic tryptophan fluorescence intensity after the addition of lipid and F _o is the intrinsic tryptophan fluorescence intensity before the addition of lipid. The recombinant PRK1‐(HR1) protein versus phospholipid‐binding curve (equilibrium type) was further analyzed using the nonlinear regression program (SigmaPlot 4.0®), where two linear segments were drawn based upon the best‐fit curve, to determine the lipid‐protein binding molar ratio.25 The lipid‐protein saturation values can be determined by a double reciprocal graphic analysis (F/F _o)⁻¹ versus (${\overline{C}}_S$)⁻¹.25 The fractional saturation and Kd values were calculated as described previously.25, 37

MALDI‐TOF MS

Protein samples were premixed with the matrix solution consisting of 10 mg/mL sinapinic acid (α‐cyano‐4‐hydroxycinnamic acid was used as matrix for cardiolipin preparations) in 70% acetonitrile and 0.05% (v/v) aqueous TFA solution. Aliquots (0.5–1 μL) of the matrix solutions were applied to the MALDI‐TOF target sample disks, followed by 0.5–1 μL of aliquots of the recombinant protein or their Lys‐C digests. The sample–matrix mixtures were then lyophilized. The instrument was operated in the positive ion‐linear mode with a mass accuracy error of <0.5% in the mass range of the polypeptides to be investigated. The mass values obtained in the linear mode, with calibration based on ribonulease A. The accelerating voltage used was in the range of 20–25 kV and a delayed extraction condition of 600–750 ns was applied to the sample analysis; the spectra obtained were the average of 100 laser shots.

Analysis of the sites of PRK1‐(HR1) and phospholipid interactions by limited protease digestion

The protein and lipid mixture was incubated at 36°C for 20 min prior to the addition of endo Lys‐C to the final concentration, 4.2 ng/μL. Protease digestions of PRK1‐(HR1) at 36°C, in the presence or absence of phospholipid, were continued for the times indicated in the figure legends of individual experiments; the digestions were stopped by placing the digest on the ice and adding leupeptin to the final concentration, 77 ng/μL (Auspep, Australia). Digests were analyzed by Coomassie blue and immunoblotting analyses. The mass analyses of the predicted Lys‐C fragments generated from the HR1 sequences were carried out using the “PeptideCutter” program as described.38

Immunoblotting analysis

Proteins were separated on 10%–20% SDS‐PAGE gels, and then transferred to nitrocellulose membranes at 4°C in the buffer: 25 mM Tris–HCl pH 7.6, 192 mM glycine, and 20% ethanol. The nitrocellulose membrane was then blocked with 5% (w/v) skim milk proteins in 1× TBS‐T buffers: 20 mM Tris–HCl pH 7.6, 137 mM NaCl, and 0.1% (v/v) Tween 20. The membrane was first probed with anti‐PRK1/PKN‐(1‐150) at room temperature. The membrane then washed with 1× TBS‐T wash buffer for three times followed by the addition of the secondary antibody horseradish peroxidase‐conjugated goat anti‐rabbit (Silenus). The immunoreactive bands were detected and developed by enhanced chemiluminescence reagents using Kodak BioMax™.

Supporting information

Appendix S1: Supporting Information