The NMR solution structure of subunit G (G61-101) of the eukaryotic V1VO ATPase from Saccharomyces cerevisiae

Subunit G is an essential stalk subunit of the eukaryotic proton pump V₁V_O ATPase. Previously the structure of the N-terminal region, G_1-59, of the 13 kDa subunit G was solved at higher resolution. Here solution NMR was performed to determine the structure of the recombinant C-terminal region (G_61-101) of subunit G of the Saccharomyces cerevisiae V₁V_O ATPase. The protein forms an extended α-helix between residues 64 and 100, whereby the first five- and the last residues of G_61-101 are flexible. The surface charge distribution of G_61-101 reveals an amphiphilic character at the C-terminus due to positive and negative charge distribution at one side and a hydrophobic surface on the opposite side of the structure. The hydrophobic surface pattern is mainly formed by alanine residues. The alanine residues 72, 74 and 81 were exchanged by a single cysteine in the entire subunit G. Cysteines at positions 72 and 81 showed disulfide formation. In contrast, no crosslink could be formed for the mutant Ala74Cys. Together with the recently determined NMR solution structure of G_1-59, the presented solution structure of G_61-101 enabled us to present a first structural model of the entire subunit G of the S. cerevisiae V₁V_O ATPase.     

Low resolution structure of subunit b (b (22-156)) of Escherichia coli F(1)F(O) ATP synthase in solution and the b-delta assembly

The first low resolution solution structure of the soluble domain of subunit b (b _22–156) of the Escherichia coli F₁F_O ATPsynthase was determined from small-angle X-ray scattering data. The dimeric protein has a boomerang-like shape with a total length of 16.2 ± 0.3 nm. Fluorescence correlation spectroscopy (FCS) shows that the protein binds effectively to the subunit δ, confirming their described neighborhood. Using the recombinant C-terminal domain (δ_91–177) of subunit δ and the C-terminal peptides of subunit b, b _120–140 and b _140–156, FCS titration experiments were performed to assign the segments involved in δ–b assembly. These data identify the very C-terminal tail b _140–156 to interact with δ_91–177. The novel 3D structure of this peptide has been determined by NMR spectroscopy. The molecule adopts a stable helix formation in solution with a flexible tail between amino acid 140 to 145.     

Assembly of subunit d (Vma6p) and G (Vma10p) and the NMR solution structure of subunit G (G1-59) of the Saccharomyces cerevisiae V1VO ATPase

Understanding the structural traits of subunit G is essential, as it is needed for V₁V_O assembly and function. Here solution NMR of the recombinant N- (G_1-59) and C-terminal segment (G_61-114) of subunit G, has been performed in the absence and presence of subunit d of the yeast V-ATPase. The data show that G does bind to subunit d via its N-terminal part, G_1-59 only. The residues of G_1-59 involved in d binding are Gly7 to Lys34. The structure of G_1-59 has been solved, revealing an α-helix between residues 10 and 56, whereby the first nine- and the last three residues of G_1-59 are flexible. The surface charge distribution of G_1-59 reveals an amphiphilic character at the N-terminus due to positive and negative charge distribution at one side and a hydrophobic surface on the opposite side of the structure. The C-terminus exhibits a strip of negative residues. The data imply that G_1-59-d assembly is accomplished by hydrophobic interactions and salt-bridges of the polar residues. Based on the recently determined NMR structure of segment E_18-38 of subunit E of yeast V-ATPase and the presently solved structure of G_1-59, both proteins have been docked and binding epitopes have been analyzed.     

Structural and functional analysis of the coupling subunit F in solution and topological arrangement of the stalk domains of the methanogenic A1AO ATP synthase

The first low-resolution shape of subunit F of the A₁A_O ATP synthase from the archaeon Methanosarcina mazei Gö1 in solution was determined by small angle X-ray scattering. Independent to the concentration used, the protein is monomeric and has an elongated shape, divided in a main globular part with a length of about 4.5 nm, and a hook-like domain of about 3.0 nm in length. The subunit-subunit interaction of subunit F inside the A₁A_O ATP synthase in the presence of 1-ethyl-3-(dimethylaminopropyl)-carbodiimide EDC was studied as a function of nucleotide binding, demonstrating movements of subunits F relative to the nucleotide-binding subunit B. Furthermore, in the intact A₁A_O complex, crosslinking of subunits D-E, A-H and A-B-D was obtained and the peptides, involved, were analyzed by MALDI-TOF mass spectrometry. Based on these data the surface of contact of B-F could be mapped in the high-resolution structure of subunit B of the A₁A_O ATP synthase.     

Interaction between subunit C (Vma5p) of the yeast vacuolar ATPase and the stalk of the C-depleted V1 ATPase from Manduca sexta midgut

Projection maps of a V₁-Vma5p hybrid complex, composed of subunit C (Vma5p) of Saccharomyces cerevisiae V-ATPase and the C-depleted V₁ from Manduca sexta, were determined from single particle electron microscopy. V₁-Vma5p consists of a headpiece and an elongated wedgelike stalk with a 2.1×3.0 nm protuberance and a 9.5×7.5 globular domain, interpreted to include Vma5p. The interaction face of Vma5p in V₁ was explored by chemical modification experiments.     

The <Emphasis Type="Italic">b</Emphasis><Subscript>arg36</Subscript> contributes to efficient coupling in F<Subscript>1</Subscript>F<Subscript>O</Subscript> ATP synthase in <Emphasis Type="Italic">Escherichia coli</Emphasis>

In Escherichia coli, the F₁F_O ATP synthase b subunits house a conserved arginine in the tether domain at position 36 where the subunit emerges from the membrane. Previous experiments showed that substitution of isoleucine or glutamate result in a loss of enzyme activity. Double mutants have been constructed in an attempt to achieve an intragenic suppressor of the b _arg36→ile and the b _arg36→glu mutations. The b _arg36→ile mutation could not be suppressed. In contrast, the phenotypic defect resulting from the b _arg36→glu mutation was largely suppressed in the b _{arg36→glu,glu39→arg} double mutant. E. coli expressing the b _{arg36→glu,glu39→arg} subunit grew well on succinate-based medium. F₁F_O ATP synthase complexes were more efficiently assembled and ATP driven proton pumping activity was improved. The evidence suggests that efficient coupling in F₁F_O ATP synthase is dependent upon a basic amino acid located at the base of the peripheral stalk.     

The vacuolar (H+)-ATPase: subunit arrangement and in vivo regulation

The V-ATPases are responsible for acidification of intracellular compartments and proton transport across the plasma membrane. They play an important role in both normal processes, such as membrane traffic, protein degradation, urinary acidification, and bone resorption, as well as various disease processes, such as viral infection, toxin killing, osteoporosis, and tumor metastasis. V-ATPases contain a peripheral domain (V₁) that carries out ATP hydrolysis and an integral domain (V₀) responsible for proton transport. V-ATPases operate by a rotary mechanism involving both a central rotary stalk and a peripheral stalk that serves as a stator. Cysteine-mediated cross-linking has been used to localize subunits within the V-ATPase complex and to investigate the helical interactions between subunits within the integral V₀ domain. An essential property of the V-ATPases is the ability to regulate their activity in vivo. An important mechanism of regulating V-ATPase activity is reversible dissociation of the complex into its component V₁ and V₀ domains. The dependence of reversible dissociation on subunit isoforms and cellular environment has been investigated. Qi and Wang contributed equally to this work.     

Spectroscopical identification of residues of subunit G of the yeast V-ATPase in its connection with subunit E

A critical point in the V₁ sector and entire V₁V_O complex is the interaction of stalk subunits G (Vma10p) and E (Vma4p). Previous work, using precipitation assays, has shown that both subunits form a complex. In this work, we have analysed the N-terminal segment of subunit G (G_1–59) of the V₁V_O ATPase from Saccharomyces cerevisiae by using nuclear magnetic resonance (NMR) spectroscopy. Analyses of ¹H-¹⁵N heteronuclear single quantum coherence (HSQC) spectra of G_1–59 in the absence and presence of the N-terminal peptides E_1–18 and E_18–38 as well as the produced and purified C-terminal segment (E_39–233) shows specific interactions only with the peptide fragment E_18–38. The binding of this peptide occurs via the residues M₁, V₂, S₃, and K₅ as well for V₂₂, S₂₃, K₂₄, A₂₅ and R₂₆ of G_1–59. The specific E_18–38/G_1–59 binding has been confirmed by fluorescence correlation spectroscopy data. The E_18–38 peptide has been studied by CD spectroscopy and NMR. The 3D structure of this peptide adopts a stable helix-hinge-helix formation in solution. A model structure of the E_18–38/G_1–59 complex reveals the orientation of E_18–38 relative to G_1–59 via salt-bridges of the polar residues and van der Waals forces at the very N-terminus of both segments.     

Microbial transformation of ginsenoside Rb<Subscript>1</Subscript> by <Emphasis Type="Italic">Acremonium strictum</Emphasis>

Preparative-scale fermentation of ginsenoside Rb₁ (1) with Acremonium strictum AS 3.2058 gave three new compounds, 12β-hydroxydammar-3-one-20 (S)-O-β-d-glucopyranoside (7), 12β, 25-dihydroxydammar-(E)-20(22)-ene-3-O-β-d-glucopyranosyl-(1→2)-β-d-glucopyranoside (8), and 12β, 20 (R), 25-trihydroxydammar-3-O-β-d-glucopyranosyl-(1→2)-β-d-glucopyranoside (9), along with five known compounds, ginsenoside Rd (2), gypenoside XVII (3), ginsenoside Rg₃ (4), ginsenoside F₂ (5), and compound K (6). The structural elucidation of these metabolites was based primarily on one- and two-dimensional nuclear magnetic resonance and high-resolution electron spray ionization mass spectra analyses. Among these compounds, 2–6 are also the metabolites of ginsenoside Rb₁ in mammals. This result demonstrated that microbial culture parallels mammalian metabolism; therefore, A. strictum might be a useful tool for generating mammalian metabolites of related analogs of ginsenosides for complete structural identification and for further use in pharmaceutical research in this series of compounds. In addition, the biotransformation kinetics was also investigated.     

Introduction: A Close Look at the Vacuolar ATPase

The vacuolar ATPases (V-type ATPases) are a family of ATP-dependent ion pumps and found in two principal locations, in endomembranes and in plasma membranes. This family of ATPases is responsible for acidification of intracellulare compartments and, in certain cases, ion transport across the plasma membrane of eucaryotic cells. V-ATPases are composed of two distinct domains: a catalytic V₁ sector, in which ATP hydrolysis takes place, and the membrane-embedded sector, V₀, which functions in ion conduction. In the past decade impressive progress has been made in elucidating the properties structure, function and moleculare biology. These knowledge sheds light also on the evolution of V-ATPases and their related families of A-(A₁A₀-ATPase) and F-type (F₁F₀-ATPases)ATPases.     

The adenine pocket of a single catalytic site is derivatized when the bovine heart mitochondrial F1-ATPase is photoinactivated with 4-amino-1-octylquinaldinium

The bovine heart mitochondrial F₁-ATPase (MF₁) is reversibly inhibited in the dark by 4-amino-1-octylquinaldinium (AOQ) with an I_0.5 value of 48 μM. When irradiated in the presence of AOQ, MF₁ is photoinactivated with an apparent K_d of 12 μM. About 1.1 mol of [³H]AOQ were incorporated per mol of MF₁ on complete photoinactivation. Fractionation of a cyanogen bromide digest of MF₁ photolabeled with [³H]AOQ followed by fractionation of peptic digests of partially purified cyanogen bromide fragments led to isolation of two CNBr/peptic fragments labeled with³H. Sequence analysis of the labeled peptides revealed that one contained residues 423–441 of the β subunit. A gap in position 2 of the sequence indicates that βPhe424 is derivatized. The phenyl side-chain of this residue is part of a pocket that binds the adenine moiety of ATP or ADP at catalytic sites. The other peptide, which was labeled to a greater extent, contained residues 342–358 of the β subunit, but in this case, no gap was found in the sequence indicating that the derivatized amino-acid side-chain might not have survived the conditions of automatic Edman degradation. This peptide contains βTyr345, the side-chain of which is also a component of the pocket that binds the adenine moiety of ATP or ADP to catalytic sites. However, for the reason stated, there is no direct evidence that βTyr345 is labeled in this peptide.     

Effects of different excitation and detection spectral regions on room temperature chlorophyll <Emphasis Type="Italic">a</Emphasis> fluorescence parameters

The effects of different spectral region of excitation and detection of chlorophyll (Chl) a fluorescence at room temperature on the estimation of excitation energy utilization within photosystem (PS) 2 were studied in wild-type barley (Hordeum vulgare L. cv. Bonus) and its Chl b-less mutant chlorina f2 grown under low and high irradiances [100 and 1 000 μmol(photon) m⁻² s⁻¹]. Three measuring spectral regimes were applied using a PAM 101 fluorometer: (1) excitation in the red region (maximum at the wavelength of 649 nm) and detection in the far-red region beyond 710 nm, (2) excitation in the blue region (maximum at the wavelength of 461 nm) and detection beyond 710 nm, and (3) excitation in the blue region and detection in the red region (660– 710 nm). Non-photochemical quenching of maximal (NPQ) and minimal fluorescence (SV₀), determined by detecting Chl a fluorescence beyond 710 nm, were significantly higher for blue excitation as compared to red excitation. We suggest that this results from higher non-radiative dissipation of absorbed excitation energy within light-harvesting complexes of PS2 (LHC2) due to preferential excitation of LHC2 by blue radiation and from the lower contribution of PS1 emission to the detected fluorescence in the case of blue excitation. Detection of Chl a fluorescence originating preferentially from PS2 (i.e. in the range of 660–710 nm) led to pronounced increase of NPQ, SV₀, and the PS2 photochemical efficiencies (F_V/F_M and F_V′/F_M′), indicating considerable underestimation of these parameters using the standard set-up of PAM 101. Hence PS1 contribution to the minimal fluorescence level in the irradiance-adapted state may reach up to about 80 %.     

ATP Synthases With Novel Rotor Subunits: New Insights into Structure,Function and Evolution of ATPases

ATPases with unusual membrane-embedded rotor subunits were found in both F₁F₀ and A₁A₀ ATP synthases. The rotor subunit c of A₁A₀ ATPases is, in most cases, similar to subunit c from F₀. Surprisingly, multiplied c subunits with four, six, or even 26 transmembrane spans have been found in some archaea and these multiplication events were sometimes accompanied by loss of the ion-translocating group. Nevertheless, these enzymes are still active as ATP synthases. A duplicated c subunit with only one ion-translocating group was found along with “normal” F₀ c subunits in the Na⁺ F₁F₀ ATP synthase of the bacterium Acetobacterium woodii. These extraordinary features and exceptional structural and functional variability in the rotor of ATP synthases may have arisen as an adaptation to different cellular needs and the extreme physicochemical conditions in the early history of life.     

Expression,purification and characterization of pectate lyase A from <Emphasis Type="Italic">Aspergillus nidulans</Emphasis> in <Emphasis Type="Italic">Escherichia coli</Emphasis>

Pectate lyase A (PelA) of Aspergillus nidulans was successfully expressed in Escherichia coli and effectively purified using a Ni²⁺-nitrilotriacetate-agarose column. Enzyme activity of the recombinant PelA could reach 360 U ml⁻¹ medium. The expressed PelA exhibited its optimum level of activity over the range of pH 7.5–10 at 50°C. Mn²⁺, Ca²⁺, Fe²⁺, Mg²⁺ and Fe³⁺ ions stimulated the pectate lyase activity, but Cu²⁺ and Zn²⁺ inhibited it. The recombinant PelA had a V _max of 77 μmol min⁻¹ mg⁻¹ and an apparent K _m of 0.50 mg ml⁻¹ for polygalacturonic acid. Low-esterified pectin was the optimum substrate for the PelA, whereas higher-esterified pectin was hardly cleaved by it. PelA efficiently macerated mung bean hypocotyls and potato tuber tissues into single cells.     

Effect of meal size on postprandial metabolic response in Chinese catfish (Silurus asotus Linnaeus)

The effect of relative meal size (0.5–24% body mass) on specific dynamic action (SDA) was assessed in Chinese catfish (Silurus asotus Linnaeus) (30.90±1.30 g) at 25.0°C; the cutlets of freshly killed loach without viscera, head and tail were used as a test meal. There was no significant difference in either SDA duration or peak oxygen consumption (VO₂) among low meal size ranges. But both increased linearly as meal size increased from 2 to 24% without reaching a plateau. Factorial metabolic scope was 5.92 in fish fed with 24% body mass, the highest documented feeding metabolic scope value in fish till now. The Peak VO₂ of satiated meal size groups (175.85±10.55 mg O₂ h⁻¹) was above 80% of maximum metabolic rate during locomotion recovery process (215.48±7.07 mg O₂ h⁻¹). The relationship between energy expended on SDA (E) and energy ingested (I) was described as: E=0.0000432I ²+0.140I+2.12. The lowest value of SDA coefficient appeared at 2% body mass group.     

Cloning a Truncated Fragment (stx2a<Subscript>1</Subscript>) of the Shiga-Like Toxin 2A<Subscript>1</Subscript> Subunit of EHEC O157:H7: Candidate Immunogen for a Subunit Vaccine

Shiga toxins consist of enzymatically active A and B subunit multimers. The A subunit of shiga-like toxins can be proteolytically cleaved into two parts, A₁ and A₂, with A₁ being responsible for toxic activity. Antibody neutralizing the A₁ subunit of shiga toxin may protect against infection of the enterohemorrhagic Escherichia coli (EHEC O157:H7). It was difficult to express the full-length A₁ subunit of shiga toxin 2 (stx2A₁) in a previous study. We have now analyzed the full-length of stx2A₁ using bioinformatics software. The data show that the carboxyl terminal (of ~15 amino-acid residues) has strong hydrophobicity and low antigenicity. We cloned and expressed a truncated fragment of stx2A₁ (15 amino-acid residues of the carboxyl terminal being removed), designated stx2a₁, which can evoke a humoral immune response. Anti-Stx2a₁ antibodies can neutralize the native shiga toxin 2 both in vivo and in vitro, which suggests that Stx2a₁ serves as a candidate immunogen for a subunit vaccine that can also be used as the antigen to screen phage anti-shiga toxin antibody libraries. L. Liu and H. Zeng contributed equally to this study.     

Analysis of the expressed heavy chain variable-region genes of Macaca fascicularis and isolation of monoclonal antibodies specific for the Ebola virus' soluble glycoprotein

The cynomolgus macaque, Macaca fascicularis, is frequently used in immunological and other biomedical research as a model for man; understanding it's antibody repertoire is, therefore, of fundamental interest. The expressed variable-region gene repertoire of a single M. fascicularis, which was immune to the Ebola virus, was studied. Using 5′ rapid amplification of cDNA ends with immunoglobulin (Ig)G-specific primers, we obtained 30 clones encoding full-length variable, diversity, and joining domains. Similar to the human V_H repertoire, the M. fascicularis repertoire utilized numerous immunoglobulin heavy variable (IGHV) gene fragments, with the V_H3 (41%), V_H4 (39%), and V_H1 (14%) subgroups used more frequently than the V_H5 (3.9%) or V_H7 (1.7%) subgroups. Diverse immunoglobulin heavy joining (IGHJ) fragments also appeared to be utilized, including a putative homolog of JH5β gene segment identified in the related species Macaca mulatta, Rhesus macaque, but not in humans. Although the diverse V region genes in the IgG antibody repertoire of M. fascicularis had likely undergone somatic hypermutations (SHMs), they nevertheless showed high nucleotide identity with the corresponding human germline genes, 80–89% for IGHV and 72–92% for IGHJ. M. fascicularis and human V_H genes were also similar in other aspects: length of complementarity-determining regions and framework regions, and distribution of consensus sites for SHMs. Finally, we demonstrated that monoclonal antibodies (mAbs) specific for an Ebola protein could be obtained from M. fascicularis tissue samples by phage display technology. In summary, the study provides new insight into the M. fascicularis V region gene repertoire and further supports the idea that macaque-derived mAbs may be of therapeutic value to humans.     

Singlet oxygen formation and chlorophyll a triplet excited state deactivation in the cytochrome b6f complex from Bryopsis?corticulans

We have attempted to investigate the correlation between the detergent-perturbed structural integrity of the Cyt b ₆ f complex from the marine green alga Bryopsis corticulans and its photo-protective properties, for which the nonionic detergents n-octyl-β-d-glucopyranoside (β-OG) and n-dodecyl-β-d-maltoside (β-DM), respectively, were used for the preparation of Cyt b ₆ f, and the singlet oxygen (¹O₂*) production as well as the triplet excited-state chlorophyll a (³Chl a*) formation and deactivation were examined by spectroscopic means. Near-infrared luminescence of ¹O₂ * (~1,270 nm) on photo-irradiation was detected for the β-OG preparation where the complex is mainly in oligomeric state, but not for the β-DM one in which the complex exists in dimeric form. Under anaerobic condition, photo-excitation of Chl a in the β-DM preparation generated ³Chl a* with a lower quantum yield of Φ_T ~ 0.02 and a longer lifetime of ~600 μs with respect to those as in the case of β-OG preparation, Φ_T ~ 0.12 and 200–300 μs. These results prove that the enzymatically active and intact Cyt b ₆ f complex on photo-excitation tends to produce little ³Chl a* or ¹O₂ *, which implies that the pigment–protein assembly of Cyt b ₆ f complex per se is crucial for photo-protection. F. Ma and X.-B. Chen contributed equally to this work.     

Deriving room temperature excitation spectra for photosystem I and photosystem II fluorescence in intact leaves from the dependence of <Emphasis Type="Italic">F</Emphasis><Subscript>V</Subscript>/<Emphasis Type="Italic">F</Emphasis><Subscript>M</Subscript> on excitation wavelength

The F ₀ and F _M level fluorescence from a wild-type barley, a Chl b-less mutant barley, and a maize leaf was determined from 430 to 685 nm at 10 nm intervals using pulse amplitude-modulated (PAM) fluorimetry. Variable wavelengths of the pulsed excitation light were achieved by passing the broadband emission of a Xe flash lamp through a birefringent tunable optical filter. For the three leaf types, spectra of F _V/F _M (=(F _M − F ₀)/F _M) have been derived: within each of the three spectra of F _V/F _M, statistically meaningful variations were detected. Also, at distinct wavelength regions, the F _V/F _M differed significantly between leaf types. From spectra of F _V/F _M, excitation spectra of PS I and PS II fluorescence were calculated using a model that considers PS I fluorescence to be constant but variable PS II fluorescence. The photosystem spectra suggest that LHC II absorption results in high values of F _V/F _M between 470 and 490 nm in the two wild-type leaves but the absence of LHC II in the Chl b-less mutant barley leaf decreases the F _V/F _M at these wavelengths. All three leaves exhibited low values of F _V/F _M around 520 nm which was tentatively ascribed to light absorption by PS I-associated carotenoids. In the 550–650 nm region, the F _V/F _M in the maize leaf was lower than in the barley wild-type leaf which is explained with higher light absorption by PS I in maize, which is a NADP-ME C₄ species, than in barley, a C₃ species. Finally, low values of F _V/F _M at 685 in maize leaf and in the Chl b-less mutant barley leaf are in agreement with preferential PS I absorption at this wavelength. The potential use of spectra of the F _V/F _M ratio to derive information on spectral absorption properties of PS I and PS II is discussed.     

The α2δ subunit augments functional expression and modifies the pharmacology of CaV1.3 L-type channels

The auxiliary Ca_Vα₂δ-1 subunit is an important component of voltage-gated Ca²⁺ (Ca_V) channel complexes in many tissues and of great interest as a drug target. Nevertheless, its exact role in specific cell functions is still unknown. This is particularly important in the case of the neuronal L-type Ca_V channels where these proteins play a key role in the secretion of neurotransmitters and hormones, gene expression, and the activation of other ion channels. Therefore, using a combined approach of patch-clamp recordings and molecular biology, we studied the role of the Ca_Vα₂δ-1 subunit on the functional expression and the pharmacology of recombinant L-type Ca_V1.3 channels in HEK-293 cells. Co-expression of Ca_Vα₂δ-1 significantly increased macroscopic currents and conferred the Ca_V1.3α₁/Ca_Vβ₃ channels sensitivity to the antiepileptic/analgesic drugs gabapentin and AdGABA. In contrast, Ca_Vα₂δ-1 subunits harboring point mutations in N-glycosylation consensus sequences or the proteolytic site as well as in conserved cysteines in the transmembrane δ domain of the protein, reduced functionality in terms of enhancement of Ca_V1.3α₁/Ca_Vβ₃ currents. In addition, co-expression of the δ domain drastically inhibited macroscopic currents through recombinant Ca_V1.3 channels possibly by affecting channel synthesis. Together these results provide several lines of evidence that the Ca_Vα₂δ-1 auxiliary subunit may interact with Ca_V1.3 channels and regulate their functional expression.