Introduction

Because of their property to promptly gain and lose electrons, copper (Cu) ions serve as catalytic centers of numerous proteins involved in a variety of enzymatic processes (1,2). Despite this essential role, Cu ions, when present in excess, can have detrimental effects because of their proclivity to engage in chemical reactions or by competing with other metal ions for enzyme-active sites (3-5). Therefore, specialized pathways have evolved for the signaling, transport, trafficking, and sequestration of Cu ions within cells to keep the delicate balance between essential and toxic levels (6-9). The importance of maintaining an exquisite Cu equilibrium is underscored by the existence of human inherited disorders of Cu homeostasis (Menkes and Wilson's diseases) and numerous forms of anemia linked to Cu deficiency (e.g., microcytic hypochromic anemia) (10,11). In Menkes disease, the export of Cu in extrahepatic tissues is defective, thereby rendering many Cu-dependent enzymes nonfunctional, which results in deleterious clinical manifestations (12,13). In Wilson's disease, Cu is not incorporated into key proteins in the liver, and the transport of Cu from the liver to the bile or alternative route is defective, thereby causing severe hepatotoxicity as well as brain and kidney tox-icity (12). As to anemia, it has been shown that iron (Fe) mobilization relies on the high-affinity Cu-uptake system because of the involvement of essential Cu-containing protein(s) in Fe uptake (10). Over the past few years, positional cloning and molecular genetics have culminated in the isolation of cDNAs corresponding to Menkes (MNK) (14-16) and Wilson's (WND) (17,18) disease genes. The mammalian gene involved in the development of microcytic hypochromic anemia (Heph) has also been found (19). These findings have given molecular biologists the information needed to identify yeast orthologs to these human disease-related proteins. Studies of yeast genes have uncovered additional yet unidentified loci encoding new components that function in Cu transport and other transition-metal-ion homeostatic pathways (6,8,9,20,21). In this chapter, we focus on recent advances that have been made with respect to molecular mechanisms of Cu transport, Cu detoxification, and regulatory response to Cu ions using the best features of two yeasts, Saccharomyces cerevisiae and Schizosaccharomyces pombe.*

*In this chapter, we used the nomenclature for Saccharomyces cerevisiae and Schizosaccharomyces pombe genes and proteins. S. cerevisiae wild-type genes are capitalized and italicized (e.g., CTR1). S. pombe wild-type genes are in regular text, italicized, with a superscripted "+" at the end (e.g., cuf1+). S. cerevisiae and S. pombe protein nomenclature is the same and is indicated with a capital letter at the beginning followed by regular letters (e.g., Ctrl, Cufl).

From: Handbook of Copper Pharmacology and Toxicology Edited by: E. J. Massaro © Humana Press Inc., Totowa, NJ

2. BUDDING YEAST AS A MODEL ORGANISM FOR EUKARYOTIC CELLS 2.1. Copper Acquisition

2.1.1. High-Affinity Cu-Ion Transport

Studies in bakers' yeast Saccharomyces cerevisiae have led to the identification of many components of the high-affinity Cu-transport pathway (Fig. 1). When grown under Cu-starvation conditions, following reduction of Cu2+ to Cu1+ by Fre plasma membrane Cu2+/Fe3+ ion reductases (22-27), Cu ions are transported into yeast cells by two separate high-affinity plasma membrane-transport proteins, Ctrl (28,29) and Ctr3 (30,31). The first high-affinity Cu transporter, denoted Ctrl, was identified using a genetic approach to identify proteins involved in Fe acquisition (29,32). This finding that Fe mobilization relies on the high-affinity Cu-transport system is explained largely by the involvement of a Cu-containing oxidase, Fet3, in Fe uptake (20,33). Therefore, ctrlA mutants (in which the CTR3 gene is silent) are defective in the transport of both Cu and Fe, whereas fet3A mutants are defective only in Fe acquisition. Furthermore, ctrlA mutants display other phenotypes linked to Cu deficiency, including the inability to grow on respiratory carbon sources (owing to the lack of Cu incorporation into cytochrome-c oxidase, which is essential for respiration), impaired superoxide dismutase (SOD) activity (owing to lack of Cu incorporation into Cu,Zn-Sod1), and poor cellular growth under conditions of Cu scarcity (28,29). Importantly, all of these phenotypes can be corrected by the addition of exogenous Cu ranging from 40 to 100 pM, which largely exceeds the Km of the high-affinity uptake system (28). The CTRl open reading frame (ORF) encodes a protein of 406 amino acids harboring 2 (possibly three) potential membrane-spanning regions (29). Its action in Cu uptake at the plasma membrane is highly specific for Cu over other metal ions with an apparent Km of 1-4 pM Cu (29). The amino terminus of Ctrl harbors eight copies of the sequence Met-X2-Met-X-Met, a putative extracellular Cu-binding motif found also in other proteins that participate in Cu transport (7,8,34,35). Furthermore, at the cell surface, the Ctr1 protein is modified, being glycosylated (28). Moreover, the active form of Ctr1 assembles as dimeric or multimeric complexes with itself (28). In response to Cu repletion, Ctr1 is regulated at multiple levels. In addition to the regulation at the level of gene transcription (see Section 2.1.4.), at least two other modes of regulation are known to occur at the protein level for Ctr1. When cells are exposed to Cu concentrations between 0.1 to 1.0 pM, the Ctr1 protein undergoes endocytosis (36). In response to Cu concentrations of 10 pM or more, the Ctr1 is specifically degraded (36). This proteolysis is independent of that internalization by endocytosis of Ctr1. In fact, cells defective in Ctr1 endocytosis because of mutations in the general endocytosis system (end3 or end4) or in general vacuolar hydrolysis (pep4) are still able to trigger Ctr1 degradation in response to elevated (>10 pM) Cu concentrations (36). Relatively little is understood about the molecular details of how Ctr1 undergoes two forms of posttranslational regulation as a function of changing environmental Cu levels.

Because of its ability to suppress all phenotypes linked with an inactivated CTRl gene, the CTR3-encoded high-affinity Cu-transport protein was isolated as a dominant extragenic suppressor (30). In several commonly used S. cerevisiae laboratory strains, CTR3 gene expression is inactivated by the presence of a Ty2 transposable element (30). When present, the transposon lies between the TATA box and the start sites for transcription, extinguishing CTR3 gene expression. Interestingly, S. cerevisiae strains that express both transporters (Ctr1 and Ctr3) exhibit a clear growth advantage under Cu-starvation conditions as compared to strains expressing only either CTRl or CTR3 (30). Although the Ctr3 protein supports high-affinity Cu transport in a manner similar to Ctr1, its primary structure is distinct. Ctr3 is a 241-amino-acid cysteine-rich integral plasma membrane protein. Furthermore, Ctr3 is predicted to possess three transmembrane domains, an extracellular hydrophilic amino-terminus, and a cytosolic tail (3l). Among the 11 cysteine residues found throughout the Cutransport protein, Cys-16 within the amino-terminal portion, Cys-48 and Cys-51 within the first trans-

CuREs

Fig. 1. Identified components for Cu transport in S. cerevisiae. When cells are grown during Cu scarcity, cell surface Fe3+/Cu2+ reductases (e.g. Frel) reduce Cu2+ to Cu1+ prior to uptake. Once reduced, Cu1+ is transported by two high-affinity Cu permeases, Ctrl (existing as a dimer or multimer at the plasma membrane) and Ctr3 (existing as a trimer at the plasma membrane). Within the cell, Cu is mobilized by specific chaperones. Atxl carries Cu to Ccc2 in a post-Golgi vesicle in which Cu is docked to Fet3. CCS delivers Cu to cytosolic Cu, Zn Sodl. Cox17 shuttles Cu to cytochrome-c oxidase, with the assistance of Scol and Coxll. In response to Cu-starvation conditions, the nuclear transcription factor Macl binds as a dimer to and activates the expression of high-affinity Cu-uptake genes including CTR1, CTR3, and FRE1/7, whereas in response to Cu-replete conditions, Macl is rapidly released from the Cu-response elements (CuREs).

membrane domain, and Cys-l99 found into the third transmembrane domain are necessary for Cu acquisition (31). Furthermore, these cysteine residues, except Cys-l99, are also critical for proper assembly of three Ctr3 molecules as trimer, which is the competent form of the transporter at the plasma membrane (31). Importantly, the carboxy-terminal region of Ctr3 from residues 40-233 represents a module, which is highly conserved among all eukaryotic cell-surface Cu transporters that have been identified so far, except for the S. cerevisiae Ctrl (8,31,37). Although both CTR1 and CTR3 genes are regulated at transcriptional level by cellular Cu status (see Section 2.l.4.), Ctr3, unlike Ctrl, is not regulated at the protein level either by endocytosis or degradation in the presence of high Cu concentrations (31). Currently, there is no information suggesting a functional interdependence between the two high-affinity Cu transporters, Ctrl and Ctr3, at the cell surface.

2.1.2. Cu Chaperones

Once inside the cell, free-Cu ions are virtually undetectable (38). Under Cu-limiting conditions, Cu ions are transiently associated to small Cu-binding proteins, denoted Cu chaperones, which possess the ability to distribute the transition metal as a cofactor to intracellular destinations (39-42). To date, three distinct Cu chaperones Atxl (43-45), CCS (also termed Lys7) (46), and Cox17 (47,48) have been identified and found to be involved in different Cu delivery pathways. Atxl carries Cu from the cytosol to the secretory pathway by specifically docking with Ccc2 (45,49-51), which is the MNK/WND disease yeast homologue protein (52). Subsequently, Ccc2 donates the bound Cu received from Atxl to the Cu-dependent ferroxidase Fet3 (33). Once Cu loaded, Fet3 initiates translocation in concert with another component, called Ftrl, which is an Fe permease (53). This Fet3/Ftrl complex proceeds through the secretory system to the plasma membrane. At the cell surface, Fet3 can convert Fe2+ to Fe3+ in a Cu-dependent oxidation reaction and then allows the transport of these Fe3+ ions across the membrane with the combined action of Ftrl (53). Therefore, Fe mobilization relies on the high-affinity Cu-uptake system. Although preliminary results in mammalian cells support this intimate connection between both high-affinity Fe and Cu transport systems, the molecular mechanisms are poorly characterized (54,55). Interestingly, it has been shown that GEF1 (56), encoding a putative voltage-regulated chloride channel, is required for Cu loading of Fet3 by Ccc2 within the secretory apparatus (57,58). Gefl would provide a counterbalancing charge effect that facilitates Cu transfer to the trans-Golgi network. Interestingly, other proteins were isolated from different organisms, including human (HAHl or ATOXl) (43,59), mouse (mAtxl) (60), rat (rAtoxl) (61), dog (Atoxl) (62), sheep (SAH) (63), Caenorhabditis elegans (Cucl) (64), and Arabidopsis thaliana (CCH) (65,66) that would possess orthologous function as the S. cerevisiae Atxl chaperone. Studies have provided insights into the mechanisms by which Atxl carries Cu to Ccc2 (50,51,67). Structural characterization has shown that the metal-binding motif, MXCXXC, can bind one Cu atom. This motif forms a secondary structure, called "open-faced P-sandwich," which prevents facile release of Cu yet allowing Cu exchange with a specific protein (38,41,68). Based on in vitro and two-hybrid results demonstrating a direct interaction between Atxl and its partner Ccc2, it has been proposed that Cu transfer occurs through the formation and decay of a series of two- and three- coordinate Cu ligand bonds in which Cys and ionized amino acids are involved (40,41,51). Although it is still unknown how the Atxl metallochaperone acquires its Cu, it is plausible that Atxl interacts with components of the high-affinity Cu-uptake machinery at the cell surface.

Another Cu chaperone, CCS, is able to directly activate cytosolic apo-Sodl under Cu-limiting conditions, which is consistent with the notion that CCS delivers Cu to Sodl in the metal-transfer process (69,70). CCS possesses three distinct domains, whereas only a single domain forms the Atxl and Coxl7 Cu chaperones. Although Domain I displays a strong similarity to Atxl, its substitution with Atxl does not confer the ability to the carboxyl-terminal CCS-Domain II/III region to deliver Cu to Sodl with an overall efficiency similar to that observed for the full-length CCS (70). It has been proposed that the first domain would serve during the process of Cu acquisition from a yet to be identified donor (41). Furthermore, this domain would be implicated to hold Cu with Domain III through cysteine-thiolate bonds. Domain II bears a striking percentage of homology to its target (Sodl). Furthermore, it has been shown that this domain interacts with Sodl to form either a heterodimeric or heterotetrameric complex (71,72). This complex formation between the two proteins is necessary to allow Cu to be released from CCS and inserted into Sodl (73,74). The carboxyl-terminal 30 amino acids of CCS (Domain III) is critical for activating the Sodl enzyme because of its essential role in the release of Cu from the chaperone (70,75). Interestingly, human and mouse homologs of the yeast CCS chaperone have been identified (46,76). Importantly, it has been demonstrated that mice lacking the CCS gene show dramatic reductions in scavenging toxic reactive oxygen species, thereby corroborating with the notion that CCS is required for intracellular Cu delivering to Sodl enzyme for making it active (76).

The Cox17 Cu chaperone shuttles between the cytosol and the intermitochondrial membrane space, where it provides Cu to cytochrome- c oxidase, which is the multisubunit terminal enzyme in the respiration chain (47,48). Like the yeast Atxl and CCS Cu chaperones for which homologs have been identified in humans, a human homolog of Cox17 has been found by functional complementation of a yeast cox17Anull mutant (77,78). In yeast, genetic evidence suggests that at least two intermediates, Coxll and Scol, are also required at the assembly site for the insertion of Cu from Cox17 into the cytochrome-c oxidase (79-81). Although Coxl7 is a very small protein of similar size to Atxl, they share no homology at the amino-acid sequence level. Of the seven cysteine residues found in Coxl7, only mutations in three of these residues affect the ability of the protein to bind Cu (82). Indeed, these three cysteine residues (Cys23-Cys24-X-Cys26) form a Cu-binding motif and have been shown to coordinate three Cu ions with another residue that has yet to be identified (82). Interestingly, the Cu clusters of Coxl7 are similar to those in the Cu metalloregulatory transcription factor Acel and the MT Cupl in being pH stable and luminescent (83). Although Cu-metallated Coxl7 exhibits reminiscent biophysical properties to CuAcel and CuCupl, the susceptibility of its Cu clusters to ligand-exchange reactions is much higher. Coxl7 is detected as a dimer in the cytosol, whereas the protein is found as a tetramer when isolated from mitochondria (83). This differential oligomeric assembly of Coxl7 may play a role in its ability to translocate from the cytosol to the mitochondrial intermembrane space. Clearly, the mechanism by which a chaperone delivers Cu ions appears to be associated, in part, with the nature of the target protein and its intracellular location.

2.1.3. Low-Affinity Cu-Ion Uptake

Under conditions of Cu sufficiency, a low-affinity Cu uptake system is active (29,84). Although this latter system exhibits an apparent Km of 30-40 pM Cu, the specific components involved in such activity remain to be clarified (6). Fet4 (85,86), Ctr2 (87), and Smf (88,89) proteins have been reported to contribute in Cu-ion transport with low-affinity across the plasma membrane. Although these transmembrane proteins may play a role in Cu uptake, the loci encoding these proteins are not expressed according to Cu need (90). Recently, the Fet4 protein has been studied with more details with respect to Cu transport (91). In fact, low-affinity Fe-uptake mediated by Fet4 is competitively inhibited by Cu ions, suggesting that Cu may represent an additional substrate for this low-affinity Fe transporter. Moreover, the level of Cu transport in ctrlA TY2::ctr3 fet4A cells is severely reduced (91). Metal-ion-uptake kinetic measurements revealed that Fet4 supports Cu uptake with an apparent Km value of 35 pM Cu (91). Furthermore, the Cu taken into the cell by Fet4 requires Cu reduction by the cell-surface metalloreductases Frel and Fre2, which are multimembrane-spanning NADPH- and FAD-binding proteins able to solubilize Fe3+ and Cu2+ to ferrous iron (Fe2+) and cuprous copper (Cul+), respectively (26). It is thought that such reduction prior to uptake renders ferric and cupric ions from the environment more bioavailable for the transporters at the cell surface (92). Although Fet4 appears to transport Cu with low affinity, one caveat to these results is the fact that Fet4 was overexpressed for most of the experiments performed; thus, the involvement of the endogenous protein in Cu uptake may be different. The CTR2 gene from S. cerevisiae was isolated based on homology to a putative plant Cu transporter gene, denoted COPT1 (87). Sequence comparison analyses revealed that Ctr2 shares sequence similarities to other cell-surface Cu transporters such as Ctrl and Ctr3 from S. cerevisiae (8), Ctr4 and Ctr5 from S. pombe (37,93), and hCtrl from human (94-97), thereby suggesting a role for Ctr2 in Cu acquisition. Moreover, overexpression of Ctr2 in yeast is linked with an increased sensitivity to Cu toxicity. Conversely, an S. cerevisiae ctr2A strain is more resistant when Cu is in excess of physiological requirements (87). Whether Ctr2 is involved in Cu accumulation is still unclear. As structurally related Cu transporters do exist from different organisms, the elucidation of the Ctr2 function should provide additional information on how Cu is mobilized. Recently, based on indirect observations, Smf proteins have been considered to participate in Cu uptake (88). In S.

cerevisiae, these proteins that belong to the Nramp transporter family is composed of three members: Smfl, Smf2, and Smf3 (98). When cells are starved for manganese, Smfl is localized at the plasma membrane, whereas Smf2 is detected to intracellular vesicles. Under manganese-replete conditions, both proteins move to the vacuole (98). Importantly, Cu ions, unlike manganese ions, are unable to trigger such trafficking of Smfl and Smf2. Interestingly, Smf3 is found to the vacuole membrane (98). Furthermore, only Fe can negatively regulate Smf3 at the protein level, whereas no other metal ions including manganese, copper, and zinc are capable of such regulation. Although Smf proteins can potentially recognize numerous metal ions, their biological specificity may come from their regulation, which would minimize their contribution in Cu transport.

2.1.4. Transcriptional Regulation of the High-Affinity Cu-Ion-Transport Genes

A hallmark of the genes-encoding components of the high-affinity Cu-uptake machinery including CTR1, CTR3, and FRE1/7 is the fact that they are transcriptionally expressed according to Cu need (8). The transcription of CTR1, CTR3, and FRE1/7 is induced under Cu-starvation conditions, whereas inactivation of the expression of these genes occurs under Cu-replete conditions (27,99,100). This regulation is mediated by cis-acting promoter elements, denoted CuREs (Cu-response elements) with the consensus sequence 5'-TTTGC(T/G)C(A/G)-3' (27,99). The presence of two copies of CuRE in each of the CTR1, CTR3, and FRE1/7 promoters is necessary for Cu repression and Cu-starvation activation of gene expression (27,99,101). The CuREs are arranged as either inverted or direct repeats (27,99). Furthermore, the center-to-center distances between CuREs observed for each promoter predict that they lie on opposite faces of the DNA (99). The transcription factor for regulating the expression of genes-encoding components involved in Cu transport through the CuREs is Macl (102). The MAC1 gene was initially discovered as a dominant allele, called MAC1up1, that fosters robust expression of the high-affinity Cu-uptake genes even in the presence of elevated Cu concentrations (99,101,102). As would be expected, mac1A cells display a number of phenotypes linked to defective Cu transport, including poor growth on low Cu medium, inability to grow on nonfermentable carbon sources, and a deficient Cu,Zn-Sod1 activity with a concomitant sensitivity to redox cycling drugs, which can all be biochemically ascribed to Cu insufficiency because exogenous Cu restores the normal phenotype (100,102). Macl is a 417-amino-acid protein harboring several basic residues within its amino-terminal region (l-20l), whereas its carboxyl-terminal region encompassing amino acids 202-4l7 exhibits more overall acidic residues (103-105). Precisely, the Macl amino-terminal l59 residues constitute the DNA-binding domain of the protein (106). Within this domain, the first 40 residues of Macl exhibit a strong homology to the minor groove DNA-binding domain found in the Acel and Amtl Cu metalloregulatory transcription factors that activate MT gene expression, but very little homology outside of this region (102,107,108). The carboxyl-terminal region of Macl harbors two Cys-rich repeats, denoted REP-I (also identified as Cl) and REP-II (also identified as C2) (103,109). The REP-I motif, Cys264-X-Cys266-X4-Cys27l-X-Cys273-X2-Cys276-X2-His279, is essential for Cu-sensing because substitutions created in either all or individual conserved Cys and His exhibit elevated and unregulated Cu transporter gene expression (100,109,110). Regarding the REP-II motif, Cys322-X-Cys324-X4-Cys329-X-Cys33l-X2-Cys334-X2-His337, its partial or complete disruption alters Macl ability of trans-activating target gene expression (103,109,110). Based on immunofluorescence studies, the Macl protein harbors two potential nuclear localization signals (105). The first one resides in a region between residues 70 and 287, and the second one was found in the carboxyl-terminal l28 amino acids of Macl, which contain the REP-II motif. Currently, it is unclear how these nuclear localization signals function. Under low-Cu conditions, it is known that Macl binds to CuREs as a dimer (101,105) by making contacts in both major [with 5'-GC(T/C)C(A/G)-3' sequence] and minor (with 5'-TTT-3' sequence) grooves (104). Interestingly, it has been shown that Cul+ ions can bind directly to Macl (111). Under elevated Cu concentrations, Macl is released from the CuREs in vivo (99) and it is suggested that the presence of Cu ions trigger intramolecular conformational changes to inactivate the Macl transactivation domain (105,111). Furthermore, at Cu concentrations above 10 |M, the Macl protein is degraded, therefore ensuring a complete deactivation of the Cu-transport machinery under conditions of Cu excess (109).

2.2 Copper Detoxification

Yeast genes-encoding proteins that function to prevent the accumulation of Cu to toxic levels into cells are transcriptionally regulated by Cu in the opposite direction from the high-affinity Cu-uptake genes. The transcription of CUP1- and CRS5-encoded MTs and SOD1-encoded Cu,Zn-Sod1 are induced in response to Cu excess (>10 |M) (90,112,113). MTs are known to counteract metal cytotoxicity by sequestering avidly the excess of Cu (7). Regarding Sodl, the enzyme prevents Cu-medi-ated damages by competing for the surplus of metal ions, thereby acting as a buffer for Cu (114). Increased synthesis of these three proteins in response to Cu are controlled at the transcriptional level by the Acel Cu metalloregulatory transcription factor (MRTF) (also denoted Cup2) (115-117). Furthermore, the promoter element necessary for Cu-inducible transcription of the CUP1, CRS5, and SOD1 genes is denoted either MRE (metal-regulatory element) or UASCu (Cu-responsive upstream activation sequence) and is composed of the consensus sequence, 5'-HTHNNGCTGD-3' (D = A, G or T; H = A, C or T; N = any residue), whereas the GCTG is termed the core sequence, and the region 5' to the core is called the T-rich element (108,118,119). The CUP1 promoter harbors five MREs and is strongly induced in response to Cu, whereas the CRS5 and SOD1 promoters contain a single MRE and are only modestly induced by Acel (7,120). The core sequence 5'-GCTG-3' of the MRE is recognized by Acel in the major groove, whereas the AT-rich 5' region is contacted in the minor groove (118). Although the Acel Cu MRTF was found to bind DNA as a monomer, it has been shown that the protein possesses a bipartite DNA-binding domain, explaining its ability to make contacts with both the minor and major grooves (108). The first subdomain (residues l-40) of the DNA-binding domain of Acel was found to bind a single Zn2+ atom through the Cysn, Cysl4, Cys23, and His25 residues (121). Interestingly, at the carboxyl-terminal side of this amino-terminal 40-residue segment, called the Zn module, a motif denoted (K/R)GRP was identified for AT-rich minor-groove interaction (118). The second subdomain of the DNA-binding domain (residues 4l-ll0), named Cu-regulatory domain (108), harbors nine Cys residues found as follows: one Cys-X2-Cys motif, three Cys-X-Cys motifs, and a single Cys residue. All of these Cys residues are necessary for Cu response, except for the last Cysl05 (108). The arrangement of the Cys residues is predicted to coordinate four Cu*+ atoms through cysteine sulfur bonds (122). Upon Cu activation, Acel undergoes a conforma-tional change due to the formation of a tetra-Cu cluster within the amino terminal Cu-regulatory domain via critical cysteinyl thiolates to make an active detoxifying factor (7,108,123). This enables the cupro-Acel protein to interact with the MRES to rapidly foster transactivation of target gene expression via its carboxyl-terminal domain, which is highly negatively charged (124). The discovery of the C. glabrata ACE1 gene ortholog, called AMT1, has also contributed to validate a number of properties predicted for Acel with regard to Cu sensing (125). Although at the level of the Amtl primary structure some differences exist with Acel, the Zn module and the Cu-regulatory domain in the amino-terminal region of both proteins are highly conserved (126,127).

0 0

Post a comment