Taking a long, hard look at calmodulin's warm embrace

Proteolytic footprinting titrations for estimating ligand-binding constants and detecting pathways of conformational switching of calmodulin

To dissect the chemical basis for interactions controlling regulatory properties of macromolecular assemblies, it is essential to explore experimentally the linkage between ligand binding, conformational change, and subunit assembly. There are many advantages to using techniques that will probe the occupancy of individual binding sites or monitor conformational responses of individual residues, as described here. Proteolytic footprinting titrations may be used to infer binding free energies for ligands interacting with multiple sites or domains and to detect otherwise unrecognized "silent" interdomain interactions. Microgram quantities of pure protein are required, which is low relative to the hundreds of milligrams needed for comparable discontinuous equilibirum titrations monitored by NMR. By running comparative studies with several proteases, it is easy to determine whether resulting titration curves are consistent, independent of the protease used and therefore representative of the structural response of the protein to ligand binding or other differences in solution conditions (pH, salt, temperature). The results from multiple techniques (e.g., NMR, fluorescence, and footprinting) applied to aliquots from the same discontinuous titration may be compared easily to test for consistency. Classic methods for determining thermodynamic and kinetic properties of calcium binding to calmodulin include filter binding and equilibrium or flow dialysis (employing the isotope 45Ca), spectroscopic studies of stopped-flow fluorescence, calorimetry, and direct ion titrations. A cautionary note is that many different sets of microscopic data would be consistent with a single set of macroscopic constants determined by classic methods. This was well illustrated in Fig. 9. Thus, while it is important to compare results with those obtained by classic binding methods, they are, by definition, incapable of resolving the microscopic constants of interest. Thus, there is only one "direction" for comparison. Quantitative proteolytic footprinting titrations applied to studying calmodulin provided the first direct quantitative estimate of negative interactions between domains. Although studies of site-knockout mutants had suggested interactions between domains, this approach gave the first evidence for the pathway of anticooperative interactions between domains by showing that helix B responds structurally to calcium binding to sites III and IV in the C-domain. Despite two decades of study of calmodulin and the application of limited proteolysis studies to the apo and fully saturated forms, this finding emerged only when titration studies were undertaken as described. This highlights the general observation that while the behavior of the intermediate states in a cooperative switch are the key elements of the transition mechanism, they are the most difficult to observe. The unexpected finding that the isolated domains are nearly equivalent in their calcium-binding properties (Fig. 23 B) leaves us with many of the questions we had at the start: How does the sum of two nearly equivalent domains result in a molecule that switches sequentially rather than simultaneously? But it underscores why it is not yet possible to understand similar proteins by sequence gazing alone.

Regulation of a recombinant pea nuclear apyrase by calmodulin and casein kinase II

A cDNA encoding a pea nuclear apyrase was previously cloned. Overexpressions of a full-length and a truncated cDNA have been successfully expressed in Escherichia coli BL21(DE3). The resulting fusion proteins, apyrase and the C-terminus (residues 315-453) of apyrase, were used for calmodulin (CaM) binding and phosphorylation studies. Fusion protein apyrase but not the C-terminus of apyrase can be recognized by polyclonal antibody pc480. This suggested that the motif recognized by pc480 was located in the N-terminal region of apyrase. The recombinant apyrase protein also showed an activity 70 times higher than that of endogenous apyrase using ATP as a substrate. The recombinant apyrase has a preference for ATP more than other nucleoside triphosphate substrates. CaM can bind to recombinant apyrase, but not to the C-terminus of apyrase. This implies that the CaM-binding domain must be in the first 315 amino acids of the N-terminal region of apyrase. We found that one segment from residue 293 to 308 was a good candidate for the CaM-binding domain. This segment 293 FNKCKNTIRKALKLNY 308 has a basic amphiphilic-helical structure, which shows the predominance of basic residues on one side and hydrophobic residues on the other when displayed on a helical wheel plot. Using the gel mobility shift binding assay, this synthetic peptide was shown to bind to CaM, indicating that it is the CaM-binding domain. Both recombinant apyrase and the C-terminus of apyrase can be phosphorylated by a recombinant human protein kinase CKII. Phosphorylation does not affect CaM binding to recombinant apyrase. However, CaM does inhibit CKII phosphorylation of recombinant apyrase and this inhibition can be blocked by 5 mM EGTA.

Fluorescent tags of protein function in living cells

A cell's biochemistry is now known to be the biochemistry of molecular machines, that is, protein complexes that are assembled and dismantled in particular locations within the cell as needed. One important element in our understanding has been the ability to begin to see where proteins are in cells and what they are doing as they go about their business. Accordingly, there is now a strong impetus to discover new ways of looking at the workings of proteins in living cells. Although the use of fluorescent tags to track individual proteins in cells has a long history, the availability of laser-based confocal microscopes and the imaginative exploitation of the green fluorescent protein from jellyfish have provided new tools of great diversity and utility. It is now possible to watch a protein bind its substrate or its partners in real time and with submicron resolution within a single cell. The importance of processes of self-organisation represented by protein folding on the one hand and subcellular organelles on the other are well recognised. Self-organisation at the intermediate level of multimeric protein complexes is now open to inspection. BioEssays 22:180-187, 2000.

Dynamic redistribution of calmodulin in HeLa cells during cell division as revealed by a GFP-calmodulin fusion protein technique

It has been suggested by many studies that Ca2+ signaling plays an important role in regulating key steps in cell division. In order to study the down stream components of calcium signaling, we have fused the gene of calmodulin (CaM) with that of green fluorescent protein (GFP) and expressed it in HeLa cells. The GFP-CaM protein was found to have similar biochemical properties as the wild-type CaM, and its distribution was also similar to that of the endogenous CaM. Using this GFP-tagged CaM as a probe, we have conducted a detailed examination of the spatial- and temporal-dependent redistribution of calmodulin in living mammalian cells during cell division. Our major findings are: (1) high density of CaM was found to distribute in two sub-cellular locations during mitosis; one fraction was concentrated in the spindle poles, while the other was concentrated in the sub-membrane region around the cell. (2) The sub-membrane fraction of CaM became aggregated at the equatorial region where the cleavage furrow was about to form. The timing of this localized aggregation of CaM was closely associated with the onset of cytokinesis. (3) Using a TA-CaM probe, we found that the sub-membrane fraction of CaM near the cleavage furrow was selectively activated during cell division. (4) When we injected a CaM-specific inhibitory peptide into early anaphase cells, cytokinesis was either blocked or severely delayed. These findings suggest that, in addition to Ca2+ ion, CaM may represent a second signal that can also play an active role in determining the positioning and timing of the cleavage furrow formation.

Sequential assignment of 1H, 15N, 13C resonances and secondary structure of human calmodulin-like protein determined by NMR spectroscopy

Human calmodulin-like protein (CLP) is closely related to vertebrate calmodulin, yet its unique cell specific expression pattern, overlapping but divergent biochemical properties, and specific target proteins suggest that it is not an isoform of calmodulin. To gain insight into the structural differences that may underlie the difference target specificities and biochemical properties of CLP when compared to calmodulin, we determined the sequential backbone assignment and associated secondary structure of 144 out of the 148 residues of Ca2+-CLP by using multinuclear multidimensional NMR spectroscopy. Despite a very high overall degree of structural similarity between CLP and calmodulin, a number of significant differences were found mainly in the length of alpha-helices and in the central nonhelical flexible region. Interestingly, the regions of greatest primary sequence divergence between CLP and calmodulin in helices III and VIII displayed only minor secondary structure differences. The data suggest that the distinct differences in target specificity and biochemical properties of CLP and calmodulin result from the sum of several minor structural and side-chain changes spread over multiple domains in these proteins.

Imaging the spatial dynamics of calmodulin activation during mitosis

BACKGROUND

Calcium is an important and ubiquitous signalling ion. In most cell types, changes in intracellular calcium concentrations are sensed by calmodulin, a signal transduction protein that regulates cell function through its interactions with kinases and phosphatases. Calcium signals show complex spatiotemporal patterning, but little, if anything, is known about the patterns of calmodulin activation inside cells.

RESULTS

We have measured calmodulin activation continuously during mitosis in living cells with a new probe, a fluorescent adduct of calmodulin termed TA-calmodulin. We found that calmodulin was activated locally and episodically in the nucleus and mitotic spindle. The pattern of calmodulin activation was different from the pattern of calcium signals and could not be predicted from the pattern of calcium increase. Calmodulin activation was essential for mitotic progression: both entry into mitosis and exit from mitosis were blocked by a novel peptide that bound to calmodulin with high affinity and so prevented the interaction of calmodulin with its target proteins.

CONCLUSIONS

These data suggest that calmodulin regulates mitotic transitions and demonstrate the utility of fluorescent adducts for studying protein activation in living cells with good temporal and spatial resolution.

CALMODULIN AND CALMODULIN-BINDING PROTEINS IN PLANTS

Calmodulin is a small Ca2+-binding protein that acts to transduce second messenger signals into a wide array of cellular responses. Plant calmodulins share many structural and functional features with their homologs from animals and yeast, but the expression of multiple protein isoforms appears to be a distinctive feature of higher plants. Calmodulin acts by binding to short peptide sequences within target proteins, thereby inducing structural changes, which alters their activities in response to changes in intracellular Ca2+ concentration. The spectrum of plant calmodulin-binding proteins shares some overlap with that found in animals, but a growing number of calmodulin-regulated proteins in plants appear to be unique. Ca2+-binding and enzymatic activation properties of calmodulin are discussed emphasizing the functional linkages between these processes and the diverse pathways that are dependent on Ca2+ signaling.

Identification of novel human tumor cell-specific CaMK-II variants

CaMK-II (the (type II) multifunctional Ca2+/CaM-dependent protein kinase) has been implicated in diverse neuronal and non-neuronal functions, including cell growth control. CaMKII expression was evaluated in a variety of human tumor cell lines using RT-PCR (reverse transcriptase coupled polymerase chain reaction). PCR primers which flanked the CaMK-II variable domain were used so that all possible variants of the four mammalian CaMK-II genes (alpha, beta, gamma and delta) could be identified. 8 distinct CaMK-II isozymes were identified from human mammary tumor and neuroblastoma cell cDNA, each of which represented a variant of beta, gamma or delta CaMK-II. They included 2 beta isozymes (beta e, beta 'e), 4 gamma isozymes (gamma B, gamma C, gamma G, gamma H) and 2 delta isozymes (delta C, delta E) This is the first report of human beta and delta CaMK-II sequences. A panel of human cell types was then screened for these CaMK-II isozymes. As expected, cerebral cortex predominately expressed alpha, beta and delta A CaMK-II. In contrast, tumor cells, including those of neuronal origin, expressed an entirely different spectrum of CaMK-II isozymes than adult neuronal tissue. Tumor cells of diverse tissue origin uniformly lacked alpha CaMK-II and expressed 1-2 beta isozymes, at least 3 gamma isozymes and 1-2 delta isozymes. When compared to undifferentiated fibroblasts, beta e, beta'e, gamma G and gamma H were preferentially expressed in tumor cells. CaMK-II immunoblots also indicated that neuroblastoma and mammary tumor cells express isozymes of CaMK-II not present in their non-transformed cell or tissue counterpart. The identification of these new, potential tumor-specific CaMK-II variants supports previous indications that CaMK-II plays a role in growth control. In addition, these results provide insight into both splice variant switching and variable domain structural similarities among all CaMK-II isozymes.

Mutations which block the binding of calmodulin to Spc110p cause multiple mitotic defects

We have generated three temperature-sensitive alleles of SPC110, which encodes the 110 kDa component of the yeast spindle pole body (SPB). Each of these alleles carries point mutations within the calmodulin (CaM) binding site of Spc110p which affect CaM binding in vitro; two of the mutant proteins fail to bind CaM detectably (spc110-111, spc110-118) while binding to the third (spc110-124) is temperature-sensitive. All three alleles are suppressed to a greater or lesser extent by elevated dosage of the CaM gene (CMD1), suggesting that disruption of CaM binding is the primary defect in each instance. To determine the consequences on Spc110p function of loss of effective CaM binding, we have therefore examined in detail the progression of synchronous cultures through the cell division cycle at the restrictive temperature. In each case, cells replicate their DNA but then lose viability. In spc110-124, most cells duplicate and partially separate the SPBs but fail to generate a functional mitotic spindle, a phenotype which we term ‘abnormal metaphase’. Conversely, spc110-111 cells initially produce nuclear microtubules which appear well-organised but on entry into mitosis accumulate cells with ‘broken spindles’, where one SPB has become completely detached from the nuclear DNA. In both cases, the bulk of the cells suffer a lethal failure to segregate the DNA.

Discontinuous equilibrium titrations of cooperative calcium binding to calmodulin monitored by 1-D 1H-nuclear magnetic resonance spectroscopy

Calmodulin binds up to four calcium ions cooperatively in response to cellular signaling events. To understand the functional energetics of calcium activation of calmodulin, it is important to monitor individual Ca(2+)-binding sites and other positions at partial degrees of saturation. This study is the first use of 1-D proton NMR to monitor the equilibrium Ca(2+)-binding properties of calmodulin. Protein concentrations required for NMR experiments (approximately 1 mM) are approximately 1000-fold greater than the Kd values for calcium binding to calmodulin, preventing a direct continuous equilibrium titration of calmodulin. Thus, dialysates of calmodulin in buffers of experimentally determined [Ca2+]free were prepared to conduct discontinuous equilibrium titrations at both 92 and 152 mM KCl. For the C-terminal domain, the normalized area of the delta-protons of Y138 defined calcium binding isotherms. For N-terminal domain resonances (F16C delta H, T26C alpha H, D64C alpha H, and F65C delta H), the calcium-dependent change in chemical shift defined isotherms. These are the first residue-specific studies to monitor the energetics of Ca2+ binding to the N-terminal domain in wild-type holo calmodulin. Calcium binding to both domains appeared cooperative and binding affinity decreased in higher KCl. Isotherms resolved from the side chain resonances of F16 and F65 had a lower median ligand activity and a slightly higher degree of cooperativity than isotherms resolved from the backbone resonances of D64 and T26. Salt-dependent changes in apparent intradomain cooperativity differed for the domains: at higher salt, delta Gc increased for the C-terminal domain while remaining constant or decreasing for the N-terminal domain.

Activation-dependent and activation-independent localisation of calmodulin to the mitotic apparatus during the first cell cycle of the Lytechinus pictus embryo

We have used confocal microscopy and a fluorescent calmodulin probe to examine the mechanism of localisation of calmodulin during the first cell cycle of the sea urchin zygote. Using fluorescein-calmodulin, calmodulin can be observed within the nucleus and interphase astral microtubule arrays as cells approach mitosis. During mitosis, calmodulin redistributes to the mitotic apparatus and to condensed chromosomes. Quantitative analysis with reference to a control dye (fluorescein-dextran) shows that the distribution of calmodulin is specific. We used a competitive inhibitor of calcium-dependent calmodulin binding (Trp-peptide; Torok & Trentham (1994) Biochemistry 33, 12807-20) to test whether the cell cycle localisation of calmodulin was due to its binding to targets on activation. The Trp-peptide eliminates localisation of calmodulin within the nucleus. However, microtubule localisation persists in the presence of the Trp-peptide. These data show that calmodulin can localise by calcium (and hence activation)-dependent as well as calcium-independent mechanisms. This suggests that distinct mechanisms of localisation may be involved in the regulation of the differential functions of calmodulin, at least during the cell cycle.

Gain-of-function mutations in a human calmodulin-like protein identify residues critical for calmodulin action in yeast

A human epithelial cell-specific transcript (NB-1) encodes a calmodulin-like protein (hCLP), which is identical in length and 85% identical in amino acid sequence to authentic human calmodulin (hCaM). Although hCaM shares only 60% amino acid sequence identity with yeast calmodulin (CMD1 gene product), hCaM was able to substitute functionally for Cmd1 in yeast cells. In contrast, hCLP was unable to support either spore germination or vegetative growth in Cmd1-deficient yeast cells, even when stably expressed at a level at least an order of magnitude above that of hCaM. Thus, hCLP provides an indicator protein for discerning those residues that are critical for calmodulin function in vivo. In addition to 20 conservative amino acid replacements, hCLP differs from hCaM (and other vertebrate calmodulins that are able to complement a cmd1 null mutation) by only three nonconservative substitutions. Site-directed mutagenesis was used to convert these three positions back to residues more typical of those found in authentic calmodulins and to prepare all possible combinations of these three mutations, specifically: three single mutants (R58V, R112N, and A128E), three double mutants (R58V A128E, R112N A128E, and R58V R112N), and the triple mutant (R58V R112N A128E). The triple mutant and one of the double mutants (R58V A128E) were able to restore an apparently normal growth rate to a cmd1 delta strain, indicating that the altered hCLPs have acquired the ability to behave as functional calmodulins in yeast. The other two double mutants were able to support growth of Cmd1-deficient cells only weakly, but cells expressing the R112N A128E mutant grew noticeably better than those expressing the R58V R112N mutant. Remarkably, one single mutant (A128E), but not the other two single mutants, was also reproducibly able to support weak growth of a cmd1 delta strain. The properties of these gain-of-function, or neomorphic, mutations implicate E128, and to a lesser extent V58, as residues critical for calmodulin action in vivo. Molecular modeling of these positions within the structure of a Ca(2+)-calmodulin.peptide complex indicates that E128 projects directly into the central cavity occupied by the bound peptide. Thus, E128 may contribute a contact that is vital for the interaction of Cmd1 with one or more of the targets that are essential for yeast cell growth.

Nuclear calmodulin responds rapidly to calcium influx at the plasmalemma

We have studied the rate and extent of calcium binding to calmodulin in neuronal cytosol and nucleus during brief calcium influx across the plasmalemma. Rat sensory neurones were whole-cell patch clamped using a pipette containing a fluorescent analogue of calmodulin that reports when it has bound calcium. Cytosolic and nuclear signals were separated using a confocal microscope. Plasmalemmal calcium influx during a one second depolarization that activated L type calcium channels caused large fractions of calmodulin in both the cytosol and nucleus to bind calcium. Thus, contrary to previous predictions, nuclear processes that require the calcium:calmodulin complex will be activated readily by even brief cell stimulation.

Calmodulin: effects of cell stimuli and drugs on cellular activation

The activity, localization and cellular content of CaM can be regulated by drugs, hormones and neurotransmitters. Regulation of physiological responses of CaM can depend upon local Ca(2+)-entry domains in the cells and phosphorylation of CaM target proteins, which would either decrease responsiveness of CaM target enzymes or increase CaM availability for binding to other target proteins. Despite the abundance of CaM in many cells, persistent cellular activation by a variety of substances can lead to an increase in CaM, reflected both in the nucleus and other cellular compartments. Increases in CaM-binding proteins can accompany stimuli-induced increases in CaM. A role for CaM in vesicular or protein transport, cell morphology, secretion and other cytoskeletal processes is emerging through its binding to cytoskeletal proteins and myosins in addition to the more often investigated activation of target enzymes. More complete knowledge of the physiological regulation of CaM can lead to a greater understanding of its role in physiological processes and ways to alter its actions through pharmacology.