This entire article is copyrighted 1994, American Society of Plant Physiology, reproduced with explicit permission.
This edition was reproduced from OCR scans, so apologies to readers and authors for any missed errors, about which you may notify me by e-mail.
Click for References page.
A MAP by Any Other Name Would Still Bind to Microtubules
Recently, we published data to indicate that a homolog of elongation factor-l-alpha (EF-1a) interacts with microtubules (MTs), causing them to bundle in vitro (Durso and Cyr, 1994a). Furthermore, this in vitro interaction could be modulated by the addition of Ca2+ plus calmodulin (Ca2+/CaM). In his letter, Morejohn raises a number of questions about this work, as well as two general points that warrant a more general discussion. First, what defines a microtubule-associated protein (MAP)? Second, how should we experimentally examine and critically evaluate data regarding the interaction between soluble proteins and the cytoskeleton?
Numerous laboratories have been working on identifying microtubule-associated proteins (MAPs) in plants with the aim of understanding how these proteins affect the behavior of cellular MTs. Morejohn presents one opinion of how to define a MAP, namely as "proteins confirmed to be associated with MTs in cells fixed prior to their extraction. "This definition does not, however, represent an invariant standard used by all workers in the field. Moreover, the papers cited as substantiating this definition do not actually do so. Sherline and Schiavone (1977) and Sheterline (1978) do, in fact, utilize an immunocytochemical approach to localize MAPs to MTs, but they make no claim for this method being the definitive technique for MAP identification.
Defining "MAP" is problematic because it is a descriptive term which has evolved over the years. A recent review by Cleveland (1993) emphasizes the constant evolution of the definition and points out some of the historical pitfalls that have occurred as a consequence of adopting too narrow a definition for a MAP. Moreover, one of the most commonly cited review articles (Olmsted, 1986) on MAPs states that "MAPs [are] a collection varied molecules that have been defined on the basis of their binding and/or putative interaction with microtubules" (our emphases). The dogmatic application of only one set of criteria to define a MAP runs the risk of arbitrarily ranking proteins in importance. Currently, it is not uncommon to use the term MAP to describe any protein for which evidence exists that it associates with MTs. The evidence may be biochemical, immunocytochemical, or genetic. Of course, corroborative data using two or more approaches (like biochemistry and immunocytochemistry) provide more compelling evidence that a given protein functions in the cell to affect MT activity.
Although a large number of proteins have been classified as MAPs there are only two, kinesin and dynein, for which it is known with any degree of certainty how their presence affects the functioning of MTs in vivo. As an object lesson in the dangers of adopting an inflexible definition for a MAP, Cleveland (1993) points out that "kinesin ... among the most interesting microtubule related components, would fail to qualify under [the] early definition of MAPs." The ultimate challenge that confronts researchers in the field of cytoskeletal research is to discover the physiological significance of MAP function in vivo.
Regarding the question of how to study the interaction of soluble proteins with MTs, consider that MTs are dynamic elements that exist in equilibrium with soluble tubulin. Recent in vivo measurements indicate that the interphase cortical MTs of plant cells possess the most dynamic MTs reported for any eukaryote (Hush et al., 1994; Yuan et al., 1994). Presumably, any protein in association with MTs must likewise be in equilibrium between the soluble, cytosolic phase and the insoluble MT element. When a cell is ideally fixed for examination by immunofluorescence, all of its constituents are immobilized in their relative positions. However, the ability to detect any moiety at the subcellular level requires a high signal-to-noise ratio. When fixed cells are treated with antitubulin antibodies, diffuse cytoplasmic fluorescence (presumably from soluble tubulin) is observed along with brighter fluorescent MT elements. To increase the signal-to-noise ratio, most immunofluorescence protocols incorporate a detergent extraction step to remove the unpolymerized, soluble tubulin. Morejohn states strongly that the detergent step must be done after fixation; he does the reader a disservice by overstating facts and by inappropriately citing work in support of his opinion. There is one report (which is not cited by Morejohn) where the effects of detergent upon protein redistribution have been closely examined (Melan and Sluder, 1992). This work presents caveats but, in our opinion, does not negate the value of utilizing detergents for cytoskeletal immunolocalizations (either before or after fixation). As with any technique, immunolocalization has its limitations, and alternative methodologies (e.g., fluorescent analog cytochemistry) will certainly be useful in studying MAP/MT interactions. In the meantime, immunolocalization studies using properly controlled detergent application will undoubtedly continue to provide valuable information. Progress will not be served by halting all immunolocalization studies that employ detergents; instead, to aid in interpretation, data can be provided on the relative intensity of the fluorescing signal, the percentage of cells showing the reported localization patterns, and the number of different ways the cells can be prepared and still retain the reported staining pattern.
Among Morejohn's specific objections to our paper is our designation of the carrot EF-1a homolog, pp5O, as a MAP. Although we disagree with his narrow definition, his criteria are met for EF-1a by studies in sea urchin and by reports on a plant EF-1a homolog from tobacco cells. Our article explicitly discussed a tobacco homolog of EF-1a that localizes to MTs of cultured tobacco cells (Hasezawa and Nagata, 1993). We could not show this for carrot cells because the antibody used in the tobacco study was not available to us and because the antibody we did use for our protein gel blots was not appropriate for immunocytochemical localizations. The latter problem is not surprising, because it is inappropriate to posit a priori that an antibody that recognizes an epitope on a denatured protein transferred to a nitrocellulose membrane should also recognize, or have access to, the same epitope in cytological specimens. These considerations led us to begin raising our own antibodies against the carrot EF-1a (pp5O). Perhaps Morejohn's inability to localize the anti-wheat EF-1a antibody to the cytoskeleton (including actin) of maize cells warrants the same consideration, especially because an EF-I a homolog also localizes to actin in tobacco cells (Nagata et al., 1994).
Our article (Durso and Cyr, 1994a) reported experiments that have already examined Morejohn's concern that charge differences between tubulin and EF-1a lead to spurious associations in vitro. We reported that EF-1a does not bind to control columns made with BSA. Importantly, BSA and bovine tubulin have similar charges under physiologically relevant conditions (e.g., Durso and Cyr, 1994a).
Morejohn points out that tubulin and EF-1a have opposite charges and that EF-1a is an abundant cellular protein. Tubulin is also an abundant cellular protein (e.g., Cyr et al., 1987). Because there is no evidence (to our knowledge) for the separate intracellular compartmentalization of these two proteins, their electrostatic interaction within the cell is feasible. This is one reason we proposed (not concluded) that EF-1a is associated with MTs in vivo.
Throughout his critique, Morejohn alludes to the notion of specificity, a term with variable meaning, typically reflecting opinion, particularly when it is not defined. It is misleading to imply that electrostatic protein-tubulin interactions are "nonspecific," because even for MAPs that conform to Morejohn's criteria, such interactions are a typical basis for their interactions with tubulin (e.g., Wiche et al., 1991; Hugdahl et al., 1993).
Low ionic strength in our MT binding assays also concerns Morejohn. Our paper reported results of such assays atone salt concentration and the elution 6f tubulin binding proteins by a single 25-325 mM NaCl step (to obtain a partially purified protein subfraction that was used for CaM affinity chromatography). We did not report, but have performed, MT binding assays and tubulin affinity chromatographic analyses at various NaCl concentrations. These analyses indicate that the interaction of EF-I a with MTs/tubulin persists at high ionic strengths (e.g., above 150 mM NaCl).
Morejohn points out that the characterization of EF-1 &s interaction with CaM is in a very early stage; more extensive experiments are certainly required. Nevertheless, we believe that Morejohn's depiction of CaM's mode of action is too narrow. CaM is emerging as an unexpectedly and remarkably versatile protein. It has been described as "promiscuous": "Calmodulin counts its regulatory targets in double figures, and there is no consensus primary sequence in targets that bind CaM (Torok and Whitaker, 1994). So the meaning of Morejohn's phrase, "authentic CaM binding site, holds little meaning for us.
Although it is true that we did not perform solution binding experiments, Kaur and Ruben (1994) have found that, in solution, EF-1a can be reversibly crosslinked to Ca2+/CaM. Moreover, similar to what we found, these authors reported that both purified rabbit and trypanosome EF-1a bind to Ca2+/CaM that is immobilized on a chromatographic resin. They also found that EF-1a electrophoretically transferred to nitrocellulose binds biotinylated CaM in a Ca2+dependent manner. Morejohn's suggestion that the specificity of the interaction be examined using CaM antagonists is appreciated, but prominent authorities in CaM research warn that .results obtained by the application of these compounds to biological systems should be interpreted with caution" (Roberts et al., 1986; see also Roberts and Harmon, 1992).
It is unclear to us what else Morejohn might mean by "specific" vs. "nonspecific binding." Quite simply, we use "CaM binding protein" in the same way that many other investigators do. In any event, our method included measures against what we perceived as nonspecific interactions. Because CaM has amphipathic biochemical properties (e.g., it binds to both phenylsepharose and DEAE anion exchange resins), CaM affinity columns were (1) loaded with proteins in the presence of 0,3 M NaCl to reduce "nonspecific" ionic interactions and (2) washed with 1 mM CHAPS detergent to elute proteins weakly bound by hydrophobic interactions. Moreover, all steps prior to elution included plentiful free Ca2+,, the removal of which was required for elution of adsorbed proteins. To us, this protocol seems to be a rigorous test of specificity.
Morejohn points out the atypical stoichiometry for EF-1a's binding to MTs, but we question the assumption that the Stoichiometry of EF-l-alpha's binding to MTs should be similar to that of mammalian neuronal MAPs. In fact, it seems inappropriate to use neuronal MAPs as a basis of comparison with any other type of MAPs especially those from another kingdom because neuronal MAPs themselves are strikingly heterogeneous and, moreover. are obtained from what is arguably the most specialized eukaryotic cell (Olmsted, 1986; Wiche et al., 1991). Furthermore, neuronal MAPs are elongated, fibrous, flexible, and extensible proteins that appear as large as -350 kD and can be observed by electron microscopy to be 50-250 nm long (Wiche et al., 1991). They cover a considerable area on a MT's wall, leading to a substoichiometric ratio of, for example, 1 [neuronal MAP]: 6 [MT subunits). By contrast, EF-1a is a much smaller protein (-50 kD, or about half the MW of a MT subunit) that is a globular monomer in solution (e.g., Demma et al., 1990). Thus, a very different binding stoichiometry would not seem unreasonable.
Regarding the Ca2+/CaM modulation of EF-1a-induced bundles of taxol-stabilized MTs, Morejohn again highlights the need for more work. But some clarification is required. Morejohn claims that we used "an undetermined amount of EF-1a: but in fact we reported in the Methods that we used "0.17 mg mL-1 non-tubulin protein." Indeed, Morejohn could not have calculated the stoichiometry of EF-1a's binding to MTs without this figure.
Morejohn also argues that EF-l-alpha's effect on MT bundling could be due to nonspecific competition between CaM and tubulin for EF-1a binding or by Ca2+ mediated MT depolymerization. However, we reported that no effects on bundles were observed in the presence of equivalent concentrations of CaCl2 alone or CaM alone. In addition, we have elsewhere (Durso and Cyr, 1994b) shown a dark-field micrograph of EF-1a-induced carrot MT bundles in the presence of 0.5 mM Ca2+; that is, the bundles persist in the presence of Ca2++. These bundles were dissipated upon the addition of only 1.6 ELM CaM, conditions that would seem more agreeable to Morejohn.
We have argued here that the abundance of EF-1a and tubulin/MTs in the cytosolic compartment suggests that their interaction is likely. This thesis can easily accommodate CaM as well. The major portion of CaM activity is in soluble fractions of plant cells, and CaM associates with MT arrays in plant cells (reviewed in Roberts et al., 1986). For similar reasons, this thesis readily extends to glycolytic enzymes as well-the next point Morejohn erroneously discards.
Regarding the binding of glycolytic enzymes to MTs, Morejohn points out similarity to the EF-1a situation: Both EF-1a and glycolytic enzymes are abundant, basic, cytosolic proteins. Morejohn, however, then declares that these enzymes' "binding to MTs in vitro is probably an artifact, because GAPDH is not associated with MTs in cells" This is indeed what Balaban and Goldman (1990) and Bramblett et al. (1989) found, but their studies were done in trypanosomes, in which, as these authors make explicitly clear, glycolytic enzymes are located in membrane-bound organelles called glycosomes. So, in this exceptional case that is unique to the kinetoplastidae, the enzymes are not in the same cellular compartment as MTs.
In other organisms, by contrast, glycolytic enzymes are cytosolic, and indeed, the possible interactions between glycolytic enzymes and tubulin/MTs are being investigated in several laboratories (e.g., Lehotzky et al., 1993; Aithal et al., 1994). The reader should be aware, however, that these studies focus primarily on how the interactions affect glycolytic enzyme activities, and only secondarily on how the interactions regulate MT organization and activity (e.g., Knull and Walsh, 1992).
As Morejohn points out in his concluding paragraph, "research on the plant MT cytoskeleton" is accumulating at an "increased pace and growing complexity." He also points out that discoveries in our field require careful study, and indeed the possibility that EF-1a plays a significant role in the organization and regulation of MTs in plant cells is no exception. However, given the excitement about new discoveries in the field, it would seem premature to dismiss findings based on unsubstantiated opinions. Quite simply, as far as EFla and the cytoskeleton go, there is evidence that it associates with the major components of the cytoskeleton in plant cells. And, there are precedents in non-plant fields that also support this contention (see Durso and Cyr, 1994b). Moreover, a recent report (Shiina et al.,1994)indicates that EF-1a may harbor MT severing, as well as bundling, activity. Preliminary experiments indicate that the EF-1a homolog that we isolated does not sever MTs, either in vivo or in vitro, suggesting that plants may differentially regulate these activities. Morejohn's statement that MT severing is "antithetical" to MT bundling is unfounded because Shiina et al. (1994) report MT bundling, as well as severing, at low EF-1a concentrations. The authors also present immunolocalization data showing EF-1a to associate with unsevered MTs.
In summary, the aim of MT research is to understand how these essential elements function in the cell. This goal will require a more complete understanding of all molecules that interact with MTs and will entail numerous studies, from many laboratories, using a variety of approaches. The field is too young for the arbitrarily dismissal of facts. Critical evaluation of all data, with an open mind, is clearly in order and will undoubtedly lead to a greater understanding in this exciting area of plant cell biology.
Neil A. Durso, Richard J. Cyr
©1996 Neil A Durso, III
Free host = GeoCities