Coordination of posttranslational modifications of bovine brain α-tubulin: Polyglycylation of Δ2 tubulin

Microtubules participate in a large number of intracellular events including cell division, intracellular transport and secretion, axonal transport, and maintenance of cell morphology. They are composed of tubulin, a heterodimeric protein, consisting of two similar polypeptides α and β. In mammalian cells, both α- and β-tubulin occur as seven to eight different genetic variants, which also undergo numerous posttranslational modifications that include tyrosination-detyrosination and deglutamylation, phosphorylation, acetylation, polyglutamylation, and polyglycylation. Tyrosination-detyrosination is one of the major posttranslational modifications in which the C-terminal tyrosine residue in α-tubulin is added or removed reversibly. Although this modification does not alter the assembly activity of tubulin in vitro, these two forms of tubulin have been found to be distributed differently in vivo and are also correlated with microtubule stability (Gunderson, G. G., Kalnoski, M. H., and Bulinski, J. C. (1984) Cell 38, 779–789). Thus, the question arises as to whether these two forms of tubulin differ in any other modifications. In an effort to answer this question, the tyrosinated and the nontyrosinated forms of the α1/2 isoform have been purified from brain tubulin by immunoaffinity chromatography. matrix-assisted laser desorption/ionization-time of flight mass spectrometric analysis of the C-terminal peptide revealed that the tyrosinated form is polyglutamylated with one to four Glu residues, while the Δ2 tubulin is polyglycylated with one to three Gly residues. These results indicate that posttranslational modifications of tubulin are correlated with each other and that polyglutamylation and polyglycylation of tubulin may have important roles in regulating microtubule assembly, stability, and function in vivo.

In addition to the existence of different genetic variations, ␣and ␤-tubulin also undergo a number of covalent modifications that include tyrosination-detyrosination (34 -42) and deglutamylation of the ␣-tubulin C terminus or the formation of ⌬2 tubulin (␣-tubulin lacking both the Glu and the Tyr residues from the C terminus) (43) and acetylation at Lys 40 (44,45); ␤ III -tubulin undergoes phosphorylation at a serine residue (46); both ␣and ␤-tubulin also undergo polyglutamylation and polyglycylation, in which glutamyl or glycyl units are attached as side chains through the ␥-carboxyl of a Glu residue near the C terminus (46 -56). Polyglycylation of tubulin has hitherto been observed in flagellar and ciliary microtubules (48), and could conceivably play a role in the unusual stability and morphology of these microtubules. Except for acetylation and phosphorylation, most of these covalent modifications occur near the C termini of ␣and ␤-tubulin.
It is not known exactly how different posttranslational modifications affect the function of tubulin and microtubules in vivo. Although tyrosinated and nontyrosinated tubulin do not appear to differ in the assembly (36,40), these two forms of tubulin are known to be distributed differently in the interphase microtubules and are believed to form separate populations of microtubules. Microtubules enriched in detyrosinated tubulin appear to be much more stable than those enriched in tyrosinated tubulin (42). Thus, it seems possible, that the tyrosinated and the nontyrosinated forms of tubulin may differ in other covalent modifications that may give rise to different subsets of microtubules in vivo.
In an effort to study how different covalent modifications may contribute to the structure and function of a tubulin isoform, tubulin dimers were separated on an immunoaffinity column that contained a monoclonal antibody to tyrosinated ␣1/2 tubulin (56). My previous results demonstrated that the tubulin dimers containing tyrosinated form of ␣1/2 assemble poorly in the presence of glycerol and Mg 2ϩ . In this manuscript, I have analyzed the covalent modifications of tyrosinated and nontyrosinated ␣1/2 by MALDI-TOF 1 mass spectrometry. The results show that tyrosinated ␣1/2 is polyglutamylated with one to four Glu residues, the tetraglutamylated form being the predominant one. On the other hand, the ⌬2 form of ␣1/2 that assembles normally is polyglycylated with one to three glycyl units, the biglycylated form being the major one. These results indicate that there is a correlation between tyrosination/detyrosination and deglutamylation of tubulin on one hand and polyglutamylation/polyglycylation on the other hand. This correlation among posttranslational modifications of tubulin is a novel finding. Since tyrosination/detyrosination appears to play a role in microtubule stability, these results raise the possibility that polyglycylation and polyglutamylation may also be involved in microtubule stability and function. These results also indicate that polyglycylation is not restricted to tubulin destined to form axonemal microtubules.
Preparation of Tubulin and Microtubule-associated Proteins-Microtubules were isolated from bovine brain cortex by one cycle of assembly and disassembly as described before (17). Tubulin was purified from microtubules by phosphocellulose chromatography. MAP2 and tau were purified from microtubule proteins by ultrogel chromatography. All purifications were carried out in Mes-Na (pH 6.4), 1 mM EGTA, 0.1 mM EDTA, 0.5 mM MgCl 2 , and 1.0 mM GTP.
Preparation of ␣-Tubulin Isoforms-Different ␣-tubulin isoforms were purified by immunoaffinity chromatography using an anti-␣ column as previously described (56). Briefly, tubulin was first passed through the immunoaffinity column containing the monoclonal antibody AYN.6D10. The unbound fraction (Fraction A) was pooled and concentrated in 8 M glycerol and was frozen. After elution of the unbound fraction, the column was washed, and the bound fraction was eluted with a gradient of 0 -1 M NaCl. Of the two peaks, the first is designated as Fraction B, while the second peak is Fraction C. All these fractions were desalted immediately after the elution, dialyzed against 8 M glycerol in assembly buffer and were stored frozen at -80°C. Prior to an experiment, glycerol was removed from the protein samples by repeated desalting in Centricon spin columns.
Purification of ␣-Tubulin-Separation of ␣-tubulin was performed as described previously (56). Briefly, tubulin subunits were first separated on preparative SDS-PAGE, and the bands were visualized by staining with Coomassie Brilliant Blue, the ␣-tubulin band was excised, and tubulin was electroeluted. Contaminating SDS was removed from the eluted protein by acetone precipitation.
HPLC Separation of ␣-Tubulin C-terminal Peptide-Purified ␣-tubulin in 0.1 M Tris-HCl (pH 9.2) was digested with endoproteinase Lys-C (17 g/ml) at 30°C for 12-16 h. The digested protein was subjected to reversed-phase HPLC using a Phenomenex Jupiter C18 column (4.6 ϫ 250 mm) as described previously (56). Solvent A was 0.1% trifluoroacetic acid in water, and Solvent B was 0.085% trifluoroacetic acid in 70% acetonitrile. Chromatography was performed using a gradient of 0 -60% solvent B in 60 min at a flow rate of 0.5 ml/min. The C-terminal peptide was identified by sequencing the first five amino acid residues. Sequence analysis was performed as described previously (56).
MALDI-TOF Mass Spectrometry-MALDI-TOF analysis was performed at the Protein Core Facility, Department of Biochemistry, Albert Einstein College of Medicine, Bronx, New York. HPLC-purified C-terminal peptides were diluted in a saturated solution of ␣-cyano-4hydroxycinnamic acid in 50% acetonitrile/water and were spotted directly. The samples were examined by MALDI-TOF in a positive and linear mode. Spectra were obtained using a Voyager RP Biospectrometry Work Station (Perseptive Biosystems, Framingham, MA), using a 300-nm laser at an accelerating voltage of 28 kV. Internal calibration was performed using an ACTH fragment obtained from Sigma. About 50 -120 laser scans were performed for each spectrum.
Other Methods-Amino acid analysis and protein sequencing were performed at the Protein Core Facility of the Department of Biochemistry as described before (56). Protein measurements, SDS-PAGE, and immunoblotting were performed as described before (56).

Separation of ␣-Tubulin
Fractions-Bovine brain tubulin contains two ␣-tubulin isoforms, ␣1/2 and ␣4. Using an anti-␣tubulin immunoaffinity column I have previously fractionated bovine brain tubulin into three functionally active forms, fractions A, B, and C (56). Immunoblot analysis showed that this fraction is not recognized by the monoclonal antibodies to tyrosinated tubulin. Sequence analyses have shown that Fraction A contains nontyrosinated forms of ␣1/2 that may include ⌬2 tubulin. Fraction B is a mixture of the nontyrosinated forms of ␣1/2 and ␣4. Fraction C is essentially the tyrosinated form of ␣1/2. Since fractions A and C are primarily the nontyrosinated and the tyrosinated forms of ␣1/2, these constituted an appropriate system for structural comparisons. Thus, it was interesting to study the posttranslational modifications on these two fractions by MALDI-TOF analyses.
Mass Spectrometric Analysis of the C-terminal Peptides of the ␣-Tubulin Isoforms-Bovine brain tubulin contains two ␣-tubulin isoforms, ␣1/2 and ␣4, and each of the isoforms have three posttranslational variations, namely the tyrosinated (Tyr form), the detyrosinated (Glu form), and the ⌬2 form (lacking both the Tyr and Glu residues from the C terminus). The expected m/z values for the protonated forms (MH ϩ ) are given in Table I.
To study the posttranslational status of the purified tubulin fractions, the C-terminal peptide was isolated by preparative SDS-PAGE, followed by digestion with endoproteinase Lys-C as previously described (56). The C-terminal peptide was separated by reversed phase HPLC on a C18 column and was identified by sequencing the N-terminal five residues (DYEEV) since this sequence does not exist anywhere other than the C-terminal. The fractions that were confirmed to have the C-terminal sequence were examined by MALDI-TOF. Fig. 1A shows the HPLC profile of the endoprotease Lys-Cdigested Fraction A. Sequence analysis shows that the C-terminal peptide is present in two peaks: a major peak a followed by a minor peak b (A). The peaks a and b were both found to contain the sequence DYEEVGVDSVEGEGEEEGE, which corresponds to the ⌬2 form of ␣1/2. The underlining of the Glu residues (Glu 445 , Glu 447 , and Glu 449 ) signifies that signals for those residues in the sequencing chromatogram were poor, indicating that the residues might be modified by posttranslational modification. MALDI-TOF analysis revealed that peak a exhibits a single major peak at m/z value of 2170 and several minor peaks at 2227.8, 2322.66, 2387.29, 2451.7, and 2516.22 (Fig. 1C). Peak b exhibits peaks at 2113.82, 2149, 2170, and 2227.8. The peaks at 2113 and 2170 differ by 57, which is the mass value of a glycyl unit (2170Ϫ2113 ϭ 57), and the peaks at  The analysis for tyrosinated ␣1/2 is summarized in Fig. 2. As shown in Fig. 2A, the C-terminal peptide eluted as a single peak. Sequence analysis showed that this peptide contains the sequence DYEEVGVDSVEGEGEEEGEEY, which is the sequence of the tyrosinated form of ␣1/2. Here also the signals for the underlined Glu residues (Glu 445 , Glu 447 , Glu 449 , and FIG. 1. MALDI-TOF analysis of the C-terminal peptide of nontyrosinated tubulin. A, HPLC purification of the C-terminal peptide from nontyrosinated tubulin (unbound fraction). Bovine brain PC-tubulin was passed through an immunoaffinity column containing covalently bound anti-␣-tubulin monoclonal antibody AYN.6D10. The unbound fraction was pooled and subjected to SDS-PAGE on a preparative gel to separate ␣and ␤-tubulin. The ␣-tubulin band was cut out from the stained gel, and the protein was electroeluted and purified by acetone precipitation. 15 g of ␣-tubulin was digested with endoproteinase Lys-C at room temperature for 16 h. The digest was subjected to reversed phase HPLC on a C-18 column. The peaks were sequenced for the first five residues, and the C-terminal peptides were identified. Peaks marked a and b are the C-terminal peptides. B and C are the mass spectrometric analysis of the peaks a and b, respectively. Notice that the major component of peak a has a m/z value of 2170 and a minor component at 2227; peak b, exhibits m/z values of 2113 and 2170. The MALDI peaks marked by 1G, 2G, and 3G correspond to posttranslational addition of one, two, and three glycyl units, respectively, to the ⌬2 form of ␣1/2 tubulin (m/z ϭ 2056).

FIG. 2. MALDI-TOF analysis of the C-terminal peptide. A,
HPLC purification of the C-terminal peptide from tyrosinated ␣1/2 tubulin. Bovine brain PC-tubulin was chromatographed on the AYN.6D10 column, tyrosinated ␣1/2 tubulin was purified as described under "Experimental Procedures," and the ␣-tubulin was isolated. 15 g of ␣-tubulin was digested with endoproteinase Lys-C as described before. The digest was subsequently subjected to reversed phase HPLC as done for Fig. 1 and the C-terminal peptide was identified. B shows the MALDI-TOF analysis of the C-terminal peptide. Notice that the major peak at 2866.57 corresponds to the tetraglutamylated form of tyrosinated ␣1/2 tubulin, while the minor peaks at 2350, 2479, 2608.5, 2737.54 correspond to species formed by sequential addition of zero to three Gly residues to the tyrosinated ␣1/2 tubulin. The peak at 2350 corresponds to tyrosinated ␣1/2 with no glycyl residue. The peaks at 2479, 2609, 2737, and 2866 are derived from the addition of one, two, three, and four Glu units (m/z 129) to 2350 (2479 ϭ 2350 ϩ 129; 2608 ϭ 2350 ϩ 2 ϫ 129; 2737 ϭ 2350 ϩ 3 ϫ 129; 2866 ϭ 2350 ϩ 4 ϫ 129). Thus this fraction seems to be polyglutamylated with one to four Glu residues, and the tetraglutamylated is the major one.
Glu 450 ) were poor, indicating that the residues may be modified by posttranslational modification.
Amino Acid Analysis-In an effort to confirm the presence of posttranslationally added Gly residues, amino acid analyses were performed on the C-terminal peptides. Since the peptide contains one Ser residue, the amount of Gly residues were calculated per mole of Ser present in the peptide. Analysis of ⌬2 tubulin revealed that this fraction (Fraction A) contains 6.2 moles of Gly per mole of Ser. Similar analysis for the tyrosinated tubulin showed that it contained 4 moles of Gly per mole of Ser. These results clearly demonstrate that the ⌬2 tubulin is modified with more than two posttranslationally added Gly residues, while the tyrosinated tubulin does not contain any posttranslationally added Gly residue. DISCUSSION In my previous study, different ␣-tubulin isoforms were separated in functionally active forms. The fraction that did not bind the affinity column was found to be the nontyrosinated form of ␣1/2, while the tyrosinated ␣1/2 was eluted from the column using a salt gradient. These two fractions exhibited remarkable differences in their assembly properties. The nontyrosinated fraction assembled fairly well, while the tyrosinated fraction did it poorly. Thus, it was tempting to study whether these two fractions differ in their posttranslational modifications. In this study, I have studied the posttranslational status of affinity-purified ␣-tubulin isoforms by MALDI-TOF mass spectrometry.
Mass spectrometric data for the nontyrosinated ␣1/2 exhibited the major peak at 2170 with some minor peaks at 2113 and 2227. The peaks at 2170 clearly indicate that this must be smaller than the Glu form of ␣1/2. This fraction must therefore be the ⌬2 form of ␣1/2. The minor peak at 2113 differs from the mass of the ⌬2 form by 57. The peaks at 2113, 2170, and 2227 are formed by the addition of one, two, and three glycyl units (57 mass units) to the ⌬2 form of ␣1/2. Amino acid analyses of the C-terminal peptide also confirmed the presence of more than two Gly residues per mole of the peptide.
The analysis of Tyr-␣1/2 revealed that this fraction exhibited a major peak at 2867 and several minor peaks at 2057, 2350, 2479, 2608, and 2737. All these peaks except at 2057 clearly originate from tyrosinated ␣1/2 and correspond to the modifications with one, two, three, and four glutamate units. The peak at 2057 corresponds to the ⌬2 form of ␣1/2. The MALDI-TOF signals clearly indicate that the tetra-glutamylated form is the major species.
It should be mentioned that both isoforms of ␣1/2 exhibited several unidentified minor species in the mass spectrograph. The peak at 2149 seems to be derived from the ⌬2 form of ␣4 by the addition of a mass unit of 156, which is the mass of an arginine residue. On the other hand, the peaks at 2323 and 2387 seem to be derived from the ⌬2 forms of ␣4 and ␣1/2 by the addition of a mass unit of 200. Though it may be tempting to conclude that these modifications might be due to the addition of a glutamate (129 mass unit) and an alanine residue (71 mass unit), extensive mass spectrometric analyses are required to confirm this. It is not known whether the glycylating enzyme can use any other amino acid (such as alanine, serine, or histidine) as its substrate. Purification of this enzyme will be essential to shed more light on this.
It is interesting to note that the two forms of ␣1/2 differ remarkably in their assembly properties in the presence of Mg 2ϩ (56). The detyrosinated ␣1/2 assembled quite normally, while the tyrosinated form did it poorly. Although it was predicted in my earlier work that these two forms might differ in their posttranslational modifications, the present MALDI-TOF studies clearly show that the major difference is in the glycylation/glutamylation status of the isoform. If one compares the major peaks for both the fractions, one can see that the detyrosinated form is mainly the biglycylated ⌬2 form of ␣1/2, while the tyrosinated form is mainly the tetraglutamylated form of tyrosinated ␣1/2. It is conceivable that these two forms will differ significantly in the charge distribution at the C terminus.
Since it is believed that the C termini of ␣and ␤-tubulin is involved in the binding of MAPs to tubulin (59 -61), polyglycylation and polyglutamylation may represent modifications by which the assembly and disassembly of microtubules may be regulated in vivo.
These results are fairly consistent with the recent study by Vinh et al. (54), which shows that both dynamic cytoplasmic microtubules as well as stable axonemal microtubules, can be glycylated on each of the last four C-terminal glutamate residues of Glu 437 , Glu 438 , Glu 439 , and Glu 441 in the ␤-tubulin sequence 427 DATAEEEGEFEEEGEQ 442 . In both dynamic and stable microtubules, the majority of the ␤-tubulin contains six posttranslationally added glycine residues: two glycine residues on both Glu 437 and Glu 438 and one glycine residue on both Glu 439 and Glu 441 .
In this context it should be mentioned that Redeker et al. (55) did not observe glycylation in brain tubulin. They have studied the posttranslational modifications of ␣-tubulin from unfractionated PC-tubulin (tubulin purified by phosphocellulose chromatography) by purifying the C-terminal peptide on an arginine-Sepharose column. On the other hand, my affinity column separates the ⌬2 form of ␣1/2 from its tyrosinated form. It is not known whether ␣4 also is glycylated. Sequencing studies have identified ␣4 tubulin in a fraction (Fraction B) that contained both ␣1/2 and ␣4 as the detyrosinated forms. MALDI-TOF analysis of this fraction yielded many unknown species, which makes me believe that ␣4 may be subject to posttranslational modifications other than polyglycylation and polyglutamylation. To get a clear picture on the posttranslational modification of ␣4, it must be separated from ␣1/2 prior to massspectrometric analysis.
At this point it is not clear about the number of glutamate residues modified by polyglycylation. I have previously suggested that there might be more than one Glu residue that may be modified (56). The sequencing chromatogram identified residues Glu 445 , Glu 447 , and Glu 449 , which gave poor signals indicating that these residues are probably modified by polyglycylation. Thus it seems likely that the residues Glu 445 , Glu 447 , and Glu 449 are each modified by the addition of a glycyl unit.
The tubulin molecule is subject to a large number of posttranslational modifications including phosphorylation, acetylation, tyrosination/detyrosination, deglutamylation, polyglutamylation and polyglycylation (21). In general the functional significance of these modifications is not well understood. Tyrosination and detyrosination are thought to be involved, respectively, in the destabilization and stabilization of microtubules (57), and acetylation appears to stabilize microtubules also (58). It has been speculated that polyglutamylation of ␣ and phosphorylation of the ␤ III isotype of tubulin may regulate the interaction of microtubules with MAPs (59,60). Recent studies have demonstrated that polyglutamylation of tubulin modulated the interaction of tubulin with MAP1A, MAP1B, MAP2, and tau (61). Deglutamylation removes ␣-tubulin from the tyrosination/detyrosination cycle and also leads to increased microtubule stability (43), while polyglycylation, hitherto found only in axonemal microtubules, may play a role in the formation and stabilization of those unusual microtubules (62). Since the MAPs-binding domain on tubulin is localized on the region of helices H11 and H12 on the electron crystallographic structure of tubulin (63), it is possible that polyglutamylation or polyglycylation can modulate the conformation of the binding domain drastically.
The results reported here have several implications for these issues. First they indicate that polyglycylation can occur in non-axonemal microtubules, thereby reopening the question of the function of this modification. As far as is known, in mammalian sperm only ␤-tubulin is polyglycylated (49). Studies with Tetrahymena ciliary tubulin suggest that, although both ␣and ␤-tubulin are polyglycylated, only the polyglycylation of ␤ is important in axonemal function (61). Thus the finding reported here that mammalian brain ␣1/2 is polyglycylated is a novel observation and raises the possibility that polyglycylation of ␣-tubulin serves a unique function. Second, these results show that tyrosinated and deglutamylated (⌬2) tubulins differ in more than just two residues, and that one form is polyglutamylated and the other is polyglycylated. This is the first time that any posttranslational modifications of tubulin have been found to be correlated with each other. Third, they raise questions regarding the mechanism of the modifications. For example, is tubulin deglutamylated before being polyglycylated or vice versa? In either case the modifying enzymes must be specific for one particular form of tubulin. Does one modification make the other irreversible? The possibility now exists that there is a pathway of sequential modifications of the tubulin molecule. Fourth, these results suggest that some as yet unknown modifications may occur to the ␣4 isotype. Studies are in progress toward investigating these questions.