Frequently, the introduction of the new termini leads to a loss of continuity of the polypeptide chain within domains that generally are considered folding units. It seemed important, therefore, to compare proteins composed of such permuted chains with their counterparts in which the polypeptide fragments are independent because they still contain the wild-type termini. How efficient is the fragment complementation, and how do the proteins with fragmented chains compare in enzymatic and physical properties with the circularly permuted and wild-type proteins? Is connectivity of the polypeptide chain within a domain essential, or can the separate fragments fold sufficiently for docking and forming a stable domain?

Circular permutation of proteins, which was pioneered by Goldenberg and Creighton (Goldenberg and Creighton 1983) using chemical methods, has been achieved subsequently through molecular genetic approaches for the construction of genes with permuted coding sequences. Studies on a variety of proteins have been summarized by Heinemann and Hahn (1995). More recent studies of circularly permuted proteins include dihydrofolate reductases (Buchwalder et al. 1992; Protasova et al. 1994; Uversky et al. 1996; Nakamura and Iwakura 1999), the catalytic (c) polypeptide chain of E. coli aspartate transcarbamoylase (aspartate carbamoyltransferase, carbamoyl phosphate: l-aspartate carbamoyltransferase, EC 2.1.3.2) (Yang and Schachman 1993a; Graf and Schachman 1996; Zhang and Schachman 1996), ribonuclease T1 (Mullins et al. 1994; Garrett et al. 1996), Dsb A (Hennecke et al. 1999), barnase (Tsuji et al. 1999), and chymotrypsin inhibitor 2 (Otzen and Fersht 1998). Studies on the SH3 domain of α-spectrin (Viguera et al. 1995, 1996) indicated that changing the order of secondary structural elements caused by the permutation does not affect the three-dimensional structure of the protein, but it does alter the protein-folding pathway. The three-dimensional structures of circularly permuted variants of several proteins (Hahn et al. 1994; Viguera et al. 1996; Pieper et al. 1997; Chu et al. 1998; Wright et al. 1998) have been determined and were shown to be very similar to their wild-type counterparts.

In most circularly permuted proteins, the newly introduced N and C termini are located in loops or turns at the surface of the molecules and have little interactions with other regions of the protein. Nonetheless, permuted monomeric proteins sometimes are not as stable as their wild-type counterparts. Whether this decreased stability is attributable to the strain caused by the formation of a covalent bond linking regions near the original N and C termini or to the flexibility and introduction of charged residues resulting from the cleavage at the other site is generally not clear. Having available both circularly permuted proteins and their counterparts composed of polypeptide fragments may aid in interpreting such findings.

Polypeptide fragment complementation studies have been used to address a number of biological problems including protein folding. Early studies using limited proteolytic or chemical cleavage of different proteins yielded fragments that can reassociate in vitro to form active proteins. Among these were ribonuclease (Richards 1958; Richards and Vithayathil 1959), staphylococcal nuclease (Taniuchi et al. 1977), cytochrome c (Taniuchi et al. 1986; Fisher and Taniuchi 1992), cytochrome b5 reductase (Strittmatter et al. 1972), human pituitary growth hormone (Li and Bewley 1976), thioredoxin c (Holmgren and Slabay 1979), alanine racemase (Galakatos and Walsh 1987), the c chain of E. coli ATCase (Powers et al. 1993), trp repressor (Tasayco and Carey 1992), and barnase (Kippen et al. 1994). These protease-generated fragments are thought to represent well-defined structural units or domains linked through a flexible, exposed peptide. Protein engineering, through gene manipulation, has also been used to generate protein fragments either separately expressed in different cells or coexpressed in vivo. Thermostable alanine racemase (Toyama et al. 1991) has been reconstituted from two fragments that are cloned into a single vector and coexpressed in the same cells. Escherichia coli iosleucyl-tRNA synthetase (Shiba and Schimmel 1992) has been reconstituted in vivo from two fragments which are cloned into separate vectors but coexpressed in the same cells.

In complementation studies on ATCase (Yang and Schachman 1993b), active enzyme was formed from polypeptide fragments of the c chains either expressed separately or coexpressed in E. coli. The observations on ATCase are of particular interest, because the disruption of the continuity of the chain occurred within the structural region generally identified as the aspartate-binding domain. Each polypeptide expressed independently was insoluble, but mixing them in 6.5 M urea followed by removal of the denaturant yielded active enzyme in good yield (Yang and Schachman 1993b). In parallel studies (Powers et al. 1993), it was shown that limited proteolysis of the catalytic (C) trimer led to the hydrolysis of a single peptide bond between residues 240 and 241, with neither disruption of the trimeric structure nor loss of enzyme activity. The enzyme molecules containing the fragmented chains, produced both by genetic manipulation and by proteolysis, lacked allosteric properties. Similarly, dodecamers containing circularly permuted c chains with N and C termini in the same region were devoid of the allosteric properties characteristic of wild-type ATCase. Subsequently, other studies of ATCase containing circularly permuted c chains with termini in diverse regions of the structure yielded enzyme exhibiting homotropic and heterotropic effects (Zhang and Schachman 1996; Beernink et al. 2001). It was of interest, therefore, to perform companion studies on ATCase variants containing fragmented c chains along with the analogous enzyme composed of intact, circularly permuted c chains. As shown here with four ATCase variants composed of fragmented c chains and the analogous molecules containing circularly permuted chains, the two types of dodecamers with new N and C termini in each of the domains of the c chains have comparable enzyme activities and thermal stabilities, exhibit the characteristic allosteric properties of wild-type ATCase, and undergo similar ligand-promoted conformational changes.

Fig. 1. Ribbon diagram of the three-dimensional structure of a catalytic chain of wild-type ATCase. The N-terminal domain (residues 1–134, in yellow) is on the left, and the C-terminal domain (residues 150–284, in blue) on the right. The two domains (more ...)

For the preparation of each ATCase variant, the pyrB gene was modified to produce the desired circularly permuted or fragmented gene needed for expression of the encoded cpc or frc chain. The plasmids also contained a modified pyrI gene (Graf and Schachman 1996), which encodes the r chain of ATCase containing a sequence tag of six His residues at the N terminus. In this way, fully assembled ATCase was produced in E. coli and readily purified. The amino acid sequences of the four different cpc and frc chains are illustrated schematically in Figure 2 , which also shows how the continuity of the polypeptide chain within the two domains is disrupted.

Fig. 2. Schematic diagram illustrating the amino acid sequence of the N-terminal (gray square) and C-terminal (black square) domains in the wild-type c chain (cwt), fragmented c chains (frc), and circularly permuted chains (cpc). Numbers refer to the positions (more ...)

Fig. 3. Detection and characterization of ATCase variants and C trimers by polyacrylamide gel electrophoresis. (A) Electrophoresis under nondenaturing conditions was performed in 7% polyacrylamide gels stained for protein using Coomassie brilliant blue G250. (more ...)

Fig. 4. Enzyme activity of holoenzymes and C trimers containing fragmented and circularly permuted c chains. Assays were performed at 30°C with saturating [14C]carbamoyl phosphate (5 mM) in 50 mM MOPS buffer at pH 7.0, containing 0.2 mM EDTA and 2 mM (more ...)

A single endotherm with a T_m of 65°C was observed for _H6ATCase_wt, whereas ATCase_wt exhibits two overlapping, relatively sharp transitions at 64°C and 68°C (Zhang and Schachman 1996). Upon the addition of excess PALA, the endotherm showed overlapping transitions at 65°C and 70°C for _H6ATCase_wt. These results indicate that the (His)₆ tag at the N terminus of the r chain has little effect on the properties of the enzyme.

As expected, C trimers isolated from _H6ATCase_wt were very similar to trimers obtained from the wild-type enzyme. Gel electrophoresis under denaturing conditions (SDS-PAGE) showed a single band with a mobility corresponding to a 34-kD polypeptide.

Table 1. Properties of ATCase variantsa

Purified _frATCase_c181 exhibited a single band upon electrophoresis in polyacrylamide gels under nondenaturing conditions (Fig. 3 A, lane 3) with a mobility slightly larger than that of _H6ATCase_wt. The other three fragmented ATCase variants (_frATCase_c99, _frATCase_c122, and _frATCase_c222) had identical or very similar mobilities to wild-type enzyme under nondenaturing conditions (data not shown). Differential scanning calorimetry showed that the four ATCase variants containing fragmented c chains had values of T_m comparable to that of the wild-type enzyme (Table 1). A single endotherm with a T_m of 63°C was observed for _frATCase_c181. The Tm increased to 66°C and 70°C in the presence of 20- and 100-fold molar excess of PALA, respectively.

Electrophoresis under denaturing conditions was used to demonstrate that the c chains in the various oligomers were fragmented and that the sizes of the constituent polypeptides were compatible with the DNA constructs used for their expression. The patterns from SDS-PAGE experiments are illustrated in Figure 3 B. The two major polypeptides from _H6ATCase_wt (lane 2), representing intact c chains and His-tagged r chains, have values of M_r corresponding to molecular weights of 34 kD and 17 kD, respectively. A minor band was observed in many SDS-PAGE experiments and is probably attributable to the in vivo removal of the His tag. Lanes 3, 4, and 5 in Figure 3 B show the polypeptide fragments in _frATCase_c99, _frATCase_c122, and _frATCase_c222, respectively. Each pattern shows a prominent band corresponding to the His-tag r chain with a minor satellite band representing a slightly smaller polypeptide. The other two bands in each pattern represent the polypeptide fragments encoded by the fragmented pyrB gene, and the mobilities were consistent with the molecular weights expected based on the DNA sequences. Mobilities of the fragments from _frATCase_c99 correspond to a 23-kD polypeptide representing the C-terminal region of the chain and the 11-kD polypeptide derived from the N terminus, whereas the 24-kD and 10-kD fragments from _frATCase_c222 represent the N- and C-terminal peptides. The 22-kD and 13-kD fragments from _frATCase_c122 correspond to the C- and N-terminal peptide fragments. The polypeptide fragments were isolated by reverse-phase high performance liquid chromatography and analyzed by electron spray mass spectrometry for molecular weight determinations. For fragment 1–180 the measured molecular weight was 19,516.2, a value very close to the calculated value of 19,517.1 (assuming the removal of the N-terminal Met residue). Mass spectrometric analysis for fragment 181–310 yielded 14,929.2 and 14,798.1, which are identical to the values 14,929.2 and 14,798.1 calculated for the fragment with and without an N-terminal Met residue. Approximately one-third of the fragments had the N-terminal Met residue and about two-thirds of the polypeptide fragments lacked that residue. The measured molecular weights of the polypeptide fragments of the c chains of _frATCase_c99 and _frATCase_c122 were virtually identical to the values calculated from the amino acid sequence with the assumption that the N-terminal Met residue was removed in vivo.

The C trimer containing the fragmented c chains was isolated from _frATCase_c181 by treatment of the holoenzyme with a mercurial (Yang et al. 1978). As seen in Figure 3 A (lane 6), gel electrophoresis experiments under nondenaturing conditions revealed a single band for _frC_c181 with a mobility similar to that for C_wt. In SDS-PAGE experiments on _frC_c181, two bands were detected corresponding to molecular weights of 20 kD and 15 kD.

Although _frC_c181 trimers tended to aggregate, thereby limiting physical studies on them, assays of enzyme activity showed that V_max and K_m for aspartate were 31 μmole/μg per hour and 28 mM, compared to 27 μmole/μg per hour and 6 mM for C_wt, respectively. Fragmented c chains (_frcC_c181) were also expressed in E. coli lacking the pyrI gene, resulting in the formation of active _frC_c181 trimers (data not shown). The C trimers from the other variants containing fragmented c chains were insoluble, precluding measurements of enzyme activity or physical properties.

These results along with analogous observations (Table 1) for _frATCase_c99, _frATCase_c122, and _frATCase_c222 demonstrate that ATCase molecules containing c chains fragmented in different regions of the structure exhibit homotropic and heterotropic properties characteristic of wild-type enzyme. Table 1 also summarizes the results for the four ATCase variants containing circularly permuted c chains. Two of these variants, _cpATCase_c122 and _cpATCase_c222, had been prepared earlier (Zhang and Schachman 1996) using a different technique for generating the permuted c chains. Except for the values of V_max, the results for the variants containing fragmented and circularly permuted c chains were similar. Values of V_max are known to vary because of storage of the enzymes and the use of different substrate preparations; hence it is difficult to attach significance to those variations. The results in Table 1 demonstrate that the ATCase variants containing fragmented and/or circularly permuted c chains are very similar in their allosteric properties, their thermal stability, and in the PALA-promoted change in quaternary structure.

The four ATCase variants containing fragmented c chains exhibited homotropic and heterotropic effects, as well as thermal stabilities, similar to those of the wild-type enzyme (Table 1). These results indicate that a lack of continuity in the polypeptide chain is readily tolerated as long as the interruption is located in a noncritical region of the structure. In earlier studies (Powers et al. 1993; Yang and Schachman 1993b), it was found that the holoenzyme containing fragmented c chains cleaved in the 240s loop lacked allosteric properties. Residues in that loop are involved in interactions between the two C trimers when the enzyme is in the T state, and that region undergoes a large conformational change upon the binding of PALA to the enzyme (Lipscomb 1994). Hence, connectivity in that region of the polypeptide chain is probably essential for stabilizing the enzyme in the compact, taut state. It is striking that all of the ATCase variants summarized in Table 1 exhibit allosteric properties despite the interruption of the polypeptide chain in either domain and in diverse regions of the structure. For three of the variants, _frATCase_c99, _frATCase_c122, and _frATCase_c222, the cleavage of the chain occurs in regions that are largely buried in the wild-type enzyme. In _frATCase_c181, the disruption is in a region involving chain–chain interactions in the same trimer and some contacts with other segments of the main chain. The similarity in behavior of the analogous variants containing fragmented or circularly permuted c chains indicates that not only are the introduction of new N and C termini tolerated readily, but also that the linkage of the wild-type terminal regions does not interfere with the conformational change essential for allostery. Although the allosteric properties of the variants were not significantly altered by the introduction of newly charged termini in the different domains, the values of K_0.5 were affected. It is noteworthy that the substantial increases in K_0.5 are almost identical for the holoenzymes containing either the fragmented or the circularly permuted chains (Table 1). These increases, based on the comparable findings for the nonallosteric C trimers (Fig. 4 D), are attributable to a decreased affinity for the substate, aspartate. Speculations regarding explanations for these increases in K_0.5 are unwarranted until evidence is available about possible structural changes resulting from the enhanced flexibility of the chains and the presence of new ion pairs.

Many proteins composed of circularly permuted polypeptide chains have slightly lower thermal stabilities than their wild-type counterparts (Buchwalder et al. 1992; Yang and Schachman 1993a; Zhang et al. 1993; Vignais et al. 1995; Garrett et al. 1996; Zhang and Schachman 1996; Pieper et al. 1997; Hennecke et al. 1999). Interpreting the decreased thermal stability of circularly permuted variants is hazardous because of the possible constraint introduced by linkage of the original termini, the introduction of flexibility in the vicinity of the new N and C termini, and possible structural changes resulting from these charged residues. For ATCase, values of T_m for the holoenzyme containing either circularly permuted or fragmented c chains are similar and only slightly lower than that of wild-type enzyme (Table 1). The comparison between the two types of variants provides evidence to support the conclusion that linkage of the original terminal regions does not cause significant destabilization of the structure.

Domains in proteins are generally considered as important folding units along the pathway of assembly of multidomain proteins, and it is frequently assumed that folding of chains is contemporaneous with synthesis of the polypeptides. The formation of a stable domain from a continuous polypeptide chain would involve a folding process based on unimolecular reactions. When the continuity of the polypeptide chain within the domains is disrupted, as in the variants described here, the formation of the domains must involve a bimolecular process. The experiments on _frATCase_c181 were particularly interesting because active _frC_c181 trimers were isolated readily by dissociation of the holoenzyme, and independently these trimers were formed in vivo upon expression of the fragmented pyrB gene in the absence of pyrI. Clearly the fragments, 1–180 and 181–310, must fold sufficiently into secondary and tertiary structures for recognition and docking by bimolecular reactions. The discontinuous chain containing the folded domain then oligomerizes to give trimers, which then associate with regulatory dimers to form stable, active holoenzymes exhibiting the allosteric properties of wild-type ATCase. Complementation between fragments that comprise structurally recognizable domains in the c chains of ATCase appears to be very efficient, and physical chemical studies with isolated fragments are necessary for a more detailed understanding of the individual reactions in the assembly process.

For the preparation of pyrB_cp181 and pyrB_fr181, Primer P1 was an oligonucleotide coding for residue 181–187 with an NdeI site in the 5′ end. Primer P2 was an oligonucleotide whose complementary sequence encodes for the residue 174–180 with a BamHI site and a stop codon in the 5' end. Primer P3a for the circular permutation mutant was a 42-mer oligonucleotide coding for the residues 300–306 and 1–7 in the sense direction. The Primers 1 and 2 for the expression of the fragmented c chain were the same as those used in the preparation of the circularly permuted gene. The third primer P3b was also a 42-mer oligonucleotide containing the coding sequence for residue 1–7 and part of the sequence between the stop codon of pyrB and the start codon of pyrI in PAX4. The PCR reactions were carried out under standard conditions with solutions containing 4 mM MgCl₂, 1× Taq polymerase buffer, 0.2 mM each dNTP, 0.2 mM primer P1 and P2, 0.01 mM primer P3, and 1 unit of Taq polymerase. The PCR products were digested with NdeI and BamHI and then ligated to PAX4a digested with the same pair of enzymes. Gene sequences were confirmed by sequencing using ABI 100.

The ATCase mutants were purified by procedures used routinely in this laboratory. Catalytic trimers of wild-type and mutant ATCases were prepared from holoenzyme by treating the enzyme with neohydrin and separating the subunits as described (Yang et al. 1978).

A Beckman Model XL-A Analytical Ultracentrifuge equipped with absorption optics was used to measure the change in sedimentation coefficient (Δs/s) of the enzyme caused by the binding of PALA. The protein concentration was 1.5 mg/mL in buffer A containing 0.1 M KCl. The ratio of PALA per active site of ATCase is 2.5 : 1 for wild type and 20 : 1 for the variants.

Differential scanning microcalorimetry was performed using the Microcal MC-2 calorimeter as described (Peterson and Schachman 1991). The protein concentration was 0.4 mg/mL in 40 mM potassium borate buffer at pH 9.0 containing 0.2 mM EDTA.

Nondenaturing PAGE was performed as described (Jovin et al. 1964), using 7% polyacrylamide gel. The gel was stained for total protein by Coomassie blue or for activity using 2 mM carbamoyl phosphate and 100 mM aspartate as described (Bothwell 1975). SDS-PAGE was performed on 17.5% polyacrylamide gel followed by staining with Coomassie blue.

This research was supported by National Institute of General Medical Sciences research grant GM 12159. We are indebted to Ying R. Yang for her help and valuable suggestions during the course of this work. We thank Dr. David King of the Howard Hughes Medical Institute, University of California at Berkeley, for determining the molecular weights of the polypeptide chains by mass spectrometry.

The publication costs of this article were defrayed in part by payment of page charges.This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.

• ATCase, aspartate transcarbamoylase
• C, catalytic trimer or subunit
• c, catalytic polypeptide chain
• R, regulatory dimer or subunit
• r, regulatory polypeptide chain
• wt as subscript, wild type
• fr as subscript, fragmented polypeptide chain
• cp as subscript, circularly permuted
• c and number following it in subscript designate the position of the amino acid residue in the wild-type catalytic chain at which the new N terminus in the fragmented or circularly permuted chain is located
• H6 as subscript, hexa-His sequence at the N terminus of the regulatory chain
• MOPS, 3-(N-morpholino)propanesulfonic acid
• PAGE, polyacrylamide gel electrophoresis
• SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis
• PCR, polymerase chain reaction
• PALA, N-(phosphonacetyl)-l-aspartate
• T_m, melting temperature corresponding to the maximum temperature in the endotherm obtained by differential scanning microcalorimetry