**5. Wild type vs. IBMPFD mutant p97: Structural characteristics**

Changes in structure as a result of amino acid mutations can lead to a global disruption of the protein folding, resulting rapid clearance by cellular stress response mechanisms, or to localized structural changes that cause complete loss of the protein function, or to subtle conformational changes that alter the function of the protein. Structural changes in mutant

IBMPFD and p97, the Structural and Molecular Basis for Functional Disruption 159

crystal structures such as LTag (SV40 large tumor antigen) (Gai et al., 2004), FstH

Conformational changes in AAA+ proteins have been probed by methods other than crystallography, which do not depend on obtaining 3-D crystals albeit at relatively lower resolutions. Such methods include cryo EM and SAXS. Cryo EM has revealed flexibility in the N-domains of *E. coli* AAA+ protein ClpA (Ishikawa et al., 2004) and SAXS experiments have shown large conformational changes in NtrC1 (Chen et al., 2010). Although in some cases conformational changes observed with different methods do not completely agree, it is widely accepted that AAA+ proteins undergo dynamic movements during their catalytic cycle. One general observation relating to structural movements among AAA+ proteins is the change in size of the axial pore where substrates enter. The "open-and-close" of the axial pore in AAA+ proteins is thought to provide the mechanical force needed to pull the

Crystal structures of wild type p97 have been solved for both the full-length protein and a truncated N-D1 fragment (absent of the D2-domain) (DeLaBarre and Brunger, 2003; Huyton et al., 2003; Zhang et al., 2000). Although both D1- and D2-domains are capable of hydrolyzing ATP (Song et al., 2003), in all the wild type p97 crystal structures determined to date, ADP was invariably found in the D1-domains, while either ADP or ADP-AlFx (transitional analog) was observed in the D2-domain (DeLaBarre and Brunger, 2003; Huyton et al., 2003). These structures share an identical N-domain conformation with the N-domains attached to the periphery of the D1 ring, and in plane with it (Fig. 1B). Unsuccessful attempts have been made to crystallize wild type p97 with other forms of nucleotides

On a different front, studies using EM and SAXS to gain structural insights into the conformational changes of p97 have revealed rather dramatic changes in the positions of Ndomains (Davies et al., 2005; Rouiller et al., 2002). In contrast to X-ray crystallography, these approaches are limited to providing molecular shapes without a clear delineation of bound nucleotides. Nevertheless, large conformational changes in the N-domains of p97 can be detected by modeling individual domains from crystal structures into these molecular envelopes to re-construct structures of p97 under various conditions, although lacking absolute certainty. In the presence of different nucleotides (ATP or ADP) and their nonhydrolysable or transitional analogs (ADP-AlFx, AMP-PNP or ATPS), the N-domains of p97 were shown to undergo some of the most dramatic movements during the ATP cycle. Although some changes observed by different methods or in different laboratories were not always compatible with each other, it is generally agreed that N-domains of p97 are conformationally flexible. EM studies of p97 complexed with the adaptor protein p47 also showed the N-domains undergoing a large conformational change in the presence of different nucleotides (Beuron et al., 2006). However, conflicting results were reported in these low-resolution studies on the direction of the N-domain movement in response to binding of different nucleotides. An intrinsic difficulty with these studies is the uncertainty concerning the nucleotide states at the AAA+ ATPase domains due to the resolution limits of these methods. Compounding these problems, it has been known that at least half of the nucleotide-binding sites in the D1-domains of p97 are pre-occupied by ADP molecules,

(Bieniossek et al., 2009) and HslU (Wang et al., 2001).

**5.2 Structural studies of wild type p97** 

substrates through the pore (Kravats et al., 2011; Zolkiewski, 2006).

trapped in the D1-domain (DeLaBarre and Brunger, 2003).

proteins can be directly visualized by X-ray crystallography, NMR (nuclear magnetic resonance), and EM (electron microscopy) or inferred indirectly by biophysical and biochemical methods such as SAXS (small angle X-ray scattering). To understand how IBMPFD mutations lead to functional change in p97, it is absolutely necessary to know what structural changes these mutations entail. However, knowing what has changed in mutants depends heavily on our baseline knowledge of the wild type proteins.


Table 1. IBMPFD mutations in p97.

#### **5.1 Conformational changes in AAA+ proteins**

Studies of AAA+ proteins have revealed conformational changes in various domains in response to changes in the environment, to substrate binding, and to various bound nucleotides. At least some of the observed conformational changes are believed to be necessary for function (Vale and Milligan, 2000). N-domains of ClpB were found in different positions in crystal structure even though all the subunits were bound with the ATP analog AMP-PNP (Lee et al., 2003); this conformational plasticity in N-domains is likely the result of different environments each subunit experienced in the crystal and may not be directly related to function. By far most of the observed conformational changes, though relatively subtle, are induced by binding of various nucleotides in the AAA+ domains. Such nucleotide-driven conformational changes have been observed for both Type I (proteins with one AAA+ domain) and Type II (proteins with two AAA+ domains) AAA+ proteins in crystal structures such as LTag (SV40 large tumor antigen) (Gai et al., 2004), FstH (Bieniossek et al., 2009) and HslU (Wang et al., 2001).

Conformational changes in AAA+ proteins have been probed by methods other than crystallography, which do not depend on obtaining 3-D crystals albeit at relatively lower resolutions. Such methods include cryo EM and SAXS. Cryo EM has revealed flexibility in the N-domains of *E. coli* AAA+ protein ClpA (Ishikawa et al., 2004) and SAXS experiments have shown large conformational changes in NtrC1 (Chen et al., 2010). Although in some cases conformational changes observed with different methods do not completely agree, it is widely accepted that AAA+ proteins undergo dynamic movements during their catalytic cycle. One general observation relating to structural movements among AAA+ proteins is the change in size of the axial pore where substrates enter. The "open-and-close" of the axial pore in AAA+ proteins is thought to provide the mechanical force needed to pull the substrates through the pore (Kravats et al., 2011; Zolkiewski, 2006).

#### **5.2 Structural studies of wild type p97**

158 Neuromuscular Disorders

proteins can be directly visualized by X-ray crystallography, NMR (nuclear magnetic resonance), and EM (electron microscopy) or inferred indirectly by biophysical and biochemical methods such as SAXS (small angle X-ray scattering). To understand how IBMPFD mutations lead to functional change in p97, it is absolutely necessary to know what structural changes these mutations entail. However, knowing what has changed in mutants

**Location References** 

Schroder et al., 2005; Watts et al., 2004)

(Rohrer et al., 2011)

2 R93C 277 C T (Guyant-Marechal et al., 2006; Watts et al., 2004)

5 R155C 463 C T (Gidaro et al., 2008; Guyant-Marechal et al., 2006;

R159H 476 G A (Haubenberger et al., 2005; van der Zee et al., 2009)

 **proteins**  Studies of AAA+ proteins have revealed conformational changes in various domains in response to changes in the environment, to substrate binding, and to various bound nucleotides. At least some of the observed conformational changes are believed to be necessary for function (Vale and Milligan, 2000). N-domains of ClpB were found in different positions in crystal structure even though all the subunits were bound with the ATP analog AMP-PNP (Lee et al., 2003); this conformational plasticity in N-domains is likely the result of different environments each subunit experienced in the crystal and may not be directly related to function. By far most of the observed conformational changes, though relatively subtle, are induced by binding of various nucleotides in the AAA+ domains. Such nucleotide-driven conformational changes have been observed for both Type I (proteins with one AAA+ domain) and Type II (proteins with two AAA+ domains) AAA+ proteins in

(Watts et al., 2004)

(Watts et al., 2004)

4 P137L 410 C T (Palmio et al., 2011; Stojkovic et al., 2009)

R155H 463 C A (Viassolo et al., 2008; Watts et al., 2004)

depends heavily on our baseline knowledge of the wild type proteins.

Ndomain

linker

D1 domain

11 T262A 784 A G (Spina et al., 2008) 12 N387H 1159 A C (Watts et al., 2007) 13 A439S 1315 G T (Stojkovic et al., 2009)

9 L198W 593 T G (Kumar et al., 2010; Watts et al., 2007)

R155P 463 C C (Watts et al., 2004) R155S 463 C A (Stojkovic et al., 2009) R155L 464 G T (Kumar et al., 2010) 6 G157R 469 G C (Djamshidian et al., 2009) 469 G A (Stojkovic et al., 2009) 7 R159C 475 C T (Bersano et al., 2009)

3 R95C 283 C T (Kimonis et al., 2008a) R95G 283 C G (Watts et al., 2004)

**Change in amino acid** 

1 I27V 79 A G

**Change in gene**

8 R191Q 572 G C N-D1

10 A232E 695 C A

Table 1. IBMPFD mutations in p97.

**5.1 Conformational changes in AAA+**

Crystal structures of wild type p97 have been solved for both the full-length protein and a truncated N-D1 fragment (absent of the D2-domain) (DeLaBarre and Brunger, 2003; Huyton et al., 2003; Zhang et al., 2000). Although both D1- and D2-domains are capable of hydrolyzing ATP (Song et al., 2003), in all the wild type p97 crystal structures determined to date, ADP was invariably found in the D1-domains, while either ADP or ADP-AlFx (transitional analog) was observed in the D2-domain (DeLaBarre and Brunger, 2003; Huyton et al., 2003). These structures share an identical N-domain conformation with the N-domains attached to the periphery of the D1 ring, and in plane with it (Fig. 1B). Unsuccessful attempts have been made to crystallize wild type p97 with other forms of nucleotides trapped in the D1-domain (DeLaBarre and Brunger, 2003).

On a different front, studies using EM and SAXS to gain structural insights into the conformational changes of p97 have revealed rather dramatic changes in the positions of Ndomains (Davies et al., 2005; Rouiller et al., 2002). In contrast to X-ray crystallography, these approaches are limited to providing molecular shapes without a clear delineation of bound nucleotides. Nevertheless, large conformational changes in the N-domains of p97 can be detected by modeling individual domains from crystal structures into these molecular envelopes to re-construct structures of p97 under various conditions, although lacking absolute certainty. In the presence of different nucleotides (ATP or ADP) and their nonhydrolysable or transitional analogs (ADP-AlFx, AMP-PNP or ATPS), the N-domains of p97 were shown to undergo some of the most dramatic movements during the ATP cycle. Although some changes observed by different methods or in different laboratories were not always compatible with each other, it is generally agreed that N-domains of p97 are conformationally flexible. EM studies of p97 complexed with the adaptor protein p47 also showed the N-domains undergoing a large conformational change in the presence of different nucleotides (Beuron et al., 2006). However, conflicting results were reported in these low-resolution studies on the direction of the N-domain movement in response to binding of different nucleotides. An intrinsic difficulty with these studies is the uncertainty concerning the nucleotide states at the AAA+ ATPase domains due to the resolution limits of these methods. Compounding these problems, it has been known that at least half of the nucleotide-binding sites in the D1-domains of p97 are pre-occupied by ADP molecules,

IBMPFD and p97, the Structural and Molecular Basis for Functional Disruption 161

Fig. 3. **Observed structural re-arrangements in the N-D1 linker and N-terminal peptide in** 

structure of the N-D1 linker undergoes a transition from a random coil to a three-turn helix as the N-domains move from the Down- to Up-conformation. A close-up view shows the

nucleotides are represented by sticks with carbon atoms in yellow, oxygen in red, nitrogen in blue, phosphorous in orange, and sulfur in gold. (B) The reordering of N-terminal peptide Leu12 to Lys20 in the ATPS-bound form (Up-conformation) is represented by stick model. The rest of the N-domain and D1-domain are shown as magenta and blue surfaces,

The above observation appears to favor the hypothesis that the Up and Down movement of N-domains is nucleotide dependent because the binding of ADP at the D1-domain of p97 results in a Down-conformation while binding of ATPS leads to an Up-conformation. However, this nucleotide-driven movement may be arguable, as these two conformations were observed in two different systems – the wild type in ADP form and the IBMPFD mutants in ATPS form. A subsequent structure determination using the same IBMPFD mutant and ADP showed that N-domains adopt the Down-conformation, just as the wildtype p97 in the presence of ADP (Tang et al., 2010), thus unequivocally confirming the dependency of N-domain conformation on the nucleotide binding states at the D1-domain.

**5.4 Small angle X-ray scattering (SAXS) studies of wild type and IBMPFD mutant p97** 

Why crystallographic studies on wild type p97 can only reveal the Down-conformation, whereas IBMPFD p97 mutants can be crystallized in both Up- and Down-conformations was a paradox. One possible interpretation was that wild type and mutant p97 differ in nucleotide binding properties. Alternatively, this could be a crystallization effect. To investigate this, we performed SAXS experiments to identify conformational changes in IBMPFD mutants in solution. The results clearly demonstrated that in solution IBMPFD mutants undergo a nucleotide-dependent N-domain conformational change that is consistent with the Up- and Down-conformations observed in the crystals (Fig. 4). By serendipity, another major finding from this experiment was that wild type p97 also

**response to binding of different nucleotides in the D1-domain** (A) The secondary

two conformations of the N-D1 linker (in green) in the ADP-bound form (Downconformation) and in the ATPS-bound form (Up-conformation) of p97 N-D1. The

respectively.

**in solution** 

which are very difficult to remove (Briggs et al., 2008; Davies et al., 2005). Clearly, highresolution structures of p97 in different nucleotide states are needed to unambiguously define the relationship of N-domain conformation with nucleotide-binding states.

#### **5.3 Crystallographic studies of IBMPFD mutant p97**

Recently, a new conformation of N-domains was observed by X-ray crystallography at 2 Å resolution with ATPS (a non-hydrolysable ATP analog) bound at the D1-domains of two IBMPFD-associated p97 N-D1 mutants, R155H and R95G; both are N-domain mutations (Tang et al., 2010). With ATPS bound at the D1-domain, the N-domain undergoes a rotational and translational movement to adopt a new position, which is in sharp contrast to the "in-plane" position with the D1-ring or Down-conformation, as observed previously in the ADP-bound form. In this new conformation, the N-domains uniformly occupy a position above the plane of the D1-ring or in the Up-conformation (Fig. 2), despite the fact that no detectable changes are seen in the N-domain itself due to either the R155H or R95G mutation (Data not shown) (Tang et al., 2010). Accompanying the transition of N-domain from the Down- to Upconformation are two prominent structural rearrangements. One is the transition in secondary structure of the linker between N- and D1-domain (N-D1 linker), going from the random coil in the Down-conformation to the three-turn -helix in the Up-conformation reminiscent of a contracted spring pulling the N-domains out of the planar conformation upon the binding of ATPS (Fig. 3A). This novel conformation of p97 demonstrated for the first time the dynamic movement of the N-domain at near atomic resolution, although observed in p97 mutants. A second change is the re-ordering of the N-terminal fragment that encompasses residues 12 to 20, which was disordered in the ADP-bound structures (Fig. 3B). This re-ordering of the Nterminal peptide apparently protects the Lys18 from limited proteolysis by trypsin seen in the ADP form of p97 (Fernandez-Saiz and Buchberger, 2010).

Fig. 2. **Changes in N-domain conformation in response to binding of different nucleotides to the D1 domain** Ribbon diagrams showing the two N-domain conformations of p97 N-D1 obtained from crystal structures. The N-domains, D1-domains and the IBMPFD mutations are colored in magenta and blue ribbons and in yellow balls, respectively.

which are very difficult to remove (Briggs et al., 2008; Davies et al., 2005). Clearly, highresolution structures of p97 in different nucleotide states are needed to unambiguously

Recently, a new conformation of N-domains was observed by X-ray crystallography at 2 Å resolution with ATPS (a non-hydrolysable ATP analog) bound at the D1-domains of two IBMPFD-associated p97 N-D1 mutants, R155H and R95G; both are N-domain mutations (Tang et al., 2010). With ATPS bound at the D1-domain, the N-domain undergoes a rotational and translational movement to adopt a new position, which is in sharp contrast to the "in-plane" position with the D1-ring or Down-conformation, as observed previously in the ADP-bound form. In this new conformation, the N-domains uniformly occupy a position above the plane of the D1-ring or in the Up-conformation (Fig. 2), despite the fact that no detectable changes are seen in the N-domain itself due to either the R155H or R95G mutation (Data not shown) (Tang et al., 2010). Accompanying the transition of N-domain from the Down- to Upconformation are two prominent structural rearrangements. One is the transition in secondary structure of the linker between N- and D1-domain (N-D1 linker), going from the random coil in the Down-conformation to the three-turn -helix in the Up-conformation reminiscent of a contracted spring pulling the N-domains out of the planar conformation upon the binding of ATPS (Fig. 3A). This novel conformation of p97 demonstrated for the first time the dynamic movement of the N-domain at near atomic resolution, although observed in p97 mutants. A second change is the re-ordering of the N-terminal fragment that encompasses residues 12 to 20, which was disordered in the ADP-bound structures (Fig. 3B). This re-ordering of the Nterminal peptide apparently protects the Lys18 from limited proteolysis by trypsin seen in the

define the relationship of N-domain conformation with nucleotide-binding states.

**5.3 Crystallographic studies of IBMPFD mutant p97** 

ADP form of p97 (Fernandez-Saiz and Buchberger, 2010).

Fig. 2. **Changes in N-domain conformation in response to binding of different** 

mutations are colored in magenta and blue ribbons and in yellow balls, respectively.

**nucleotides to the D1 domain** Ribbon diagrams showing the two N-domain conformations of p97 N-D1 obtained from crystal structures. The N-domains, D1-domains and the IBMPFD

Fig. 3. **Observed structural re-arrangements in the N-D1 linker and N-terminal peptide in response to binding of different nucleotides in the D1-domain** (A) The secondary structure of the N-D1 linker undergoes a transition from a random coil to a three-turn helix as the N-domains move from the Down- to Up-conformation. A close-up view shows the two conformations of the N-D1 linker (in green) in the ADP-bound form (Downconformation) and in the ATPS-bound form (Up-conformation) of p97 N-D1. The nucleotides are represented by sticks with carbon atoms in yellow, oxygen in red, nitrogen in blue, phosphorous in orange, and sulfur in gold. (B) The reordering of N-terminal peptide Leu12 to Lys20 in the ATPS-bound form (Up-conformation) is represented by stick model. The rest of the N-domain and D1-domain are shown as magenta and blue surfaces, respectively.

The above observation appears to favor the hypothesis that the Up and Down movement of N-domains is nucleotide dependent because the binding of ADP at the D1-domain of p97 results in a Down-conformation while binding of ATPS leads to an Up-conformation. However, this nucleotide-driven movement may be arguable, as these two conformations were observed in two different systems – the wild type in ADP form and the IBMPFD mutants in ATPS form. A subsequent structure determination using the same IBMPFD mutant and ADP showed that N-domains adopt the Down-conformation, just as the wildtype p97 in the presence of ADP (Tang et al., 2010), thus unequivocally confirming the dependency of N-domain conformation on the nucleotide binding states at the D1-domain.

#### **5.4 Small angle X-ray scattering (SAXS) studies of wild type and IBMPFD mutant p97 in solution**

Why crystallographic studies on wild type p97 can only reveal the Down-conformation, whereas IBMPFD p97 mutants can be crystallized in both Up- and Down-conformations was a paradox. One possible interpretation was that wild type and mutant p97 differ in nucleotide binding properties. Alternatively, this could be a crystallization effect. To investigate this, we performed SAXS experiments to identify conformational changes in IBMPFD mutants in solution. The results clearly demonstrated that in solution IBMPFD mutants undergo a nucleotide-dependent N-domain conformational change that is consistent with the Up- and Down-conformations observed in the crystals (Fig. 4). By serendipity, another major finding from this experiment was that wild type p97 also

IBMPFD and p97, the Structural and Molecular Basis for Functional Disruption 163

molecules associated with the Mg2+ ion (Fig. 5). Perhaps due to better diffraction resolution, in the mutant structures of p97, a Mg2+ ion is present in the nucleotide-binding site of every subunit. The Mg2+ ion is at the center of an octahedral *mer*-triaquo complex with the additional three oxo ligands coming from the side chain of the highly conserved Thr252 and from the - and -phosphates. The acidic residues of the DEXX sequence (Asp304 and Glu305) in the Walker B motif make hydrogen bonds with two of the water molecules in the Mg2+ coordination sphere and, additionally, Asp304 stabilizes Thr252. As expected, most of the changes in the nucleotide-binding environment are a consequence of the introduction of the

Fig. 5. **ATPS binding vicinity of the D1-domain** The nucleotide-binding pocket is located at the subunit interface. The two subunits are in different colors, green and gray. The ATPS molecule is shown as a stick model with carbon atoms in purple, oxygen in red, nitrogen in blue, phosphorous in magenta, and sulfur in yellow. The ATPS molecule is enclosed in a difference electron density cage contoured at the 2.5 σ level. The Mg2+ ion is shown as a

Structurally observed differences in the N-domain conformation of p97 strongly suggest a change in nucleotide binding affinities between wild type and IBMPFD mutants, even though the binding environment for nucleotide seems unperturbed in mutant p97. The fact that mutant p97 can be crystallized in the presence of ATPS suggests a few competing hypotheses: one of which is the possibility of IBMPFD mutants acquiring a higher affinity for ATPS, leading to an alteration in ATPase activity. Alternatively, IBMPFD mutations could lead to a reduced ADP binding affinity. From structural studies, we can readily infer that the D1 nucleotide-binding site is seriously affected by IBMPFD mutations. However, how mutations at the N-D1 interface influence the D1 or D2 ATPase sites is not clear. Measuring the binding affinities of various nucleotides toward either D1 or D2 ATPase sites and the ATPase activities of the protein will provide biochemical indications as to how

**6. Wild type vs. IBMPFD mutant p97: Biophysical and biochemical** 

green ball with three coordinating water molecules in red.

IBMPFD mutations might affect the biochemical properties of p97.


**characteristics** 

undergoes a similar nucleotide-driven conformational change as observed in IBMPFD mutants (Fig. 4) (Tang et al., 2010). Therefore, the lack of success in crystallizing the Upconformation of wild type p97 suggests the presence of an intrinsic conformational heterogeneity or asymmetry in the N-domains of the homo-hexamer.

Fig. 4. **Nucleotide-driven conformational changes in solution observed by SAXS** Distance distribution functions, p(r), of p97 N-D1 normalized to a common total probability for wild type and mutant N-D1 fragments in the presence of ADP (solid line) ADP and ATPS (dashed line). The calculated distribution (Glatter, 1980) is shown of the left based on the crystal structure in the absence of bound solvent molecules.

#### **5.5 ADP and ATPS binding at the D1-doamin**

Although the binding of ATPS to the D1-domain triggers a dramatic movement of the Ndomain, the immediate vicinity of the D1 nucleotide-binding site shows only limited perturbations (Tang et al., 2010). When C atoms of the wild type and mutant N-D1 are superimposed, the adenosine moieties of the bound nucleotides align very well and the immediate environment around the adenosine moiety shows little change. By contrast, the phosphate groups in the alignment between ADP and ATPS forms differ, even though the same set of residues are involved in contacting the - and -phosphate in both the wild type and mutant structures. The -phosphate in the ATPS structure is stabilized by the ionic interaction with a magnesium ion (Mg2+, see below), by hydrogen bonds with Gln348 and Lys251, by Arg359, an Arg finger residue from a neighboring subunit, and by two water

undergoes a similar nucleotide-driven conformational change as observed in IBMPFD mutants (Fig. 4) (Tang et al., 2010). Therefore, the lack of success in crystallizing the Upconformation of wild type p97 suggests the presence of an intrinsic conformational

Fig. 4. **Nucleotide-driven conformational changes in solution observed by SAXS** Distance distribution functions, p(r), of p97 N-D1 normalized to a common total probability for wild type and mutant N-D1 fragments in the presence of ADP (solid line) ADP and ATPS (dashed line). The calculated distribution (Glatter, 1980) is shown of the left based on the

Although the binding of ATPS to the D1-domain triggers a dramatic movement of the Ndomain, the immediate vicinity of the D1 nucleotide-binding site shows only limited perturbations (Tang et al., 2010). When C atoms of the wild type and mutant N-D1 are superimposed, the adenosine moieties of the bound nucleotides align very well and the immediate environment around the adenosine moiety shows little change. By contrast, the phosphate groups in the alignment between ADP and ATPS forms differ, even though the same set of residues are involved in contacting the - and -phosphate in both the wild type and mutant structures. The -phosphate in the ATPS structure is stabilized by the ionic interaction with a magnesium ion (Mg2+, see below), by hydrogen bonds with Gln348 and Lys251, by Arg359, an Arg finger residue from a neighboring subunit, and by two water

crystal structure in the absence of bound solvent molecules.

**5.5 ADP and ATPS binding at the D1-doamin** 

heterogeneity or asymmetry in the N-domains of the homo-hexamer.

molecules associated with the Mg2+ ion (Fig. 5). Perhaps due to better diffraction resolution, in the mutant structures of p97, a Mg2+ ion is present in the nucleotide-binding site of every subunit. The Mg2+ ion is at the center of an octahedral *mer*-triaquo complex with the additional three oxo ligands coming from the side chain of the highly conserved Thr252 and from the - and -phosphates. The acidic residues of the DEXX sequence (Asp304 and Glu305) in the Walker B motif make hydrogen bonds with two of the water molecules in the Mg2+ coordination sphere and, additionally, Asp304 stabilizes Thr252. As expected, most of the changes in the nucleotide-binding environment are a consequence of the introduction of the -phosphate.

Fig. 5. **ATPS binding vicinity of the D1-domain** The nucleotide-binding pocket is located at the subunit interface. The two subunits are in different colors, green and gray. The ATPS molecule is shown as a stick model with carbon atoms in purple, oxygen in red, nitrogen in blue, phosphorous in magenta, and sulfur in yellow. The ATPS molecule is enclosed in a difference electron density cage contoured at the 2.5 σ level. The Mg2+ ion is shown as a green ball with three coordinating water molecules in red.

#### **6. Wild type vs. IBMPFD mutant p97: Biophysical and biochemical characteristics**

Structurally observed differences in the N-domain conformation of p97 strongly suggest a change in nucleotide binding affinities between wild type and IBMPFD mutants, even though the binding environment for nucleotide seems unperturbed in mutant p97. The fact that mutant p97 can be crystallized in the presence of ATPS suggests a few competing hypotheses: one of which is the possibility of IBMPFD mutants acquiring a higher affinity for ATPS, leading to an alteration in ATPase activity. Alternatively, IBMPFD mutations could lead to a reduced ADP binding affinity. From structural studies, we can readily infer that the D1 nucleotide-binding site is seriously affected by IBMPFD mutations. However, how mutations at the N-D1 interface influence the D1 or D2 ATPase sites is not clear. Measuring the binding affinities of various nucleotides toward either D1 or D2 ATPase sites and the ATPase activities of the protein will provide biochemical indications as to how IBMPFD mutations might affect the biochemical properties of p97.

IBMPFD and p97, the Structural and Molecular Basis for Functional Disruption 165

biphasic titration curve for ATPS, reflecting the first phase binding of the empty sites and the second phase of the pre-bound sites (Table 2) (Tang et al., 2010). Extraction by heat denaturation experiments of pre-bound ADP from the D1 sites of p97 supported the observation by ITC that the number of titratable sites is inversely related to the amount of pre-bound ADP present at the D1-domain (Tang et al., unpublished data). These findings suggest that while in wild type p97 a significant number of sites with pre-bound ADP in D1 domains of p97 are not exchangeable by a different form of nucleotides present in solution, IBMPFD mutations have altered the environment and lowered binding affinity for ADP to allow exchange, even though the change in the ADP binding site is too subtle to be detected

A consequence of lowered binding affinity for ADP in the D1-domain of IBMPFD mutants is the uniformly increased accessibility of D1 sites to various nucleotides present in solution such as ATPS. Indeed, successful crystallization of mutant p97 in the presence of ATPS is a result of this effect. On the contrary, the pre-bound ADP molecules in the D1-domains of some subunits of wild type p97 do not substitute for ATPS, in spite of having a higher affinity towards ATPS than ADP (Tang et al., 2010). Consequently, in the presence of excess ATPS in solution, there will be an admixture of ADP- and ATPS-bound D1 sites within a hexamer. Thus, the failure to crystallize wild type p97 in the presence of ATPS is a manifestation of non-uniformity in binding to nucleotides by different subunits in the D1-

By interacting with various adaptor proteins, p97 is able to play a role in a number of important cellular pathways. Therefore, alterations in adaptor protein binding by IBMPFD mutants have been investigated in both *in vivo* and *in vitro* experiments. Again, results from different groups are not completely consistent (Fernandez-Saiz and Buchberger, 2010; Manno et al., 2010). Isolated mutant p97 exhibited the same binding as wild type p97 towards adaptor proteins p47, Ufd1-Npl4, E4B and the human UFD-2 homolog. However, mutants showed impaired binding to ubiquitin ligase E4B in the presence of Ufd1-Npl4. *In vivo* pull-down experiments using HEK293 cells showed reduced binding towards the E4B and enhanced binding towards ataxin 3, thus resembling the accumulation of mutant ataxin 3 on p97 in spinocerebellar ataxia type 3 (Fernandez-Saiz and Buchberger, 2010). However, similar *in vivo* studies were done showing enhanced binding of the Ufd1-Npl4 pair by

One major contribution of p97 to cellular processes is its apparent participation in protein quality control and homeostasis, involving the ubiquitin-proteasome degradation pathway, ER associated degradation, and formation of autophagosomes. The multifaceted clinical presentation of patients with IBMPFD is consistent with the broad spectrum of p97 functions. Indeed, pathology and the cellular hallmarks such as the accumulation of inclusion bodies and rimmed vacuoles of IBMPFD can be reproduced both in cell culture and in animal models (Custer et al., 2010; Ju and Weihl, 2010; Weihl et al., 2006; Weihl et al., 2007). However, the structural and molecular basis of how p97 is involved in these different pathways and the mechanism of how p97 mutations lead to dysfunction remain elusive.

by crystal structures.

domains of hexameric p97.

**6.3 Changes in interaction with adaptor proteins** 

IBMPFD mutants but not for p47 (Manno et al., 2010).

**7. Implications concerning p97 function and disease** 

#### **6.1 Nucleotide-binding affinity and ATPase activity**

As mentioned earlier, p97 has two ATPase domains, D1 and D2; both are capable of hydrolyzing ATP. ATPase activity requires the presence of Mg2+ and is stimulated by high temperature (Song et al., 2003). While the D2-domain mediates most ATPase activity, the D1 domain contributes to heat-induced activity (Song et al., 2003). By using isothermal calorimetry (ITC) and Walker A mutants of either the D1- or D2-domains, it was shown that the binding of both domains to ATPS is similar with dissociation constants in the range of 1 M, but binding of ADP to the D1-domain is nearly 30-fold higher than that of the D2-domain, which is consistent with the higher ATPase activity of the D2-domain (Briggs et al., 2008).

Because the onset of IBMPFD is relatively late in life, the mutations in p97 are not expected to have dramatic defects in function. Indeed, IBMPFD equivalent mutations introduced into cdc48, a yeast homolog of p97, did not appear to interfere with normal cell growth (Esaki and Ogura, 2010). Although IBMPFD mutation sites are not in the immediate vicinity of the ATP-binding sites, as shown by crystal structures of p97, especially in the ATPS bound form, reports on how mutations affect ATPase activity of p97 vary. Two reports showed that mutants exhibit higher ATPase activity than the wild type to various degrees (Manno et al., 2010; Weihl et al., 2006). One paper, by contrast, reported no significant alterations in ATPase activity of four IBMPFD mutants (Fernandez-Saiz and Buchberger, 2010). Mutant p97 also displayed an even higher level of heat-stimulated ATPase activity (Halawani et al., 2009).

ITC measurements in N-D1 fragments of p97 showed that instead of predicted higher affinity for ATPS in the D1-domain, all mutants showed lowered affinity, up to five-fold weaker, towards ADP when compared to the wild type (Table 2) (Tang et al., 2010). More interestingly, all mutants displayed biphasic titration profiles toward ATPS, suggesting two distinct binding sites, one high and one low affinity site.


ITC with ATPS for IBMPFD mutants showed biphasic titration curves and data were fitted with a 2 site model. The *K*d values for mutants are derived from fitting to the first phase.

Table 2. Dissociation constants (*K*d) and binding stoichiometry (N) of wild type and mutant p97 N-D1 fragments for ATPS and ADP determined by ITC.

#### **6.2 Pre-bound ADP in D1 nucleotide-binding sites**

A unique characteristic of p97 was demonstrated by a urea denaturation experiment; the D1 sites are occupied by a significant portion of pre-bound ADP, which is difficult to release without denaturing the protein (Davies et al., 2005). ITC experiments with wild type p97 confirmed that a fraction of D1 sites are not accessible to nucleotide titration (Briggs et al., 2008). Consistent with the lowered *K*d values for ADP binding in IBMPFD mutants, D1 sites were shown to be significantly more accessible to ADP titration by ITC and displayed a

As mentioned earlier, p97 has two ATPase domains, D1 and D2; both are capable of hydrolyzing ATP. ATPase activity requires the presence of Mg2+ and is stimulated by high temperature (Song et al., 2003). While the D2-domain mediates most ATPase activity, the D1 domain contributes to heat-induced activity (Song et al., 2003). By using isothermal calorimetry (ITC) and Walker A mutants of either the D1- or D2-domains, it was shown that the binding of both domains to ATPS is similar with dissociation constants in the range of 1 M, but binding of ADP to the D1-domain is nearly 30-fold higher than that of the D2-domain, which is consistent with the higher ATPase activity of the D2-domain (Briggs et al., 2008).

Because the onset of IBMPFD is relatively late in life, the mutations in p97 are not expected to have dramatic defects in function. Indeed, IBMPFD equivalent mutations introduced into cdc48, a yeast homolog of p97, did not appear to interfere with normal cell growth (Esaki and Ogura, 2010). Although IBMPFD mutation sites are not in the immediate vicinity of the ATP-binding sites, as shown by crystal structures of p97, especially in the ATPS bound form, reports on how mutations affect ATPase activity of p97 vary. Two reports showed that mutants exhibit higher ATPase activity than the wild type to various degrees (Manno et al., 2010; Weihl et al., 2006). One paper, by contrast, reported no significant alterations in ATPase activity of four IBMPFD mutants (Fernandez-Saiz and Buchberger, 2010). Mutant p97 also displayed an even higher level of heat-stimulated ATPase activity (Halawani et al.,

ITC measurements in N-D1 fragments of p97 showed that instead of predicted higher affinity for ATPS in the D1-domain, all mutants showed lowered affinity, up to five-fold weaker, towards ADP when compared to the wild type (Table 2) (Tang et al., 2010). More interestingly, all mutants displayed biphasic titration profiles toward ATPS, suggesting two

Wild type 0.89 ± 0.28 0.12 ± 0.01 0.88 ± 0.18 0.35 ± 0.06 R95G 0.13 ± 0.02 0.56 ± 0.01 2.27 ± 0.11 0.62 ± 0.08 R155H 0.13 ± 0.01 0.61 ± 0.01 4.25 ± 0.54 0.72 ± 0.18 ITC with ATPS for IBMPFD mutants showed biphasic titration curves and data were fitted with a 2-

Table 2. Dissociation constants (*K*d) and binding stoichiometry (N) of wild type and mutant

A unique characteristic of p97 was demonstrated by a urea denaturation experiment; the D1 sites are occupied by a significant portion of pre-bound ADP, which is difficult to release without denaturing the protein (Davies et al., 2005). ITC experiments with wild type p97 confirmed that a fraction of D1 sites are not accessible to nucleotide titration (Briggs et al., 2008). Consistent with the lowered *K*d values for ADP binding in IBMPFD mutants, D1 sites were shown to be significantly more accessible to ADP titration by ITC and displayed a

*K*d (M) N *K*d (M) N

N-D1 p97 ATPS ADP

site model. The *K*d values for mutants are derived from fitting to the first phase.

p97 N-D1 fragments for ATPS and ADP determined by ITC.

**6.2 Pre-bound ADP in D1 nucleotide-binding sites** 

**6.1 Nucleotide-binding affinity and ATPase activity** 

distinct binding sites, one high and one low affinity site.

2009).

biphasic titration curve for ATPS, reflecting the first phase binding of the empty sites and the second phase of the pre-bound sites (Table 2) (Tang et al., 2010). Extraction by heat denaturation experiments of pre-bound ADP from the D1 sites of p97 supported the observation by ITC that the number of titratable sites is inversely related to the amount of pre-bound ADP present at the D1-domain (Tang et al., unpublished data). These findings suggest that while in wild type p97 a significant number of sites with pre-bound ADP in D1 domains of p97 are not exchangeable by a different form of nucleotides present in solution, IBMPFD mutations have altered the environment and lowered binding affinity for ADP to allow exchange, even though the change in the ADP binding site is too subtle to be detected by crystal structures.

A consequence of lowered binding affinity for ADP in the D1-domain of IBMPFD mutants is the uniformly increased accessibility of D1 sites to various nucleotides present in solution such as ATPS. Indeed, successful crystallization of mutant p97 in the presence of ATPS is a result of this effect. On the contrary, the pre-bound ADP molecules in the D1-domains of some subunits of wild type p97 do not substitute for ATPS, in spite of having a higher affinity towards ATPS than ADP (Tang et al., 2010). Consequently, in the presence of excess ATPS in solution, there will be an admixture of ADP- and ATPS-bound D1 sites within a hexamer. Thus, the failure to crystallize wild type p97 in the presence of ATPS is a manifestation of non-uniformity in binding to nucleotides by different subunits in the D1 domains of hexameric p97.

#### **6.3 Changes in interaction with adaptor proteins**

By interacting with various adaptor proteins, p97 is able to play a role in a number of important cellular pathways. Therefore, alterations in adaptor protein binding by IBMPFD mutants have been investigated in both *in vivo* and *in vitro* experiments. Again, results from different groups are not completely consistent (Fernandez-Saiz and Buchberger, 2010; Manno et al., 2010). Isolated mutant p97 exhibited the same binding as wild type p97 towards adaptor proteins p47, Ufd1-Npl4, E4B and the human UFD-2 homolog. However, mutants showed impaired binding to ubiquitin ligase E4B in the presence of Ufd1-Npl4. *In vivo* pull-down experiments using HEK293 cells showed reduced binding towards the E4B and enhanced binding towards ataxin 3, thus resembling the accumulation of mutant ataxin 3 on p97 in spinocerebellar ataxia type 3 (Fernandez-Saiz and Buchberger, 2010). However, similar *in vivo* studies were done showing enhanced binding of the Ufd1-Npl4 pair by IBMPFD mutants but not for p47 (Manno et al., 2010).
