`
`Tissue deposition of soluble autologous proteins as insoluble
`| amyloidfibrils is associated with serious diseases including systemic
`amyloidosis, Alzheimer’s disease, and transmissible spongiform
`encephalopathy, but the mechanisms of amyloid fibrillogenesis
`are poorly understood’, Although the diverse human proteins
`that can form amyloidfibrils in vivo have unrelated sequences and
`tertiary folds,
`they can all polymerize into fibrils with similar
`ultrastructural appearance and identical
`tinctorial properties”.
`Furthermore, the core structure of all amyloid fibrils consists of
`B-sheets with the strands perpendicular to the long axis of the
`fibre**. Knowing which conformational rearrangements converge
`on the samefinal fold is important for understanding the determi-
`nants of protein structure, and may enable the development of
`rational approachesto the treatment of amyloid diseases. However,
`not previously been isolated, were produced in the baculovirus
`although a lot is known about the mutations and substitutions
`responsible for hereditary and acquired amylodosis
`(see,
`for
`expression system, and the correct mass of each purified, recombi-
`nant protein was demonstrated by electrospray ionization mass
`example, refs 5-14), there is little detailed information about the
`spectrometry(ESI-MS) (Table 1). They were all enzymatically
`relationship between structure and folding in amyloid proteins.
`active, although the Asp67His variant had a higher Ky and lowerka, |
`The two known natural mutations in the human lysozyme gene
`both cause autosomal dominanthereditary amyloidosis’®. Affected
`than the wild type and the Ile56Thr variant (Table 1).
`individuals are heterozygous for
`single base changes which
`The native folds of the two amyloidogenic variants determined by
`encode non-conservative amino-acid substitutions, [le56Thr and
`X-ray crystallography both resemble that of the wild-type protein”
`(Fig. 1a), and all have the four correct, intact disulphide bonds.
`Asp67His, respectively, and the amyloid fibrils consist exclusively of
`the variant protein’® (see below). The structure, dynamics and
`However,substitution ofAsp 67 by histidine destroys the network of
`
`folding of c-type lysozymes and the related a-lactalbumins have hydrogen bondsthatstabilizes the B-domain,resulting inalarge,
`been studied comprehensively'***. The identification of lysozyme as
`concerted movementof the B-sheet and the long loop within the B-
`an amyloidogenic protein was therefore of particular interest.
`domain away from each other, distortion of the active site, and an
`Here we present a detailed analysis of the structure, stability,
`overall displacement of backbone atomsin the vicinity of residues
`48 and 70 by as muchas 11 A (Fig. 1a). The crystal structure of the
`conformational dynamics andfibrillogenic properties of the amy-
`loidogenic lysozyme variants which links the formation of amyloid
`Ile56Thr variant does not show such changes, indicating that the
`with the folding behaviourof proteins.
`movements in the B-domain are not the direct cause of amyloido-
`genicity. However, closer inspection of the two structures demon-
`strates that subtle, but structurally significant, changes at
`the |
`interface region between the a- and B-domains occur in both
`variants (Fig. 1b, c). fle 56 is a pivotal residue for the structural
`integrity of the lysozymefold in thatit links the two domains;its
`importance is emphasized by its high conservation in the lysozyme
`sequences’”. An increased B-factor (11.7 A’) was found for the Cx
`Amgen Exhibit,2047
`Apotex Inc. et al. v. Amgen Inc. et al., IPR2016-01542
`Page 1
`
`
`
`Ky and Kea are given as mean (s.d.).
`
`Amyloidogenic variants have native folds
`Wild-type lysozyme and the amyloidogenic variants'*, which have
`
`§ Present addresses: Dipartimentodi Biochimica, Universita di Pavia, Via Taramelli 3B, 27100 Pavia,Italy
`{V.B.); Department of Biochemistry and Molecular Biology, University of Leeds, Leeds LS2 9JT, UK
`(S.E.R.).
`
`NATURE! VOL 385127 FEBRUARY 1997
`
`
`
`aes
`
`instability,unta
`unfolding and
`aggregation of human
`lysozymevariants
`underlying amyloid fibrillogenesis
`
`  
`
`David R. Booth*|, Margaret Sunde?|, Vittorio Bellotti*s, Carol V. Robinson:, Winston L. Hutchinson’,
`Paul E. Frasers, Philip N. Hawkins*, Christopher M. Dobson, Sheena E. Radford: {, Colin C. F. Blake;
`& Mark B. Pepys*
`
`* Immunological Medicine Unit, Royal Postgraduate Medical School, Hammersmith Hospital, London W12 ONN, UK
`+ Laboratory ofMolecular Biophysics and t New Chemistry Laboratory, Oxford Centre for Molecular Sciences, University of Oxford, Oxford OX1 3QT, UK
`§ Centre for Research in Neurodegenerative Diseases, University of Toronto, Toronto M5S 3H2, Canada
`\| These authors contributed equally to this work.
`
`Tissue deposition of soluble proteins as amyloid fibrils underlies a range of fatal diseases. The two naturally occurring
`humanlysozymevariants are both amyloidogenic, and are shownhere to be unstable. They aggregate to form amyloid
`fibrils with transformation of the mainly helical native fold, observed in crystal structures, to theamyloid fibril cross-6
`fold. Biophysical studies suggest that partly folded intermediates are involvedin fibrillogenesis, and this may be
`relevant to amyloidosis generally.
`
`
`
`Table 1 Molecular masses and enzyme characteristics of wild-type and
`variant lysozymes
`Protein
`Electrospray ionization
`Enzyme properties
`mass spectrometry
`Observed M,
`Predicted ,
`
`Kea (Ms~')
`
`Ky (uM)
`
`
`
`
`Amgen Exhibit 2047
`Apotex Inc. et al. v. Amgen Inc. et al., IPR2016-01542
`Page 1
`
`

`

` articles Figure 1 a, Overlay of ribbon diagrams representing the structures of wild-type
`
`
`
`atom ofresidue 56 in the [le56Thrvariant, relative to that of Ile 56 in
`the wild-type protein (6.4 A’). This is presumably due to the now
`hydrophilic side chain being in an unfavourable hydrophobic
`environment, even though its hydrogen-bonding potential seems
`to be at least partly satisfied by a hydrogen bond(length 3.3 A) to
`one of the water molecules found in both the wild-type and the
`variant structures. In the Asp67His variant, the changes in the
`conformations of the ®-sheet and long loop are transmitted down
`to residue 56, resulting in a new orientation fortheside chain and an
`increase in the B-factor of the Ca atom ofthelatter (9.7 A’). This
`suggests that the crucial interface region between the a- and B-
`domainsis less constrained in both variants than in the wild-type
`protein. This common feature of both structures implies that it
`could be an important factor in their amyloidogenic properties.
`Substitution ofIle 55 in hen lysozyme(the correspondingresidue to
`Tle 56 in humanlysozyme) by threonine also reduces the stability of
`the protein and generates a tendency to aggregate”, further sup-
`porting the viewthatthis residueis essential for the maintenance of
`the lysozymefold.
`
`Fibril formation occursin vitro
`In contrast to the reversible thermal denaturation of wild-type
`lysozyme from both natural and recombinant sources, the amyloi-
`dogenic variants were inactivated by heating (Fig. 2). The variants
`were also less stable than wild-type lysozyme, with unfolding
`transition midpoints reproducibly 10°C or more below that of
`
`|
`
`
`
`|
`
`human lysozyme (grey) and the soluble form of Asp67His lysozyme (coloured
`from blue at the N terminus to red at the C terminus). Red arrowsindicate the
`relative movementin the positions of residues 45-54 and 67-75 in the Asp67His
`variant compared with thosein the wild-type protein. The four native disulphide
`bondsin the wild-type protein and both variant structures are shownin yellow.
`Inset, plot showing the displacementof the residues of the Asp67His(solid line)
`and lleS6Thr (brokenline) variants from their positions in the wild-type protein. b,
`Ribbon diagram of the B-domain of the wild-type protein, showing thecritical role
`of Asp 67 in the network of hydrogen bondsthat stabilizes this domain. ¢, Ribbon
`diagram illustrating the same region in the Asp67His variant and the disruption of
`the domain that occurs when the aspartateat position 67 is replaced by histidine
`and the hydrogen-bonding networkis destroyed.
`
`the wild-type protein. Furthermore,both variantseventuallylostall
`activity when incubated at pH 7.4 at the physiological temperature
`of 37°C, whereas the wild-type protein retained full activity under
`these conditions (data not shown).
`The amyloidogenic lysozymevariants also aggregated on heating,
`unlike the wild-type protein. The rate and extent of aggregation
`varied with protein concentration and the expression batch, and
`althoughthe aggregates stained with Congo red”, they generally did
`not give the green—red birefringence in polarized light that is
`pathognomonic of amyloid. Nevertheless, negatively stained elec-
`tron micrographsrevealed rigid, non-branchingfibres of indeter-
`minate length and approximately 8-10nm diameter, with the
`typical appearance of amyloid fibrils. Fibrils were also seen in
`electron micrographs of the sediment that formed spontaneously
`at 4°C in concentrated solutions of both Ile56Thr and Asp67His
`variants (Fig. 3). Strikingly, one preparation of heated Asp67His
`lysozyme contained fibres that stained with Congo red and did
`display the diagnostic green birefringence, confirming the capacity
`of the variant
`to form the classical amyloid structure in vitro
`independently of any other component.
`Fourier-transform infrared spectroscopy (FTIR) of recombinant
`Asp67His lysozyme heated under conditions in which fibrils form
`demonstrated a predominance of B-structure and a loss of helical
`structure relative to the wild-type protein (Fig. 4). The FTIR
`spectrum also indicated the persistence of some helical structure
`in the heated sample. This couldarise from residualsoluble forms of
`NATUREI VOL 385! 27 FEBRUARY 1997
`
`788
`
`Page 2
`
`Page 2
`
`

`

`a
`
`A Wild type
`
`Wle56Thr\a
`
`
`
`articles
`
`Figure 3 Electron micrograph of sediment from the lle56Thr lysozyme, after
`standing at 1 mgm ° for 14 days at 4°C in 10MM HEPES, 1M LiCl, pH 8.0. Scale
`bar, 100 nm.
`
`
`
`30
`
`40
`
`60
`50
`Temperature (°C)
`
`70
`
`80
`
`Figure 2 Melting
`lysozymes.
`
`temperatures of wild-type and amyloidogenic variant
`
`100
`
`80
`
`40
`
`20
`
`
`
`
`
`Residualactivity(%)
`
`
`
`
`
`Table 2 Recovery of enzymatically active lysozyme from ex vivo Asp67His lysozyme amyloid fibrils
`V, on gel filtration (ml)
`Lysozyme recovered (pg)
`Protein by Azgo
`Active enzyme
`
`Electrospray ionization mass spectrometry
`Observed M.
`Predicted M,
`14,715 for Asp67His lysozyme
`| Ex vivo Asp67His lysozymefibrils
`Aspé67His
`|
`ith MetSO
`6M
`
`HCl
`solubilToca
`ym
`
`
`
`
`
`Ex
`vivo
`Asp
`is
`lysozy'|
`
`solubilized in 6 M guanidinium-HCl, pH 6.7,
`0.1% 2-mercaptoethano!
`V. given as mean (s.d.} from 3 experiments.
`
`the lysozyme variant, from persistence of helical structure in the
`fibril, or both.
`
`Fibrillogenesisis reversible
`Our identification of a second Asp67His family’, apparently
`unrelated to the original kindred’, provided the opportunity to
`study Asp67His lysozyme amyloid fibrils. The X-ray fibre diffrac-
`tion pattern (not shown) contains distinctive reflections at 4.6—
`4.8 A on the meridian and at 8-14 A on the equator of the image,
`indicating that the underlying ordered structure is a B-sheet in
`which the constituent B-strandsareat right anglesto thefibre axis*.
`Such cross-B structures are characteristic of amyloid and havealso
`been described in the glutamine repeats that are associated with
`several neurodegenerative diseases. including Huntington’s disease,
`and which cause oligomerization of proteins”. The fibre diffraction
`pattern of ex vivo Asp67His lysozymefibrils containsno reflections
`attributable to helical structure, suggesting that, if helices persist
`after transformationofthe soluble protein to the fibrillar form, they
`are not regularly ordered.
`As previously reported for Ile56Thr fibrils’*, 85% of the total
`protein in water-extracted” Asp67His fibrils ran in reduced SDS—
`PAGE in the same position as intact monomeric lysozyme; the
`remainderconsisted of oligomeric lysozyme aggregates andtraces of
`uncharacterized high-molecular-weight material that is seen in all
`ex vivo amyloid fibrils. However, like the fibrils from the Ile56Thr
`case'*, the Asp67His lysozymefibrils could not be dissociated to a
`form detectable by ESI-MS using either acetonitrile/acetic acid
`mixtures or up to 100% formic acid. We therefore solubilized some
`of the ex vivo Asp67His fibrils by denaturation in 6M guanidine
`HCl, isolated the lysozymebygel filtration in the same denaturing
`conditions, and attempted to refold it by dialysis into water at pH
`NATURE VOL 385127 FEBRUARY 1997
`
`in which natural wild-type lysozymeis stable.
`3.8, a solvent
`Although somereaggregation occurred during dialysis, the recov-
`ered lysozymewasdetectable by ESI-MSwith a mass corresponding
`to intact, monomeric, Asp67His variant (Table 2). The exclusive
`presence of variant lysozymein either Asp67His or [le56Thr(ref.
`15) ex vivo amyloid fibrils indicates that their pathological aggrega-
`tion does not engage wild-type lysozyme in vivo, presumably
`because the wild type has greater stability than the variants.
`Remarkably, the refolded Asp67His variant lysozyme was enzy-
`matically active, in contrast to the absence of any activity in the
`original fibril preparation, using soluble penta-N-acetyl-B-chito-
`pentaoside substrate. Although the variant protein must have
`undergone major conformational change im vivo to form character- |
`istic cross-B amyloid fibrils, it remained able, after unfolding, to
`renature spontaneously into the active enzyme. However, when the
`disulphide bonds within the chain were reduced with 2-mercap-
`toethanol during solubilization and unfolding, no lysozymeactivity
`was observed and the protein could not be detected by ESI~MS |
`(Table 2).
`
`Stabilized molten globule intermediate
`Wehave used circular dichroism to monitor the unfolding beha-
`viour of the two lysozymevariants under conditions in which they
`form fibrils in vitro. The results (Fig. 5a—d) showed that both
`variants were less thermostable than the wild-type protein, with
`midpoints of denaturation approximately 12°C lower than that of
`the wild-type protein at pH 5.0. More importantly, however, the
`unfolding transition of the two amyloidogenic variants, although
`reversible under conditions in which fibril formation did not occur,
`was not cooperative. This resulted in a partly folded state being
`significantly populated near the midpoint of unfolding. This state |
`789
`
`Page 3
`
`Page 3
`
`

`

`articles
`
`
`
`Table 3 Crystallographic data statistics
`Parameters
`
`Structure determination
`Resoiution (A)
`Data completeness (%)
`Io for all hkt
`17a at high-resotution limit
`Observations
`Unique reflections
`Space group
`Unit cell (A)
`g-angle
`Reym (%)
`Structure refinement
`Resolution (A)
`Non-H atoms
`Water molecules
`R-factor(reflections)
`.
`Average B-value
`R.m.s.d. bond lengths(A)
`R.m.s.d. bond angles (°)
`R.m.s.d. dihedral angles(°)
`R.m.s.d. improper angles (°)
`
`Recombinant
`wild type
`
`.
`30-18A
`89.9
`124
`3.7
`51,285
`10,221
`P2422)
`56.62 x 60.88 x 33.79
`
`8.3
`
`.
`8-1.8A
`1,029
`38
`21.3 (10,216)
`13.11
`0.013
`1.674
`24.132
`1.463
`
`Asp67His variant
`
`lieS6Thr variant
`
`30-1.75A
`90.9
`85
`34
`41,333
`11,071
`P2,
`37.34 x 31.86 x 51.50
`102.66°
`10.3
`
`8-18A
`1,031
`115
`22.8 (10,216)
`13.54
`0.015
`1.814
`23.076
`1.666
`
`30-18A
`92.8
`11.2
`3.6
`49,524
`10,578
`P2,2,2,
`56.80 x 60.89 x 33.70
`
`9.2
`
`8-18A
`1,028
`43
`211 (10,613)
`16.76
`0.013
`1.643
`24.601
`1.391
`
`
`
`QoO
`S
`2
`2
`<<
`
`Heated
`Aspé7His
`lysozyme
`
`
`Native Asp67His
`lysozyme
`
`1,550
`1,650
`1,750
`Wave number (cm ~1)
`
`Figure 4 FTIR spectra, offset for comparison, of soluble Asp67His lysozyme, and
`after heating to inducefibril formation. The dominant absorption band centred at
`1,655 cm~'
`in the untreated samplereflects the large a-helical componentin the
`soluble, native protein. The shift to absorption at about 1,630cm~' after heating
`indicates an increase in B-sheet content. The shoulderat 1,655 cm 7 ' inthe heated
`sample demonstrates the persistence of some helix.
`
`had substantial helical secondary structure but lacked persistent
`tertiary interactions (Fig. 5b, d). Such behaviouris quite different
`from the cooperative unfolding displayed by the wild-type protein
`under these conditions (Fig. 5a, c), but is similar to the thermal
`unfolding of wild-type human lysozyme under conditions of
`extremely low pH, where the protein unfolds through a partially
`structured intermediate with circular-dichroism properties similar
`to those identified here for the variants at pH 5.0 (ref. 30). More-
`over, species with similar properties have been identified on the
`kinetic or equilibrium folding pathways of other lysozymes and a-
`lactalbumins”!*’. At the midpoint of thermal denaturation,
`the
`partly folded amyloidogenic intermediates bound the hydro-
`phobic dye 1-anilino-naphthalenesulphonic acid (ANS) (Fig. 5f);
`this is one of the major characteristics of the previously character-
`ized lysozymeand a-lactalbumin molten globules”. The Nle56Thr
`variant also bound ANSat 20°C, although it generated weaker
`
`indicating the
`its midpoint of unfolding,
`fluorescence than at
`presence of exposed hydrophobic regions even at this temperature.
`
`Transient unfolding
`The conformational dynamics of the wild-type and variant proteins
`in solution at 37 °C were investigated by using ESI-MS to monitor
`the exchange of the labile amide and side-chain hydrogens with
`solvent deuterons (Fig. 6). The hydrogen exchange kinetics of the
`two amyloidogenic variants were remarkable in that there was very
`little protection from exchange (Fig. 6);
`in contrast, about 55
`hydrogens were strongly protected from exchange in the wild-
`type protein under these conditions (Fig. 6). The lack of protection
`of the variants cannot be explained simply by their thermal
`destablization relative to the wild-type protein; a chemically mod-
`ified hen lysozyme, which lacks a single disulphide bridge, has a
`midpoint of unfolding 24°C lower than the wild-type protein, but
`still shows significant protection against hydrogen exchange”.
`Rather, these results suggest that the alterations in the domain
`interface ofthe Ile56Thr and Asp67Hisvariants reduce the stability
`and cooperativity of the native fold such that both the amplitude
`and frequencyof the native-state fluctuationsare increased, even at
`37 °C, to an extentthat allows solvent wateraccess to the interior of
`the protein. The degreeofprotection ofthe variants is similar to that
`previously observed by ESI-MS in the well-characterized, partly
`folded, molten globule state of a-lactalbumin”’. We suggest, there-
`fore, that the aggregation-prone, partly folded forms are present in
`dynamic equilibrium with the native protein at significant concen-
`trations, even under conditions where the native state is thermo-
`dynamically stable, and could be important determinants of the
`amyloidogenic properties of the variants and the slow deposition of
`fibrils observed at 4°C.
`The previously characterized kinetic and equilibrium partly
`folded intermediates of lysozymes and the a-lactalbuminsall have
`persistent structure in the a-domain but lack stable, native-like
`structure in the @-domain””"”’, Based on the presentresults, we propose
`a model for lysozyme fibrillogenesis in which association of the
`partly folded forms of the variants occurs through the unstable
`B-domain (Fig. 7). In support of this, a peptide corresponding to
`the B-sheet region of hen lysozyme has been shown to form
`extensive intermolecular B-structure™*. The developmentof stable
`B-structure through such intermolecular association could thenact
`as a template for the progressive recruitment of polypeptide chain
`into the nascent
`fibril, with the growth of hydrogen-bonded
`B-structure providing the context® for the deposition of poly-
`
`790
`
`NATURE! VOL 385/27 FEBRUARY 1997
`
`Page 4
`
`
`
`Page 4
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`100 4 t=90'AttinalWild type
`
`t=0
`
`Figure 6 Kinetic profiles of hydrogen exchange at pH 5.0, 37°C, for wild-type
`human lysozyme (circles) and lle56Thr variant(triangles) monitored by ESI-MS.
`
`The exchangeprofile for the Asp67His variant is very similar to that shown here
`for the le56Thr variant. The plain blackline is the simulated curve predicted®for a
`
`completely unstructured peptide with the sequence of human lysozyme at pH 5.0
`and 37°C. The twoinserts represent mass spectra obtained forthe wild-type and
`
`variant protein before exposure to D,O (f= 0), 90min after the initiation of
`exchange(f = 90), and after heating to 70°C for 15min to facilitate compiete
`
`exchange(trina|). Data were corrected for the residual 10% H.O.
`
`ened14,700 14,900 15,000
`Mr
`xz A —_Az—
`T
`100
`Time (mins)
`
`g
`o
`g3S 50
`2B
`2oa
`
`|
`
`0 4
`
`T
`0
`
`14,700
`
`_Y(-~—_1d
`44,900
`15,000
`vr
`
`
`
`lleS6Thr
`t=0
`
`
`
`—*—
`T
`200
`
`ms)
`
`NATURE! VOL 385127 FEBRUARY 1997
`
`791
`
`Page 5
`
`
`
`
`
`
`
`articles
`
`
`
`200
`210
`220
`230
`240
`
`Wavelength (nm}
`
`midpoint of thermal denaturation, 7(f}. The midpoint of thermai denaturation
`was 74°C for wild-type human lysozymeand 62°C for the Asp67His and lle56Thr
`variants. Fluorescence intensity (arbitrary units) is shown for ANS-containing
`buffer solution (solid line), wild-type lysozyme (©), Asp67His lysozyme(ml) and
`tle56Thr lysozyme(A). (6), motar ellipticity.
`
`~ oO
`oO
`
`a
`
`2 S
`
`4
`
`s
`€ua
`aL
`£oO
`m 8
`
`12
`
`200
`
`210
`
`220
`
`230
`
`240
`
`os°2x
`
`2S
`
`(8y(degcm?dmol!res-!)
`
`
`
`Fluorescenceintensity
`
`Wavelength (nm)
`
`Figure 5 Thermal denaturation of wild-type and Asp67His human lysozymes.Far-
`UV (a) and near-UV (¢) CD spectra for wild-type human lysozyme and far-UV (b}
`
`and near-UV (d) CD spectra for the Asp67His variant lysozyme,all obtained in
`
`water at pH 5.0 and collected at 20 (©), 60
`(Cl), 70 (@) and 95°C (a). Binding
`
`of 1-anilino-naphthalenesulphonic acid (ANS) to the proteins at 20 °C (e) and atthe
`
`
`
`
`Page 5
`
`

`

`Further assembly of protofilaments
`
`Figure 7 Proposed mechanism for lysozyme amyloid fibril formation. Blue, B-
`| sheet structure: red, helical structure; dotted lines, undefined structure. A parily
`folded, moiten globule-like form of the protein (II), distinct from the native (i} and
`denatured(Ill) states, self-associates through the -domain (IV)to initiate fibril
`formation. This provides the template for further deposition of protein and for the
`development of the stable, mainly B-sheet, core structure of the fibril (V). The
`undefined regions in V represent the possibility that not all of the polypeptide
`sequenceis involvedin the cross-B structure. The nature of this residual structure
`in V is not known, and the figure fs not intended to represent any defined
`secondarystructural type (see text).
`
`
`
`|
`
`=
`
`[2
`
`Vv
`vv
`
`ll 1
`
`—
`
`ill
`
`wwn
`wn
`
`SRwen
`|
`vo
`pe ICON
`vil PPPS
`ISIS ro,
`OSS EE
`
`peptide chainin the stable cross-B fold. The FTIR data indicate that
`fibrillogenesis involves an increase in B-sheet structure; conversion
`| of a- to B-structure will be easier in the molten globule state than
`|
`the native state because of the much lower cooperativity of the
`| unfolding process”.
`
`Mechanism of amyloid fibril formation
`All amyloid fibrils have similar morphological and tinctorial char-
`acteristics and are predominantly B-sheetstructures, indicating that
`a conformational change, involving a helix-to-sheet transition in
`some proteins, occurs during fibril formation’. As in lysozyme
`amyloidosis, amino-acid substitutions responsible for amyloid
`formation in immunoglobulin light chain’ and transthyretin
`variants" affect the stability of the proteins and their tendency to
`aggregate. It has been suggested that molten globule states are
`critical in protein folding and related structural transitions**’’. We
`proposethat transient population of the amyloidogenic proteins in
`a molten globule-like state that Jacks global cooperativity is an
`important feature of the conversion from the soluble to the fibrillar
`form. The structure of a domain of the prion protein PrP(121-
`231)** also demonstrates that residues for which mutations are
`associated with prion disease are involved in maintenance of the
`hydrophobic core. It has also been suggested that the conversion of
`the cellular form to the infectious form, which involves helix-to-
`sheet conversion, may beinitiated by the B-sheet elements of the
`native structure’, Thereis no evidence for infectivity of other types
`of amyloidosis or for conversion of non-amyloidogenic wild-type
`proteins by exposure to amyloidogenic variants. Nevertheless, the
`mechanism we have described for lysozyme amyloidosis (Fig. 7),
`proceeding from the soluble forms of amyloidogenic precursor
`
`792
`
`
`
`|
`
`proteins through a transient population of intermediates with the
`structural characteristics of molten globules, and on to intermol-
`ecular B-sheet association, may occurgenerally in the amyloidoses.
`Note added in proof: After submission of this manuscript Funahashi
`et al.” reported that
`the crystal structure of He56Thr variant
`lysozyme is similar to that of wild-type human lysozyme, as
`shown here. Their physicochemical studies also demonstrate
`reduced protein stability and altered folding kinetics, strongly
`
`supporting the idea that partly folded intermediates play an
`
`
`
`importantrole in lysozymefibril formation.
`
`Methods
`
`Lysozyme expression. Human wild-type and Asp67His variant lysozyme
`cDNAs were amplified from macrophage RNA of the Asp67His proband.
`The 5’ primer, CTTGGATCCCTAGGCACTCTGACCTAGCAGT,contained a
`BamHIsite and targeted sequencein the untranslated region of the cDNA. The
`3’ primer, NNNNNNTCTAGATTACACTCCACAACCTTG,contained an Xbal
`site and 6 random nucleotidesatits 5’ endtofacilitate cleavage”. The Ile56Thr
`variant sequence was obtained by in vitro mutagenesis of wild-type cDNA
`(pAlter system, Promega). The three cDNAswerecloned into the BamHI/Xbal
`sites of pBacPAK8,transfected into Sf9 cells, and recombinant baculoviruses
`wereselected and amplified (Clontech). Lysozyme was detected” in medium
`frominfected cells; spinner cultures ofHi5 and Sf9 cells, infected at multiplicity
`of infections from 1 to 15, yielded 2-20 mg!~!.
`Isolation and characterization of recombinant lysozymes. Lysozymes were
`isolated by cation-exchange chromatography (Macroprep S$, BioRad) and
`FPLC gelfiltration (Superose 12, Pharmacia), and gavesingle bands on reduced
`SDS 8-18% gradient PAGE (Pharmacia ExcelGel) stained with silver. Lysozyme
`enzyme kinetics were determined with penta-~N-acetyl-B-chitopentaoside”’.
`For electron microscopy, protein diluted in water was placed on a formavar-
`coated grid and negatively stained with 2% sodium phosphotungstate.
`Crystal-structure determination. All crystals were grown by vapourdiffusion
`at 20°C; wild-type lysozyme from 30 mM sodium phosphate, 2.5 M NaCl, pH
`4.9, Asp67His from 0.1 M ammonium sulphate, 30% PEG 8000 and Ile56Thr
`from 0.16M ammonium sulphate, 24% PEG 8000. Dropsinitially contained
`equal volumesof protein (10 mgml~' in 10mM HEPES, 0.4-0.5 M LiCl, pH
`7.1-8.0) and reservoir buffer. X-ray data were collected at 15°C using a MAR
`RESEARCHimageplate, mounted on a Rigaku rotating anode X-ray generator.
`Data processing and reduction were performed with the DENZO and SCA-
`LEPACK programs”. The CCP4 suite of programs” was used for map
`calculation and coordinate analysis. The structures were refined using cycles
`of restrained molecular dynamics and positional refinement in X-PLOR 3.1
`(ref. 44) and manual fitting with the interactive graphics programme O*.
`Crystallographic data statistics are given in Table 3. Recombinant wild-type
`human lysozymedata werecollected as a control set. The initial 2F, — F, map
`for the Ile56Thr variant structure was produced with phases calculated from
`the refined model of recombinant wild-type human lysozyme’’. The structure
`of Asp67His lysozyme was solved by molecular replacement using X-PLOR”.
`The solution indicated changes in the conformation of the B-domain (regions
`45-54 and 66-75). The structure was rebuilt manually and refined with cycles
`ofmolecular dynamics and modelbuilding guided by interpretation ofelectron
`density maps. Figures were gencrated with O* and a version of MOLSCRIPT*
`modified by R. Esnouf.
`Thermal stability. Melting temperatures were determined by placing wild-
`type and variantlysozymesat 2.5 1g ml~’ in 20mM HEPES, 100 mM LiCl, pH
`7,23, containing 1% (w/v) bovine serum albumin,for 15 min at the tempera-
`tures shown,then cooling to 21°C for 15 min, before enzyme assay” at 21°C.
`Infrared spectroscopy.Infrared spectra of proteins (~5 mg ml~') dissolved
`in D,O buffer containing 20 mM Tris-HCl, pH 7.0 (uncorrected for deuterium
`effects) were collected (Bruker IFS-55 spectrometer, 2cm 7 ' resolution) before
`and after heating to induce unfolding andfibril formation.
`Recovery of active lysozyme from ex vivo lysozyme amyloid fibrils.
`Amyloid fibrils were isolated by water extraction from amyloidoticliver tissue
`of a patient with the Asp67His lysozyme gene mutation who underwent
`emergencyliver transplantation following spontaneous rupture oftheliver.
`Lyophilized fibrils were incubated for 72 h in 6 M guanidine HCl, pH6.7, with
`or without 0.1% (w/v) 2-mercaptoethanol,
`then centrifuged to remove
`
`NATURE] VOL385/27 FEBRUARY 1997
`
`Page 6
`
`Page 6
`
`

`

`
`
`insoluble material. The supernatants were fractionated by FPLC gel filtration
`(Superose 12, Pharmacia) eluted with the correspondingsolvent; the main peak
`in each case, composed of monomeric lysozyme, was pooled and dialysed
`against H,O, pH 3.8. Lysozyme enzymeactivity was quantified”, and the
`molecular massof the solubilised protein was determined by ESI-MS”. The
`elution volume( V.) of lysozyme recoveredin the presence of mercaptoethanol
`wassignificantly lower than that obtained without reduction, corresponding to
`the expected larger volume of unfolded lysozyme without disulphide bridges,
`confirming that reduction had occurred. This material did not recover enzyme
`activity and gave no signalin the mass spectrometer, presumably becauseofits
`propensity to aggregate. Natural ex vivo wild-type lysozyme dissolved in 6M
`guanidine-HCl, pH 6.7, and run on the same columnas a control, eluted with
`the same V, as the Asp67His lysozyme from the amyloidfibrils.
`Circular dichroism. Spectra were collected at 1-nm intervals using a JASCO
`J720 spectropolarimeter with 1 mm and 10 mm path-length quartz cuvettes in a
`temperature-controlled housing, over the wavelength ranges 190-250 nm and
`250-360 nm respectively, The protein samples were at 0.22 mg ml7' in H,0,
`pH 5.0, based on jg) for 1m path length = 25.5.
`in 50mM sodium
`ANS binding and fluorescence.Proteins at 0.2 mgml~'
`acetate, pH 5.0, with final ANS concentration 0.1 mg ml~’, were analysed in a
`Perkin Elmer luminescence spectrometer LS50B with a temperature-controlled
`cell; excitation wavelength, 385nm; total fluorescence emission monitored
`between 400 and 600 nm.
`Deuterium exchange and mass spectrometry. Proteins were washed
`extensively in water, pH 3.8, then equilibrated in water, pH 5.0, at 20 uM for
`determination of mass spectra’’. Hydrogen exchange wasinitiated by a 10-fold
`dilution from protein in H,O at pH 5.0 and 37°C into D,O at pH 5.0
`(uncorrected for deuterium effects) and 37°C.
`Received 12 September 1996; accepted 27 January 1997.
`1. Pepys, M. B. in Samter’s Immunologic Diseases (eds Frank, M. M., Austen, K. E, Claman, H. N. &
`Unanue, E. R.) 637-655 (Little, Brown and Company,Boston, 1994).
`2, Tan,S. ¥. & Pepys, M. B. Amyloidosis. Histopathology 25, 403-414 (1994).
`3. Glenner, G. G. Amyloid deposits and amyloidosis—theB-fibrilloses. I. N. Engl. J. Med. 302, 1283-
`1292 (1980).
`4. Blake, C. C. F & Serpell, L. C. Synchrotron X-ray studies suggest that the core of the transthyretin
`amyloid fibril is a continuous B-sheet helix. Structure 4, 989-998 (1996).
`5. Fraser, P. E. et al. Fibril formation by primate, rodent and Dutch-hemorrhagic analogues ofAlzheimer
`B-protein. Biochemistry 31, 10716-10723 (1992).
`6. Goldfarb, L. G., Brown,P., Haltia, M., Ghiso, J. & Frangione, B. Synthetic peptides correspondingto
`different mutated regionsof the amyloid gene in familial Creutzfeldt-Jakob disease show enhanced in
`vitro formation ofmorphologically different amyloidfibrils. Proc. Natl Acad. Sci. USA 90, 4451-4454
`(1993).
`7. Abrahamson, M. & Grubb,A.Increased body temperature accelerates aggregation of the Leu-68-Gln
`mutant cystatin C, the amyloid-formingprotein in hereditary cystatin C amyloid angiopathy. Proc.
`Natl Acad. Sci. USA 91, 1416-1420 (1994).
`8. Hurle, M. R., Helms, L. R, Li, L., Chan, W. & Wetzel, R. A role for destabilizing amino acid
`replacementsin light-chain amyloidosis. Proc. Nat! Acad. Sci. USA 91, 5446~5450 (1994).
`9. Maury, C. P. J., Nurmiaho-Lassila, E.-L. & Rossi, H. Amyloid fibril formation in gelsolin-derived
`amyloidosis. Definition of the amyloidogenic region and evidenceofaccelerated