HAL Id: pastel-00004880
https://pastel.archives-ouvertes.fr/pastel-00004880
Submitted on 27 Mar 2009
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Lorenzo Marcucci
To cite this version:
Lorenzo Marcucci. A mechanical model of muscle mechanics. Life Sciences [q-bio]. Ecole
Polytech-nique X, 2009. English. �pastel-00004880�
1 Mus le physiology and early modeling 1 1.1 Mus le physiology. . . 1 1.2 Me hani alexperiments . . . 7 1.3 Me hani almodeling . . . 14 1.3.1 Hill1938 model . . . 14 1.3.2 Huxley 1957 model . . . 19
1.3.3 Huxley and Simmons1971 model . . . 22
2 Power Stroke 29 2.1 Introdu tion . . . 29
2.2 Re ent Models. . . 30
2.3 New model of a power stroke. . . 38
2.3.1 Generalshape of the energy . . . 38
2.3.2 Double paraboli approximation . . . 41
2.4 Deterministi ase: globalminimum . . . 43
2.4.1 One ross-bridge . . . 43
2.4.2
N
ross-bridges . . . 452.5 Sto hasti ase:
N
ross-bridges atnite temperature . . . 482.6 Variable power stroke size . . . 50
2.7 Negativeslope of the
T
2
(δ)
urve . . . 572.7.1 Inhomogeneity in a hain of sar omeres . . . 62
2.7.2 Distributionof the atta hment positions . . . 73
2.8 Rate of fast tensionre overy . . . 76
2.9 Dis ussion . . . 85
3 The atta hment-deta hment pro ess 87 3.1 Introdu tion . . . 87
3.3.2 Prost et al. model . . . 98
3.3.3 Cooperative and non- ooperative motors . . . 101
3.4 Dire t simulationof a set of sto hasti equations . . . 107
3.5 Thermalrat het . . . 109
3.6 Cooperative Magnas omodel . . . 111
3.6.1 Governing equations . . . 111
3.6.2 Ben hmark problem:
K = 0
. . . 1133.6.3 Cooperative Magnas omodel with
K
6= 0
. . . 1164 Full ross-bridge y le 123 4.1 Introdu tion . . . 123
4.2 Numeri al implementation of the powerstroke model . . . 125
4.2.1 Isometri ase . . . 125
4.2.2 Length lamp devi e . . . 127
4.3 Whole y le models . . . 130
4.3.1 Extended Huxley and Simmonspotential . . . 130
4.3.2 Chemi al-Spring-Motormodel (CSM) . . . 133
4.3.3 Spring-Chemi al-Motormodel (SCM) . . . 136
4.3.4 Spring-Chemi al-Motormodel with a ba kbone (SCM1) . . 140
4.3.5 Dis ussion . . . 148
5 Con lusions 151 A Appendix 155 A.1 Brownianmotion . . . 155
A.2 Probability . . . 155
A.3 The Langevin Equation. . . 157
A.4 Diusion of aparti le ina uid . . . 159
A.5 The Fokker-Plank equation. . . 161
A.6 High fri tionlimit . . . 163
A.6.1 The Fokker-Plank equation inthe high-fri tion limit. . . 163
A.6.2 Canoni al distribution . . . 165
A.6.3 The First passage time . . . 166
A.7 Kramers' approximation . . . 167
Thisthesiswasdevelopedatthe Laboratoirede Mé aniquedes Solides(LMS)and
mainly funded by the Marie Curie Resear h Training Network and partially by
So iété Fx-Conseil. Many people have ontributed in dierent ways to this work,
I would liketo thank all ofthem and inparti ular:
My advisor Prof. Lev Truskinovsky for the dis ussions whi h we have had
duringallthese threeyears andforhisen ouragingtorea hastronger understand
of the mathemati alte hniques used in this work.
Dr. Jean-Mar Allain for the interest expressed on the development of this
work and for hishelpful suggestions.
Prof. ChaouquiMisbahfora epting tobethe Presidentof my thesis
ommit-teeaswellasProf. FrançoisGalletandProf. Vin enzoLombardiwhi hhavefound
the time to read and report on my thesis do ument. I would also like to thank
Prof. Marie-Christine Ho Ba Tho, Prof. Cé ile Sykes and Prof. Pas al Martin
for a epting to parti ipate to my thesis ommittee and for their suggestions to
improve the nal versionof the thesis.
All the people of the Laboratorie de Mé anique des Solides where I spent the
lastthree years. I met many ni e persons, with ahigh professionality. My spe ial
thanksgotoDire torBernardHalphenforhissupportbeforeandafterthedefense
of the thesis.
Prof. Grégoire Allaire, team leader at the E ole Polyte hnique of the
Multi-s ale modellingand hara terisation for phase transformationsin advan ed
mate-rials(MULTIMAT) proje t inthe Marie CurieResear h Training Network.
I would like to express my gratitude to Prof. Vin enzo Lombardi and Prof.
GabriellaPiazzesi for having shared with me part of their tremendous amount of
knowledgeonmus le me hani sfromthe very beginningofthis work. Iremember
espe ially our rst meeting and the time whi h they have given to my questions
despite the dieren es in our respe tive approa hes to the same subje t. I have
Skeletal mus le ontra tion is a broad domain of s ien e that overs many areas,
from biophysi s and hemistry to me hani s. The foundations of the theory of
mus le ontra tion were built 50 years ago, when it was understood that it is
myosin ross-bridge,linking adja ent myosin and a tin laments,that generates
for e and motion. Sin e that time many experimentaladvan es have been made.
Theseadvan eshavenot been alwaysmat hed byimprovementsinthe buildingof
mathemati almodels.
Mathemati al approa hes tomus le ontra tion are mainlybased onthe ideas
proposed inthe Huxley 1957 model[4℄ andHuxley and Simmons1971model[10℄,
that dominated the eld for the past half entury. Althoughthey do not a ount
for all observed phenomena, these models still represent the paradigm of hoi e.
The two models of Huxley an be seen as omplementary sin e the Huxley 1957
model des ribes the atta hment-deta hmentpro ess and the events related tothe
slow time s ale, while the Huxley and Simmons 1971 model des ribes the power
stroke pro ess and the events related tothe fasttime s ale.
In this Thesis we shall follow some re ent insight and explore the possibility
to bring together these two type of pro esses and to obtain a unied model that
is able to des ribe the whole ross-bridge y le. Before the uni ation we rst
modify the existing models to ast them into afully me hani alframework. Both
Huxley 1957 model [4℄ and Huxley and Simmons 1971 model [10℄, present ad ho
assumptionsregardingthe hemi alrate onstantsthatdrivethepro esses. Similar
assumptions were made in all re ent models to t the experimental data at the
expense of maintainingthe linkwith me hani s.
In Chapter 1 we des ribe the physiology of mus les and their me hani al
be-havior, as well as the orresponding experimental pro edures. There we alsogive
the detailsof the Huxley 1957 model and Huxleyand Simmons1971 model whi h
are importantfor the originaldevelopment in the subsequent pages.
en ounters problems in mat hing the observed time s ale of tension relaxation
when a realisti value of the stiness of the myosin head is taken. After a review
of how the more re ent models, whi h in orporate one or more aspe ts of the
original Huxley and Simmons 1971 model, deal with these problems, we present
our modi ation of the theory whi h pla es the power stroke me hanism entirely
in a me hani al framework. The novelty of our approa h from the perspe tive of
me hani s is that we deal with the me hani al behavior of a multi-stable system
in a Brownian domain, where the ee ts of thermal u tuations are important.
We obtain an analyti al des ription of the behavior of our model at equilibrium
and during the transients and show how the resulting modi ation of the Huxley
and Simmons 1971 model helps one to avoid the intrinsi problems of this model
indi ated above. Finally we show that our model gives a new meaning to the
powerstroke step,whi hisin quantitativeagreement withallre entexperimental
observations.
In Chapter 3, we turn to the atta hment-deta hment pro ess and review from
the new,fullyme hani alpointofviewtheHuxley1957 model. Weshowthat this
model anbeviewedasbelongingtoa lassofmodels ofBrownianrat hets. These
models,rst developed inthe early
′
90s
,haveanimportantroleinthe des ription
of mole ularmotorsofwhi hthemyosin II isanexample. Weareinterestedinthe
Brownian rat hets theory be ause it allows one to have a ompletely me hani al
interpretation of the mus le ontra tion pro ess. We present dierent types of
rat hetsrepresentingthepro essofATPhydrolysis. Wemodifyoneofthesemodels
by in luding ooperative ee ts and adapt it to the des ription of the slow time
phase of the ontra tionphenomenon. Wealsodevelop and test inthis Chapter a
numeri alalgorithmtosolvethe oupledsystemofsto hasti dierentialequations
whi h islater used for our numeri al experiments.
In Chapter 4, we ombine the power stroke model from Chapter 2 with the
model of a ooperative Brownian rat het developed for the simulation of the
atta hment-deta hmentpro essinChapter 3. Wepresentdierentways oflinking
together thetwomodelsandstudybothadvantagesandlimitationsofea hversion
of the unied model. We nally ome up with a model apableof providing fully
me hani al des ription of all four stages of the bio hemi al Lymn-Taylor y le of
mus le ontra tion. The resultingmodel still has drawba ks and we present some
and present a general dis ussion and on lusions. In Appendix we review some
mathemati al results regarding sto hasti dierential equations whi h we used in
Mus le physiology and early
modeling
1.1 Mus le physiology
The movements of a mus le on a ma ros opi s ale appear as the result of the
on erted a tion of millions of elemental units that work in unison. The most
studied mus les are alled skeletal mus les be ause these mus les are atta hed
to the skeleton. The ontra tion of skeletal mus les is under voluntary ontrol.
They belong to the lass of striated mus les whi hare omposed of long, parallel,
ylindri al bers. Ea h of these bers is a multinu leate ell, of
1
− 100 mm
inlengthand
10
− 100 µm
indiameter. Fibers ontain myobrils,also ylindri alinshape with a diameter of
0.5
− 2 µm
. Myobrils are made of repeated segmentsea habout
2.5 µm
inlength, that are alled sar omeres (Fig. 1.1) [95℄.Sar omere is the smallestelementof a mus le that an ontra t. Being pla ed
inseries,sar omeresgenerate the ontra tionofthewholemus le. Ea hsar omere
isformed by anarray of lamentsof twodierent types, whi hintera t with ea h
other: a thinner lament, ontaining the protein A tin, and a thi ker lament,
ontainingthe proteinMyosin. Thesar omere anbedividedinzones: inFig. 1.2
wesee alongitudinaland atransverse viewofit. In theregionwherethe laments
overlap, six thin laments are lo ated around ea h thi k lament (Fig. 1.2b).
Thinlamentsare an horedtothe Z-disk(fromgermanzwis hen, between) whi h
onne ts adja ent sar omeres. Thi k laments are an hored to M-line (Mittel,
middleofthesar omere,notshowedinFig. 1.2)andalsototheZ-diskviaanelasti
elementthe giant proteintitin. Theserepeating stru tures, (A=anisotropi atthe
Figure 1.1: Mus le'sanatomi almi rostru ture
generate the typi alstriatedstru ture that givesthe name tothis type of mus les
[19℄. A longitudinal view as it appears on a ele tron mo rograph is given in Fig.
1.3.
The sliding-lament hypothesis was proposed fty years ago. It assumes that
during ontra tionthethinlamentmovespastthethi kone,sothatboththe
sar- omere, andthemus le,shortenwithout hangingthelengthofthetwostru tures.
ThehypothesiswasbasedonthepapersofHughHuxleyandJeanHanson[2℄(using
aphase ontrast lightmi ros opy)and ofAndrewHuxley andRolfNiedergerke [3℄
(using aspe iallydeveloped interferen elightmi ros ope)both publishedin1954.
Both works showed that when the mus le ontra ts the laments keep a onstant
length, andthe on lusion wasmade thatthey must slide duringshortening. This
hypothesis has not been immediately a epted: the then urrent view was that
myosin was a long negatively harged polypeptide without mu h stru ture that
shorten down due to the addition of
Ca
2+
Figure 1.2: Longitudinal and transverseviewof asar omere: (a)Longitudinal viewof 3
sar- omeres(sket h). (b)Transverse viewat 3dierentse tion (sket h). ( ) Transverse viewof 3
sar omeres(mi ros ope). From[97℄
Figure 1.3: Longitudinalviewofasar omereas seeninele tronmi rgraph. From[98℄
the two laments intera t through the ross-bridges (later we use notation Xb);
these are the globular portions, or heads, that emergein regularly repeating
ou-ples from the thi k lament formed by the polymerisation of the dimeri protein
myosin II (Fig. 1.4). Ea h head has a site with an anity for a tin, and a site
with an anity for a high energy mole ule, alled ATP (adenosintriphosphate).
The rst site bounds an a tin monomer while the se ond site an ATP mole ule
whi h a ts as the fuel for the mus le motor. ATP is hydrolyzed by myosin in
ADP (adenosindiphosphate) and orthophosphate whi h subsequently are
disso i-ated with release of hemi alenergy [9℄.
AsimpliedmodelofXb y leisshown inFig. 1.5,whereone anseefourmost
important states in whi h Xb an exist. When atta hed to a tin (state 2 in the
gure),ea hXbusesitspotentialenergytopullthea tinlamentthroughapower
stroke (state3)whi h,a ordingto rystallographi studies, onsistsinatiltingof
Figure 1.4: Myosin lament stru ture. (a) Myosinmole ule ( ouple). (b) Bundle of oupled
myosinmole uleswhi hgeneratethethi klament. From[100℄
pla e in one dire tion (plus dire tion), but due to the antiparallel arrangement
of the two halves of the sar omere, the two Z-disks are pulled towards the enter
of the sar omere, redu ing its length. In this sense, the half-sar omere, the zone
between one Z-disk and the next M-line, an be seen asthe smallest element that
an ontra t. Togoba ktoitsoriginal onguration(state1)theXbneedsanother
ATP to deta h from a tin and start another y le (state 4). It then binds to a
new a tive site onthe a tin lament (state 2) and the whole pro ess starts again
[64℄. This inner working is des ribed in the bio- hemi al Lymn-Taylor model of
a ross-bridge y le [9℄. The y le in Fig. 1.5 is a simplied four-states model
that omits anumber of intermediate states, nevertheless it des ribes the essential
steps of the pro ess. An important general observation is that mus le needs ATP
for both the ontra tionand the relaxation;the unphysiologi al depletion of ATP
below a ertain on entration will prevent the deta hment of the heads from the
a tin lament, whi h auses rigor mortis [19℄.
Thestru tureofthehead anberesolvedwithapre isionofonenanometer[33℄,
[87℄. It has been proved that the relative displa ement of the laments is mainly
taking pla e during the power stroke (state 3). It is a hieved by a rotation of the
distal part (C-terminal) of the head that a ts like a lever arm. This me hanism
gives tothe whole approa hthe nameof swinging lever armtheory.
Regulationofthe ontra tionisdue tothe fa tthat Xbs an bound a tinonly
Figure 1.5: SimpliedCrossBridge(Xb) y le(seetext). From[44℄
thatisatta hedtotheproteintropomyosin andlieswithinthegroovebetween the
twostrandofthe a tinlamentinmus letissue. Inarelaxedmus le,tropomyosin
preventtheintera tionofmyosinwiththeatta hmentsiteona tin,thuspreventing
ontra tion. Whenthe mus le ellisstimulated, al ium hannelsopen inthe
sar- oplasmi reti ulumandrelease al iumintothesar oplasm. Someofthis al ium
ions atta h totroponin, ausinga onformational hange that moves tropomyosin
out of the way so that the Xbs an atta h to a tin and produ e mus le
ontra -tion. The ions
Ca
2+
are stored inthe sar oplasmati reti ulum (SR) surrounding
the myolaments. The a tion potential originated at the neuromus olar jun tion
triggers the release of al ium from the SR almost syn hronously everywhere by
indu inganin rease inCa2+ permeability of the SR membrane. The ontra tion
ismaintained untilthe nerve ontinues to re; when the train of a tionpotentials
stops,the Ca2+ permeabilityfalls, whilethe Ca2+ pumpbringsba k the al ium
intotheSR.The de reaseof al ium on entrationbelowthe thresholdina tivates
Figure 1.6: Ex itation ontra tion oupling: s hemati des ription. From[97℄
Wehavegiven hereabriefoverviewofthe omplexeventsthat leadtothe
on-tra tionofskeletalmus les. A omplete des riptionshouldtake into onsideration
how the ele tri al signal generates the bio hemi al pro esses, with all their
om-plexity, whi h nallyleads to the me hani al for eor displa ement insar omeres.
Wewouldliketoemphasizethatinthisworkweshall onsideronlytheme hani al
aspe ts ofthe ontra tion,avoidingthe des riptionofproteinsintera tion through
hemi al rate onstants. Despite some limitations in the physi al interpretation
of the nal model, dis ussed at the end of the thesis, this approa h allows one
to produ e a fully me hani almodel of the ontra tion inthe sar omere, opening
the way to the onstru tion of arti ial mus le type ma hines. Moreover, as we
shall show the new approa h improves, in some aspe ts, the predi tive power of
1.2 Me hani al experiments
There exist dierent experimental approa hes to the study of the me hani s of
mus le ontra tion(see[76℄,[75℄,[78℄andreferen estherein). Thete hnologyused
in these experiments has been often highly innovative, leading to te hnologi al
spin-o. We have already mentioned dierent mi ros opy te hniques. Another
te hniqueistheinvitromotilityessay,wheresinglemyosinmole ulesatta hedtoa
beadtrappedby alaser beamareused tomeasure thegeneratedfor e. Dierently,
the IVMA measures the speed of sliding of a tin laments, atta hed to a bead,
gliding on a bed of myosins. Then the syn hrotron radiation (an intense X-ray
sour e) was developed to study the Xb movements in situ in whole mus le or
single bres. Finally protein rystallography was applied toinvestigate the power
stroke inthe myosin mole ule atatomi resolution (see Fig. 1.7) [76℄.
Figure1.7: Stru tureofmyosinS1from hi kenskeletalmus le. From[33℄
In thisThesis weapproa hthe modelingof skeletalmus les ontra tionfroma
me hani alpoint of view. Therefore, we shall be mainlyinterested in aparti ular
setofexperimentsperformedonthemus lebers ormyobrils. Theseexperiments
have ommon aspe ts with the usual me hani al measurements aimed at testing
the behavior of passivematerials [19℄.
A mus le responds to a single stimulus with a single transient rise in tension,
alled twit h. Two stimuli, generated after a suitable interval of time, produ e
identi al for e transients. When the se ond twit h starts before the rst one is
over, the se ond one develops a larger peak tension. Witha train of stimulations
the for e rea hes a steady state value, alled unfused tetanus, and hara terized
frequen y themean for erises toanalmost onstantvalue: this situationis alled
tetanus. The required frequen y depends onthe type of mus les and onthe
tem-perature (50-60 Hz in mammalian mus les at body temperature, not used in the
gure) [19℄. An experiments in whi h we are interested have been made in the
state oftetanus, that an beviewed as asteady state ondition.
Figure 1.8: For egenerated atdierentstimulationfrequen y. 1pps orrespond tothesingle
twit h,at 80pps isrea hedthetetanus. From[97℄
The me hani al experiments, either on a ber or on a myobril, are usually
performedwith one end ofthe spe imen xed and the otherlinked toaleverwith
a at h me hanism and a transdu er of for e (Fig. 1.9). In this Se tion we shall
explain indetail the three major proto ols used in these type of experiments and
present theirmain results.
For e-length urves
Whenthe at hme hanismisxed,themus leundergoesanisometri ontra tion.
Byimposingtetanizationwiththeendsxedone anregisterthetensiongenerated
by the mus le. Byvarying the initiallengthof the mus le beforethe tetanization,
a for e-length urve an be onstru ted[95℄.
InFig. 1.10weshowthe s hemati for elength urveforthetotalfor eandfor
its two omponents: a tive for eand passive for e. Passive for eis the resistan e
generated by elasti omponentsin parallel tothe ontra tile element,it be omes
relevant when the sar omeres are overstret hed. The passive resistan e is almost
zero until a ertain riti al elongation of the sar omere, and then in reases fast
showing nonlinearelasti ity. Subtra tingthe passivefor efromthetotal isometri
Figure 1.9: Experimental devi es (a,d),experimental urves(b, ) andone versionof theHill's
model(e)(seeSe tion1.3.1). (a)Whenthe at hme hanismisa tingthemus le anbetetanized
ata onstantlength,leftpartof( ),rea hingthetetanusat
T
0
in(b). Whenthesystemisreleased(d),theisotoni ontra tionagainsta onstantload,
T
in(b),generatesthelength-time urveintherightpartof( ). From[19℄
Dierent types of mus les have dierent passive responses and so dierent total
for es, but the a tive for e-length urve, for most of them, shows the same
non-monotone behavior [7℄. This behavior (Fig. 1.11) is in agreement, with the fa t
thatthetwolamentsmustoverlaptogeneratefor e. Infa t,themaximuma tive
for eisgeneratedwhentheoverlapbetweenthe twolamentsisoptimal,i.e. when
allthe Xbssee ana tinsite where they an bound and,at the sametime, there is
nointerferen e between the two halfparts ofasar omere. When theinitiallength
in the passive state is su h that some Xbs, the ones near the M-line, donot have
any a tive site toatta h,the a tive for estarts tode rease linearlywith the total
number of Xbs available to intera t with a tin. For shorter initial lengths than
those orresponding to the plateau of the for e-length urve, two opposite a tin
lamentsstart tointerfere withea hother,that again ontributestoade rease of
for e[7℄.
Animportantthingtonoteisthat thefor e-length (T-l) urve is reated point
by point: rstwex alengthinthe passivestate,thenwetetanizethe mus leand
registerthefor ewhi hdevelopsinisometri ontra tion. TheT-l urverepresents
therefore a series of isometri a tivations at dierent initial passive lengths. In
parti ular this urve does not represent the response of the mus le to quasi-stati
Figure1.10: Total,passiveanda tivefor easafun tionofthelengthfortwodierenttypesof
mus les. Thepassivefor eisanalyzedstret hingthemus leinthepassivestate. Thetotalfor e
is analyzedtetanizingthemus lefrom a onstantpassivelength. Thea tivefor e isderivedby
subtra tion. From[19℄
Figure 1.11: Upper gure: A tivetension generated by isometri tetanization from dierent
passivelengthsofthesar omere. Lowergure: relativepositionsofthethinlament(bla kline)
For e-velo ity experiments
Thedynami albehaviorofskeletalmus les isusuallystudiedinadierent typeof
experiments[1℄, [67℄,[66℄,[93℄aimedat onstru tingthefor e-velo ity urve. This
urverelatestheload imposedtoa ontra tingmus letothe velo ityatwhi hthe
mus le shortens. It an be obtained, still point by point,within the experimental
setup dis ussed before [5℄: a mus le is tetanized at a xed passive length, then
the leveris released, while a onstant load is applied. The length of the mus le is
plottedagainst time (Fig. 1.12). As soon astension isredu ed, the mus le length
de reases: this typi ally fast response shows the presen e of an elasti element
whose shortening takes pla e before aslower time s ale dynami s of the Xb y le
gets a tivated. After this fast transient, the mus le starts to shorten at onstant
velo ity. Repeating the experiment with dierent loads, one an onstru t the
urve plottedin Fig. 1.13.
Figure 1.12: Shorteningvs. time urves, forone load(A)and fordierentloads (B). Length
hange axes refers to shortening. The os illating regime is due to the me hani al apparatus.
From[5℄
As we an see, thereis amaximum velo ity
v
0
that the mus le ontra tion an rea h under free (unloaded) shortening; this velo ity is independent of the lengthof the mus le in the passive state. There is also a load against whi h the mus le
undergoesanisometri ontra tionat
v = 0
,thisvalueisprovidedbytheT-l urve.Applying a onstant load greater than this value gives the for e velo ity urve in
lengthening (or e entri ) portion. This latter range ismu h less known than the
shortening range be ause of the higher dispersion of the experimental points. A
Figure1.13: For e-velo ity urve. Inshorteningthebehavior anbedes ribedbyanhyperbola
whi hinterse ts theabs issaatthemaximumunloadedspeed(normalized) andtheordinateat
the isometri tetani tension. The slope has adis ontinuity in the isometri point
v = 0
. Aplateau isobservedathigh lengtheningvelo ities. From[99℄
toinnity. Moreoverthereisadis ontinuity inthe derivativeofthe
F
− v
urveatthe isometri point
v = 0
: inthe e entri portionthederivativeofthe urveissixtimegreaterthanintheshorteningportion. Inorderto onstru tthefor e-velo ity
urvethesteadyshorteningstatemustberea hed,whi hhappensinatypi altime
s ale oftens ofmillise onds[93℄. Thetransientthatpre edes this statetakespla e
in a typi al time s ale of some millise onds and an be analyzed alsoin aslightly
dierent me hanism, whi h we introdu e inthe next Se tion.
Experiments on fast for e re overy
There is athird type of experiments with whi hwe shall mainly o upy ourselves
in this Thesis. Imposing on a tetanized mus le a small in rement, say negative,
of length
δ
generates a hange in tension as shown in Fig. 1.14(a) (see [10℄ andreferen es below). Thereis an instantaneous (hundreds of mi rose onds) de rease
in tension to a new value alled
T
1
, just as it would be if the thin and thi k lamentswereatta hedtoea hotherby elasti springs. Almostimmediatelyafterthis elasti stress drop,the tensionrisesand thenforsome time(millise ondstime
s ale) remains lose to a plateau level ( alled
T
2
) before nally re overing fully the value it had had before the length hange (tens of millise onds time s ale).The hanges in length in these experiments are very small, about 4-10 nm per
su hthat itis realisti to assume that the number of atta hed Xbs remainsxed.
Imposing dierent length in rements, one obtains the relationship between the
imposed lengthin rements and the tensions
T
1
andT
2
shown in Fig. 1.14(b).(a) (b)
Figure 1.14: Fast re overyexperiments. (a)A rapidsmallshortening isappliedto themus le
(uppertra e)andtheresultingtensionhistoryismeasured(lowertra e). (b)The urves
T
1
andT
2
vs. the imposed length in rement fortwodierentvalues ofinitial length, normalizedwith respe ttothehigherisometri tensionT
0
. Symbolsaredenedin thetext. From[19℄Animportantunderstandingthatderivesfromthisexperimentisthatthevalues
of
T
1
at various shortenings lay pra ti ally on a straight line. Another important result is that the rate of re overy of tension hanges with the step imposed in ahighly non linear manner (see Fig. 1.15). It tends to in rease in an exponential
way from positive lengthsteps tohigher negative steps.
The behavior exhibited by a mus le in this set of experiments is an
impor-tantsour e of informationabout mus le me hani s, be ause atleast the fasttime
response produ ing the fun tions
T
1
(δ)
andT
2
(δ)
appears to be independent of the atta hment-deta hment pro ess. Sin e the pioneering paper of Huxley andSimmons[10℄ these experimentshave been repeated by many groups [14℄ [41℄ [43℄
[67℄.
Thefor e-length,thefor e-velo ityandthe
T
1
andT
2
vs. step-length urvesare themostimportantexperimentalresultsthatdeal dire tlywiththeme hanismsofmus le ontra tion. Wehavegivenreferen es tosomere entexperimentsrevealing
for instan e the history dependen e in the me hani al response when mus le is
stret hed after tetanization [52℄. Nevertheless, in what follows, we shall fo us on
the explanationof onlythe mainexperimentalfa tsthat are onsideredtobe well
Figure1.15:Rate onstant
r
ofqui ktensionre overyfollowingalengthin rementofmagnitudey
. Estimatedasln(3)/t
1
/3
wheret
1
/3
isthetimeforre overyfromT
1
to(2T
2
+ T
1
)/3
. From[10℄1.3 Me hani al modeling
In this Se tionwe introdu eseveralbasi modelsaimed atexplainingthe
me han-i al behavior of mus les. They are: the Hill 1938 model, the Huxley 1957 model
and the Huxley and Simmons 1971 model. These models represent the basis on
whi h the majority of more re ent models are based. Some of this more re ent
models are reviewed later inthe Thesis.
1.3.1 Hill 1938 model
An analyti alexpression for the on entri part ofthe for e-velo ity urve was
ob-tainedbyHillin1938[1℄. Heusedhisownexperimentsfo usedontheenergeti sof
mus le ontra tionagainsta onstantfor e. Firsthe observed thatwhen the
mus- le is allowed to shorten, it liberates more energy (thermal and me hani al) than
during isometri ontra tion. He divided the totalenergy rate
E
intothreeterms:the maintenan e heat rate (
A
) liberatedby amus le inisometri ontra tion, theshortening heat rate (
H
), that is the total heat liberated during the ontra tionminus
A
, and the rate of work done (W
) equal toF
· v
whereF
is the onstantdis ussion of the for e velo ity urve. Hill wrote the energy balan ein the form:
E = A + H + W
⇒ E − A = H + W.
(1.1)Bya very pre isemeasurementof the rst term
A
and of the totalenergy rateE
, Hillobserved empiri ally the relation:H + W = b(F
0
− F ).
(1.2)In the right hand side of (1.2) we see the dieren e between the for e
F
appliedto the mus le and the maximal for e
F
0
exerted by it in an isometri ontra tion whenv = 0
. Independently Hill observed thatH
depends linearly onthe velo ityof ontra tion,
H = av
. In this way we have:H + W = av + F v = b(F
0
− F ).
(1.3) Byrearranging terms in(1.3), Hill obtained:(a + F )(v + b) = b(a + F
0
).
(1.4)Figure 1.16: For evelo ityrelation. The ir les representtheexperimental observation(frog
mus le),theline orrespondsto the urve(1.4). From[52℄
Inthe
F
−v
spa eequation(1.4)des ribesahyperbolawithasymptotes−a
and−b
(seeFig. 1.16)[46℄. Ittstheexperimentalpointsverywell(usingappropriates values ofa
andb
) for a large variety of mus les. In the free ontra tionF = 0
,the velo ity be omes maximal
v
max
and it has been observed that for many type of mus les a ross spe ies and temperatures [52℄:a
F
0
=
b
v
max
Figure 1.17: Hill1938 model withwith aseriesof apassivespringand a ontra tileelement,
bothinparallelwithase ondpassivespring. From[19℄
With these results, Hill proposed a model (see Fig. 1.17) where the a tive
mus leis represented by anelasti elementSE inseries witha ontra tileelement
CEwhosefun tionistolinktheappliedfor etothevelo ity,inabla kbox manner.
Su essively, to a ount for passive elasti ity, anelasti element PE was addedin
parallel with the CE and the SE(Fig. 1.17).
In the passive state the CE an be stret hed without any resistan e. During
the ontra tion, the total for e generated by the system is
F = k
p
u + k
s
(u
− w)
. Herek
s
(u
− w) = f
CE
[w
′
, l
0
]
,k
s
isthe stiness of the SE andk
p
of the PE,u
the totaldispla ement,w
thedispla ementoftheCE,f
CE
isthefor einthe ontra tile elementwhi hdependsontherateof hangeofthedispla ementw
′
= v
. A ording to observations made by Hill, the CE exerts afor e of the typef
CE
=
0
w 6
˙
−F (l
0
)
b
a
F (l
0
)b + a ˙
w
− ˙
w + b
−F (l
0
)
b
a
6
w 6 0
˙
1.5F (l
0
)
− 0.5
F (l
0
)b
′
− a
′
˙
w
˙
w + b
0 6 ˙
w 6 F (l
0
)
b
′
a
′
1.5F (l
0
)
F (l
0
)
b
′
a
′
6
w
˙
that a ounts also for the e entri ontra tion. The isometri for e
F
0
has a dependen e on the initial length of the mus lel
0
, as shown by the for e length urve. Thefor eis equal tozero forlarge negativevaluesofw
′
(shortening),while
it an be dire tly obtained from (1.4) for smaller ontra tion velo ities. In the
e entri region,
w
′
> 0
, the values are taken to mat h the behavior observed experimentally(see Fig. 1.13).time dependen e of the for e for dierent given proto ols of stret hes. Thus if we
introdu eaparameter
β
toa ountforthe on entrationof al ium,asintherstand third part of Fig. 1.18, weobtain:
F (l
0
, t)
− k
p
u = βf
CE
"
1 +
k
p
k
s
˙u
−
F (l
˙
0
, t)
k
s
, l
0
#
(1.6)where
0 < β < 1
modulatesthefor einthe ontra tileelementf
CE
. Twoexamples of loading programs are presented in Fig. 1.18. The elongation, equal for bothexperiments,is given by arampthat in reases the lengthofthe mus le, maintains
it onstant and then shortens it to the initial state. The a tivation parameters
were dierent. The experimental observations obtained for the given elongation
history is shown in Fig. 1.19. The predi tions of the model are in Fig. 1.20: the
two responses are rather similar.
For 50 years Hill 1938 model dominated the eld. In this period many ideas
havebeenaddedtothemodelinordertoa ommodatenewlydis overedfa ts[52℄.
Originallyquitesimplethe model be amemore and more ompli atedand lostits
appeal;howeverthe simplestversion isstilltoday usedtosimulatethe me hani al
behaviorof mus les.
Figure 1.18: Two dierent experimental pro edures for Hill's 1938 model. The response is
Figure1.19: ExperimentalresultsfortheelongationhistoryshowninFig. 1.18. From[13℄
Figure 1.20: Response of the Hill's 1938 model for twodierentpro edures I (upper partof
The main reason for the sear h of dierent on epts in mus le modeling was
the following: Hill's model does not provide insights into the me hanism of the
produ tion of for e. Its bla k box nature is su ient to give a good t to the
experimental urves, but it does not provide a tool for the understanding of the
me hanismsthatoperatesatthe mi ro-s aleswhi harenot visibleinthestandard
me hani alexperiments.
1.3.2 Huxley 1957 model
Before1954,mosttheoriesofmus le ontra tionwere basedontheideathat
short-eningandfor eprodu tionweretheresultofsomekindoffoldingor oilingoflarge
proteinmole ules. In1954, HEHuxleyand JHansen[2℄aswellasAFHuxleyand
RM Niedergerke [3℄ demonstrated that mus le ontra tion is not asso iated with
any hangeoflength insidethe mi rostru ture. These authorspostulated thatthe
for eis generated through the intera tion of a tin and myosin laments.
Based on this understanding, AF Huxley developed in 1957 a new theory of
mus le ontra tion [4℄. The thi k myosin lament is assumed to be xed inspa e
whilethethinlamentisassumed toslideparalleltomyosinwith onstantvelo ity
v
. The movementisgeneratedby ame hani alstru ture(thatisnowknown tobe theXb)that an o upy dierentpositionsalongtheba kbone ofa tin,and whosemovement is limited by an elasti element (Fig. 1.21(a)). The model postulates
that thenumberof a tiveXbs is onstant and onsiders onlythe fulla tivationof
the mus le (tetani response).
Figure1.21: Huxley1957model. (a)ThemyosinheadMiselasti ally oupledtotheba kbone.
Theintera tion between thelaments an beestablished whenM rea h theatta hmentsite A
The stru ture in question an atta h itself only to spe i sites on the a tin
lament. When it is atta hed, then there is a for e between a tin and myosin,
whi h depends on the position of Xb. To al ulate the total for e generated by
the mus le one needs to know the total number of atta hed Xbs at ea h position
x
relative tothe referen e position of the stru ture, at every timet
.As a result of thermal u tuations Xbs atta h to the a tin in a range of axial
position. They exert a for eif they rea h the atta hed position where the elasti
elementisstret hed;noti ethatasour eofasymmetryisneeded togenerateanet
for e in one parti ular dire tion [4℄. It is assumed that the probability
f
that adeta hedXb an atta hand theprobability
g
thatanatta hed Xb andeta h,arefun tions ofthe variable
x
, asshowed inFig. 1.21(b). The atta hmentprobabilityf (x)
is assumed to belinear inx
and is zeroboth beyond amaximum distan eh
, and forx < 0
(the Xb an not atta h toan a tivesite whenthe elasti elementisompressed). The deta hment probability fun tion
g(x)
is alsolinear for positivex
, the probability in reases even beyondh
, and islarge and onstant for negativex
. Ifn(x, t)
is the fra tion of the total populationof atta hed Xbs whose distan e fromthe a tivesite isx
attimet
,then itstimeevolution anbefound fromarstorder kineti equation [4℄:
∂n(x, t)
∂t
− v
∂n(x, t)
∂x
= (1
− n(x, t))f(x) − n(x, t)g(x).
(1.7) Huxleylimitedtheanalysistosteadystate ase,whenthesolutionis onstantintime,sothersttermintheleftiszero. Theequation(1.7)allowsthe omputation
of
n(x)
at dierentv
: at zero velo ityn(x)
rea hes the onstant valuef /(f + g)
.Athighervaluesofvelo itytherearetwofa torsthatredu eitsvalue: rst thereis
less timefortheXbstoatta h,se ondthe Xbsare broughtfastertowards negative
values of
x
. The predi tions of the model (1.7) are illustrated in Fig. 1.22. Theanalyti al solutionis given by:
n(x, v) =
f
1
f
1
+ g
1
[1
− exp(−φ/v)] exp(
xg
2
v
)
x 6 0
f
1
f
1
+ g
1
1
− exp
x
2
h
2
− 1
φ
v
0 6 x 6 h
0
x > h
(1.8) whereφ =
f
1
+ g
1
2
h.
Usingthissolutionone anwriteanexpli itexpressionforthefor evelo ity
Figure1.22: RelativedistributionofXbsatvariousvelo itiesa ordingtoHuxley1957model.
From[4℄
k
, generating a for e proportional to its displa ementkx
. Then the total tension an be writtenas:T (v) = ρ
R
−∞
∞
kxn(x, v)dx =
ρk
f
1
+ g
1
2
h
2
2
1
−
v
φ
1
− e
−φ/v
1 +
1
2
f
1
+ g
1
g
2
v
φ
(1.9)where
ρ
standsfor thedensity ofXbsperunit volume. Optimizingthe parameterstottheHill'sdata,Huxleyobtainedanex ellenttasinFig. 1.23. Theisometri
tension be omes
C(f /(f + g))
, where C depends on the number of Xbs presentin the segment of mus le under onsiderationand on the other parameters of the
model, for instan e the elasti onstant.
Inadditiontothe on entri part ofthefor evelo ity urvethe modelpredi ts
alsootherfeatures of the mus le response, even if onlyqualitatively. Forinstan e,
the model predi ts the e entri part of the for e velo ity urve, showing both a
dierent slope of the urve at the isometri point and an asymptoti behavior of
the for e at high velo ities. Both values however are highly overestimated. The
model alsooverestimates the rate of heat releaseduring lengthening,however this
Figure 1.23: Huxley's predi tion for the for e velo ity urve (line) and experimental data
(points). From1.23
during lengthening there is a me hani al breakdown of the Xbs, whi h deta h
withoutATP release. Manymorere entdevelopmentshavebeen donealongthese
lines, see for instan e [21℄, [22℄, [27℄, [58℄, [64℄. Overall, the Huxley 1957 model
represents an improvement over the Hill 1938 model be ause it gives a pre ise
mathemati aldes ription of the mi ros opi events behind the bla k box behavior
postulated by Hill.
1.3.3 Huxley and Simmons 1971 model
The experimental response of mus les to rapid length in rements, des ribed in
Se tion1.2, annotbeeasilyexplainedbythe1957Huxley'smodel. Thepioneering
experiments of this type, made in [10℄, have lead to the development of another
importantmathemati almodel: Huxleyand Simmons'modelof 1971. This model
isnot anexpansionof theHuxley 1957model,but isaquitedierentmodel whi h
deals only with for e generated by the atta hed Xbs. In parti ular it does not
take into a ount the deta hment pro ess. What brings the ne essity of a new
model is the fa t that the rapid re overy of for e takes pla e in the millise onds
time s ale,whi hisdi ulttoexplaininthe frameworkofthe slower
atta hment-deta hment pro ess, relatedto the time s ale of tenthof a se ond. The approa h
used by Huxley and Simmons, whi h we shall des ribe below, is the predominant
onrmation from the measurements of the axial motions of the myosin heads at
angstrom resolution by X-ray interferen e te hnique [66℄.
The parti ularme hanismsuggested byHuxley andSimmonsforthe stru ture
ofthe Xbsisshown inFig. 1.24. First ofall,they assumed thatthe Xb ontains a
linearelasti spring linked to the head ofthe myosin. When atta hed tothe a tin
lament, the head of the myosin an be in two states, and an swit h from one
state to another in a jumpfashion. The ratio of the rates of jumps are ontrolled
by the relative energy of the two states. The energy
U
h
of the head is a double wellfun tionof onguration oordinatex
,it isplottedinFig. 1.25 togetherwiththe paraboli energy of an elasti element. The swit hing an stret h or relaxthe
elasti element, sowe an referto the states as alow for e generating state and
a high for e generating state. The total potential energy
U
tot
, given by the sum of the potential energy of the elasti elementU
s
and the potential energy of the hemi alstateU
h
, is plottedin Fig. 1.26.Figure 1.24: Huxley and Simmons 1971 model. The myosin head S-1 is linked to the thi k
lamentthroughanelasti elementS-2and hastwostable positions. From[19℄
The model analyzes the distribution of Xbs inea hof the energy wellin order
toobtain the total for e generated by the mus le. Be ause inea h halfsar omere
the Xbsarearranged inparallelbetween ana tinlamentand therelativemyosin
lament, the total generated for e is the sum of the for es generated by ea h Xb.
It is assumed that when the mus le is isometri allytetanized, the two states have
the same total energy.
Due tothe linearity of the elasti element:
U
s
=
1
2
Kl
2
=
K
2
(y
0
±
h
2
+ y)
2
(1.10)where
l
is the stret h whi h an be written as the sum ofy
0
± h/2
(wherey
0
is a point lo ated at equal distan e from both wells,h
is the distan e between theFigure 1.25: Thepotentialenergiesin HuxleyandSimmons1971model. (a) Elasti energyin
isometri ontra tion(
y = 0
), after stret h(y > 0
)and afterreleaseofthe mus le(y < 0
). (b)Congurationalenergyofthehead,twostablestates1and2arepresent. From[19℄
Figure1.26: Thetotalenergy
U
tot
= U
h
+ U
s
inHuxleyandSimmons1971modelasafun tiontwopotentialenergy wells, orrespondingtothetwostates1and 2. Theheightsof
the potential energy barriers
E
1
andE
2
in Fig. 1.25, are assumed to be the same for both wells. Sin e in the state of isometri ontra tion the two minima of thetotal potentialenergy havethe same levelwhen
y = 0
,the totalnumber ofXbs inthe two ongurations is the same. When a length in rement is imposed,
y
6= 0
,there isa hange in
U
s
(upperpart of Fig. 1.25) and therefore inthe total energyU
tot
= U
s
+ U
h
,as shown in Fig. 1.27.Before giving the mathemati al details, we des ribe briey how the model
works. The hange in the total length
y
rst ae ts the tension in the linearspring, and is therefore responsible for the
T
1
for e observed in the experiments. After the step, the levels of the energy in the two minima be ome dierent, anda hange in the total number of Xbs in ea h state is generated. This adjustment
pro essfollowskineti spostulatedforthejumppro ess,andtakespla einaslower
time s ale than the time s ale responsible for the
T
1
response. The nal steady state isresponsible for the value of for eT
2
.To ompute the for es
T
1
andT
2
, we needto know the relative number of Xb,n
1
andn
2
inea hwell,n
1
+n
2
= 1
. Undertheassumptionthatthestateofdetailed balan eisrea hed,the rate onstantsk
+
,des ribing transitionsfromposition 1to position 2,andk
−
, des ribing transitionfrom2 to1, are related through:k
+
k
−
= C exp
(B
12
− B
21
)
k
B
T
,
(1.11)where
T
isthe absolutetemperature,k
B
the Boltzmann onstant,C
a onstantto be determinedandB
12
andB
21
the a tivationenergies for passingfrom state1 to state 2 and vi e versa. In Fig. 1.27, we also see thatk
−
is onstant sin eB
21
is a xed quantity independentof the tensioninthe elasti element. Therefore we anwrite
B
21
= E
1
andB
12
= E
1
+ ∆U
tot
, where∆U
tot
is given by∆U
tot
= (E
2
− E
1
) +
1
2
K
"
y
0
+ y +
h
2
2
−
y
0
+ y
−
h
2
2
#
=
(1.12)kh(y
0
+ y) + (E
2
− E
1
)
Sin e inisometri ontra tionthe two states have the same energy:
∆U
tot
|
y=0
= 0
⇒ −Khy
0
= (E
2
− E
1
),
(1.13) the relation(1.11) be omes:k
+
k
−
= exp
−Khy
k
B
T
.
(1.14)Figure1.27:ThedierentbehaviorofHuxleyandSimmons1971model: totalenergyinstret h
(
y > 0
, where the onguration 1 is energeti ally preferred) and in release (y<0, where theonguration2isenergeti allypreferred)modes. From[10℄
The dierential equation des ribing the number of Xbs in the state 2 during
transientsis:
dn
2
(t)
dt
= k
+
n
1
(t)
− k
−
n
2
(t) =
−(k
+
+ k
−
)n
2
(t) + k
+
.
(1.15) Due to the hypothesis of equal energies of the states duringisometri ontra tionweobtain that
n
2
(0) = 1/2
. We an now solve (1.15)and write:n
2
(t) = n
∞
2
+ (0.5
− n
∞
2
) exp[
−t(k
+
+ k
−
)]
(1.16)where:
n
∞
2
=
k
+
k
+
+ k
−
.
One an see that the fra tion of Xbs instate 2, starts at one-half and rises to the
value
n
∞
2
exponentially with ratek
+
+ k
−
. To ompute the steady state tension, only the ratioof the rates onstants (1.14)is needed. The transientof tension anbe writtenas:
T (y, t) = n
1
(t)K
y
0
+ y
−
h
2
+ n
2
(t)K
y
0
+ y +
h
2
= K[y
0
+ y + (n
2
(t)
− 0.5)h].
The hara teristi values of tensionpredi ted by this modelare
T
1
(y) = T (y, 0) =
K(y
0
+ y)
andT
2
(y) = T (y,
∞) = K[y − 0 + y + (n
∞
2
− 0.5)h]
.In order to ompute these fun tions, one needs the elasti onstant
K
. As itwas not known at that time, Huxley and Simmons used the data on the rate of
re overy
r(y)
(Fig. 1.15). They obtained a t:r(y) = r
0
(1 + e
−αy
)
with
r
0
= 0.2 ms
−1
and
α = 0.5 nm
−1
. As we have seen, the same urve an be
predi ted fromthe model:
r(y) = k
+
+ k
−
= k
−
(1 + e
−Kh/k
B
T y
).
(1.17)Huxleyand Simmonsused this formulatoobtainthe values ofboth
k
−
andKh =
α k
B
T
. Then they ould ompute the equilibriumfor e:T
2
(y) =
αk
B
T
h
y
0
+ y
−
h
2
tanh
αy
2
(1.18)whi hisshowninFig. 1.28. Theresulting urves
T
1
(y)
andT
2
(y)
exhibitthesameFigure1.28: Predi tion oftheHuxleyandSimmons1971model. From[10℄
value for the stiness be omes
K
≃ 0.2 pN/nm
, however a value of one order ofmagnitude greater [35℄ [38℄ [36℄ [80℄ [92℄ was proven later. Weshall ome ba k on
the importan e of this value extensively in Chapter 2, however we mention here
that the more realisti value
K = 2 pN/nm
ae ts dramati allythe predi tion ofthe model. First,we see immediatelyfrom(1.17)that the
r(δ)
dependen e willbehighly overestimated, be ause it depends on
K
exponentially. Se ond, theT
2
(δ)
urve is more ine ted with the higher value of the parameterK
, thus it shows anegative slopeat
δ = 0
whi h isin ontrast withthe experimental result.There were several re ent attempts toimprove the quantitative predi tions of
the theory[18℄[43℄[48℄[59℄[73℄[96℄,howevernothingfundamentallynewhas been
added to the model. The most attra tive feature of the Huxley and Simmons
1971 model is thatit attempts tolink bio hemistrytome hani s. Notonlyin the
more re entmodelsthis linkwasnotimproved, but,onthe ontrary, itwasalmost
lost. In the next Chapter we shall give abriefreview of some of these models and
Power Stroke
2.1 Introdu tion
The spe ial hara ter of the me hani al response of skeletal mus les des ribed
in Chapter 1, allows one to lassify them as a tive materials, be ause they an
adapt to external stimuli. The advan es in te hnology are often linked to the
developmentofsu hmaterialsthat anprovidea tivefun tioning,likesensingand
a tuation. In the past, a tiveresponse was a hieved through organizing elements,
with passive response at the mi ro-level, into omplex stru tures with multiple
equilibrium states. However modern te hnologies require that su h me hanisms
fun tionatmi rometerandevennanometers ales,sotraditionalsolutionsbe ome
una eptable, and there is a demand for materialswhere the omplex behavior is
realized already at the mole ularlevel. An example of su h materialsis given by
shapememoryalloys,wherethemulti-stabilityofthesystematthemole ularlevel
isduetophasetransformationwhi hdoesnotrequirediusion,and anbeindu ed
bystress,temperatureorele tro-magneti eld. Theanalysisandmodelingofsu h
a tive materials has rea hed a level of pre ision that one would want to a hieve
in the des ription of skeletal mus les,given some similarity of the behavior of the
two types of systems. The similarity is based on the idea of multi-stability of the
mi ros opi elementsof the system.
As we have seen in Chapter 1 the model of Huxley and Simmons (HS71) an
des ribe fast response of skeletal mus les assuming the presen e of bi-stable
ele-ments with double well energy. In this model the energy lands ape is degenerate
be ause the wells are innitely narrow. This leads to a des ription in terms of
a jump pro ess, whi h requires the knowledge of hemi al rate onstants. In the
not needed: the Xb swit hes between the states instead of ontinuously moving
between them. This makes the pre ise analyti al omparison of this model with
me hani al models of shape memory alloys di ult. Despite these apparent
dis-tin tions, the main ingredients in both types of models are similar whi h leaves
a possibility to link the Huxley and Simmons model to the ontinuum theory of
martensiti transformations ina tive materials.
We re all that, the main dieren e between the multi-stableand onventional
linear elasti elements is that the energy of the former is non- onvex. As it was
shown inthe pioneering work of Eri ksen [11℄, this non onvexity isof
fundamen-tal importan e for the interpretation of the behavior experimentally exhibited by
shape memory alloys, whi h is related to the presen e of multiple stable
mi ro- ongurations. Eri ksen onsidered the behavior of a ontinuum 1-D problemfor
a material with a non- onvex energy under slowly varying load showing that a
mathemati al model based on bi-stability an explain hysteresis. After that, a
thorough study of the problem was performed, in parti ular a pre ise des ription
of themi ros opi eventswasobtained bydis retizing the1-D ontinuum problem
and viewingit asa hain of bi-stable elements [11℄, [16℄, [29℄, [62℄.
In this Chapter we reformulate the original Huxley and Simmons 1971 model
in this pre ise mathemati al framework. We show that this reformulation an
produ e a pi ture whi h avoids some drawba ks of the original HS71 model. We
start by briey des ribing the way in whi h other re ent models have dealt with
these drawba ks. Then we introdu e our new me hani al model aimed rst at
modelingthe power stroke only and present a quantitative analysis of this model
in ludingthe omparisonwith experimental urves.
2.2 Re ent Models
Already in 1978 [17℄ it was realized that the Huxley and Simmons 1971 model
an not predi t orre t time s ale of tension relaxation, if a realisti value of the
stinessoftheelasti elementisused. Thequantitativeresolutionofthisandother
problemsoftheHS71model, alreadymentionedinSe tion1.3.3,willbegivenlater
in this Chapter, while now we would like to briey review the main approa hes
used to ir umventthese problems. In parti ular we show that the way hosen by
the authors of the re ent models to deal with the drawba ks of the HS71 model,
leads toalmost omplete lossof ouplingbetween the two aspe ts of the problem:
un orre t time s ale predi ted by the HS71, however here we shall fo us only on
this aspe t of the problem.
Eisenberg and Hill model
An earlymodi ation of the originalHS71 model was proposed in 1978 by
Eisen-berg and Hill [17℄. The model was extended in 1980 on amore quantitativebasis
[18℄. Itisbasedontheobservationthatby assumingtwoverynarrowenergywells,
Huxley and Simmons made impli itly the hypothesis that the transition between
the states takes pla e only after a Xb had olle ted the total amount of energy
needed to over omethe barrier.
Eisenberg and Hill proposed to make the wells wider in order to allow the
transitiontostart atlowerenergy. They alsolinked the for egeneratedby the Xb
in ea h state with the rst derivative of the free energy, instead of the stret h of
the elasti element, that has now been formally eliminated. Without the elasti
element, the oordinate
x
of the Xb is ontrolled by the imposed length of thesar omere,a tuallythe oordinate
x
has be omeameasure ofaxialpositionoftheparti ular a tin site at whi h the Xb is atta hed. The origin
x = 0
was hosenin su h a way that the Xb in the pre-power stroke state is in itsresting position.
At every value of
x
the Xbs an be in four dierent states whose free energieswere postulated to have an
x
-dependen e shown in Fig. 2.1. HereAM
‡
DD
is
the a tin-myosin omplex in the pre-power stroke state and
AMD
is thea tin-myosin omplex in the post-power stroke state, (phases 2 and 3 in Fig. 1.5).
Similarly
M
∗∗
D
and
M
‡
D
are two deta hed states, the refra tory state and the
non-refra torystate,respe tively(phases4and 1inFig. 1.5). Thedierentangles
of the lever arm in the two atta hed states are ree ted by the dierentpositions
oftheminimainthe relativeenergies. The hangeofstateisassumedtobeajump
pro ess, allowing the Xb to follow the entire Xb- y le as shown by the arrows in
Fig. 2.1.
In dening the rate onstants of the individualjump pro esses, Eisenberg and
Hill used the hypothesis of detailed balan e whi h imposes one ondition for the
ratioofea hpair ofrate onstants. Inthe mostgeneralformthis ondition anbe
writtenas:
k
jm
(x)/k
mj
(x) = exp
G
j
(x)
− G
m
(x)
k
B
T
(2.1)Theelasti elementispresentinthis model throughthe
x
-dependen e of theFigure 2.1: Freeenergiesfortwoatta hedstatesandtwodeta hedstatesintheEisenbergand
Hillmodel. Thestate
M
∗∗
D + D
isshiftedwithrespe tto
M
∗∗
D + T
byanamountofenergy
givenbytheATPhydrolyzation. ThearrowsshowapossibleXb y le. From[17℄
the shape of the energy barrier between the states at a given
x
. That leaves forea h transition one of the rate onstants
k
ij
as a free parameter. In the Huxley and Simmons 1971 model the dependen ies of both rate onstants on the stepx
were ompletely dened by the shape of the total energy. Instead in Eisenberg
and Hill model these onstants retain some freedom (used in Fig. 2.2), sin e only
one ondition, equations (2.1), is imposed. To our knowledge, the EH78 model
wasrst toabandon the me hani altransparen y of the HS71 model even though
some features of the HS71 model were preserved. In a sense the bio hemi al
in-terpretation of dynami s has over ome the me hani al basis of the HS71 model.
The freedom left by the EH78 model was used to hoose the
k(x)
dependen iesphenomenologi ally inorder to tthe experimentalobservations almost perfe tly.
Piazzesi and Lombardi model
The model of Piazzesi and Lombardi developed in 1995 [43℄, deals with the entire
y leoftheXb,andisabletopredi tboththefor evelo ity urveandthebehavior
of the mus le subje ted to rapid in rementsin the total length as well asthe ux
of energy and the e ien y of the ontra tion. There are two deta hed states,
D1 and D2, and three atta hed states, A1, A2 and A3 in Fig. 2.3 A. Moreover
there are two distin t paths in whi h Xbs an split ATP to generate for e. Two
a tive states, A1 and A2, are ommonfor the two paths, as shown in Fig. 2.3(a).
From A2 there is a long path, whi h ontains a se ond a tive state A3 beforethe
deta hmentof thehead D1. Thispath an ompete with the short one, wherethe
head deta hes immediatelyafterA2into D2. The long path generateslarger for e
but it has a lower rea tion rate, about 20/se ond. The rea tion rate for the short
path is about 100/se ond. The orresponding rate onstants satisfy the detailed
balan e equation (2.1). As in the Eisenberg and Hill model, the fun tions
k
jm
(x)
andk
mj
(x)
depend onthe imposed step, and this dependen e an be hosen tot the experimental data. The spe i hoi e of the authors is shown in Fig.2.3(b-e). The system of dierentialequations governing the distribution of the number
of Xbs in various states,
a
1
,a
2
,a
3
andd
1
,d
2
, generalizes the equation proposed by Huxley in 1957. Without going in all details we just mention that the systemonsistsof the equationsof the type:
∂a
1
(x, t)
∂t
= k
1
(x)d
1
(x) + k
−2
(x)a
2
(x) + k
6
(x)d
2
(x)
−
(k
−1
(x) + k
2
(x) + k
−6
(x))a
1
(x, t)
− v
∂a
1
(x, t)
∂x
(2.2)for ea h of the vepossible states and issolved numeri ally.
Small length in rements bring the Xb in the region where the short y le is
favored and, being rapid, it an explain the behavior observed experimentally at
moderate shortening velo ities. For higherlength step the long y le is preferred,
leading to the possibility to t the
11 nm
power stroke. In this way the modelis able to predi t both for e-velo ity urve ( omputed onsidering the onstant
urvature ofthe freeenergy
G
i
(x)
asinthe EH78modeland showninFig. 2.4(a)) and step inlength type experiments (Fig. 2.4(b)).Huxley and Tideswell model
Huxleyand Tideswell proposed in1996 amodel whi h wasexpli itlydeveloped to
Figure2.3: Generals hemeandrate onstantsforthePiazzesiandLombardimodel. A:(Left)
S hemeof thePL95model, A=atta hed, D=deta hed. (Right)Basi freeenergyof ea hstate.
(a) (b)
Figure 2.4: Comparison between simulations(lines) and experimental results (points) of the
PL95model. (a) For e-velo ity urve. (b)ModelingofT1andT2. From[43℄
on the same idea as the HS71 model, however new features were added. Among
them, the most important for our approa h are that, rst, three atta hed states
A
1
,A
2
, andA
3
are onsidered, with two onstant power strokes between them of5.4 nm
and4.5 nm
. Se ond, the rate oe ients for ea h of the four possiblerea tions
A
1
→ A
2
,A
2
→ A
1
,A
2
→ A
3
,andA
3
→ A
2
, aregivenbythe expressionA/(1 + exp[B(x
− C)])
where
x
isrelatedtothestret hoftheelasti elementandA
,B
andC
aredierentfor ea h transition, so that 12 parameter have tobe spe ied. Observe that as in
the previous models, there is an
x
-dependen e of the rate onstants. In order torespe t the ondition(2.1) some onstraintsare added,however, noexpli it shape
of the free energywaspres ribed, whi h meansweakerme hano hemi al oupling
than in HS71 model. Finally, to mimi the in ommensurability of the spa ing of
thea tivesitesonthea tinlamentwithrespe t tothespa ingofthe headsalong
the myosin lament, ve populations of Xbs were equally spa ed relative to the
a tivesites. The authorsobserved that withthe lastassumptionthe tension
T
2
(δ)
ends up averaged over the range of spa ing of the a tive sites (see also Se tion2.7.2). Under the assumptionthatthis spa ingisequal to
5.5 nm
,the diameterofthe a tin monomer, the omputed
T
2
(δ)
urve ontains a at region aroundδ = 0
even withK = 2 pN/nm
. This solves the rst in oheren e of the HS71 modelDespite this su ess the single stroke size of
5.5 nm
was too small to justifyother experimentalobservations,and forthis reason theauthors were for etoadd
ase onda tivestateinthemodel. Thebehaviorofthemodelwasanalyzed
numer-i ally,and arathergoodtofthe experimentaldataregardingtherate ofre overy
was rea hed. The results are shown in Fig. 2.5, together with unsatisfa tory
pre-di tions of HS71 model. This solved even the se ond problemof the original1971
model.
Figure 2.5: Re ipro al of half-time of tension re overy vs. imposed length step. Solid line:
experimental values from [14℄. Points: Huxley and Tideswell model predi tion. Dotted line:
Huxleyand Simmonsmodelpredi tion
Weemphasize,however, thatthepri eofalargernumberofpossiblestateswas
a largernumberof freeparameters, and that the pre iserelationbetween the rate
onstants andthe shapeof thetotal energyused inHS71model hasbeenpartially
lost. We shouldmention though that,the hypothesis that the rate onstants vary
exponentiallywiththeworkdoneinstret hingoftheelasti element,waspreserved
(see [39℄). Asthe authorshaveobserved, thisrepresentsa dieren ebetween their
model and the Eisenberg and Hill model, where the authors used, as a typi al
rea tion rate, an arbitrary fun tion adjusted to obtain agreement.
Smith et al. model
Asalastexample,wedis ussaveryre entandthemost omplete modelofmus le