Thermophilic Enzyme or Mesophilic Enzyme with Enhanced

Jul 8, 2017 - On the other hand, the T130K species exhibits the same affinity to ligands and the same thermodynamic parameters of complex formation as...
2 downloads 14 Views 4MB Size
Subscriber access provided by UNIVERSITY OF MICHIGAN LIBRARY

Article

Thermophilic Enzyme or Mesophilic Enzyme with Enhanced Thermostability: Can We Draw a Line? Xiaomin Jing, Wilfredo Evangelista Falcón, Jerome Yves Baudry, and Engin H. Serpersu J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.7b04519 • Publication Date (Web): 08 Jul 2017 Downloaded from http://pubs.acs.org on July 10, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry B is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Thermophilic Enzyme or Mesophilic Enzyme with Enhanced Thermostability: Can We Draw a Line? Xiaomin Jing1,^, Wilfredo Evangelista Falcon1,3,^, Jerome Baudry1,3,*, Engin H. Serpersu2,%,*   1:  Department  of  Biochemistry  and  Cellular  and  Molecular  Biology,       2:    Graduate  School  of  Genome  Science  and  Technology,  the  University  of  Tennessee  and  Oak   Ridge  National  Laboratories,  Knoxville,  Tennessee  37996     3:  UT/ORNL  Center  for  Molecular  Biophysics,  Oak  Ridge  National  Laboratory,  Oak  Ridge,  TN.                     ^

 These  authors  contributed  equally    

*

Corresponding  authors  

%

 Current  Address,  National  Science  Foundation,  4201  Wilson  Blvd.  Arlington,  VA  22230      

Address  for  correspondence:  Engin  Serpersu,  National  Science  Foundation,  4201  Wilson  Blvd.   Arlington,  VA  22230  ([email protected])  and  Jerome  Baudry,  BCMB  Department,  M407 Walters Life Sciences 1414 Cumberland Avenue , Knoxville, TN 37996 ([email protected])  

1    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 26

Abstract The  aminoglycoside  nucleotidyltransferase  4'  (ANT)  is  a  homodimeric  enzyme  that  modifies  the  C4'-­‐OH   site    of  aminoglycoside  antibiotics  by  nucleotidylation.    A  few  single  and  double  residue  mutants  of  this   enzyme   (T130K,   D80Y   and   D80Y/T130K)   from   Bacillus   stearothermophilus   show   increased   thermostability.  This  paper  investigates  how  such  residue  replacements,  which  are  distant  from  active   site  and  monomer-­‐monomer  interface,  result  in  various  changes  of  the  thermostability  of  the  enzyme.   In   this   work,   we   show   that   thermodynamic   properties   of   enzyme–ligand   complexes   and   protein   dynamics   may   be   indicators   of   thermophilic   behavior.     Our   data   suggests   that,   one   of   the   single   site   mutants   of   ANT,   D80Y,   may   be   a   thermophilic   protein   and   the   other   thermostable   mutant   T130K   is   actually  a  more  heat-­‐stable  variant  of  the  mesophilic  WT  with  a  higher  Tm.    Our  data  also  suggest  that   T130K  and  D80Y  adopt  different  global  dynamics  strategies  to  achieve  different  levels  of  thermostability   enhancement,  and  that  the  differences  between  the  properties  of  the  species  can  be  described  in  terms   of   global   dynamics   rather   than   in   terms   of   specific   structural   features.   Thermophilicity   of   the   D80Y   comes  at  the  cost  of  less  favorable  thermodynamic  parameters  for  ligand  binding  relative  to  WT.  On  the   other   hand,   the   T130K   species   exhibits   the   same   affinity   to   ligands   and   the   same   thermodynamic   parameters   of   complex   formation   as   the   WT   enzyme.     These   observations   suggest   that   a   quantitative   characterization   of   ligand   binding   and   protein   dynamics   can   be   used   to   differentiate   thermophilic   proteins  from  their  simply  more  heat  stable  version  of  mesophilic  counterparts.            

2    

ACS Paragon Plus Environment

Page 3 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Introduction     Enzymes   from   thermophilic   organisms   play   an   important   role   in   biotechnology   and   industry.   Their   thermostability  and  optimal  activity  at  high  temperatures  makes  them  more  suitable  for  many  processes   compared   to   their   mesophilic   counterparts.   Thermophilic   proteins   may   adopt   various   strategies   to   achieve   thermal   adaptation.   Therefore,   there   has   been   many   studies   aimed   to   understand   molecular   mechanisms   that   increase   thermal   stability   of   proteins.     However,   the   molecular   and   thermodynamic   basis   of   thermophilicity   of   proteins   remains   largely   undetermined.   Earlier   work   suggested   that   thermophilic  variants  of  enzymes  have  several  differences  from  mesophilic  enzymes,  such  as  increased   polar   interactions1-­‐4   or   more   Hydrogen   bonds   or   increased   hydrophobic   interactions   due   to   a   better   packing  of  the  hydrophobic  core5-­‐8  or  some  combination  of  these  properties.  In  addition,  other  studies   also  suggested  that  distributed  effects  and  dynamic  properties  of  proteins  may  play  significant  roles  in   attaining   thermostability   9-­‐12   and   indeed   justification   of   thermophilicity   based     solely   on   amino   acid   sequences  do  not  allow  a  clear  rationalization  of  thermophilic  behavior13-­‐15.          

Figure  1.  Crystal  Structure  of  D80Y16  with  bound  ligands.  MgATP  (purple)  and  kanamycin  A  (red)  are   bound  to  the  active  site,  which  is  formed  at  the  interface  of  monomers.  Residues  D80  (green)  and  T130   (yellow)  are  shown  as  ball  and  stick  model.       Attempts   at   engineering   thermostable   proteins,   using   either   evolutionary   strategies   or   structure-­‐ based   approaches,   have   worked   on   a   case   by   case   basis.     Most   of   the   current   efforts   appear   to   be   directed  towards  the  improvement  of  thermostability17-­‐19.  These  types  of  approaches  may  yield  enzymes   with   higher   thermostability   for   practical   applications;   however,   as   indicated   in   20   they   may   not   be   thermophilic   proteins.   Instead,   they   may   be   mesophilic   proteins   with   increased   melting   temperatures   (Tm)20-­‐22.  In  this  work,  we  attempted  to  show  that  increased  Tm  alone  does  not  render  a  protein  to  be   considered   as   “thermophilic”   and   there   are   molecular   properties   that   are   specific   for   thermophilic   proteins.    By  using  single  mutant  variants  of  the  aminoglycoside  nucleotidyltransferase(4ʹ′),  we  show  that   dynamic  properties  of  apo-­‐enzymes  and  thermodynamics  of  enzyme-­‐ligand  interactions  are  among  the   parameters   that   allow   discrimination   between   thermophilic   proteins   and   those   simply   with   increased   Tm.    In  this  work,  for  simplicity,  we  will  use  the  term  “thermostable”  to  denote  the  variant  with  higher   Tm  but  otherwise  have  identical  protein  dynamics  and  thermodynamics  of  ligand-­‐protein  interaction  to   3    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 26

the  mesophilic  WT  enzyme  and  reserve  the  term  “thermophilic”  for  the  variant  that  show  significantly   different  properties  in  these  aspects  that  we  attribute  to  be  indicators  of  thermophilic  behavior.     The   aminoglycoside   nucleotidyltransferase   (4ʹ′)   (ANT)   catalyzes   the   transfer   of   the   AMP   group   from   ATP  to  the  4'-­‐OH  site  of  aminoglycosides  (Fig  S1),  which  results  in  elimination  of  their  effectiveness  as   antibiotics.   The   enzyme   was   originally   isolated   from   S   aureus   as   a   mesophilic   protein23.   Thermostable   variant  was  isolated  from  Bacillus  stearothermophilus24.    Of  the  two  thermostable,  single  mutants  of  the   enzyme,  T130K  displays  dynamic  and  thermodynamic  properties  identical  to  that  of  the  mesophilic  WT   while,   the   other   mutant,   D80Y,   has   significantly   different   properties.     It   is   not   clear   how   a   single   residue   replacement,   distant   from   active   site   and   the   monomer-­‐monomer   interface   (Figure   1),   yield   global   stabilization   of   enzyme   at   elevated   selection   temperatures.     This   work   describes   thermodynamic   and   computational  data  leading  to  differentiation  of  thermophilic  variant  from  more  heat  stable  mesophilic   variant  of  ANT.       Experimental  and  Computational  Methods         Chemicals  and  Reagents     High-­‐performance   Ni-­‐Sepharose   resin   and   IPTG   were   obtained   from   GE   Healthcare   (Pittsburg,   PA)   and    Inalco  Spa  (Milan,  Italy)  respectively.  Ion  exchange  matrix,  Macro  Q,  was  purchased  from  Bio-­‐Rad   Laboratories   (Hercules,   CA).   Dr.   E.   Fernandez   of   the   University   of   Tennessee   generously   provided   the   purified   thrombin.   Aminoglycosides   and   all   other   chemicals   were   purchased   at   the   highest   purity   available   from   Sigma,   Aldrich   (St.   Louis,   MO).     As   a   standard   procedure   in   this   laboratory,   aminoglycosides   were   purchased   as   sulfate   salts   and   desulfated   by   ion   exchange   chromatography   for   use  as  base.       Site-­‐directed  mutagenesis     T130K  variant  cloned  in  a  pET15b  vector  was  used  as  the  template  in  PCR  amplification  to  create  wild   type   and   double   variant   clones.   For   D80Y   variant,   wild   type   clone   was   used   as   the   template   in   PCR   reaction.  All  the  mutagenesis  was  performed  using  the  Quikchange  Lightning  site-­‐directed  mutagenesis   kit   (Stratagene,   La   Jolla,   CA).   PCR   products   were   transformed   into   competent   XL10   cells.   pET15b   plasmids   with   wild   type/each   variant   were   isolated   by   Miniprep   (Qiagen,   Germantown,   MD)   and   confirmed   by   sequencing   check   (Molecular   Biology   Resources   Facility,   University   of   Tennessee,   Knoxville).  Positive  clone  was  transformed  into  E.coli  BL21  (DE3)  Gold  competent  cells.    Positive  clones   were  confirmed  by  sequencing  check.  Purity  of  wild  type  enzyme  and  variants  (>95%)  were  visualized  by   10%  SDS-­‐PAGE  and  further  validated  by  analytical  ultracentrifugation  tests.       Overexpression  and  purification     Overexpression   and   purification   of   wild   type   enzyme,   D80Y   and   double   variants   are   performed   similarly   with   T130K   variant   as   previously   described25,   except   cell   lysis   was   performed   by   three   cycles   of   French   press.   Enzymes   remained   active   for   at   least   a   month   at   -­‐20°C.   Protein   concentrations   were   determined   by   absorbance   at   280nm   using   extinction   coefficient   of   49390   M-­‐1cm-­‐1   for   wild   type   enzyme   and   T130K   and   50880   M-­‐1cm-­‐1   for   all   other   variants.   Concentrations   of   proteins   were   reported   as   monomer.       Analytical  ultracentrifugation     4    

ACS Paragon Plus Environment

Page 5 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

  Beckman   XL-­‐1   analytical   ultracentrifuge   was   used   to   perform   sedimentation   velocity   experiments.   400   μL   samples   were   loaded   into   double-­‐sector   cells   in   an   An-­‐50Ti   rotor.   Sample   temperature   was   equilibrated   at   25°C   for   an   hour   before   the   run.   To   achieve   linearity   protein   concentrations   were   detected   at     230   nm   for   low   concentrations   (1-­‐3   μM)   while   measurements   at   280   nm   was   used   for   concentrations   5-­‐20   μM.     For   enzyme   at   concentration   higher   than   40   µM,   the   interference   optical   system   was   used   for   detection.   All   the   scans   were   collected   at   a   rotor   speed   of   50,000   rpm   at   25°C.   SEDFIT   (version   12.44)26     was   used   to   fit   sedimentation   data   by   using   continuous   (c(s))   distribution   model.   Protein   partial   specific   volume,   buffer   density   and   buffer   viscosity   were   calculated   using   SEDNTERP27.     The  enzymes  were  dialyzed  extensively  in  a  buffer  solution  of  50  mM  PIPES  pH  7.5  containing  0  or  100   mM   NaCl   and   used   in   experiments.     All   ligands   were   also   dissolved   in   the   same   dialysis   buffer.   The   weight-­‐average  sedimentation  coefficients  (sw(S))  were  determined  at  each  concentration  by  integrating   the   peaks   from   the   c(s)   distributions   using   SEDFIT26.   The   monomer-­‐dimer   self-­‐association   model   in   SEDPHAT28    was  used  to  determine  dissociation  constants  for  the  dissociation  of  dimers  into  monomers   by  using  sw(S)  as  a  function  of  protein  concentration.     Isothermal  titration  calorimetry     Titrations   of   substrate   binding   were   performed   as   previously   described29.   Titrations   of   dimer   dissociation  were  performed  similarly,  except  each  titration  consisted  of  4  injections  of  65  μL  and  were   separated   by   360   seconds.   Apo-­‐enzyme   with   concentration   above   dissociation   constant   was   titrated   into   buffer.   As   a   result,   a   certain   fraction   of   dimeric   enzyme   dissociated   into   monomeric   form,   accompanied   with   the   observed   heat   (ΔH0).   Known   dissociation   constants   of   dimers   for   each   variant,   determined  by  AUC,  allowed  calculation  of  fractions  of  dimer  and  monomer  in  both  syringe  and  cuvette,   which,  in  turn,  allowed  calculation  of  the  heat  of  dimer  dissociation.       Circular  Dichroism     CD   was   performed   on   a   Jasco   J-­‐815   spectrometer   using   a   cuvette   with   a   path   length   of     2   mm.   Spectra  were  recorded  several  times    with  a  rates  of  1  and  1.5°C/min  increments    in  temperature  while   monitoring  the  changes  at  222  nm.  Protein  concentration  was  18uM  (monomer)  for  all  the  runs.       Molecular  Dynamic  simulations                Systems   were   constructed   based   on   the   crystal   structure   of   D80Y   species   (Protein   Data   Bank   ID:   1KNY)   using   Molecular   Operating   Environment   (MOE,   version   2012,   Chemical   Computing   Group,   Ltd,   Montréal,  Canada).  The  co-­‐crystalized  ligand  and  cofactor  were  removed  from  the  model  such  that  the   apo  form  of  the  wild  type,  and  T130K  species  were  built.  Each  structure  was  explicitly  solvated  in  a  TIP3P   water   in   a   cubic   box   of   8   nm   x   8   nm   x   8   nm.   Periodic   Boundary   Conditions   in   all   directions   were   applied   with  electrostatic  type  Fast  smooth  Particle  Mesh  Ewald,  PME.  A  18,000  steps  energy  minimization  was   performed  using  the  steepest  decent  algorithm  and  maximum  force  equal  to  10  KJ  mol-­‐1  nm-­‐1.                Molecular   Dynamics   simulations   were   carried   out   using   the   Gromacs   4.6.130-­‐31   simulation   engine   and   the   AMBER-­‐f99sb32   force   field.   For   each   species   and   temperature   studied   here,   a   50   ns   equilibration   with  a  2fs  integration  timestep  was  performed  in  the  NPT  thermodynamic  ensemble,  and  1  bar  pressure   using   the   Nosé-­‐Hoover   and   Berendsen   weak   coupling.   Finally,   a   100   nanoseconds   production   was   run   using  the  Panirello-­‐Rahman  algorithm.  Atomic  coordinates  were  saved  on  disk  every  5  ps.   5    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 26

           The   trajectories   obtained   from   the   production   time   were   analyzed   using   Principal   Component   Analysis   from   Gromacs   tools,   eigenvectors   and   eigenvalues   were   calculated   to   identify   the   main   dynamic  modes  for  the  three  species,  WT,  T130K,  and  D80Y  at  three  different  temperatures  300K,  322K,   and  330K.         Results       A  couple  of  single  mutations  on  ANT  increased  the  thermostability  of  the  enzyme  without  affecting   the   catalytic   activity   of   the   enzyme   where   each   mutation   provided   a   different   degree   of   thermostability   33 .      Thermal  denaturation  of  wild  type  and  thermostable  variants,  determined  by  CD,  confirmed  these   observations  (Figure  S2).      The  melting  temperatures  increase  in  the  order  of  wild  type  (40.9  ±  0.5°C),   T130K   (49.1±0.6°C),   D80Y   (56.2±0.2°C)   and   D80Y/T130K   (62.6±0.1°C).   While   D80Y   displays   a   higher   thermostability  relative  to  T130K,  the  double  variant  T130K/D80Y  shows  an  additive  effect  of  individual   mutations   on   Tm.     The   additivity   of   the   increments   in   Tm   values   is   consistent   with   two   different   mechanisms   providing   an  increase   in   thermostability   to   the   double   variant.     Thermal   denaturation   of   all   four  enzymes  was  irreversible  and  aggregation  was  observed  post  denaturation.  However,  identical     denaturation   profiles   were   obtained   with   each   variant   at   different   heating   rates,   which   indicates   that   thermal  denaturation  of  enzymes  were  not  perturbed  by  kinetics  of  aggregation34.        Subunit-­‐subunit  interaction  is  affected  by  single  amino  acid  replacements               ANT  is  a  homodimeric  enzyme  with  ligand  binding  sites  at  the  interface  of  monomers.  Even  though   the   mutation   sites   are   away   from   the   subunit-­‐subunit   interface,   they   affected   the   dimer   formation   as   detected  by  analytical  ultracentrifugation  (AUC).    As  shown  in  Figure  2,  monomer-­‐dimer  equilibrium  was   concentration   dependent   with   all   variants   and   there   was   a   progressive   increase   in   tendency   for   dimerization  with  increasing  thermostability.    Dissociation  constants  (Kd)    of  dimers  for  all  enzymes  were   determined  as  previously  described29.  The  curves  shown  in  Figure  S3  represent  the  best  fit  to  the  data   points  for  all  the  tested  enzymes.    The  dissociation  constants    for  homodimers  of  wild  type,  T130K,  D80Y   and  D80Y/T130K    were  estimated  to  be  within  the  range  of  28-­‐32  μM,  1.6-­‐1.8  μM,  0.8  -­‐  0.9  μM  and  0.40   -­‐   0.5   μM   respectively.   In   all   cases,   Kd   increased   steadily   from   the   most   thermophilic   to   mesophilic   variants   reaching   to   a   ~70-­‐fold   difference   between   the   mesophilic   WT   and   the   most   thermostable   variant  D80Y/T130K.         Our   previous   study   on   T130K   showed   that   increasing   salt   concentration   favored   the   formation   of   dimer   indicating   that   hydrophobic   interactions   are   the   major   cause   of   dimerization29.   Similar   sedimentation  velocity  experiments  were  performed  for  wild  type,  D80Y  and  D80Y/T130K  variants  in  the   absence  of  added  salt.    On  the  average,  there  was  an  almost  an  order  of  magnitude  increase  in  Kd  with   all  enzymes  indicating  that  hydrophobic  interactions  are  the  main  contributors  of  dimer  formation  with   all  variants.       Dimer   dissociation,   detected   by   isothermal   titration   calorimetry   (ITC),   showed   an   endothermic   behavior  with  all  four  enzymes  in  the  presence  and  absence  of  100  mM  NaCl.  The  heat  of  dissociation   varied  within  3.1  -­‐  5.5  kcal/mol  for  all  variants  and  didn’t  appear  to  follow  the  order  of  thermostability.     We   note   that   these   determinations   were   based   on   dilution   of   the   enzyme   from   a   concentrated   stock   to   the  ITC  chamber  and  errors  were  larger  for  D80Y  and  D80Y/T130K  due  to  low  Kd  for  dimer  dissociation.       6    

ACS Paragon Plus Environment

Page 7 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

 Antibiotic  selection  separates  two  thermostable  variants  carrying  a  single  mutation                  Thermodynamic   properties   of   enzyme–ligand   complexes   with   all   variants   were   determined   by   ITC.   Tobramycin   and   neomycin   were   used   as   representatives   of   kanamycins   and   neomycins   respectively   (Figure  S1).  The  formation  of  the  aminoglycoside-­‐enzyme  complexes    are  typically  results  in  changes  in   pKas   of   functional   groups   in   the   enzyme   and   in   the   ligand   causing   a   net   proton   uptake   or   release35-­‐38.   Therefore,    buffer  solutions  with  different  heats  of  ionization  (ΔHion)    were  used  in  titrations       Figure   2.   Size   distribution   of   ANT   and   its   variants   determined   by   sedimentation   velocity   experiments.    

For  all  plots,  red,  blue  and  green  lines  represent  5,  10  and  20  μM  protein  respectively.  Panels  are  WT,   T130K,  D80Y  and  T130K/D80Y  from  top  to  bottom  respectively.     to  determine  the  intrinsic  binding  enthalpies  (∆Hint)  and  the  change  in  net  protonation  (∆n)  for  wild  type   and  thermostable  variant  of  ANT  as  described  in  detail  earlier36.  Figure  3  shows  an  example  of  such  data   as  a  plot  of  observed  enthalpy  (ΔHobs)  vs.  ΔHion  where  the  intercept  on  y-­‐axis  is  ΔHint  and  ∆n  is  obtained   from   the   slope   of   the   line.   Experiments   performed   with   T130K   yielded   identical   results   to   those   published  earlier29.    As  it  can  be  seen  in  Figure  3,  slopes  of  lines  determining  the  value  of  Δn  display  a   clear  separation  between  the  pairs  of  D80Y/double  mutant  and  WT/T130K  when  neomycin  is  the  ligand.  

7    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 26

    Figure  3.    Effect  of  the  heat  of  ionization  on  the  observed  enthalpy  of  binding.    The  binding  enthalpy  of   neomycin  to  WT  (○),  T130K  (●),  D80Y  (▫)  and  D80Y/T130K  (◊)  is  plotted  against  the  heat  of  ionization  of   buffers  (PIPES,  HEPES  and  Tris  with  2.7,  4.9  and  11.4  kcal/mol  of  heat  of  ionization  respectively).     The   dissociation   constants   for   enzyme–neomycin   complexes   showed   remarkable   differences;   while   the  WT  and  T130K  have  nanomolar  affinity,  binding  of  neomycin  to  D80Y  and  D80Y/T130K  was  weaker   by   almost   three   orders   of   magnitude.   Large   differences   were   also   visible   in   the   other   thermodynamic   properties   of   enzyme–ligand   complexes   (Table   1).   Binding   of   neomycin   to   WT   and   T130K   was   highly   exothermic  with  similar  ΔHint  values  but  the  binding  enthalpy  of  neomycin  to  D80Y  and  D80Y/T130K  was   much  less  exothermic  with  less  unfavorable  entropy  relative  to  WT/T130K  pair.  Overall,  WT  and  T130K   showed   highly   similar   thermodynamic   properties   of   complex   formation   with   neomycin,   which   were   significantly  different  than  D80Y  and  D80Y/T130K.       Table  1.    Thermodynamic  parameters  for  the  formation  of  ANT–Neomycin  complexesa    

Wild  Type  

T130K  

D80Y  

T130K/D80Y  

         ΔHint  (kcal/mol)  

-30.0 ± 0.8  

-32.8 ± 0.6  

-17.2 ± 0.6  

-15.6 ± 0.4  

990 ± 200  

990 ± 150   -8.4 ± 0.2  

                   KD  (nM)  

40 ± 6  

24 ± 6  

           ΔG  (kcal/mol)  

-10.1 ± 0.1  

-10.4 ± 0.1  

-8.3 ± 0.3  

         TΔS  (kcal/mol)  

-19.9

-22.4

-8.9

-7.2

                         Δn  

1.5 ± 0.1  

1.95 ± 0.1  

1.0 ± 0.1  

0.95 ± 0.1  

  a

Standard  error  of  the  mean,  based  on  three  trials,  are  shown.    Least  square  fits  from  ΔHobs  vs  ΔHion  plots   were  used  to  determine  errors  in  the  intrinsic  enthalpy  and  in  net  protonation.     8    

ACS Paragon Plus Environment

Page 9 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Table  2.    Thermodynamic  parameters  for  the  formation  of  ANT–Tobramycin  complexesa    

Wild  Type  

     T130K  

     D80Y  

         ΔHint  (kcal/mol)  

-19.5 ± 0.9  

-19.3 ± 1.4  

-13.9 ± 2.1

                     KD  (µM)  

2.5 ± 0.4  

1.1 ± 0.2  

1.7 ± 0.6  

2.0 ± 0.7  

           ΔG  (kcal/mol)  

-7.7 ± 0.1  

-8.1 ± 0.1  

-8.0 ± 0.2  

-7.8 ± 0.2  

           TΔS  (kcal/mol)  

-11.8

-11.2

1.1 ± 0.1  

1.4 ± 0.2  

                         Δn  

T130K/D80Y     -13.3 ± 0.8  

-5.9 0.8 ± 0.3  

-5.5 0.9 ± 0.10  

a

Standard  error  of  the  mean,  based  on  three  trials,  are  shown.    Least  square  fits  from  ΔHobs  vs  ΔHion  plots   were  used  to  determine  errors  in  the  intrinsic  enthalpy  and  in  net  protonation.     Contrary   to   the   binding   of   neomycin,   the   binding   affinity   of   tobramycin   to   all   variants   was   similar   (Table   2).   However,   just   like   in   the   complex   formation   with   neomycin,   the   enthalpy   of   formation   of   WT-­‐ and  T130K–tobramycin  complexes  were  more  exothermic  than  D80Y  and  D80Y/T130K  variants  without  a   clear  separation  in  Δn  values.  Furthermore,  the  change  in  heat  capacity  (ΔCp),  as  determined  from  the   slopes   of   lines   of   temperature   dependence   of   the   binding   enthalpy,   displayed   much   smaller   range   of   variation   with   D80Y   as   compared   to   those   observed   with   the   WT   and   T130K   when   different   aminoglycosides  are  used  (Figure  4).    

  Figure  4.  Temperature  dependence  of  binding  enthalpy.    Data  for  each  enzyme  are  shown  with  4   aminoglycosides  representing  the  widest  range  of  binding  affinity  to  each  variant.  Top  panel,  WT:   neomycin  (●),  tobramycin  (▪),  kanamycin  B  (Δ)  and  ribostamycin  (○);  Middle  panel,  T130K:  neomycin  (●),   paromomycin  (◊),  tobramycin  (▪)  and  kanamycin  A  (▫);  Lower  panel,  D80Y  neomycin  (●),  kanamycin  A  (▫)   kanamycin  B  (Δ)  and  ribostamycin  (○).  Scale  of  y-­‐axis  is  matched  to  aid  in  visual  comparison.      Distinct  dynamic  properties  of  T130K  and  D80Y.     The  effect  of  mutations  on  the  dynamics  properties  of  WT,  T130K,  and  D80Y  was  characterized  from  the   100  ns  MD  trajectories  at  three  different  temperatures,  300K,  322K,  and  330K  (melting  temperature  of   D80Y).   The   trajectories   were   analyzed   via   Principal   Component   Analysis   (PCA)   to   identify   the   main   9    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 26

motions   of   the   protein   species   at   each   temperature39,   as   previously   shown   by   us   40.   Analysis   of   300K   data  suggests  that  the  global  dynamics  of  D80Y  is  dominated  by  the  first  dynamic  mode  (Figure  5-­‐right   panel),  whereas  at  322K  and  330K  the  first  two  modes  contribute  both  significantly  to  the  global  protein   dynamics.    The  dynamics  of  the  WT  and  T130K  species  are  mainly  due  to  the  contribution  of  the  first  two   modes   at   all   temperatures   studied   here   as   shown   in   Figure   5   (left   and   middle   panels),   although   the   first   mode  of  motion  contributes  more  at  330K.  The  orientation  and  amplitude  of  the  first  two  modes  for  WT,   T130K,   and   D80Y   at   300K   are   shown   by   vectors   in   Figure   6.   The   origin   of   these   vectors   indicate   the   region  of  the  protein  undergoing  motion  in  those  particular  vectors’  direction.  The  left  panels  of  these   figures  show  the  motion  due  to  the  first  mode  of  the  protein.  The  first  mode  corresponds  to  the  same   dominant   movement   in   the   three   species:   an   open/close   “breathing”   motion.   The   difference   between   the  dynamics  of  the  three  species  at  different  temperatures  originates  from  the  second  dynamics  mode,   right  panels  of  Figure  6,  which  becomes  significantly  contributing  in  the  D80Y  variant  only  at  322K  and   330K.    These  results  suggest  that  the  first  mode  of  motion  has  the  same  direction  at  all  temperatures  for   all  the  three  species    (left  panels  of  Figures  S4  and  S5).  As  it  occurs  at  300K,  the  difference  seems  to  arise   from   the   second   mode,   whose   vectors   have   different   origins   and   directions   depending   on   the   species     (right  panels  of  Figures  S4  and  S5).    

    Figure  5.    Principal  modes  of  motion  of  ANT  projected  onto  the  first  25  eigenvalues  calculated  from  PCA   in  MD  trajectories  at  different  temperatures;  Left  to  right  are  WT,  T130K,  and  D80Y,  respectively.  Filled   circles  (blue),  squares  (green)  and  diamonds  (red)  represent  300K,  322K  and  330K  respectively.  Insets   show  expanded  regions  of  the  several  initial  modes  of  motion.     Discussion     D80Y  and  T130K  adopt  different  structural  strategies  in  enhancing  enzyme  thermostability       Thermostability  of  proteins  can  be  increased  by  residue  replacements  to  introduce  specific  stabilizing   interactions,  which  may  produce  a  thermostable  protein.  However,  such  studies  may  generate  simply  a   variant  of  mesophilic  protein  with  higher  Tm  rather  than  being  a  thermophilic  variant.  In  this  work,  we   determined  dynamic  and  thermodynamic  properties  of  two  single  mutants  of  ANT,  both  of  which  have   increased  thermostability  however  only  one  of  them,  D80Y,  appeared  to  be  thermophilic  enzyme.  Data   shown   in   this   work   revealed   significant   clues   on   the   identification   of   molecular   properties   that   separate   thermophilic   protein   from   a   thermostable   mesophilic   counterpart.   In   this   case,   T130K   shows   a   7°C   increased   Tm   relative   to   WT,   however,   its   dynamic   properties   and   thermodynamics   of   enzyme–ligand   complexes   are   identical   to   the   mesophilic   WT.   D80Y   variant,   on   the   other   hand,   has   14°C   higher  Tm   and   has  different  protein  dynamics  and  thermodynamics  parameters  of  enzyme–ligand  interactions.     10    

ACS Paragon Plus Environment

Page 11 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Moreover,   findings   from   MD   simulations   point   out   to   a   temperature-­‐   and   sequence   of   mode-­‐    

dependence  of  the  ANT’s  protein  dynamics.  The  three  protein  species  characterized  here  all  exhibit  the       Figure  6.    The  arrows,  obtained  from  PCA,  represent  the  first  (left  panel)  and  second  mode  (right  panel)   of  motions  respectively.  Top  to  bottom  are:      WT,  T130K,  and  D80Y.  C-­‐α  atoms  of  Asp80  and  Thr130  are   represented  by  orange  and  purple  spheres,  respectively.     same  first  mode  of  motion  at  the  three  temperatures  tested  here.  However,  the  second  mode  of  motion   is   different   in   various   species   and   temperatures.     The   WT   and   T130K   species   exhibit   very   similar   magnitude  of  contributions  from  the  second  mode,  but  the  regions  affected  by  this  second  mode  and  its   directions  are  different  between  the  two  species,  as  shown  in  Figures  6  and  5S.  On  the  other  hand,  D80Y   exhibits  different  relative  contributions  from  the  first  and  second  mode  with  respect  to  the  other  two   species  at  300K  (Figure  5).  At  330K,  this  variant  shows  a  significantly  different  second  dynamics  mode,   both  in  magnitude  and  direction,  when  compared  to  that  of  WT  or  T130K    (Figure  S5).  These  differences   suggest   that   mesophilic/thermophilic   properties   can   be   characterized   by   global   dynamic   properties,   rather  than  just  by  structurally-­‐localized  differences  between  species.       11    

ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 26

Crystal  structures  of  WT  and  T130K  are  not  available,  however,  the  structure  of  D80Y  is  available16.   CD  spectra  (210-­‐280  nm)  suggests  that  structures  of  all  variants  are  similar  and,  furthermore,  the  effect   of  salt  on  homodimer  formation  also  indicates  that  similar  interactions  are  driving  the  dimer  formation   in  all  mutants.  Therefore,  all  interpretations,  made  based  on  the  available  structure  of  D80Y,  which  we   believe,  are  reasonably  justified.       Both  residues  T130  and  D80  are  exposed  to  solvent,  are  away  from  the  monomer-­‐monomer  interface   and   the   active   site.   Yet,   the   mutation   of   these   residues   increases   Tm.   The   crystal   structure   of   D80Y   shows  that  the  130th  residue  is  positioned  at  the  beginning  of  the  fifth  major  alpha-­‐helical  region.  E127   and   D133   are   within   4.5Å   of   T130.   Thus,   the   replacement   of   T   with   K   may   create   strong   polar   interactions  with  either  E127  or  E133.  Positions  of  these  residue  are  shown  in  a  close  up  of  the  structure   in  Figure  7.  Thus,  it  is  likely  that  the  new  interactions  raise  the  melting  temperature  of  the  mutant  by   affecting  local  interactions  and  structure  without  affecting  the  overall  dynamic  properties  of  the  enzyme   and  thermodynamics  of  enzyme-­‐ligand  interactions.    

  Figure  7.    Close-­‐up  of  T130  and  its  surrounding  residues.  T130  (blue),  E127  and  D133  (purple)  are  shown   in  ball  and  stick  manner  with  several  other  neighboring  residues.  Kanamycin  A  is  shown  in  white.  

  Figure  8.    Close-­‐up  of  the  position  of  residue  80.  The  surroundings  of  position  80  is  shown  in  the  crystal   structure  of  D80Y16  .  D80Y  is  shown  in  yellow.    Kanamycin  A  is  shown  in  white.     There   are   no   positively   charged   residues   within   4.5Å   of   D80.   One   of   the   neighboring   residues   of   D80,   N78,   forms   hydrogen   bonds   with   E76   and   E67,   which   are   the   key   residues   involved   in   electrostatic   12    

ACS Paragon Plus Environment

Page 13 of 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

interactions  with  antibiotics.  (Figure  8)      In  addition,  D80  toY80  replacement  may  also  benefit  from  the   internal  hydrophobic  packing  of  enzyme  by  the  introduction  of  an  aromatic  residue  at  this  site.  Thus,  it   appears   that   D80   acts   a   node   in   the   connectivity   network   of   residues   of   ANT   and   have   significant   role   in   determining  protein  dynamics.  This,  in  turn,  affects  alters  thermodynamics  of  ligand  binding  and  makes   it  thermodynamically  less  favorable.    The  behavior  of  the  double  variant  D80Y/T130K  suggests  that  D80Y   mutation   has   the   dominant   effect   while   the   effect   of   mutation   at   T130   is   limited   to   increasing   Tm   confirming  that  thermostability  achieved  by  different  mechanisms  by  these  variants.       We  should  also  mention  that  thermophilic  behavior  of  D80Y  is  contrary  to  current  paradigm,  which   suggests   that   thermophilic   proteins   contain   more   polar   residues   relative   to   their   mesophilic   counterparts.   Replacement   of   the   ionic   side   chain   of   aspartate   with   phenolic   side   chain   of   tyrosine   in   D80Y   contrasts   this   notion   and   suggests   that   alteration   of   dynamic   properties   of   the   protein   may   be   more  important  in  achieving  thermophilicity.  This  observation,  again,  suggests  that  comparisons  based   on   protein   sequences   alone   need   to   be   interpreted   cautiously   to   differentiate   molecular   properties   of   thermophilic  proteins  from  mesophilic  ones.     Various  structural  and  dynamic  features  of  thermophilic  and  mesophilic  enzymes  were  compared  in   numerous   studies.   These   efforts,   however,   may   be   complicated   by   the   fact   that   they   included   all   thermostable   enzymes,   some   of   which   may   simply   converted   to   thermostable   versions   of   mesophilic   enzymes   by   the   introduction   of     stabilizing   interactions   such   as   disulfide   bonds   or   salt   bridges.   Furthermore,   these   comparisons   included   proteins   that   have   differences   ranging   from   a   few   residues   to   tens   of   residues   in   their   primary   sequences.   Some   of   these   variations   may   introduce   changes   that   are   not   necessarily   specific   for   thermophilicity.     In   this   work,   such   differences   are   minimized   to   a   single   residue  replacement  between  the  variants.     Ligand  binding  to  the  thermophilic  variants  are  enthalpically  less  favored  relative  to  WT  and  T130K.   However,   relatively   less   unfavorable   entropic   contribution   renders   ligand   binding   still   thermodynamically   favorable   (ΔG