• Keine Ergebnisse gefunden

Principles  of  size-­‐exclusion  chromatography

C)                                   top  five  unique  and  symmetric  complexes  of  perlucin   (without  calcium  ions)

3.4. Size-­‐exclusion  chromatography  of  perlucin

3.4.1.   Principles  of  size-­‐exclusion  chromatography

Back  in  1959  Porath  and  Flodin  (Porath  &  Flodin  [1959],  Porath  [1960])  introduced  “a   method  for  desalting  and  group  separation”  of  macromolecular  solutions,  which  they   called   “gel   filtration”.   They   filled   columns   with   liquid-­‐soaked   beads   of   cross-­‐linked   dextran  under  wet  conditions.  Dextran  is  a  polymer  of  glucose  that  can  be  cross-­‐linked   with  epichlorohydrin  in  small  beads  (see  e.g.  Porath  [1960],  Flodin  [1967]  and  Arshady   [1991a]  for  a  review  on  the  manufacturing  of  beaded  gel  filtration  media).  Porath  and   Flodin   (Porath   &   Flodin   [1959])   applied   a   solution   of   polysaccharides   of   different   molecular  weights  and  a  monosaccharide  on  top  of  a  bed  of  hydrated  dextran  beads  as   described  above.  After  the  solution  had  entered  the  bed  the  column  was  continuously   purged  with  distilled  water  at  approximately  2  𝑚𝑚𝑚𝑚/𝑚𝑚𝑚𝑚𝑚𝑚.  The  eluated  solution  from  the   column  was  fractionated  and  analysed  for  carbohydrate  content.  It  turned  out  that  the   components  of  the  saccharide  solution  were  separated  and  those  molecules  with  larger   molecular   weight   were   found   in   the   fractions   corresponding   to   a   smaller   elution   volume.  

This  observed  effect  –  the  separation  of  components  with  different  molecular  weight  in   such   a   manner   that   heavier   substances   have   a   lower   elution   volume   compared   to   lighter   molecules   –   can   be   explained   by   the   following   basic   mechanism   (see   e.g.  

Tiselius   et   al.   [1963]   or   Hagel   [2011]   as   well   as   Yau   et   al.   [1979]   for   a   general   introduction  to  SEC).  Depending  on  the  “effective  size”  (“effective”  in  this  context  shall   denote   those   size   of   the   molecules   that   takes   effect   in   SEC)   of   the   molecules   in   the   solvent  that  is  used  during  the  chromatographic  analysis,  different  volumes  inside  the   polymer  beads  are  accessible  for  them.  The  smaller  the  molecules  the  larger  the  space   inside  the  beads  consisting  of  (cross-­‐linked)  polymers  they  can  occupy  and  vice  versa.  

This   separation   principle   is   illustrated   schematically   in   Fig.   3.4.1.   This   easily   drawn   picture   of   the   operation   principle   of   SEC   assumes   that   an   equilibrium   between   the   concentrations   of   the   molecules   inside   the  polymer   beads   and   in   the   void   volume   is   established  fast  with  respect  to  the  solvent  flow  rate.    

 

Fig.   3.4.1.   Principle   of   size-­‐exclusion   chromatography.   A   dissolved   sample   A   consisting   of   several  constituents  of  different  sizes  (red,  green  and  blue  dots)  is  brought  into  a  tube  between   a  buffer  solution  reservoir  B  and  a  SEC  column  C.  Through  continuous  pumping  of  the  buffer   solution  the  sample  is  applied  on  top  of  the  SEC  bed  (orange  circles).  The  beads  consist  of  a   (cross-­‐linked)   polymer   network   schematically   illustrated   by   black   entangled   lines   inside   the   orange  circles.  The  polymer  beads  consist  of  pores  of  different  sizes.  Due  to  the  different  size  of   the  sample  molecules  (red,  green,  blue  dots)  different  volumes  inside  the  polymer  beads  are   accessible  for  these  molecules.  This  is  indicated  in  the  figure  as  follows.  The  red  particles  are   large   and   cannot   enter   the   polymer   beads.   The   green   particles   are   smaller   than   the   red   particles  and  can  enter  parts  of  the  polymer  beads.  In  contrast  the  smallest  blue  particles  of  the   injected  sample  can  access  most  of  the  space  inside  the  polymer  beads.  This  leads  to  a  size-­‐

dependent   separation   of   the   injected   sample.   The   different   molecules   leave   the   column   in   different  elution  volumes  𝑉𝑉!.  When  the  separated  molecules  leave  the  SEC  column  they  can  be   detected   by   a   photometer   D   operating   at   the   appropriate   wavelength.   The   output   of   the   photometer  is  a  graph  of  the  absorbance  dependent  on  the  elution  time  or  volume.  Finally  the   solution  that  passed  the  photometer  is  fractionated  in  E.  An  important  parameter  in  SEC  is  the   (extra-­‐particle)  void  volume  𝑉𝑉!.  The  void  volume  is  shown  in  this  figure  in  dark  grey.  It  is  the   space   inside   the   SEC   column   that   is   not   occupied   by   polymer   beads   (orange).   Particles   that   cannot  penetrate  the  polymer  beads  (or  interact  with  them)  leave  the  SEC  column  in  an  elution   volume  that  is  equivalent  to  the  void  volume.  Note  that  in  this  illustration  the  buffer  solution   pump  and  the  valve  for  the  sample  injection  is  not  shown  for  simplicity.  Figure  prepared  with   Inkscape  (http://inkscape.org).  Similar  figures  can  be  found  e.g.  in  Yau  et  al.  [1979]  as  well  as   Hagel  [2011].    

 

This  assumption  was  tested  with  at  least  two  approaches  (see  also  Yau  et  al.  [1979],   chapter  2.3.  therein).  Little  et  al.  (Little  et  al.  [1969])  found  no  flow  rate  dependency  of  

the  elution  volume  of  acetonitrile  and  polysterene  in  the  flow  rate  range  between0.1   and   12.5  𝑚𝑚𝑚𝑚/𝑚𝑚𝑚𝑚𝑚𝑚.   Yau   (Yau   [1969])   incubated   dry   SEC   media   directly   with   different   solutions  of  molecules  of  different  molecular  weight  each  for  24  ℎ.  Then  he  determined   the  concentration  of  the  corresponding  molecules  in  the  volume  outside  of  the  swollen   SEC   media.   The   concentrations   of   the   molecules   in   solution   before   and   after   the   addition   of   SEC   media   were   measured.   A   relationship   between   these   concentrations   and   the   elution   volume   of   the   corresponding   substance   on   the   corresponding   SEC   media  could  be  derived  if  the  ideal  case  of  instantaneous  equilibration  during  SEC  was   assumed.   For   Yau   the   experimental   values   obeyed   the   expected   relationship   sufficiently  to  conclude  that  “the  exclusion  effect  plays  the  primary  role  in  GPC  peak   separation”  (Yau  [1969],  p.  491).  However  Yau  noted  that  other  processes  can  affect   the   separation   of   macromolecules   on   a   SEC   column   as   well.   In   particular   lateral   diffusion  can  have  an  influence  on  the  separation  depending  on  the  substances  to  be   separated   and   the   experimental   conditions.   The   effect   of   diffusion   can   be   expected   being   more   pronounced   at   higher   flow   rates   and   substances   with   low   diffusion   coefficients.  

To   allow   a   comparison   of   the   results   of   SEC   experiments   with   different   column   geometries   not   the   elution   volume  𝑉𝑉!  of   a   substance   is   usually   stated   but   its  

“distribution  coefficient”  (also  termed  “partition  coefficient”  or  “exclusion  coefficient”)   𝐾𝐾!  (see  e.g.  Gelotte  [1960]).  It  is  given  by  

 

𝐾𝐾! =𝑉𝑉!− 𝑉𝑉!

𝑉𝑉!   (3.4.1.a)  

 

and   includes   the   (extra-­‐particle)   void   volume  𝑉𝑉!  and   the   accessible   inner   (intra-­‐

particle)  pore  volume  𝑉𝑉!  of  the  SEC  media.  Introducing  the  total  liquid  volume  𝑉𝑉!  inside   a  SEC  column  the  distribution  coefficient  can  be  written  as  (see  e.g.  Hagel  [2011])    

𝐾𝐾! =𝑉𝑉!− 𝑉𝑉!

𝑉𝑉!− 𝑉𝑉!   (3.4.1.b)  

 

Note  that  usually  𝑉𝑉!  is  not  equal  to  the  geometrical  SEC  column  volume  𝑉𝑉!  due  to  the   volume   occupied   by   the   cross-­‐linked   hydrated   polymers   of   the   SEC   media.   The   distribution  coefficient  of  a  substance  can  be  interpreted  as  the  fraction  of  the  volume  

of   the   SEC   media   that   is   available   to   the   corresponding   substance   during   the   SEC   (Laurent   &   Killander   [1964]).   Under   ideal   conditions   the   distribution   coefficient   has   values  between  0  (substance  is  completely  excluded  from  the  SEC  media)  and  1  (the   SEC  media  is  completely  accessible  to  the  substance).  

 

Obviously   there   are   at   least   two   important   parameters   relevant   to   the   separation   of   substances  with  SEC.  One  parameter  is  the  pore  shape  inside  the  polymer  beads  that   constitute  the  media  inside  the  SEC  column.  The  other  parameter  is  the  “effective  size”  

of   the   molecules   to   be   analysed   under   the   conditions   during   the   SEC.   Knowledge   of   both  parameters  would  be  necessary  to  derive  a  theoretical  relationship  between  the   distribution  coefficient  of  a  given  substance  and  a  molecular  property  like  molecular   weight.  

The   pore   size   distribution   of   dry   SEC   media   can   be   determined   experimentally   for   example   with   mercury   intrusion   porosimetry   (see   Grimaud   et   al.   [1978]   for   an   example   study   and   Adamson   [1990]   for   a   brief   introduction   to   mercury   intrusion   porosimetry).   Many   different   pore   shape   models   for   SEC   media   can   be   found   in   the   literature,  for  example:  simple  pore  geometries  like  cylinders  or  spheres  (e.g.  Yau  et  al.  

[1979]  or  Knox  &  Scott  [1984]),  a  normal  distribution  of  the  accessible  volume  fraction   (Ackers   [1967]),   SEC   media   as   a   network   of   randomly   oriented   rods   (Laurent   &  

Killander  [1964]),  or  the  accessible  volume  is  designed  as  the  space  between  random-­‐

sized  touching  spheres  (Knox  &  Scott  [1984]).  Hagel  et  al.  (Hagel  et  al.  [1996])  stated   that  “many  materials  have  a  rather  complicated  structure  which  may  not  be  described   accurately  by  a  few  parameters”  (p.  34).  They  used  experimentally  obtained  SEC  data   to  calculate  “apparent  size-­‐exclusion  pore  dimensions”  (p.  42)  for  a  given  SEC  media.  

These  dimensions  were  obtained  from  a  fit  of  the  experimental  data  to  the  following   model.   The   authors   assumed   on   the   one   hand   that   the   SEC   media   consists   of   pores   (cylindrical,  spherical  or  slab  geometry)  with  a  normal  distribution  of  radii  (the  mean   and   the   standard   deviation   were   fit   parameters)   and   on   the   other   hand   that   the   molecules  to  be  separated  are  hard  spheres.  Dependent  on  the  chosen  pore  geometry   (cylindrical,   spherical   or   slab   geometry)   different   apparent   pore   dimensions   were   calculated  for  the  same  set  of  experimental  values.  

It   seems   that   for   the   SEC   media   used   in   this   thesis   –   cross-­‐linked   copolymer   of   allyldextran  and  bisacrylamide  (Sephacryl  S-­‐100  HR,  Hagel  et  al.  [1989],  see  also  Fig.  

10  in  Arshady  [1991b]  for  a  chemical  structure)  –  a  simple  pore  geometry  cannot  be  

inferred  from  scanning  electron  microscopy  images  given  by  Hagel  et  al.  (Hagel  et  al.  

[1989],  Fig.  2  therein  for  Sephacryl  S-­‐500  HR).  

A  determination  of  the  apparent  pore  dimensions  as  described  by  Hagel  et  al.  (Hagel  et   al.   [1996])   was   beyond   the   scope   of   this   thesis.   In   principal   this   would   require   a   sufficient  number  of  well  characterized  dextran  standards.  

 

The  reason  why  dextran  standards  are  used  in  several  different  studies  as  calibration  –   because  they  are  flexible  polymers  –  leads  to  the  second  important  parameter  involved   in   the   separation   of   substances   with   SEC:   the   “effective   size”   of   the   molecules   to   be   separated.  

It  seems  to  be  reasonable  to  assume  that  the  “effective  size”  is  the  hydrated  volume  of   the   substance.   The   hydrated   volume  𝑉𝑉!  of   a   molecule   of   this   substance   can   be   expressed  as  (Cantor  &  Schimmel  [1980],  Chapter  10-­‐2)  

 

𝑉𝑉! = 𝑀𝑀

𝑁𝑁!   𝜈𝜈!+ 𝛿𝛿!"  𝜈𝜈!" ≈  𝑀𝑀

𝑁𝑁! 𝜈𝜈! + 𝛿𝛿!"  𝜈𝜈!   (3.4.2.)  

 

Here  𝑀𝑀  is   the   molecular   weight   of   the   substance,  𝜈𝜈!  is   the   specific   volume   of   the   hydrated  substance,  𝛿𝛿!"  is  the  ratio  of  the  masses  of  the  solvent  bound  to  the  hydrated   substance  and  the  substance  itself  and  𝜈𝜈!"  is  the  specific  volume  of  the  bound  solvent.  

Assuming   that   the   substance   is   diluted   and   its   concentration   does   not   affect   its   hydration  then  the  sum  in  the  brackets  in  3.4.2.  can  be  approximated:  𝜈𝜈!  is  the  partial   specific  volume  of  the  substance  and  𝜈𝜈!  the  specific  volume  of  the  (pure  bulk)  solvent.  

The   partial   specific   volume   of   a   substance   expresses   the   change   of   the   volume   of   a   (liquid)  system  when  a  certain  mass  of  the  substance  is  added  to  the  solvent  (see  also   IUPAC   Gold   Book,   http://goldbook.iupac.org/P04422.html,   last   access   07/03/14).  

Some   illustrative   values   for   proteins   quoted   from   Cantor   and   Schimmel   (Cantor   &  

Schimmel  [1980])  are  given.  The  partial  specific  volume  of  proteins  is  in  the  order  of   0.73  𝑐𝑐𝑚𝑚!/𝑔𝑔  and  the  hydration  is  in  the  order  of  0.4   𝑔𝑔  H!O 𝑔𝑔  protein.  

There   are   at   least   two   easier   approaches   to   determine   the   hydrated   volume   of   a   solvated  molecule.  One  exploits  the  diffusion  of  molecules  in  solution.  If  the  diffusion   constant  𝐷𝐷  of   a   molecule   in   a   given   solution   with   viscosity  𝜂𝜂  is   known,   one   can   calculate  its  frictional  coefficient  with  the  Einstein-­‐Smoluchowski  relation.  In  a  second  

step   Stokes   law   can   be   used   to   calculate   the  (Stokes)   radius  𝑟𝑟!"  of   a   sphere   with   the   same  diffusion  and  friction  coefficient  

 

𝑟𝑟!" = 3 4  𝜋𝜋 𝑉𝑉!

!!

= 𝑘𝑘!  𝑇𝑇 𝐷𝐷

1

6  𝜋𝜋  𝜂𝜂   (3.4.3.)  

 

Here  𝑘𝑘!  and  𝑇𝑇  are   the   Boltzmann   constant   and   the   Temperature.   An   ellipsoidal   molecule   shape   can   be   considered   by   shape   parameters   (see   for   example   Cantor   &  

Schimmel  [1980]).  

Following   Cantor   and   Schimmel   (Cantor   &   Schimmel   [1980],   Chapter   12-­‐1)   another   possibility   to   determine   the   hydrated   volume   exploits   the   intrinsic   viscosity  [𝜂𝜂]  of   a   solvated  substance.  It  is  defined  as  

 

𝜂𝜂 = lim!→!

𝜂𝜂𝜂𝜂!− 1  

𝑐𝑐   (3.4.4.)  

 

If  a  (macro)molecular  substance  is  solvated  in  a  liquid  with  viscosity  𝜂𝜂!  then  one  can   expect   an   concentration  𝑐𝑐  dependent   increase   in   the   solution´s   viscosity  𝜂𝜂(𝑐𝑐).   The   limiting   value   –   at   infinite   dilution   –   of   the   relative   change   in   viscosity   divided   by   concentration   is   the   intrinsic   viscosity.   Note   that   𝜂𝜂  has   the   units   of   the   inverted   concentration  𝑐𝑐,  e.g.  volume  per  mass.  This  volume  can  be  associated  with  the  hydrated   volume  of  the  solute  

 

𝜂𝜂 = σ  𝑉𝑉!𝑁𝑁!

𝑀𝑀   (3.4.5.)  

 

where  σ  is  a  shape  factor  considering  the  geometry  of  the  molecule.  It  is  σ ≥ 2.5  where   the  lower  bound  corresponds  to  a  spheric  molecule.  

In  terms  of  SEC  both  parameters  –  the  Stokes  radius  as  well  as  the  intrinsic  viscosity   [𝜂𝜂]  –   were   proposed   amongst   others   as   parameter   that   describes   the   separation   of   different   substances   (see   Hagel   [2011]).   Potschka   (Potschka   [1987])   proposed   the   viscosity  radius  obtained  from  the  intrinsic  viscosity  as  universal  calibration  principle.  

Frigon   et   al.   (Frigon   et   al.   [1983])   observed   that   dextrans   and   native   proteins   with  

similar  value  of  the  product   𝜂𝜂 ⋅ 𝑀𝑀  had  similar  elution  volumes.  Less  agreement  was   found   when   Stokes   radii   were   used.   Dubin   and   Principi   (Dubin   &   Principi   [1989])   found   the   use   of   𝜂𝜂 ⋅ 𝑀𝑀  only   meaningful   for   flexible   polymers   and   globular   proteins   since  rod-­‐like  molecules  show  an  elution  volume  deviating  from  the  one  expected  from   the   intrinsic   viscosity.   It   might   be   important   that   the   latter   two   studies   used   silica   based  SEC  media  whereas  Potschka  used  agarose-­‐based,  silica-­‐based  and  cross-­‐linked   polymer  SEC  fillings.  

 

It   could   be   observed   that   the   elution   volume   of   proteins   can   depend   on   the   ionic   strength  (for  example  Frigon  et  al.  [1983])  and  pH  value  (for  example  Golovchenko  et   al.  [1992])  of  the  buffer  solution  as  well  as  on  the  explicit  column  material  (for  example   Agrawal  &  Goldstein  [1965]).  

The  ionic  strength  and  pH-­‐value  dependency  of  the  elution  volume  of  proteins  during   SEC  can  be  attributed  at  least  to  two  effects.  As  summarised  by  Hagel  (Hagel  [2011])   between   the   samples   and   the   SEC   media   electrostatic,   van   der   Waals   and   repulsive   interactions  can  occur.  Those  interactions  can  be  influenced  by  the  choice  of  the  buffer   solutions   pH   value   and   ionic   strength   (see   also   Ruckenstein   &   Lesins   [1986]).  

Additionally   the   buffer   solution   composition   can   influence   the   conformation   of   proteins.  A  drastic  example  is  the  pH  dependent  interaction  between  β-­‐lactoglobulin   (β-­‐Lg)  molecules:  at  pH  2.6  it  is  a  monomer  and  at  near  neutral  pH  values  it  forms  a   dimer   (Uhrínová   et   al.   [2000]).   Additionally   between   pH   6   and   8   conformational   changes  –  e.g.  a  change  of  the  solvent  accessible  surface  area  –  occur  in  the  protein  (Qin   et  al.  [1998]).  

Finally  it  must  be  pointed  out  that  some  proteins  can  interact  with  the  SEC  media  if  it  is   composed  of  structures  similar  to  native  ligands  of  the  protein.  Agrawal  and  Goldstein   (Agrawal  &  Goldstein  [1965])  observed  that  the  lectin  concanavalin  A  can  bind  to  the   cross-­‐linked  dextran  of  a  SEC  medium.  This  must  be  kept  in  mind  since  perlucin  as  a  C-­‐

type  lectin  might  show  a  similar  behaviour.  As  it  is  stated  above  the  SEC  media  used  in   this   thesis   consist   also   of   a   cross-­‐linked   dextran   and   it   was   reported   by   Hagel   et   al.  

(Hagel   et   al.   [1989])   that   a   lectin   from   lentil   show   affinity   for   this   particular   SEC   medium.  Such  behaviour  might  be  beneficial  for  purification  and  separation  of  proteins   from  protein  mixtures  but  is  definitely  unwanted  for  size  estimation.  

 

The  information  provided  so  far  is  only  a  very  brief  summary  of  the  fundamentals  of   the  separation  principle  of  SEC  and  the  influence  of  the  buffer  solution.  Nonetheless  it   is  considered  to  be  sufficient  in  the  scope  of  this  thesis.  It  will  become  clear  in  the  next   sections   that   the   main   challenge   was   the   determination   of   suitable   buffer   solution   conditions  for  the  perlucin  elution  from  the  SEC  media.  The  systematic  variation  of  the   buffer   conditions   had   to   be   left   for   future   research.   The   provided   background   is   therefore   considered   sufficient   to   estimate   the   apparent   size/molecular   weight   of   perlucin.  Four  proteins  of  known  molecular  weight  were  used  as  reference  substances.  

As   it   was   shown   for   example   by   Andrews   (Andrews   [1964])   the   elution   volume   of   native  proteins  during  SEC  can  be  correlated  to  the  molecular  weight  of  those  proteins   –  at  least  under  the  conditions  given  in  this  study.  In  a  certain  range  –  that  depends  in   particular  on  SEC  media  –  the  elution  volume  varied  nearly  linear  with  the  logarithm  of   the  molecular  weight.  

In   the   following   section   the   results   of   the   SEC   with   the   reference   substances   and   perlucin   are   shown.   Detailed   information   on   the   used   reference   substances   and   the   experimental  set-­‐up  will  be  given  in  the  next  section.