CLIN CHEM 391 4852 1993
`
`Ionic Binding Net Charge and Donnan Effect of Human Serum Albumin as a
`Function of pH
`
`Niels FoghAndersen Poid Jannik Bjerrum and Ole SiggaardAndersen
`
`The ionic activities and total molalities of sodium potas
`sium calcium lithium and chloride in a solution of human
`serum albumin were measured at different values of pH
`between 4 and 9 The same quantities were measured
`simultaneously in a protein free electrolyte solution in
`membrane equilibrium with the albumin solution Taking
`liquid junction potential and bias from unse
`the residual
`lectivity of the electrodes into account we determined the
`own bound and net charges of albumin Chloride was
`low pH and calcium at high pH The
`amply bound at
`ions bound to albumin opposed the
`varying charge of
`effect of acid or base on the net charge All
`ions were
`distributed across the membrane according to the same
`electric potential difference which equalled the Donnan
`The high concordance
`between
`observation
`potential
`and theory favors the Donnan
`theory and furthermore
`implies that the electrodes are as accurate in a solution
`with albumin as in a protein free solution
`
`Addftlonal Keyphrases artificial kidney
`motic pressure
`ion selective electrodes
`ride
`lithium
`
`sodium
`
`potassium
`
`dialysis
`
`calcium
`
`colloid os
`chlo
`
`The colloid osmotic pressure of albumin in plasma
`balances
`the
`intravascular
`pressure
`hydrostatic
`thereby maintaining a normal plasma volume About
`half of the colloid osmotic pressure is due to the Donnan
`effect The permeable cations and anions are unevenly
`distributed intra and extravascularly according to the
`total molality
`mmolkg of water on the vascular side The contribu
`tion of the Donnan effect
`to the colloid osmotic pressure
`ircouoid is proportional to the squared molality of im
`to the
`permeable charge and inversely proportional
`molality of salt
`
`Donnan theory 1 with the larger
`
`irconoid = R T Kmaibumin
`
`Znet2Orneetn
`
`PH20 1
`
`where R is the gas constant T is the temperature
`is
`the osmotic coefficient m is molality Znet
`is the molality
`of impermeable albumin net charge and PH20 is the
`density of water
`The net charge of albumin is defined as its own charge
`plus the charge of all bound ions all of which depend on
`
`pH 2 pH and electrolyte disturbances are common in
`
`patients with abnormal plasma volume and albumin
`for example because of neonatal asphyxia hemorrhage
`shock nephrotic syndrome or lung edema so a study of
`
`Department of Clinical Chemistry Herlev Hospital DK2730
`Herlev Denmark
`Received June 29 1992 accepted August 21 1992
`
`48 CLINICAL CHEMISTRY Vol 39 No 1 1993
`
`the influence of pH on the net charge Donnan effect
`and ionic binding of albumin is clinically relevant
`Increasing the pH in a patient might be as effective for
`the colloid osmotic pressure as an albumin infusion
`Although plasma is the system used normally in
`is the interstitial
`fluid that consti
`clinical chemistry it
`tutes the actual environment of the cells and is regu
`Ion activity is not
`lated by homeostasis
`identical
`in
`fluid The Donnan effect and the
`plasma and interstitial
`factors affecting it should be known and taken into
`for the interpretation of electrolyte
`account
`results
`especially in patients with severe electrolyte pH or
`protein disturbances 3 The Donnan effect may also
`
`underlie the alkalosis of hypoproteinemia
`dosis of hyperproteinemia
`
`and the aci
`
`Materials and Methods
`The experimental setup is shown in Figure 1 We used
`an artificial kidney with a Cuprophan membrane Gam
`bro GF 120M Hollow Fiber Dialyzer with 13 m2 effec
`tive membrane area On the inner side of the mem
`brane was a 200 gL solution of human serum albumin
`for injection Nordisk Gentofte Denmark 05 L which
`had been dialyzed to remove its stabilizing caprylic acid
`The ratio of fatty acid to albumin was 1 The high
`concentration of albumin was chosen to give a high
`precision to the measurements but we must assume
`the association constants were unaffected We
`that
`concentrations of NaC1 KC1 and
`added physiological
`CaC12 plus LiC1 1 mmolL imidazole 1 mmolL and
`succinic acid 1 mmolL All reagents were analytical
`grade from Merck Darmstadt Germany The two buff
`ers were necessary to stabilize pH in the proteinfree
`the pH interval
`solutions and they were chosen to cover
`without binding calcium Extra CaCl2 was added to
`for the increased albumin binding of Ca2+
`compensate
`with increasing pH Bicarbonate was not included here
`because Pc02 could not be controlled and because bicar
`bonate interferes with the chloride electrode On the
`outer side of the membrane was a proteinfree solution
`with ions in equilibrium with the albumin solution
`The solutions were maintained at 37°C The flow of
`albumin and proteinfree electrolyte through the artifi
`cial kidney was maintained at 500 mLimin by two
`peristaltic pumps The volume was maintained by an
`adjustable clamp on the tubing leading from the artifi
`cial kidney and an increased transmembrane pressure
`was necemary with increasing pH
`pH was increased from 4 to 9 in 14 steps by adding
`sodium hydroxide Constant results indicated that equi
`librium was reached 15 min after sodium hydroxide was
`added the samples were drawn after 30 min
`
`Abraxis EX2045
`Actavis LLC v Abraxis Bioscience LLC
`1PR201701101 1PR201701103 1PR201701104
`
`

`

`37°C
`
`been identical which was not always the case We used
`the ratio between molality and electrode result in the
`protein free solutions as a correction factor for convert
`ing the electrode results into molality of free ions in the
`respective albumin solutions assuming that the activ
`ity coefficients were the same on both sides of the
`membrane
`Liquid junction potentials Es caused by diffusion
`between the bridge solution solution 1 and the test
`solution solution 2 were calculated with the Hender
`son equation
`
`Es =RTFl f1
`
`f2gi
`
`g2 ing1ig2 3
`
`f =X mi Ai zi1 g =
`where F is the Faraday constant
`X mi Ai mi is the molality of free ion i and zi
`is the
`charge number The following limiting equivalent con
`
`ductivities were used unit S cm2 mol1 ANa+ = 66
`Art+ = 92 Acr = 9605 ACa2+ = 78 Aacc2 = 68 and Au
`from NOVA had a bridge solution
`= 49 The electrodes
`of KC1 2 molkg water and those from Radiometer had
`a bridge solution of NaHCO2 sodium formate 40
`molkg water All results with ion selective electrodes in
`the albumin solutions were subsequently corrected for
`the calculated difference in liquidjunction potential
`between the protein free solution and the albumin solu
`tion
`The uneven distribution of ions as measured with the
`ion selective electrodes was reported as an electric equi
`librium potential of size and direction irrespective of
`the ionic charge number To make the results compara
`ble we expressed
`the observed
`ionic distribution
`ratios as equilibrium potentials in mV The equilib
`for each ion was calculated with the
`rium potential
`Nernst equation
`
`all
`
`EisE = R T zi1 F1 lnmiamin
`
`4
`
`where the superscripts a and w denote the albumin and
`proteinfree water solutions respectively
`The binding of each ionic species to albumin was
`calculated as the difference between the molality oftotal
`and free ion in the albumin solutions
`
`mAbound = mitotal mifree
`
`6
`
`The molality of albumins own charge without bound
`
`ions 20 was calculated from the neutrality condition
`
`because albumins own charge balances the total charge
`of all other cations and anions
`
`=
`
`X mitotal zi
`
`6
`ions bound to albumin 4d
`
`pH 4
`
`9
`
`Fig 1 Experimental
`setup
`The hollow fiber artificial kidney in the middle and the containers
`for albumin
`thermostat Two roller pumps
`and dialysis
`fluid were immersed in a water
`provided the counter current circulation with the albumin solution being on the
`Inside of the hollow fibers and the dialysis fluid on the outside The transmem
`brane pressure could be regulated by a clamp on the albumin line coming from
`samples could be taken by the syringes The
`the artificial kidney Analytical
`had the same ions as the dialysis fluid in membrane
`albumin solution
`the time of sampling
`
`equilibrium at
`
`All samples were subject to many measurements The
`activity of Na K+ Ca2+ Li Cl and pH was mea
`electrodes with some overlap
`sured by ionselective
`KNA1 K+ and Nal one CL1 0 and Na not
`caused by an excess of electrode systems We used one
`commercially available two ICAls Ca2+ and pH and
`one BMS2 pH from Radiometer Copenhagen Den
`mark and one NOVA 11 K+ Na and Lit and one
`STAT Profile 1 K+ Na Ca and pH from NOVA
`Biomedical Waltham MA
`
`concentrations of sodium and potassium
`Substance
`were measured by flame emission photometry with an
`IL 343 flame photometer
`Instrumentation Laboratory
`Lexington MA calcium and lithium were measured by
`atomic absorption spectrophotometry with a Perkin
`Elmer 403 spectrophotometer Norwalk CT and chlo
`ride was measured with a CMT10 chloride titrator from
`Radiometer The mass concentration of water pap was
`loss of 1 mL during an
`determined from the weight
`overnight drying at 105 °C The mass concentration of
`albumin was determined by the weight after drying
`subtracting the mass concentration of salt The molality
`of albumin was determined from the mass concentra
`tions of albumin and water by using a molecular mass
`for albumin of 66 000 Da
`concentrations mmolL
`The measured substance
`were converted to molality mmolkg water by dividing
`by the mass concentration of water
`
`= Cip H20
`
`2
`
`The charge molality of all
`was calculated as
`
`We observed a variable bias due to unselectivity of the
`the extremes of
`ion selective electrodes especially at
`pH If all
`ions were free in the proteinfree solutions the
`molality of total ions and electrode reading should have
`
`7
`Zbound = mbound zi
`The molality of net charge of albumin Zt equalled
`CLINICAL CHEMISTRY Vol 39 No 11993
`
`49
`
`

`

`Znet
`
`Zo
`
`+ Zbound
`
`8
`
`which is the molality of impermeable charge determin
`ing the Donnan distribution The number of ions and
`charges per albumin molecule zbend zeeen and znet
`were obtained by dividing by the molality of albumin
`
`We calculated the Dorman potential ED from
`
`the molality of net charge of albumin in the albumin
`solutions and the molality of monovalent electrolyte
`mew on the protein free side of the membrane
`Eponnan =RTF11n ZnAmealt +
`1 + Znet27nsalt2°5
`
`9
`
`The theoretical Dorman distribution ratio rDenn for
`ion i can be calculated from Ersonnim
`
`= expzi F EIR T
`
`10
`
`Results
`The results were constant after 15 min indicating
`equilibrium and did not depend on the preceding pH
`
`indicating reversibility
`The binding of chloride is shown in Figure 2 At
`physiological pH seven chloride ions were bound per
`albumin molecule The binding increased with decreas
`ing pH especially below pH 54 which is the isoelectric
`pH where Z
`= 0 and where 11 chloride ions were
`bound per albumin molecule The number of chloride
`ions bound to one albumin molecule increased to 22 at
`pH 42
`The calcium binding to albumin is shown in Figure 3
`No calcium was bound between pH 50 and 45 but the
`binding curve indicated a beginning calcium binding
`below pH 45 The calcium binding increased strongly
`with increasing pH especially above pH 65 The curve
`was Sshaped with maximum slope at physiological pH
`
`25
`
`20
`
`15
`
`10
`
`5
`
`0
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`Fig 2 Binding of chloride
`Number of chloride ions bound per albumin molecule n as a function of pH
`
`60 CLINICAL CHEMISTRY Vol 39 No 1 1993
`
`2
`
`5
`
`5
`
`0
`
`3
`4
`5
`Fig 3 Binding of calcium
`
`6
`
`7
`
`8
`
`9
`
`10
`
`of 74 where one calcium ion was bound per albumin
`molecule
`Sodium potassium and lithium were not measurably
`bound to albumin at physiological pH At pH 5 both
`sodium and potassium showed negative binding the
`total molalitybeing 510 less than the molality of free
`ions measured with ionselective electrodes The appar
`ent binding of lithium varied in an unsystematic man
`ner below the isoelectric pH probably because of the
`
`lithium electrode having low selectivity towards W
`The relative bias of the lithium electrode ranged from
`+25 to 30 in the pH interval 49 but that of the
`
`other electrodes was far less
`Figure 4 shows the charge of one albumin molecule as
`a function of pH Albumins own charge decreased to
`zero at the isoelectric pH pH 54 and became negative
`with increasing pH because of the added sodium hy
`
`Albumins Charges
`
`40
`
`30
`
`20
`
`10
`
`0
`
`10
`
`20
`
`30
`
`40
`
`6
`7
`8
`10
`3
`4
`5
`9
`Fig 4 The number of charges per albumin molecule as a function of
`PH
`Albumins own charge defined by the neutrality condition equalled the titration
`curve bound charge was the total charge of all bound ions and net charge
`was the sum of own and bound charges
`
`

`

`mV
`
`A E
`
`11
`
`PH13
`
`94st1
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`10
`
`5
`
`10
`
`5
`
`0
`
`10
`
`5
`
`0
`
`5
`
`10
`
`5
`
`0
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`10
`
`5
`
`0
`
`5
`
`10
`
`5
`
`0
`
`5
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`Fig 5 Residual
`liquid junction potential Donnan potential and membrane equilibrium potentials
`lb NOVA A Radiometer
`
`droxide The total charge of bound ions was always
`negative and it changed oppositely to albumins own
`charge as a function of pH with less chloride and more
`calcium being bound at increasing pH In this way the
`net charge of albumin changed less than its own charge
`changed At physiological pH the approximate charges
`= 17
`zbound = 6 and znet = 23 The changes per albumin
`per albumin molecule were as follows z
`molecule and pH unit at physiological pH were dzowni
`dpH = 76 dzbounddpH = 52 and dzdpH = 24
`These differential quotients are valid only at pH 74
`zuet was 14 at
`the isoelectric pH pH = 54 and
`became zero at one pH unit below the isoelectric pH
`pH 44
`The residual liquid junction potentials the calculated
`Donnan potential and the equilibrium potential deter
`mined by ionselective electrode for each ion as a func
`tion of pH are shown in Figure 5 The calculated
`residual liquidjunction potentials changed from 0 to
`1 mV in direction opposite to the calculated Eponnan
`which changed from 1 mV to 8 mV The Ems for
`each ion were 08 mV lower
`ED and the curves were parallel Table 1 gives
`
`regression data EsE vs EDonnan for each ion except
`lithium which had a larger variation Mean values
`were used for the regression when there was more than
`one electrode All correlation coefficients were >099
`the slopes were 1 but higher for chloride than for
`sodium and the SDs about
`the regression lines were
`<04 mV corresponding to a relative indeterminacy of
`<15 for a monovalent
`ion
`
`than the calculated
`
`Table 1 Regression of Elsa vs
`SD slope
`003
`004
`003
`003
`
`r
`
`0995
`0991
`0997
`0997
`
`SEE
`034 mV
`038 mV
`029 mV
`029 mV
`
`Slops
`106
`090
`112
`116
`
`Mean MI
`10 mV
`05 mV
`12 mV
`05 mV
`
`Ca2+
`Na
`
`Cl
`
`Discussion
`We have compared the ion activities over a semiper
`1911 1 A high degree of ion binding to albumin was
`meable membrane with those predicted by Donnan in
`
`observed especially of chloride and calcium so we used
`the net charge of albumin including bound ions for the
`calculations Theory and observation agreed well over a
`wide pH range which we take as a confirmation of the
`Donnan theory
`The discrepancy of 08 mV and slopes slightly differ
`ent from 1 may be due to lack of space from forbidden
`in the albumin solutions
`volumes or binding of water
`on pH were not taken
`These effects or their dependence
`into account because we had no independent way of
`determining them We were confined to using the con
`ventional molality and mass concentration of total wa
`ter Others have found
`g of water bound per gram
`of protein 4 which would equal 6 of total water
`in
`this study Because bound water by definition is not
`available for diffusible ions water binding might ex
`plain the Observed 510 negative binding of sodium at
`the lowest pH Binding of water would imply a higher
`
`CLINICAL CHEMISTRY Vol 39 No 11993
`
`51
`
`

`

`and the observed E
`
`ion binding With sodium being the most
`degree of
`free ion the calculated net charge of albumin
`abundant
`should then be more positive and the calculated Don
`nan potential should be lower improving the agreement
`between the calculated ED
`Equation 9 uses monovalent salt to calculate
`Some of the ions were actually divalent Ca2+ Taking
`this into account would further improve the agreement
`between EL9E and Erkmnan
`The observed binding of chloride and calcium to
`human albumin agrees with earlier studies The bind
`ing of chloride was studied by Scatchard et al 5 in
`1950 By dialysis conductometry and silversilver chlo
`ride electrodes and Scatchard plots they demonstrated
`two classes of binding sites at isoelectric pH one of 10
`binding sites with medium intrinsic affinity Ks = 44
`Lmol and the other of 30 binding sites with low
`intrinsic affinity KA = 11 Lmol A few binding exper
`iments at pH 32 revealed 31 chloride ions bound per
`albumin molecule in agreement with our study Chlo
`ride binding to human albumin was rediscovered by
`others 6 7 without
`reference to Scatchard Calcium
`binding has been studied by many authors We previ
`ously observed an increasing number of apparent bind
`ing sites with increasing pH which we interpreted as
`exposure of calcium binding carboxylate groups during
`
`the neutral unfolding of albumin 8 The emerging
`
`calcium binding at acid pH may similarly reflect an acid
`unfolding of albumin This study agrees with earlier
`calcium binding data We observed no albumin binding
`of sodium potassium or lithium but
`the results for
`because of interference
`lithium were less satisfactory
`
`from H+ in accord with Okorodudu et al 9 Binding of
`sodium was examined before 0 with the same result
`
`as ours
`The change in electrostatic free energy when charged
`particles come together can be used to convert binding
`data for small
`ions and proteins into intrinsic associa
`tion constants These are by definition independent of
`electric charge and they must be distinguished from the
`describing actual binding under
`apparent constants
`specific conditions The assumptions for intrinsic con
`stants are simple such as the whole net charge of
`albumin being evenly distributed on a sphere The total
`charges used for the conversion must be accurately
`known which can be difficult because several
`ions may
`be involved The net charges of albumin depicted in
`Figure 3 may help to provide more accurate
`intrinsic
`constants
`The bound charge contributes about onefourth to the
`
`net negative charge of albumin under physiological
`conditions helping to maintain the plasma colloid os
`motic pressure and plasma volume Furthermore the
`bound ions will have a stabilizing effect on plasma
`volume during disturbances of pH Instead of albumins
`own charge which will change according to the added
`acid or base the net charge of albumin will change only
`onethird because of chloride and calcium Plasma col
`loid osmotic pressure and plasma volume will decrease
`during an acidosis and increase during an alkalosis but
`binding and release of chloride and calcium will coun
`the changes
`teract
`Figge et al 11 recently studied the role of serum
`proteins in acidbase equilibria by using ultraffitration
`and pH They concluded
`the observed charge of
`that
`serum protein of 12 mmolL is solely attributable to
`serum albumin and less than hitherto assumed We did
`not include bicarbonate or magnesium but they proba
`bly bind to albumin in competition with chloride and
`calcium Our data indicate that at pH 74 and an
`albumin concentration of 06 mmolL the own charge of
`serum protein is 10 mmolL and the effective net
`charge including bound ions is 14 mmolL in close
`agreement with Figge et al
`
`References
`1 Donnan FG Theorie der Membrangleichgewichte und Mem
`branpotentiale bei Vorhandensein von nicht dialysierenden Elek
`trolyten Em Beitrag zur physikalischchemischen Physiologie Z
`Elektrochem 19111757281
`2 Tanford C Physical chemistry of macromolecules New York
`Sons 1960
`John Wiley
`3 Webster HL Colloid osmotic pressure theoretical aspects and
`background Clin Perinatol 1982950521
`4 Perutz MF Electrostatic effects in proteins Science 1978201
`118791
`5 Scatchard G Scheinberg 1H Armstrong SH Physical chemis
`try of protein solutions IV The combination of human serum
`albumin with chloride ion J Am Chem Soc 19507253540
`6 Panteghini M Bonora R Malchiodi A Calarco M Evaluation of
`
`the direct potentiometric method for serum chloride determina
`tion Comparison with the most commonly employed methodolo
`gies Clin Biochem 198619205
`7 Panteghini M Chloride binding to serum proteins Clin Chem
`1987336256
`8 FoghAndersen N The concentration of
`ionised calcium in
`plasma Measurement and some clinical applications Dan Med
`Bull 19883557582
`9 Okorodudu AO Burnett RW McComb R13 Bowers GN Eval
`uation of three first generation ion selective electrode analyzers
`for lithium systematic errors frequency of random interferences
`and recommendations
`based on comparison with flame atomic
`emission spectrometry Clin Chem 19903610410
`10 Kissel TR Sandifer JR Zumbulyadis N Sodium binding in
`human serum Clin Chem 19822844952
`11 Figge J Bossing TH Fend V The role of proteins in acid base
`equilibria J Lab Clin Med 199111745367
`
`52 CLINICAL CHEMISTRY Vol 39 No 1 1993
`
`