Anal. Chem. 1990, 62, 2107-2114
2107
Complete On-Line Determination of Biopolymer Molecular Weight via High-Performance Liquid Chromatography Coupled to Low-Angle Laser Light Scattering, Ultraviolet, and Differential Refractive Index Detection Hans H. Stuting' and Ira S. Krull* The Barnett Institute and Department of Chemistry, Northeastern University, 341 Mugar Building, 360 Huntington Avenue, Boston, Massachusetts 02115
An optkally modlfied high-performance Hquld chromatography refractlve Index detector was developed to allow complete on-line determlnatlons for blopolymer molecular weights. On-llne concentration, refractlve Index, specific refractive Index Increment (dn/dc,),, and Raylelgh factor were determlned under flow Injection analysls (FIA) and size exclusion chromatography (SEC) conditions using low-angle laser llght scatterlng, uttravklet, and modified refractlve Index detectkn. This Instrumental system is capable of determining absolute on-llne molecular weights. The error and time requirements Involved In conventha1m e t w s for proteins have been reduced. Sample quantities have been reduced from 150 to 200 mg, In conventlonal off-llne methods, to less than 2 mg for on-llne FIA and 0.5 mg for on-llne SEC, If mass absorp I tlvHles ( a ) are known. Otherwise, the detennlnatlon of a wW be the most sample-demandlng step, requiring about 3 mg of the pure protein. On-llne measurements of (dn/dc,), are In good agreement with traditional off-llne values established at Donnan equilibrium (usually within 5 % ). I n addition, thls technique provides true Injected mass as determlned by the UV detector, after chromatographic exposure where losses may occur, wh+chIs then used In the calculation of Mopolymer molecular weight.
a very significant advance over all approaches that have used off-line dnldc determinations for each biopolymer in each mobile phase. Takagi and co-workers (2,3)reported on a molecular weight estimation study of the a,&protomeric and oligomeric units of soluble (Na+,K+)-ATPaseusing SEC-LALLS. This study again used LALLS, UV, and DRI detection in series after the SEC separation of each protein subunit. Subunit composition of each component in the eluate was determined by SEC in the presence of sodium dodecyl sulfate (SDS). The results showed that the major protein components were the a,@protomer and its dimer; highly aggregated materials were present as minor components. It was possible to monitor the off-line enzymatic activity of the various subunits, although on-line detection would have been preferable for each resolved, eluting peak. These authors concluded that the combination of SEC with LALLS, UV, and DRI detection showed great promise for investigating the molecular composition of membrane proteins solubilized with a surfactant. Kameyama et al. (4)described a similar SEC-LALLS-DRI approach to estimate molecular weights of membrane proteins in the presence of SDS. LALLS and DRI data were collected, and the peak molecular weight could be calculated by using the following equation:
MP= k(output)LALLs(dn,/dcp)-2(oUtput)DRf1 INTRODUCTION Maezawa and Takagi ( I ) discussed monitoring a highperformance size exclusion chromatography (SEC) column with a low-angle laser light scattering (LALLS) photometer, a UV spectrophotometer, and a differential refractive index (DRI) detector to determine protein molecular weights (Mw) in 1983. These workers used a conventional on-line DRI detector to obtain n values and a UV detector for c values. Peak heights were used for the detector signal outputs and subsequent data manipulation where the use of standard proteins with known Mw and a gave an elegant universal calibration line. Experimental data were then compared to this universal calibration line. This approach, using three detectors in series, provided a series of measurements within several hours for each protein. The technique was said to be applicable to virtually any protein. Exceptions were proteins that absorbed in the visible region, which affected the on-line DRI and Rayleigh factor measurements. Although this DRI detector did not use the same wavelength as the LALLS detector, it apparently was able to provide useful DRI data that could be used to estimate molecular weights. Thus, no off-line dnldc measurements were required. This represented
* To whom correspondence should be addressed. 'Present address: Roche Biomedical Laboratories (RBL), Inc., 1 Roche Dr., P.O. Box 500, Raritan, NJ 08869. 0003-2700/90/0362-2107$02.50/0
(1)
where dn,/dc, is the specific refractive index increment (dn,) of the protein-SDS complex expressed in terms of the weight concentration of the protein moiety (cp),k is the instrumental system constant in reference to a standard protein, (output)LALu is the output of the LALLS photometer, and (0UtpUt)DRI is the output of the DRI instrument. These authors used a similar approach for determining membrane proteins in the presence of a nonionic surfactant (5)and amylase (6). Strictly speaking, eq 1 applies only when the refractometer is equipped with a light source that has the same wavelength as that of the LALLS detector (He-Ne laser, 633 nm). However, in these and other studies described above, the DRI detector actually had a tungsten lamp (white light extending the entire visible range, i.e. from 350 to 800 nm) as the light source. This approximation is not always valid or applicable, but suitable corrections can be made if known standards are injected with the proteins of interest. A more rigorous correction, or procedure, becomes necessary when samples and standards are of quite different species. This research concentrated on developing a method that did not need to refer to standards and that could deduce on-line (dn/dc2), from the same chromatographic experiment that would be used to determine biopolymer Mw. Previous uncertainties that we encountered, and limited quantities of expensive experimental biotechnology products, led us into this work. Our efforts concentrated on performing the com@ 1990 American Chemical Society
2108
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990
plete analysis on-line in a direct and quantitative manner. Direct measurements are preferable over that referred to a standard and with the introduction of a refractometer optimized for on-line d n measurements, this is now possible in a short period of time (a little more than the time required for the chromatography to take place). Future developments should concentrate on (1)automating the on-line (dn/dc2), calculations, such as incorporating them into the PCLALLS software package and (2) applying the experimental refractometer to gradient high-performance liquid chromatography (HPLC) applications, though this is not a task to be taken lightly. These future developments can be accompolished by computer programming or by engineering design.
THEORY The fundamental theory of LALLS pertaining to biopolymer systems has been previously described in detail (7, 8).
EXPERIMENTAL SECTION Equipment. The SEC-LALLS system was modular in design and consisted of an LDC consta-Metric I11 analytical metering pump (1/3 speed), a Rheodyne Model 7125 injection valve equipped with a 100-pL loop, a TSK SWXL-2000 or -3000 (for BSA only) size exclusion column, 30 cm X 8 mm (Supelco, Bellefonte, PA), a Chromatix KMX-6 LALLS set-up for flow analyses, an LDC spectrolllonitor-D variable UV/VIS detector monitoring 277 nm, and a modified refractoMonitor IV DRI detector, all linked to both a Soltec Model 1242 chart recorder (Soltec Corp., Sun Valley, CA) and an IBM PC-AT compatible computer (PC's Limited Model 286, Dell Computer Corp., Austin, TX) using an experimental version of the software package PCLALLS (LDC Analytical) capable of acquiring three channels of data simultaneously. This system was sometimes configured for FIA-LALLS analyses merely by decreasing the flow to a nominal 0.1 mL/min, increasing the injection loop volume to 1.00 mL, removing the chromatographic column, and having flow directly into the corresponding detectors. The off-line dn/dc determinations were performed in bulk solution, with a Chromatix Model KMX-16 laser (633 nm) differential refractometer, connected to a subambient temperature water bath (20 L, Fisher Scientific, Boston, MA). Mobile phase refractive index measurements were performed on an Abbe refractometer (Analytical Products Division/Milton Roy Corp., Rochester, NY) with the appropriate temperature control. In order to obtain the proper refractive index of the solvents at the wavelength in use throughout these experiments, the addition of a 632.8 k 0.2 nm narrow band-pass filter (Melles Griot, Rochester, NY) was installed between the traditional light source and the receiving optics. Conventional, off-line UV-vis spectral data was collected with a Spectronic 1201 UV/VIS spectrophotometer (Analytical Products Division). Chemicals. Protein standards, Le., ribonuclease A (RNase A), P-lactoglobulin A (8-lact A), and bovine serum albumin (BSA), were obtained from Sigma Chemical Co. (St. Louis, MO). Pituitary (Crescormon) and recombinant (Protropin) human growth hormones (hGH's) were obtained through Genentech (South San Francisco, CA). All samples were shipped and received in dry ice and were immediately transferred to the freezer (-5 "C), for storage. All solutions were prepared, without further sample purification, daily and kept in the refrigerator ( 5 "C) between analyses/injections. Samples were allowed to equilibrate to room temperature before chromatography taking place. Water was purified in-house with a Nanopure I1 system, and was 18 MQ-' quality (Millipore Corp., Bedford, MA). The following reagents and chemicals were used without further purification: imidazole and glycine, both 99% pure, Aldrich Chemical Co., Inc. (Milwaukee, WI); ammonium hydroxide, 30% Baker Analyzed, J. T. Baker Chemical Co. (Phillipsburg, NJ); HEPES and dialysis tubing (Protein Mw greater than 12000 retained), Sigma Chemical, Inc. (St. Louis, MO); EDTA, monobasic phosphate, dibasic phosphate, sodium chloride (all Certified ACS
reagent grade), sodium acetate (HPLC buffer salt grade), Fisher Scientific (Springfield, NJ); sodium azide, Eastman Kodak Co. (Rochester, NY); nonionic surfactant, octaethylene glycol mono-n-dodecyl ether, NIKKO Chemical Co., Ltd. (Toyko, Japan). Mobile Phases. The phosphate buffered saline (PBS) consisted of 25 mM of monobasic phosphate, 25 mM dibasic phosphate, and 150 mh4 sodium chloride, pH 7.2. The modified protein buffer (MPB) consisted of 18 mM N-(2-hydroxyethyl)piperazine-N'-2-ethanesulfonic acid (HEPES), 7 mM imidazole, 1 mM EDTA, 3 mM sodium azide, 200 mM sodium acetate, 0.5 mM nonionic surfactant, octaethylene glycol mono-n-dodecyl ether, pH = 7.0. All mobile phases were filtered under vacuum with a 0.22-pm hydrophilic Durapore membrane (Millipore, Bedford, MA) and further degassed with helium via standard laboratory vacuum filtration apparatus. Procedures. Size Exclusion Chromatography. Size exclusion chromatography utilized a bonded diol phase TSK-SWXL-2000 or -3000 (BSA only) SEC column at a constant flow rate of 0.775 mL/min, using PBS or MPB as the mobile phase. As is customary in SEC analyses, all protein solutions were prepared in the same buffer as used in the SEC experiment and analyzed the same day. The KMXB LALLS detector was set up (for both SEC and FIA) with the 5-mm flow-through cell, 6-7" annulus, 0.2-mm field stop, Go (incident intensity) was in the range of 200-600 mV with 2, 3,4 attenuation, D = -4.2 X lo4, GB(scattered intensity at initial conditions) = ranged from 150 to 350 mV, using a 0.3-s time constant. Axial dispersion, or band broadening, corrections were not performed. A minimum of five injections were performed for each SEC condition throughout these experiments and their mean values were reported. Off-Line Specific Refractive Index Increments. Off-line dn/dc values for each protein were determined with the Chromatix KMX-16, using concentrations in the range of 3-5 mg/mL. Protein dn/dc values were determined by plotting (ni - n)/ci versus ci and extrapolating to zero concentration, where ni is the refractive index at concentration ci, and n is the refractive index of the mobile phase. A minimum of five concentrations were performed per analysis. Off-Line UV Spectral Data. Off-line UV spectral data were taken with the Spectronic 1201 in the customary manner, using Beer's law, A = abc, where A is absorbance units, a is the mass absorptivity in units of (A mL)/(mg cm), b is the cell path length = 1cm, and c is the biopolymer concentration in units of mg/mL. Modification of the refractoMonitor IV. The Fresnel-type refractoMonitor IV (LDC Analytical) DRI detector was optically modified. Modification included replacing the heat blocking filter with an in-line filter (650 nm, Ah = 40 nm, Corion Corp., part number S40-650-A-Ml25,Holliston, MA), while increasing the light source voltage from 3.3 to 5.0 V (6.0 V maximum), to compensate for the decreased light (wavelength range) throughput. Detector Lag Determination. Detector lag determinations play a vital role in the final processing of SEC-LALLS derived Mw's, and especially Mn's, for narrow, monodispersed, biopolymers. This effort cannot be stressed enough, since all detectors are in series, and the data collected are off-set by a specified volume including connecting tubing and detector cell volumes. Therefore, it is mandatory to determine this lag-volumeexperimentally. The popular method developed by Lecacheux (9)was used throughout this work. Data Acquisition, Manipulation, and Processing. PCLALLS was used to acquire data in the normal mode of operation. The data were then processed (including respective detector lag volumes for the UV and the DRI), baselines were set for each detector signal, and each signal was saved. There were two files produced for every run in PCLALLS. The extension-less file contained the saved, processed, ASCII data. This file was then transferred into the Lotus 123 subdirectory, and.renamed in the following form, *.PRN. One may now use Lotus 123 and import this ASCII file in order to manipulate the data for the on-line (dn/dc2),,determination. In order for these data to be transformed into on-line (dn/dc2),,, the UV and DRI detector signals had to be quantitated. The UV signal transformation for determination of on-line concentration was based on using Beer's, A = abc. When
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990
2109
Table I. Determination of Mass Absorptivities" Off- and On-Lineb at Various Wavelengths in Phosphate Buffered Saline biopolymer
210 nm
214 nm
220 nm
off-line a 235 nm
258 nm
277 nm
280 nm
on-line a 277 nm
RNase A Protropin Crescormon fi-lact A BSA
16.2
12.3
8.63
2.47
0.301
0.562
0.484 0.901
15.2 16.7
11.8 13.4
8.76 9.94
2.35 2.40
0.489 0.425
0.604 0.82' 0.82' 0.670 0.677
0.851 0.677
0.754 0.539
NAd
" A = abc where A is the absorbance, a is the mass absorptivity in (A mL)/(mg cm), b is the path length in cm, and c is the concentration in mg/mL. Off-line, Spectronic 1201 spectrophotometer;and on-Line, FIA to spectroMonitor D detector. Value specified per Genentech (10). dNA, not aDdicable since Dure form not available.
graphically presented, y coordinate = absorbance units (A), and the n coordinate = concentration (c) in mg/mL; therefore a = slope divided by the cell path length (1cm) of this plot, or simply just the slope of the plot if using a 1cm cell path length in units of ( A mL)/(mg cm). The direct on-line calculation of change in refractive index (dn) and change in concentration (dcz)were performed by using the following equations in Lotus 123: Manipulation of retention volume data was performed straightforward by the following equation: where VRnis the retention volume for data point n, VR +1 is the retention volume for the following data point, and AQR is the change in retention volumes, in units of mL. The following equation describes the manipulation of the DRI response for on-line dn determination: dn = (DRI signalmv)x (DRI attenuationR~s)(calibration factor)/ (10mVFS detector output)(100 software multiplier) (3) where the DRI signal was the detector's mV output at some retention volume, DRI attenuation was the user selected attenuation of the detector, in units of refractive index units full scale (RIUFS), the calibration factor was the slope when the actual dn (from the detector signal) was plotted versus the theoretical dn; the 10 mV detector output was the physical signal output range from the detector; and the 100 software multiplier was the value in PCLALLS to normalize the 10 mV detector output to 1 V full scale. The resultant quantity, dn, is the RIU, which is unitless. Equation 3 yields individual dn (dni)across the peak of interest, but if these values were used to determine biopolymer (dn/dcz),, errors would result since all dn;s cannot be averaged across the peak to yield dn. One must account for this by weighing each dni with its corresponding response, in mV, and then use summation notation for overall dn, as in the following equation: (4)
The UV data treatment is similar to that of the DRI, but more detailed since it incorporates a, as is shown in the next equations, dcz = (UV signalmv)(UV attent~ation~~p~)(lO-~)/(lOmVFS detector output)(100 software multiplier)(a) = g/mL (5) where the UV signal was the detector's mV output at some retention volume, UV attenuation was the user selected attenuation of the detector, in units of absorbance units full scale (AUFS); the factor incorporates the conversion of AU to mAU, and pg/mL to g/&, the 10 mV detector output was the physical signal output range from the detector; and the 100 software multiplier was the value used in PCLALLS to normalize the 10 mV detector output to 1V full scale. The resultant quantity, dcz, is in units of g/mL. The same weighting scheme as for the DRI must be applied to the UV responses, and are accounted for in a similar manner by the following:
Therefore, (dn/dc2), can be easily determined by dividing eq 4 by eq 6 to yield the following: (7)
or simply by observing the Lotus 123 plot of VR versus on-line
(dn/dcz), and by viewing the values for ((dn/dcZ)Ji under the peak of interest. A question often arising in the HPLC of biopolymers is chromatographic recovery, since biopolymers tend to be more complex and may have other undesirable chromatographic characteristics as opposed to small biomolecules or synthetic organic soluble polymers. An issue that has been of major concern for biopolymer molecular weight determinations by LALLS is the sample concentration reaching the detector. Since on-line concentration (mass/volume) has already been determined (dc ), we can take this procedure one step further in determining totalT h e ) mass as determined at the UV detector, after chromatographic exposure. This can be easily calculated from the following equations using AV, (eq 1) and dcz (eq 4) to yield (dc,)(AC,)
lo6 = d(mass)i
(8)
where dc%= g/mL, AV, = mL, and d(mass)i needs to incorporate a factor of 10 to account for the conversion of g to pg per data slice. These d(mass)ivalues are then summed to yield determined injected mass in units of pg Cd(mass)i = determined injected mass (9) Sample recovery can be simply calculated by dividing determined injected mass (eq 9) by theoretical mass, as in the following: determined injected mass % recovery = (10) theoretical injected mass (100) The values for the on-line determined (dn/dc,), and injected mass were then transferred back into the PCLALLS software package and processed normally for the determination of biopolymer Mw.
RESULTS AND DISCUSSION Off-Line Mass Absorptivity Determination. Off-line mass absorptivities (a) for RNase A, P-lact A, and BSA were determined in PBS at 210, 214, 220, 235, 258, 277, and 280 nm using the pure biopolymer and without sample filtration. These values are listed in Table I. The exceptions to these were Protropin and Crescormon, which were not available at the time in pure form and were omitted. The values specified for these two proteins were from Genentech (10). On-Line Mass Absorptivity Determination: Calibration of the spectroMonitor D. Electrooptic calibration differences between different instruments, either off-line or on-line, need to be accounted for when quantitating data. One may chose to either (1) electronically calibrate all instruments using a standard, reference, biopolymer having a known absorptivity or (2) determine individual absorptivities using each instrument. Since we had adequate amounts of each biopolymer, we opted to use the latter method. Mass absorptivities were determined by preparing pure standard solutions, of known concentration in the buffer used later in chroma-
2110
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990
Table 11. Determination of Conventional and Dialyzed Biopolymer dn /dcz Values in Phosphate Buffered Saline via Chromatix KMX-16 conventional
dialyzed
biopolymer
(dn/dc2), mL/g
(dn/dcJ,, mL/g
RNase A Protropin Crescormon 0-lact A
0.175 0.161’ 0.160’ 0.166 0.167
0.174 0.166 0.168
BSA
(E-3 RIUl
d n at 2%
T h e o r e t i c a l do
P
i
-8-
/ A c t u a l dn
0 0 . 97 1 O.E
0.166 0.168
Concentration measured spectroscopically without filtration and adding support electrolyte in hGH formulations for conven: , tional workup. : tography, and introducing them into the system, without filtration, via FIA coupled directly to the UV detector. As expected, different coefficients were noted for the same biopolymer in the same buffer, due to differences in the instruments calibration. On-line mass absorptivities were determined for each, pure, biopolymer, with the exception of Crescormon (pure Protropin had been received from Genentech at this time), at 277 nm in PBS, as presented in Table I, and were 0.484 mL/mg for RNase A, 0.901 mL/mg for Protropin, 0.754 mL/mg for 8-lact A, and 0.539 mL/mg for BSA. These values would be used for the on-line calculation of concentration and for subsequent manipulation in (dn/dc2),. Future studies should incorporate the electronic calibration method for all instruments (UV spectrophotometers and detectors) described previously, in order to reduce sample quantity requirements. Conventional and Dialyzed Biopolymer dn /dc2’s. In order to evaluate the performance of the modified refractometer for on-line (dn/dcz),, one must first determine conventional and dialyzed biopolymer dn/dc2’s. Conventional dn/dcis were determined by solvating the protein sample (in the form supplied) and measuring its dn/dc2via the KMX-16. Since most of the proteins were “technically pure”, these analyses were straightforward,with the exception of Protropin and Crescormon. These two proteins were formulated with support buffer; hence, they were not in the pure form. Large amounts of buffer/stabilizer were added to both hGHs, which would interfere with the measurement of dn/dc2 if not accounted for. For this reason, the reference solutions (blank buffer) were matched with the components in the formulation, as close as possible, which were either glycine for Crescormon and mannitol, mono- and dibasic phosphate, for Protropin. Concentrations were determined spectrophotometrically (off-line) using the off-line a values at 277 nm previously described in Table I. Conventional dn/dcis are listed in Table 11. Typical sample quantity requirements for accurate and precise dn/dc2 values were in the range of 125-150 mg. Conventional measurements of dn/dc2 can, at times, be in error since the environment of the biopolymer may not be at Donnan equilibrium. For this reason we needed to measure (dn/dc2), via exhaustive dialysis, against the PBS buffer with determining concentration, again, spectrophotometrically (using off-line values for a). Sample requirements were increased (150-200 mg) over conventional dn/dc2 determination (125-150 mg) and resulted in similar dn/dcz values for RNase A, 0-lact A, and BSA (pure standards) and listed in Table 11. Increased dn/dc2 values were noted for Protropin and Crescormon, probably due to errors in matching the reference solutions in the conventional workup. Without question, off-line dialysis is often a required technique for reliable (dn/dcz), values; however, the laboratory times requires to obtain these values were in the 3-4 day range and considered by many to be impractical in today’s world.
O
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
C o n c e n t r a t i o n 19 NaC1/100g H201
Flgure 1. Calibration of modified DRI detector via FIA techniques.
We now had reference values to compare our experimental on-line method to. Calibration of the refractoMonitor IV. Even though the optics of the refractoMonitorIV were modified,the output still required calibration, so the corresponding mV output could be related to dn at 633 nm. Calibration was performed in a similar manner as described in the Chromatix operator’sguide ( I I ) , using the relationships developed by Kruis (12). The only difference is that the NaCl concentrationswere in the range of 0.005-0.600 g/100 g of H,O (covering a change in refractive index of 0.005 to 0.550 X when referenced to the water they were prepared by), and that the instrument constant (calibration factor) was determined by dividing the slope of the theoretical dn by the actual dn to yield a calibration factor of 1.88, as shown in Figure 1. When observing the actual dn data, one may see that the plotted data was not as linear (correlation coefficient = 0.9985) as was the theoretical dn data (correlation coefficient = 0.9999) and that deviations were present. This may be attributed either to the linearity of the electrooptics employed by the detector (13-16) or due to possible changes in the refractive index of the solutions that were handled and introduced into the FIA system. The former can be corrected for only by design, with limitation, while the latter by the analyst and controlling the environment in which the analysis is to take place (i.e. controlling the diffusion of dissolved gases, evaporation, and contamination to the FIA carrier solvent and the sample solutions). Both of these efforts are difficult to do, but might be able to produce as linear a relationship for actual dn as theoretical dn. However, further study shall prove that these problems are minor in nature. What was noticed was that the modified refractometers baseline had increased noise over that of the original optics. This can be attributed to the narrower bandwidth of light (and less intensity) that is allowed to reach the photodiode and the increasing coherence of the light causing interferenceproblems in the cell (17). Initial Quantitation Using the Modified refractoMonitor IV. Initial testing of the unit was performed by coupling FIA directly to the detector using water (Nanopure I1 grade) as the carrier. A master batch of 50 mM sodium acetate, pH 6.8, was prepared for sample dissolution in order to minimize minute batch-to-batch inconsistencies in refractive index coming from the buffer. Polyacrylic acids (PAA) were chosen since these polymers were the “purest” on hand and available in large quantity. Concentration was determined by direct weighing and dissolving in class A volumetric flasks. Conventional dn/dcz values were determined by the KMX-16, while the on-line data are shown in Figure 2. The off-line determination required 120 mg of each sample (5,8, and 11 mg/mL x 5 mL volumes) yielding only three data points from which to perform linear regression. On-line required 10 mg
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990 dn/dc (OR1 sensitivity 0.25 r
-
1OOE-6 RIUFS)
-
PAA #I
dn/dc
0 , 1 6 1 mL/g
I-
L
2111
I
-edn/dc
PAA 12 0.162 nL/g
-A-
n 0.17 0.16 0.15
0.13 0.14
L Q
,QQ
$Q
, $ Q
$3
+Q
bc.0 ~ Q QQ
Polymer Concentration lg/ml x 10-6)
- -- 5.11,300 100 g/mol PAA 12 g/mol Data b e l o w 150 ug/mL oolttad PAL. $1
from regression
7.
+DRI
Retention Volume (mL) +LALLS ,UV
os
111 I
dddc
W Chromatogram (280 nm, 0.2 AUFS = 1000mVFS) LALLS Chromatogram (loo0 rnVFS)
Figure 2. On-line specific refractive Index increment via FIA-LALLS coupled to the calibrated modified DRI detector.
Table 111. Comparison of FIA-LALLS Data Manipulation Methods Used in Determining Mw and A in Phosphate-Buffered Saline
RNase A j3-lact A BSA
0
ea
DRI Chromatogram (2.00 x 10-4 RlUFS = 1WOmVFS)
MU MU
biopolymer
*-
dnldc, (mL/g)/Mw/A2 (mL mol/g2 x lo4) conventional total on-lineb off-line" dnldc, using (dnldc,), 0.175111 4001-1.80 0.166/28500/-1.07 0.167193 7001-9.84
0.170/13800/4.21 0.175/36600/-5.95 0.165181 1001-8.85
Conventional off-line data manipulation = concentration determined by direct weighing and dissolution, conventional dnldc, by KMX-16, and on-line R, from the KMX-6. bTotal on-line data manipulation = concentration determined from on-line value of Q , dn from on-line modified DRI detector, (dn/dc2), from both the UV and DRI detectors, and on-line R, from the KMX-6. (0.05-0.5 mg/mL X 3 mL volumes), yet produced about 12 data points for regression. Omitting on-line points below 0.15 mg/mL, the y intercepts for each sample were very similar to the off-line determined values for PAA #1(0.167 vs 0.161 mL/g) and PAA #2 (0.164 vs 0.162 mL/g), despite the order of magnitude decrease in sample quantity required for the on-line determinations. These initial experimental findings were significant, which prompted us to continue with the experiments.
DATA MANIPULATION FOR MOLECULAR WEIGHT DETERMINATIONS Flow Injection Analysis. In FIA-LALLS, separation of individual components does not take place as in HPLC, and the bulk sample solution is passed through the various detectors for the determination of biopolymer Mw and AD This quantitative technique is useful only if a sample is pure and that concentration and (dn/dc2), have been previously determined, usually off-line. The FIA-LALLS technique is usually used to support SEC-LALLS data and to determine if SEC altered the sample in any way (18). It is a very simple technique and the first one we wanted to investigate by using the modified refractoMonitor IV for determination of on-line (dn/dcz),. We also investigated differences in the final Mw and Az due to the different ways of manipulating the data from using totally off-line to totally on-line methods. The FIA data for RNase A in PBS was compared to the data determined from conventional (off-line) methods and are tabulated in Table 111. Differences in the intercept and slope were evident. Off-line manipulations yielded a Mw of 11400 and Az = -1.80 X mL mol/g2 using the conventional dn/dc2 of 0.175 mL/g, while on-line yielded Mw of 13800 and Az = 4.21 X 1W mL mol/g2 using the on-line derived (dn/dc2),
( d n / d c 2 ) , = mUe
Flgure 3. SEC chromatograms and on-line (dn/dc,), for ribonuclease A in PBS, imported from PCLALLS into Lotus 123.
of 0.170 mL/g. Both Mw and A2 change according to how the data were handled. It was obvious that the totally on-line method Mw was closer to the literature value of 13 700 and that PBS is a favorable buffer/medium for RNase A due to the change in sign of A2 from negative to positive. With reference to P-lact A, treated similarly, off-line Mw = 28500 and A2 = -1.07 X lo4 mL mol/g2 using off-line dn/dcz = 0.166 mL/g, while on-line Mw = 36600 and A2 = -5.95 X lo4 mL mol/g2 using on-line (dn/dc2), = 0.175 mL/g. This showed that PBS is an unfavorable environment for P-lact A since its A2 increased (even if only slightly) in a more negative manner. Similar data were generated for BSA where off-line Mw was 93 700 and A2 was -9.84 X mL mol/g2, using dn/dc2 = 0.167 mL/g, while on-line Mw was 81 100 and A2 was -8.85 X lo4 mL mol/g2 using (dn/dc2), = 0.165 mL/g. A2 was still negative showing solution nonideality. Size Exclusion Chromatography. Since the FIA analyses were positive, we then applied this technique to SEC. As opposed to bulk solution determinations in FIA, SEC (theoretically) separates mixtures into their individual pure components (dependent on the efficiency of your column and other experimental variables). As these components are separated and elute from the chromatographic column, these bands are already at Donnan equilibrium and should be amenable to on-line (dn/dc2), determination, in addition to Mw. Ribonuclease A, Protropin, Crescormon, @-lactoglobulinA, and bovine serum albumin were investigated in phosphate buffered saline (PBS) using a high-resolution TSK-SWXL2000 or -3000 (BSA only) SEC column. In addition, Protropin and Crescormon were analyzed in the modified protein buffer (MPB) to possibly investigate mobile phase effects. Figure 3 shows the Lotus 123 plot of DRI, LALLS, UV, and on-line (dnldc,), (using the on-line value for a to determine dc2 and the calibrated refractometer for dn) versus retention volume for RNase A in PBS injecting 328 Kg of protein. One may see that the most fluctuations in the on-line (dn/dc2), values were at both extremes of the chromatographic peak (rise and fall) while the individual points attained a constant value just at, or just after, the peak apex. These fluctuations could be attributed to errors in detector lag determinations; however, this was not the case since we used the method by Lecacheux (9) which is experimentally correct. The physical reasoning behind this is that each detector responds to different physical criteria. The UV responds to concentration and a, the DRI to concentration and dn/dc2, and the LALLS to concentration and molecular weight (19). The different responses at the peak extremes from the different detectors
2112
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990
Table IV. Complete On-Line SEC-LALLS-UV-DRI Data biopolymer (buffer) RNase A (PBS)
Protropin (PBS) Protropin (MPB) Crescormon (PBS) Crescormon (MPB) 8-lact A (PBS)
BSA (PBS)
theoretical determined on-line' % injected injected mass," pg recoveryb (dn/dcz), mass, rg 164 216 328 492 2043 115 229 344 344
150 195 294 441 1630 96 192 286 287
91.5 90.3 89.6 89.6 79.8 83.5 83.8 83.1 83.4
0.172 0.171 0.170 0.170 0.175 0.175 0.175 0.175 0.165
M d 13900 13800 13800 13800 22600 22600 22800 22700
111 223 334 334
80 158 235 236
72.1 70.9 70.4 70.7
0.175 0.175 0.175 0.168
24400 24200 24200 24600
168 204 337 505 2017 78 156 203 234 2013
160 192 320 478 1853 70 137 178 206 1728
95.2 94.1 95.0 94.7 91.9 89.7 87.8 87.7 88.0 85.8
0.175 0.177 0.175 0.175 0.175 0.165 0.160 0.175 0.165 0.172
36500 36600 36500 36500
e,
I
.
a,
I S
9s
I O 0 5 ID
&OS
/ I
Retention Volume (mL)
,LALLS
,DRI
U V ,
dn/dc
UV Chromatogram (280 nm, 0.2 AUFS = 1000mVFS)
DRI Chromatogram (2.00x 10-4 RIUFS = 1000mVFS) LALLS Chromatogram (loo0 mVFS) ( d n l d c , ) , = mug
Flgure 4. SEC chromatograms and on-line (dnldc,), for Protropin in PBS, imported from PCLALLS into Lotus 123.
-
80700 81OOO 81200 81000
Determined from a using the on-line UV detector, and Lotus 123 program. * % Recovery = (determined mass/theoretical mass) X 100. Determined from calibrated DRI and UV detector outputs across the chromatographic peak. Determined via PCLALLS using on-line determined injection mass and (dn/dcz),.
may be considered large when compared to the differences at the peak apex (Le. signal due to background or baseline vs signal due to sample). Therefore, these points can either be neglected or weighted proportionally with respect to the entire response of the peak. Equations 2 through 7 were applied, and on-line (dn/dc2), = 0.170 mL/g, identical with that determined by FIA and very similar to those determined by off-line conventional (0.175 mL/g) and dialysis (0.174 mL/g) methods. An added feature of this system is that we can determine actual injected mass, as detected by the detectors, after chromatographic exposure. This was done by applying eqs 5, 8, and 9. The determined injected mass for RNase A was 294 pg. Percent recovery was determined by eq 10 and was 89.6% for RNase A for a theoretical injected mass of 328 pg. The loss of sample may have come from several, if not from all, sources, such as (1) starting material was not 100% pure, (2) human error in weighing and sample preparation, (3) incorrect injection volume, (4) membrane filtration/adsorption losses, (5) chromatographic adsorption, or (6) chromatographic degradation. A concentration study was also performed in order to determine if this system might be concentration dependent, since not all detectors have a wide, linear, dynamic range. The comprehensive results are listed in Table IV. For RNase A, as injected mass increases from 164 to 2043 pg, % recovery decreases from 91.5 to 79.8%. The on-line (dn/dc2), values do not change appreciably and were in the range of 0.170-0.175 mL/g. When the on-line (dn/dc2), and determined injected mass values were entered into the PCLALLS software package and processed for each condition, the resultant Mw was 13800 for almost every condition. The Mw for the highest concentration (2043 pg) was not determined since the LALLS detector was overloaded and the signals were beyond its designed linear capacity. These same procedures were applied to the other biopolymers. The on-line (dn/dc2), plot for Protropin in PBS yielded similar data (Figure 4). Fluctuations in (dn/dc2), at peak
L
0 28
,
-c
.b c s 6 s
I
7 s
e
ea
9
es
D IO
0s
10s
Retention Volume (mL)
,DRI
+LALLS
U V,
dn/dc
UV Chromatogram (280 nm, 0 2 AUFS = 1oOOmVFS)
DRI Chromatogram (2 00 x 1 0-4 RlUFS = 1000mVFS) LALLS Chromatogram (1000mVFS) (dn/dc,), =mug
Flgure 5. SEC chromatograms and on-line (dnldc,), for Crescormon in PBS, imported from PCLALLS into Lotus 123.
extremes were evident. In the injected mass range between 115 and 344 pg, on-line (dn/dc2), = 0.175 mL/g, and did not change. This value was higher than that determined by off-line conventional (0.161 mL/g) method but closer to the dialysis (0.166 mL/g) method. The percent recovery was between 83 and 84%, and when both the on-line (dn/dc2), and determined injected masses, for each condition, were entered into the PCLALLS software package and processed, the resultant Mw was approximately 22 700 (Table IV) which was very close to the literature value of 22125 (20). The Protropin dimer was evident, but at very low concentration with respect to the monomer (21). Percent monomer and dimer were calculated by using relative peak areas and were approximately 98.0% monomer and 2.0% dimer for the 333 pg injected mass. The dimer decreased in relative area to 1.8% for the 111 pg injected mass. Its (dn/dcz), value could not be determined due to lack of DRI signal. The limiting step in these on-line calculations was the lack of sensitivity of the modified DRI detector. Since the modification of the optics resulted in a narrower bandwidth reaching the receiving optics, sensitivity had to decrease, keeping all other components constant. Protropin in MPB yielded a lower (dn/dc2), = 0.165 mL/g over that in PBS (Table IV), but with similar % monomer, % dimer, % recovery, and Mw for the 344 pg injected mass. The decrease in (dn/dc,), is due to the physical characteristics of the MPB buffer and that it has a higher refractive index with regard to PBS, resulting in a lower difference in refractive
ANALYTICAL CHEMISTRY, VOL. 62, NO. 19, OCTOBER 1, 1990
index between the medium and the biopolymer. The physical characteristics of MPB could be that it may exhibit a small “compressing” effect on the biopolymer and this would not allow the molecule to expand to its larger hydrodynamic volume as that in PBS. These explanations could account for the small decrease in (dn/dcz), that was encountered here. The on-line (dn/dc2), plot for Crescormon in PBS is illustrated in Figure 5. In the injected mass range between 111 and 334 pg, on-line (dn/dcz), = 0.175 mL/g. This value was higher than that determined by the off-line conventional (0.160 mL/g) method but closer to the dialysis (0.168 mL/g) method. The reasons for this have to do with how the samples were handled and prepared, since concentration errors can arise quickly when working with moderately concentrated salt solutions (Le. evaporation, condensation, sample loss, etc.). The percent recovery was between 83 and 84%, and when both the on-line (dn/dcz), and determined injected masses, for each condition, were entered into the PCLALLS software package and processed, the resultant Mw was approximately 24 300 (Table IV), which was significantly higher than for Protropin in PBS. This value incorporated contributions from the Crescormon dimer (22) which was now more pronounced when compared to the Protropin dimer. The Crescormon’s were stored for a much longer period of time (several years) as opposed to the Protropin’s (3 months). Percent monomer and dimer were calculated by using relative peak areas and were approximately 90.5% monomer and 9.5% dimer for the 334 pg injected mass. Mathematically, this is correct, since if we sum the 90.5% contribution of monomer (Mw = 22 125 (20)), with 9.5% of dimer (Mw = 44 250), the total comes to 24 200, which is what we have determined. The percent dimer decreased in relative area to 8.3% for the 111 Fg injected mass. Crescormon in MPB yielded a slightly lower (dnldc,), = 0.168 mL/g over that in PBS (Table IV), but with similar percent monomer, percent dimer, percent recovery, and Mw for the 334 pg injected mass. The results are listed in Table IV. The characteristics that Protropin and Crescormon show here in SEC confirmed those previously determined (21). The on-line (dn/dcz), plot for 0-lactoglobulin A in PBS was produced in the injected mass range between 168 and 2017 pg, on-line (dn/dcz), = 0.175 mL/g. This value was higher than that determined by all off-line (0.166 mL/g) methods, while it was identical with the on-line FIA value. The percent recovery ranged from 95 to 92% depending on injected mass, and when both the on-line (dn/dcZ),,and determined injected masses, for each condition, were entered into the PCLALLS software package and processed, the resultant Mw was apparoximately 36 500 (Table IV) which was almost identical with the literature value of 36 800 Daltons (23) for the 0-lact A dimer. The on-line (dn/dc2), plot for bovine serum albumin in PBS was constructed and in the injected mass range between 78 and 2013 pg, on-line (dn/dcz), = 0.165 mL/g. This value was identical with that determined by on-line FIA and similar to that determined by all off-line (0.167-0.168 mL/g) methods. The percent recovery ranged from 89.7 to 85.8% depending on injected mass, and when both the on-line (dn/dc2), and determined injected masses, for each condition, were entered into the PCLALLS software package and processed, the resultant Mw was approximately 81000 (Table IV). Similar values for this particular lot of BSA have been previously determined (7, 8), which further support our experimental results. This particular lot of BSA was a mixture of species, and according to relative peak area determinations, consisted of approximately 83.5% monomer, 13.5% dimer, and 3% trimer. When each monomer, dimer, and trimer contribution was mathematically summed, the resulting value equaled
2113
79 000, very close to the 81 000 determined here. However, previous determinations resulted in about a 70% monomer, 30% dimer contribution. The difference can be attributed to the different batches of BSA used and to the different columns employed for the separation. What we previously used was a TSK-SW-3000 column having a mean particle size of 10-12 pm, which was now replaced by a TSK-SWXL-3000 column having a particle size of 5-7 pm, resulting in higher column efficiency and resolution. CONCLUSION In summary, this method is capable of determining direct on-line biopolymer concentration,refractive index, subsequent specific refractive index at Donnan equilibrium ((dn/dcz),), and injected mass, all after chromatographic exposure, at the detectors. The typical uncertainties in LALLS methodologies relating to chromatographic recovery and concentration issues, and using off-line determined values in an on-line system, can be removed by using this on-line method. Even though the actual measurement process and signal obtained for the determination of (dn/dcz), is not as sensitive, as with the off-line methods (more scattering of data at peak extremes), its advantages related to reduced analysis time and sample quantities can be clearly noted. ACKNOWLEDGMENT Gratitude is expressed to LDC Analytical/Thermo Instruments Systems, Riviera Beach, FL, for their support and technical collaboration in this work. Special appreciation goes to M. Munk, G. Cleaver, H. Kenny, J. Aveson, and C. A. Lukas, Jr. We thank J. Wronka of Bruker Instruments, Inc., Billerica, MA, for his generous assistance and ongoing support with all the computer data acquisition and graphics systems. We thank B. L. Karger of The Barnett Institute at Northeastern University, Boston, MA, for his interest, support, and technical suggestions. We acknowledge S.-L. Wu and W. S. Hancock of Genentech, Inc., South San Francisco, CA, for their technical collaboration, stimulating discussions, and generous donations of Crescormon, Protropin, and the SEC TSK-SWXL-2000 column. Registry No. Protropin, 82030-87-3;crescormon, 12629-01-5. LITERATURE CITED Maezawa, S.; Takagi, T. J. Chromatogr. 1983, 280, 124. Hayashi, Y.; Takagi, T.; Maezawa, S.; Matsui, H. Biochim. Biophys. Acta 1983, 748, 153. Takagi, T.; Maezawa, S.; Hayashi, Y. J. Biochem. 1987, 101, 805. Kameyama, K.; Nakae, T.; Takagi, T. Biochim. Biophys. Acta 1982, 706, 19. Maezawa, S.; Hayashi, Y.; Nakae. T.; Ishii, J.; Kameyama, K.; Takagi, T. Biochim. Biophys. Acta 1983, 747, 291. Takagi, T. J. Biochem. 1882, 89, 363. Stuting, H. H.; Krull. I.S.; Mhatre, R.; Krzysko, S. C.; Barth, H. 0. LC-GC 1989, 7 (5), 402. Mhatre, R.; Krull, I. S.; Stuting, H. H. J. Chromatogr. 1990, 502, 21. Lecacheux, D.; Lesec, J. J. Liq. Chromatogr. 1982, 5 , 217. Wu, S.-L, personal communication, 1989. Operator training manuals for the Chromatlx product line, available through LDC Analytical, Riviera Beach, FL. Kruis, A. 2.Phys. Chem., Abt. B 1936, 3 4 , 13. Frei, R. W.; Zech, K. Selective Sampie tiandling and Detection in HPLC-Part A ; Elsevier: Amsterdam, 1988. Scott, R. P. W. LC Detectors; Elsevier: Amsterdam, 1986. Vickrey, T. M., Ed. Liquid Chromatography Detectors; Marcel Dekker: New York, 1983. Yeung, E. S., Ed. Detectors for Liquid Chromatography; Wiley: New York, 1986. Munk, M., personal communication, 1989. Stuting, H. H.; Krull. I. S. Internationai GPC Symposium Proceedings 1987, 490-522. Available through Waters Chromatography Division/ Millipore Corporation, Milford, MA. Dubin, P. L., Ed. Aqueous Size Exclusion Chromatography; J. Chrom. Lib. Volume 40; Elsevier: Amsterdam, 1988. Bewley, T. A.; LI, C. H. I n Advances in Enzymology, Vol. 42; The Chemistry of Human Pituitary Growth Hormone; Meister, A,, Ed.; John Wiley and Sons: New York, 1975; p 73. Stuting, H. H.; Krull, I. S. J . Chromatogr., in press. Lewis, U. J.; Singh, R. N. P.; Peterson, S. M.; Vanderlaan. W. P. I n Growth Hormone and Related Peptides; Int. Congr. Ser. No. 381, Ex-
Anal. Chem. 1990, 62,2114-2122
2114
perta Med. Found., Humn Growth Hormone: A Family of Proteins: Pecile, A., Muller, E. E., E&.; Malan: Amsterdam, 1972; p 64. (23) Town&, R. J.; Weinbrger, L.F Tlmshff, S. N. J . Am. C M . sot. 1960, 82,3175.
RECEIVED for review March 7 , 1990. Accepted June 26, 1990.
This majority of this work was supported by LDC AnalytiInstruments Systems' funding was provided, in part, by Pfizer, Inc., and Merck Sharp & Dohme Research Laboratories, Inc. This is contribution number 427 from The Barnett Institute at Northeastern Univerisity.
Determination of Warfarin-Human Serum Albumin Protein Binding Parameters by an Improved Hummel-Dreyer High-Performance Liquid Chromatographic Method Using Internal Surface Reversed-Phase Columns Thomas C. Pinkerton* and Kenneth A. Koeplinger
Control Division, Building 259, Mail Stop 12, The Upjohn Company, Kalamazoo, Michigan 49001
The Hummel-Dreyer slzaexcluslon hlgh-performance liquid chromatographic method for the determination of proteln Mndlng parameters has been improved and autmated by use of an Internal surface reversed-phase (ISRP) column (5 cm X 4.6 mm) and a computercontrolkdmob#ophase dellvery system wRh low volume syrlnge mixlng. The high-etflclency ISRP columns, whlch are nonadsorptlve and excluslonary to serum protdns but aliow partWlng of small molecules with an Internal peptkle bonded phase, maintain htgh performance after many InJectkns d human sefum albumin (HSA), enable the use of short columns, and provlde for the resdutlon of primary llgand from protein binding dlsplacers. The modifled Hummel-Dreyer high-performance llquld chromatographlc method was demonstrated by the determlnatlon of blndlng parameters for warfarln-HSA In phosphate buffer, whkh were found to be n , = 1.0, n 2 = 2.1, K , = 3.30 X lo* W', and K , = 2.03 X lo4 M-l. The necessary sequence of chromatographic experiments was repeated 4 tlmes at 18 separate warfarin moble phase concentrations. Each automated sequence required 8 h to complete. The parameters were measured with a predsslon of less than 10% relative standard devlatlon.
INTRODUCTION Drugs can be highly bound to blood plasma proteins, and the extent of binding can have important consequences relative to the biological distribution and clearance of such drugs (1-4). This is of particular concern with toxic drugs that have narrow therapeutic indices and are bound to plasma proteins in excess of 90%. In such cases, the protein binding parameters must be determined with confidence. Further, susceptibility of the drug to displacement by other substances must be well established. When a compound binds to only one protein binding site, generally, a binding constant can be determined in a straightforward fashion with a variety of methods. However, when a small molecule binds to more than one class of binding sites on a protein, the drug bound becomes a complex function of the amount of drug present. The determination of the binding parameters can be complicated by the means of controlling the drug concentration, the number of data points collected over the drug concentration range, the procedures used for measuring the amount of drug bound, and the data analysis methodology.
Binding of warfarin to human serum albumin (HSA) is a classical example of a multiclass model which has been studied by a variety of methods, yet comparisons of reported results illustrate inconsistencies (5-9). Warfarin is an anticoagulant which decreases prothrombinogenic activity by inhibiting the vitamin K cycle (10). Warfarin can be administered in maximum doses of 15 mg per day, yielding total plasma concentrations of 3 to 8 pM in paitents 1-10 h after administration (11). Because of the toxicity of warfarin, the narrow therapeutic index, and the ease with which it can be displaced from protein by other substances, it is important to know the percent of warfarin bound to HSA under physiological conditions. Because of variations in HSA concentrations among patients and disease states, it is sometimes desirable to theoretically predict changes in unbound (free) warfarin concentration. For this to be done, warfarin-HSA binding constants must be accurately known. Human serum albumin (HSA) has two classes of binding sites for acidic drugs, such as warfarin (I, 2). It is assumed that warfarin binds to HSA in two separate binding regions in which each ligand associates independently to individual sites of a given class. This means there can be a different total number of individual binding sites within each class. The independent model assumes that binding at one site does not affect binding at any other site. The model can be represented by the following expression: n,KIPDf n2K2PDf Db = (1) 1 + K1Df 1 KzDf where Db is the concentration of drug bound to protein, Df is the free drug concentration, P is the total protein concentration, n j is the number of drug molecules bound in each binding site class, and Kiis the equilibrium association constant (i.e., affinity constant) for each class of sites. Derivation of fundamental binding relationships can be found elsewhere
+ +
(12).
The primary objective in studying any two class drugprotein binding is to determine the binding parameters n,, n2,K1,and K2 with acceptable accuracy and precision, under some set of fixed conditions (i.e., temperature, pH, ionic strength, and protein concentration). Ideally, one wishes to measure the bound drug concentration (Db) directly as a function of the free drug concentration (Of),which should be controlled as the independent variable. Since the relationship between Db and Df is nonlinear, a sufficient number of appropriately spaced data points must be taken over a drug
0003-2700/90/0362-2114$02.50/00 1990 American Chemical Society
