Abstract
Background
Almost 80 years after implementation of the industrial ethanol fractionation process, based on the pioneering work of Edwin J. Cohn's research group, it has not lost its importance in providing life-saving biotherapies to patients. The focus has shifted from albumin, which was first used to treat intensive care patients, to immunoglobulin G preparations. Nowadays the latter enables effective, life-long treatment of patients suffering from immunodeficiencies. Ethanol concentration, pH, temperature, protein, and salt concentration are diligently varied during the Cohn fractionation process to finally yield fractions in which the main plasma proteins are enriched at adequate purity. Total protein concentration of the intermediates, probably not as important as ethanol levels and pH, has nevertheless an indisputable influence on the process’ performance
Objectives
Total protein measurement is therefore a valuable in process-control for versatile process performance monitoring. For this purpose, we investigated the application of analytical digital refractometry.
Design
For our pilot study, we used six consecutive batches of nine relevant intermediates covering Takeda's whole industrial-scale Cohn Fractionation process.
Methods
Total protein concentrations of the intermediates were measured with the method of Kjehldal to obtain reference data and digital analytical refractometry, using 10-kDa ultrafiltration to obtain a protein free permeate serving as a sample-specific blank.
Results and Conclusions
Specific refractive index increment dn/dc values, allowing the calculation of the total protein concentrations by refractive index measurement, were obtained for the Cohn Fractionation intermediates investigated. Thus, analytical digital refractometry, a simple, fast, and robust technique, was shown to be fit for this purpose.
Keywords
Introduction
Analytical refractometry, introduced for the determination of serum protein in 1903 by Reiss, 1 appears to have been virtually abandoned with the emergence of the commercial processing of blood plasma into biological therapeutics. Only recently has its somehow overlooked potential for measuring high protein concentrations without pre-dilution 2 been investigated for purified albumin and IgG concentrates 3 and for plasma. 4 Surprisingly, its direct application in the plasma fractionation process has not yet been investigated, except in a study with questionable results. 5 To close this gap and introduce an accurate and fast measurement method, we calibrated Cohn fractionation intermediates based on their nitrogen content using centrifugally driven ultrafiltration to isolate a protein-free ultrafiltrate for differential refractometry. Thus, we were able to determine the specific refractive index increment dn/dc for the respective intermediate's protein composition. 3
Cohn Fractionation of Blood Plasma: an Ingenious Balance of Alcohol, pH, and Salinity
Originally devised in the early 1940s, the cold ethanol-induced precipitation of plasma proteins still constitutes the backbone of industrial plasma processing. 6 While Edwin J. Cohn and his team at Harvard University concentrated initially on the isolation of therapeutic albumin, the group of Harold F. Deutsch at the University of Wisconsin focused on the sub-fractionation of IgG from the II + III precipitate. 7 The final modification introduced by Nitschmann-Kistler in 1962 at the Swiss Red Cross Blood Transfusion Service's Central Laboratory lowered the alcohol concentration for the II + III precipitation from 25% to 19% alcohol. 8
In general, cold alcohol fractionation makes use of the minimum solubility of proteins at their isoelectric point as enhanced by lowering the dielectricity of the solvent environment and the temperature. The latter needs to be kept at sub-zero values to prevent any protein denaturation, resulting, in the worst case, in the detrimental occurrence of neoantigens, as described for plasma protein solutions as well as albumin prepared by using higher temperatures during the purification process. 9 To efficiently adjust and control the temperature, the emerging mixing heat must be effectively dissipated through a slow addition of alcohol.
About two decades after the Cohn scheme became established, cryoprecipitation was introduced to obtain a factor VIII-von Willebrand factor-enriched precipitate from pooled plasma, 10 which opened new treatment opportunities for patients with hemophilia A. Finally, anion-exchange adsorption of the cryo-poor plasma (=cryosupernatant) was added to capture the prothrombin complex coagulation factors, 11 but in general the fractionation scheme has remained the same, with some manufacturer-specific fine-tuning. Some precipitates such as Cohn I (containing mainly fibrinogen) may have lost or other fractions such as Cohn IV-4 (containing mainly α-globulins) may have lost, or even never gained, any economic importance but the growing demand for albumin and intravenous IgG is driving continuous upgrades to maximize the potential of plasma fractionation.
Analytical Characterization of Cohn Fractionation Intermediates – for in-Process Controls?
Today, immunochemical techniques such as the ELISA or 2-D electrophoresis with immunoblotting as well as mass spectrometry-based proteomics enable a virtually full characterization of the protein composition in any intermediate. As mentioned above, the unspecific in-process determination of protein and alcohol 12 appears to have drawn only moderate attention. Total protein measurement techniques such as Kjeldahl digestion and Dumas combustion 13 with the appropriate conversion factor (see Table 1)6,14–17 or even the simple biuret assay 18 will not deliver instantaneous results and require handling hazardous or environmentally critical reagents, such as sulfuric acid, copper sulfate, oxygen, and helium. UV absorbance measurement is fast, but delicate to calibrate for mixtures, and may be biased by turbidity or, worse, alcohol denaturants such as acetone or butanone. 19
Nitrogen Content of Selected Plasma Proteins and Cohn Fractions.
Remarks: N/A stands for not available, prec. for precipitate. Kjehldal factors given in brackets were calculated from published data. 5
Refractometry would most likely overcome all these drawbacks, but needs a re-calibration, as the gravimetry-based specific refractive index increment (dn/dc) data given for dried fractions 20 may not necessarily reflect the actual dn/dc values in solution. However, as seen from Table 2, the published range of dn/dc values is quite narrow.
Characteristics of Cohn Fractions Including Published dn/dc Values.
Remarks: N/A stands for not available. Kjehldal factors given in brackets were calculated from published data. 1)Although no specific data have been disclosed for cryo-poor plasma, the supernatant obtained after temperature-driven precipitation of plasma, it seems reasonable to assume characteristics similar to serum.
Materials and Methods
Six consecutive batches from commercial production of each Cohn fractionation intermediate investigated (Table 3) were analyzed. The fraction termed “Cryo + DEAE-Sup”, representing cryo-poor plasma from which the prothrombin complex factors have been adsorbed on DEAE Sephadex, constitutes the starting material for the Cohn fractionation process. The protein content was determined by Kjeldahl digestion with CuSO4 as the catalyst, assuming a nitrogen content of 16% (Kjeldahl factor = 6.25). For refractometry, a permeate providing the background value of non-protein components with influence on the refractive index, was obtained by centrifugation through 10-kDa ultrafiltration tubes (Millipore Amicon Ultra 4) at 4500 rpm for about 15 min. The refractive index at 20 °C was measured on a Rudolph Research J357 refractometer for the samples and their respective ultrafiltration permeates. To assess the applicability of the approach, the UV absorbance at 280 nm, corrected for the turbidity at 320 nm, was measured for the samples and their respective permeates after 20- or 50-fold dilution with 0.9% NaCl in 1-cm quartz cuvettes. The specific refractive index increment dn/dc was calculated according to the formula dn/dc [in mL/mg] = (nD20sample - nD20permeate)/total protein [in mg/mL, determined with Kjeldahl].
Refractometry Data for the Cohn Fractions.
Remark: The minimum and maximum concentrations obtained for the six consecutive fractions are shown, while the dn/dc values were calculated for the respective means.
Results
As expected, alcohol and its butanone denaturant precluded any valid UV absorbance calibration and measurement of total protein in Cohn fractions, as the samples readily turn turbid from protein denaturation even through short storage periods. Data on file demonstrate that the elevated turbidity could not be accurately compensated by the 320-nm measurement. Furthermore, the butanone content, contributing to the adsorption at 280 nm, may vary within the specified range, thus introducing a further noteworthy bias to this measurement.
Refractometry by contrast appeared as particularly robust regarding sample storage and alcohol/denaturant concentration. Assuming a serum-like protein composition for the intermediates “Cryo + DEAE sup” and Cohn I supernatant and consequently applying a Kjeldahl factor of 6.54 15 (see Table 3) resulted in a much closer agreement with a protein content calculated ab initio from the literature serum dn/dc value of 0.000194 mL/mg. Surprisingly, the Cohn II + III supernatant (Kjeldahl factor = 6.25) also fell within this range. The later fractions appeared to be clearly dominated by albumin.
The refractive index of the 10-kDa permeate is clearly shifted upwards by increased alcohol levels as it increases from 1.33982 at 8% to 1.35799 at 40% alcohol. Its use for a quick assessment of the alcohol concentration may however require either a calibration by a dry substance subtraction, as known for wine or extract-containing spirits, 20 or comparison with the quantitative determination of alcohol, e.g., through steam distillation, as shown by Anderle et al. 12
Discussion
Cohn Fractionation intermediates are characterized by an interesting, challenging, and unique sample matrix. This specific character is, on the one hand, caused by presence of ethanol in concentrations of up to 40% (v/v) with different types of denaturing agents added to ethanol, but also by their highly complex protein composition characteristic for human plasma. Especially the latter feature complicates the total protein measurement. Thus, total protein measurement methods including the measurement of the optical density at 280 nm or the dye-binding Bradford assay 21 demonstrate a well-known high dependency on the protein composition of the sample measured. This simply reflects differences in the amino acid composition of the proteins collectively measured as total protein. When considering the UV extinction coefficients at 280 nm for albumin (5.2) and immunoglobulin G (13.2), which differ by a factor of more than 2, this easily becomes evident. Similarly, the dye-binding capacity of albumin and human immunoglobulin is profoundly different, which is reflected by clear differences in the assay's sensitivity for the purified proteins albumin and immunoglobulin G. Consequently, using either of both methods mentioned above for eg the measurement of samples of Cohn I Supernatant and Cohn II + III Supernatant with identical total protein concentrations will provide different total protein concentrations because of the differences in protein composition, ie, the absence of immunoglobulin G in the Cohn II + III supernatant. Obviously, the data demonstrate that analytical digital refractometry is not affected to that extent by differences in protein composition as shown by the specific refractive index increment dn/dc, determined for both intermediates: 0.000193 and 0.000195 were found for the intermediates Cohn I Supernatant and Cohn II + III Supernatant, respectively. The preparation of a sample-specific blank by ultrafiltration was originally described by Neuhausen und Rioch as early as 1923. 22 For the same purpose, we here used the fast and handy method of centrifugation to obtain 10-kDa permeates, which allowed us to subtract a sample-specific blank for each sample measured. This obviously removed the influence of ethanol on the refractive index as seen for the data obtained for the neat samples and contributed to accurate and precise measurements. Timing of the results obtained is another point to be considered when it comes to in-process control analytics. Total protein measurement techniques such as Kjeldahl digestion and Dumas combustion 13 or even the simple biuret assay 18 will not deliver instantaneous results and require handling hazardous or environmentally critical reagents, such as sulfuric acid, copper sulfate, oxygen, and helium. Refractometry provides a sustainable alternative delivering total protein concentrations results close to real time.
The relatively small number of (consecutive) batches we analyzed in our pilot study and the lack of sample-specific analytical validation could be regarded as a limitation although this might be compensated by the number of purified albumin and immunoglobulin batches, we have already measured for the analytical method validation the past. 3
In summary, analytical digital refractometry was shown to work perfectly well for the in-process protein determination in plasma fractionation intermediates. Analytical method validation results have been obtained for purified human immunoglobulin G and albumin preparations in the past and confirmed the adequate performance of this approach for total protein measurement: the data demonstrated the method to be precise, accurate, robust, and linear. 3 Albumin and immunoglobulin G together constitute the main part of the total protein of the Cohn Fractionation intermediates. Furthermore, the analytical digital refractive index measurement of the validation samples was carried out in the same manner as done for the Cohn Fractionation intermediates including the measurement of the protein-free permeate as sample-specific blank. Therefore, it seems reasonable to conclude that the data obtained for the purified proteins will apply also for the intermediates. The technique is fast, simple, and robust. The measurement is easy, accurate and precise, and does not require high investment. Furthermore, the entire method can be assessed as environmentally beneficial, as it requires less power consumption and avoids using hazardous reagents, such as the sulfuric acid and NaOH in the Kjeldahl digestion, and oxygen and helium in the Dumas combustion procedure. Digital refractometers as well as small, compact centrifuges can thus be installed on production floors without risk of any hazards from reagents or consumables. The instruments may either be standalone, paper-based devices, or integrated into a LIMS environment. The permeate refractive index has potential to provide a quick estimate of the alcohol content, if appropriately calibrated by reference methods to correct for the contribution of the dissolved non-protein solids.
Footnotes
Acknowledgements
The authors thank the personal of the Ethanol Fractionation Department of Takeda Manufacturing Austria from Takeda's Vienna plant for their support in providing the samples from the Cohn fractionation and Elise Landon-Neuner for editorial support funded by Baxalta Innovations GmbH, part of Takeda. The work was funded by Baxalta Innovations GmbH, part of Takeda.
Author contribution(s)
DECLARATION OF CONFLICTING INTERESTS
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: At the time of the study, Sandra Pichler, Lucia Gnauer, Heinz Anderle, and Alfred Weber are full-time employees of Baxalta Innovations GmbH (Vienna, Austria), part of Takeda, while Karin Herbinger is a full-time employee of Takeda Manufacturing Austria. SP, LG, HA, and AW are eligible for Takeda stocks/stock options.
FUNDING
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Work was funded by Baxalta Innovations GmbH, part of Takeda.
