Vox Sang 1989;57:233-239

A Semi-Continuous Method for Purification of Factor IX Complex from Human Plasma

John P. Tharakan, Dan M . Gee, David B. Clark American Red Cross, Jerome H. Holland Laboratory for the Biomedical Sciences, Rockville, Md., USA

Abstract. We report the development of a semi-continuous method for preparation of factor IX complex from human plasma using ion exchange resins. Traditionally, stirred batch adsorption has been used due to the high pressure drops and low flow rates associated with soft gels in packed columns. Batch methods, however, typically involve higher labor costs and are more cumbersome in process environments. In the semi-continuous process, cryo-supernatant plasma is pumped through a ‘stirred column’ containing the resin. At both lab and pilot scale, higher recoveries of factor IX (FIX) were obtained at decreased total process times, compared to batch adsorption. A residence time of 15 min was found to be sufficient for capture of 95% of the FIX in the starting plasma. In the pilot plant, 550liters of plasma was passed through a 50-liter column containing 8.5 liters of resin, yielding a 68% recovery of FIX. The results suggest that the recovery of FIX depends on the mode of contact (batch, continuous or packed column) between plasma and resin.

Introduction

Human plasma is fractionated into products for ther- apeutic use by many different methodologies “1. One such fractionation scheme is shown in figurel. Here, pooled fresh-frozen plasma is first cryoprecipitated for the production of factor VIII. The cryoprecipitate-poor plasma is fractionated by ion exchange to yield factor IX complex, a mixture containing the vitamin-K-dependent coagulation factors which can then be fractionated by ion exchange or affinity chromatography into factors 11 (FII), FIX, factor X (FX) and protein C (PC). The unadsorbed cryoprecipitate- poor plasma from the initial ion exchange step can be fur- ther processed by affinity chromatography on Heparin- Sepharose to yield antithrombin 111. Finally, ethanol frac- tionation leads to the last fractionation products in the sequence, immune serum globulin and albumin.

Various processes have been developed for the isolation of factor IX complex, also known as prothrombin complex, for use in replacement therapy [5-71. In this work, we report a process for the initial ion exchange step that leads to the isolation of a FIX complex. We have developed a

semi-continuous process to facilitate contacting plasma with soft resins. This method uses a stirred column, thus eliminating the need for any handling of a resin-plasma slurry and concomitantly reducing the total process time and labor requirements. In addition, the resin is contained within one vessel throughout its use, thus minimizing its handling which may be beneficial in a process plant envi- ronment.

Materials and Methods

Plasma For the laboratory experiments, individual units of citrated cryo-

precipitate-poor plasma were obtained from the American Red Cross. Washington (D.C.) Region Blood Center and pooled to produce a uniform starting material for all the experiments. Pilot plant experi- ments were performed with a 600-liter portion of a 3,200-liter pool of citrated cryoprecipitate-poor plasma (American Red Cross).

Resins DEAE Sephadex A-50 (Pharmacia) was used in a ratio of 1.7g dry

resin per liter of plasma. The resin was swollen in 0.03M sodium chloride and then washed with distilled water. The swollen resin was

234 Thara kan/Gee/Clark

Fig. 1. Schematic for fractionation of human plasma.

Table 1. Comparison of batch and semi-continuous process

Process Batch Semi-continuous

Adsorption

Unadsorbed plasma removal

Washing

Elution

Process control

mix plasma with resin in large tank

transfer plasmairesin slurry to special filter tank to remove plasma transfer resin to column and wash with buffer

elute adsorbed proteins

difficult

mix plasma with resin in column with continuous plasma flow contained in ad- sorption step

after passage of plasma drain bed and wash with buffer elute adsorbed proteins straightforward

further washed with a high salt buffer (0.37 M sodium citrate, pH6.0) and then equilibrated in 0.07 M sodium citrate, pH 6.0 buffer.

DEAE-Sepahrose fast flow (Pharmacia) was obtained pre-swollen. For use, the resin was washed and equilibrated with 0.02M sodium citrate, pH 6.0 buffer.

Coagulation Assays FIX potency was determined by a one-stage clotting assay [8] using

FIX-deficient plasma (George King Biomedical) and APTT reagent (General Diagnostics) in a Coagamate X-2 (General Diagnostics) clot- ting instrument. Samples were diluted in 0.05 M Imidazole, 0.1 M NaCI, pH7.4 containing 0.1% (w/v) bovine serum albumin (Sigma, RIA grade) and 0.01% (v/v) Tween20 (Sigma) as described by Miekka [ 9 ] . The assay was standardized with a plasma pool of at least 10 U of fresh-frozen plasma, corresponding to a potency of 1 U/ml. FII and FX

were also assayed using a one-stage coagulation assay. The prothrom- bin time was measured using factors I1 and VII or factor X and VII- deficient plasma (Sigma) and Russells Viper Venom (RVV, Sigma).

Protein Total protein was determined from the absorbance of the sample at

280 nm measured on a Gilford Spectrophotometer. The extinction coef- ficient was assumed to be 1.4. Samples with absorbances greater than 1.0 on direct reading were diluted to bring the absorbance into the range 0.0 <A280 d . 0 .

Lab Scale Experimental Protocol and Configuration These experiments with the two resins were performed following

the general protocol of Menache et al. [7] for DEAE Sephadex. A comparison of batch and continuous processes is provided in table I while detailed descriptions of the experiments follow.

Batch Adsorption. Lab scale batch experiments were conducted with both resins, DEAE-Sephadex and DEAE-Sepharose Fast Flow (Pharmacia), to establish the times required for complete FIX ad- sorption. The DEAE Sephadex was swollen and equilibrated as de- scribed. Cryo-poor plasma was mixed with the resin in a plastic beaker at 5°C. At various time points, supernatant samples were taken through a 3-ml disposabale syringe (Becton-Dickinson) fitted with a 0.4pm filter (Costar). The syringe withdrew a slurry of plasma and resin, thus maintaining the plasma-resin density while the filter re- moved the resin, leaving a clarified sample of supernatant that was assayed for FIX. The resin was then washed with 0.07 Msodium citrate, pH 6.0, and subsequently the FIX complex was eluted with 0.2 M sodi- um citrate, pH6.0.

The experiment with DEAE Sepahrose Fast Flow was performed identically, but with different buffers. This resin is obtained pre-swollen from the manufacturer and was equilibrated as described. After plasma adsorption, the resin is first washed with 0.02M sodium citrate, pH 6.0. and subsequently the FIX complex is eluted with 0.02 M sodium citrate, 0.4 M sodium chloride, pH6.0.

Semi-Continuous Factor IX Complex Preparation 235

Fig. 2. Lab-scale experimental set-up for semi-continuous processing of plasma. Q, and Q, are the feed and effluent flow rate, respectively. The residence time in the stirred column is adjusted by controlling these flow rates.

Stirred Column Adsorption. DEAE Sepharose Fast Flow and DEAE Sephadex were used in stirred column experiments. The experi- mental apparatus is shown in figure 2. The experiments were all carried out at 5°C. The stirred column was a 44-mm ID acrylic tube with an Amicon flow adapter. It was stirred with a 25-mm-diameter, three- lobed propeller (Mixing Equipment Co.) driven by a standard lab- oratory stirring motor. Because of the large diameter of the impeller compared to the column diameter, the system was judged to be well mixed without the use of baffles. Cryosupernatant plasma was pumped into the top of the stirred column with a peristaltic pump through a filter cascade consisting of a bulk filter cartridge (AMF Cuno) followed by a 5.0-pm (Millipore) and then a 0.2-pm cartridge (Millipore). Plasma was pumped out of the bottom of the column, through the flow adapter screen at the same flow rate so that the volume in the column remained constant.

In operation, the equilibrated resin was added to the column and plasma was pumped in until the total combined volume was 80ml. At that point flow out through the bottom of the column was started. The flow rates were set to give the desired residence time in the column. Fractions were collected from the column effluent and assayed for FIX activity. Results were calculated as the amount of FIX remaining in the effluent plasma as a percentage of the initial feed plasma FIX concen- tration.

After all of the plasma had contacted the resin, the column was packed and an upper flow adapter installed, and then washed with 0.07 M sodium citrate, pH6.0 buffer until the absorbance at 280nm was less than 0.2 absorbance units. At that point the FIX complex was eluted with 0.20 M sodium citrate, pH6.0 buffer. The eluate was pooled and assayed for the various coagulation factors.

The DEAE-Sepharose Fast Flow resin was equilibrated with 0.02 M sodium citrate, pH 6.0. The experiment was run similarily, except that the equilibration, washing and elution buffers used were the same as those for the batch DEAE Sepharose experiment.

Packed Column Adsorption. Plasma adsorption on DEAE Sepha- rose Fast Flow was also studied in a packed column configuration. A

Pharmacia silanized glass column, 0.9 X 15 cm with top and bottom adaptors, was used. The resin obtained pre-swollen from the manu- facturer was packed into a column and equilibrated with 0.02 M sodium citrate, pH6.0. Plasma was passed into the packed column through a 0.2-pm (Millipore) filter. Following adsorption of the FIX complex, the packed column was washed with 0.02 M sodium citrate, pH 6.0 until the absorbance had reached the baseline. The column was eluted with 0.02 M sodium citrate, 0.3 M NaCI, pH 6.0 buffer. The bound proteins were eluted in a single peak and samples of all pools were retained for coagulation factor assays.

Pilot Scale Experimental Protocol and Configuration Batch Process. In the pilot scale batch process, 600liters of plasma

was mixed in a large tank with 935 g (dry weight) of DEAE Sephadex that had been swollen and equilibrated as described previously. The slurry was stirred for 1 h with a three-bladed propeller (Lightnin A310). The supernatant was then decanted and the resin transferred to a large polyethylene Buchner funnel where it was washed with 0.07 M sodium citrate, pH6.0. Subsequently, the resin slurry was transferred to a column. The resin was packed in the column, washed with additional 0.07 M sodium citrate, pH6.0, and the FIX complex was then eluted with 0.2 M sodium citrate, pH 6.0.

Stirred Column. A comparison of the lab and pilot scale column is provided in table 2. The laboratory stirred column process was scaled up in the pilot plant using a 35-cm-ID X 50-cm-high acrylic column (Amicon). This was configured as a stirred column as depicted in figure 3. To the main column section was added a 100-cm-high acrylic extension that was fitted with baffles to enhance mixing. The total stirred column height was 150cm.

DEAE Sephadex resin (850g) was swollen and equilibrated as described previously, and the resin was then poured into the column. Citrated cryo-poor plasma was pumped through a 0.8-pm cartridge filter (Sartorious) into the column until a total volume of 50liters was reached. The height of the slurry in the stirred column was 51 cm. At this point, inflow of plasma was stopped and the plasma resin slurry was

236 Tharakan/Gee/Clark

Table 2. Characteristics of lab and pilot scale semi-countinous columns

Scale Plasma volume, I Reactor volume, 1 Flow rate, I/min Residence time, min Effluent collection, I fractions

Lab 0.5 Pilot 550

0.08 50

0.03 2.0

25 25

0.015 16.0

Fig. 4. Batch adsorption of FIX. The normalized FIX activity in the supernatant is plotted against time. The experiments were done for two resins. D E A E Sephadex (.)and D E A E Sepharose Fast Flow (H).

Fig. 3. The pilot scale stirred column

stirred for 25 min. The bottom valve was then opened and plasma flow through the column was started at L.Ol/min giving a mean residence time of 25 min. Fractions were collected in 16-litre volumes and samples of each fraction were assayed for factors 11, IX. and X. After all the plasma had passed through the column, the resin was washed by sus- pending it in wash buffer (0.07 M sodium citrate. pH6.0), stirring for hOmin and then draining off the wash buffer. This was repeated twice, after which the upper section of the column was removed along with the impeller and baffles. The resin was then packed in the 50-cm-high section by gravity settling. The upper flow adapter was inserted into the column and washing of the packed resin continued with 0.07 M sodium citrate, pH6.0 until the absorbance was less than 0.2 at 280nm. The FIX complex was subsequently eluted with 0.2M sodium citrate. pH6.0. Eluate samples were retained to be assayed for activity of factorsll. VII. IX and X.

Results and Discussion

Batch Data from the batch adsorption of the FIX to DEAE-

Sephadex and DEAE-Sephrose Fast Flow are shown in

figure 4. The FIX activity in the supernatant as a percentage of the FIX activity of the starting material is plotted against time for both resins. Results are given as a mean of two separate experiments for the DEAE Sephadex and from one experiment with D E A E Sepharose Fast Flow. In this analysis, we assume that the amount of FIX bound to the resin is equal to the FIX activity lost by the supernatant. Using this assumption, more than 90% of the FIX is ad- sorbed within the first 10min and over 95% is adsorbed within the first 15 min. By the 25-min timepoint, essentially all the FIX has been adsorbed from the supernatant plas- ma. From the batch data, a residence time of 2Smin was estimated to be sufficient for the semi-continuous experi- ments.

Stirred Column Data from a lab scale experiment on the stirred column

adsorption of FIX on DEAE-Sephadex with a 25-min resi- dence time are shown in figure 5. The FIX activity of stirred column effluent fractions normalized as a percentage of the

Semi-continuous Factor IX Complex Preparation 237

Fig. 5. Semi-continuous adsorption of FIX on the lab scale. The activity of FIX in the stirred column effluent is normalized to the initial plasma FIX activity and plotted against effluent fraction number.

Fig. 6. Semi-continuous adsorption of FIX for residence times of 60 (W), 25 (A), 20 (0 ) and 15 (0) rnin. The effluent FIX activity norrnal- ized against the initial plasma FIX activity is plotted against volume of plasma passed.

initial feed FIX activity, is shown. The high values in the initial fractions, indicative of incomplete FIX adsorption, are due to the fractions being collected before 25 min. Since the unadsorbed plasma in these fractions had spent less than 25 min in the stirred column, the adsorption of FIX to the DEAE Sephadex was not complete as suggested by the batch experiments. Later in the process, over 95% of the FIX activity is adsorbed from the plasma at a residence time of 2Smin. The FIX activity in the supernatant effluent, normalized to initial plasma, for several different lab-scale stirred-column experiments with varying residence times is shown in figure6. The data suggest that the efficiency of FIX capture from cryo-poor plasma is not affected within the range of 15-60 min. The scatter in the FIX activity in the effluent for the various residence times is likely due to the uncertainty in the FIX assay and not to any real differences in adsorption among the various residence times.

Figure7 shows data for one experiment where the ef- fluent stream concentrations of factors I1 and X were also assayed. The values are normalized to their initial concen- trations in the starting plasma. FII was completely ad- sorbed and undetectable in the effluent fractions. Over 95% of FIX and FX were removed from the plasma. A comparison of FIX recovery between the lab and pilot scales is shown in table 3, indicating no appreciable differ- ence between the two scales.

Comparing the recovery of factors 11, IX and X and the amount of FIX remaining unadsorbed (table 4), the semi- continuous process has apparently significantly improved

Fig. 7. Semi-continuous adsorption of plasma onto DEAE Sepha- dex at the pilot scale (550liters). The residence time was 25 min. The activities of factors11 (A) , X (0) and IX (I?) in the effluent stream, normalized to the initial plasma factor activity, are plotted against effluent plasma fraction number. Fraction size was 16 liters.

the recovery of FII and FIX. The amount of FIX that remains unadsorbed, however, does not change significant- ly. This is an interesting result and subsequent experiments have shown that though the percentage of FIX removed from the plasma may remain the same, the contact mode affects the percentage of FIX recovered.

238 Tharakan/Gee/Clark

Table 3. Summary of FIX results for lab and pilot scale

Unadsorbed Ulml Wash U/ml Eluate U/ml Recovery % Scale Plasma U/ml

Lab 1.0 0.06 0.04 3.6 69

Pilot 1.0 0.03 0.14 6.1 68 (Volume, ml) (500) (500) (40) (96)

(Volume, I) (550) (550) (507) (61)

Table 4. Comparison of batch and semi-continuous recovery data at pilot scale

Parameter Semi-continuous Batch

Factor IX recovery, % 68

Factor X recovery, 7'0 57.1 Factor IX unadsorbed, % 2.7

Factor I1 recovery, % 60.4 49.1 27.6 51.2 3.96

In table5, the recovery of FIX is compared to the amount of FIX left unadsorbed for the two resins in various contacting modes. For a particular resin, the ratio of resin volume to total plasma volume is constant. However, the resin density in the reactor is dependent on the contacting mode. For a batch mode the resin density in the reactor is equal to the resin:plasma ratio. If the contacting mode is changed to a semi-continuous stirred tank, the resin density increases by an order of magnitude. If the mode is changed to a packed bed (as is possible with the more rigid DEAE Sepahrose Fast Flow), the resin density increases even fur- ther. From the table, the amount of FIX recovered in- creases with resin density. However, lowering the resin

density does not decrease the amount of FIX adsorbed. This apparently contradictory result suggests that the process of FIX adsorption and desorption is not simple. Multiple at- tachment points may lead to denaturation (and hence activ- ity loss) of the FIX molecule. Competitive binding of other proteins may also be a factor. One direction of our research is to study a simpler protein solution which does not have as many components as plasma. This may help elucidate the nature and mechanisms of FIX adsorption and desorption to ion exchange resins that lead to this result.

Conclusion

There are several features of the continuous process that are attractive compared with the batch processes. There is a decrease in labor requirements, overall process time, and the number of process vessels required because all oper- ations can be performed in one modified column. Interest- ingly, the recovery of FIX dramatically improves with the semi-continuous process. More research is required to de- termine the reasons for the increased recovery with no change in the amount of FIX adsorbed from plasma.

Table 5. Comparison of recovery and adsorption of factor IX for various scales and capture methods ~~

Resin Scale Capture method Ratio of resin to Ratio of resin to YO FIX unadsorbed 7'0 FIX recovery plasma volume reactor volume

DEAE-Sephadex lab batch 0.027 lab semi-continous 0.027

pilot batch 0.027 pilot semi-continous 0.027

(stirred column)

(stirred column)

DEAE-Sepharose lab batch 0.04 Fast Flow lab semi-continous 0.04

(stirred column) lab semi-continuous 0.05

(packed bed)

0.027 0.2 0.2 5.0

0.027 1.5 0.29 2.4

32 69

49 68

0.04 0.4 32 0.12 17.0 82

1.0 0.0 100

Semi-Continuous Factor IX Complex Preparation 239

References

1 Stryker, M. H.; Bertolini, M. J.; Hao, Y.-L.: Blood fractionation: proteins. Adv. biotechnol. Proc. 4: 275-336 (1985).

2 Mitra, G.; Lundblad, J.: Continuous fractionation of human plas- ma. Biotech. Bioeng. 20: 1037-1044 (1978).

3 Foster. P.R.; Watt, J .G.: The continuous small volume mixing fractionation process: in Curling, Methods of plasma fractionation, p. 17 (Academic Press, London 1980).

4 Deggeller, K.: Optimal use of human blood. Vox Sang. 16: 407410 ( 1969).

5 Wickerhauser. M.; Sgouris, J.T.: Development of large scale frac- tionation methods. 11. Isolation of a factor IX concentrate (pro- thrombin complex) for clinical use. Vox Sang. 22: 137-160 (1972).

6 Hystek, J . : Brummelhuis, H. G. J.; Krinjen, H. W.: Contributions to the optimal use of human blood. 11. The large scale preparation of prothrombin complex. A comparison between two methods us- ing the anion exchangers DEAE cellulose DE52 and DEAE Sepha- dex A-50. Vox Sang. 25: 113 (1973).

7 Menache, D.; Behre, H. E.: Orthner, C . L.; Nunez, H.; Anderson, H. D.; Triantaphyllopoulos, D. C.: Kosow, D. P.: Coagulation fac-

tor IX concentrate: method of preparation and assessment of po- tential in vivo thrombogenicity in animal models. Blood 64: 1220- 1227 (1984).

8 Biggs, R.: Human blood coagulation haemostasis and thrombosis. p. 164 (Blackwell Scientific, Oxford 1972).

9 Miekka, S.I.: Use of albumin and tween as stabilizers to prevent activity loss during clotting assays of factorIX and X. Thromb. Hemostasis 58: 349 (1987).

Received: November 9,1988 Revised manuscript received: April 28,1989 Accepted: May 4,1989

John Tharakan, PhD American Red Cross, PDL 15601 Crabbs Branch Way Rockville, MD 20855 (USA)