A method for the separation of delta bilirubin using Cibacron Blue affinity chromatography

ELSEVIER

Clinica Chimica Acta 239 (1995) 37-46

A method for the separation of delta bilirubin using Cibacron Blue affinity chromatography John R. Burnett*, Chew W. Lim, Graerne N. Mahoney, Michael J. Crooke Division of Chemical Pathology, Department of Laboratory Services, Wellington Hospital, Wellington, New Zealand Received 1 September 1994; revision received 9 May 1995; accepted 23 May 1995

Abstract We developed and evaluated a method for the separation of delta bilirubin (B6) by microcolumn affinity chromatography based on Cibacron Blue 3G-A-Agarose. Untreated serum was applied to affinity columns and free non-protein bound bilirubins were eluted with phosphate blfffer containing 20 g/1 Triton X-100. Retained albumin was eluted using caffeinebenzoate reagent and bilirubin associated with this fraction (B6) quantitated by the method of Jendrassik and Gr6f modified by Doumas et al (Clin Chem 1985;31:1779-1789); results correlated well with the high performance liquid chromatography (HPLC) method (n = 35, y (affinity)= 1.009x ( H P L C ) - 5.49; r = 0.959) described by Lauff et al. (J Chromatogr 1981;226:391-402). Two controls analyzed with each batch gave between-batch imprecision less than 4.0% (n = 10; Control l, mean = 20.05 /zmol/l; Control 2, mean = 74.82 /~mol/1). Within-batch imprecision was less than 3.3% for both levels. Specimens collected from 25 neonates less than 20 days of age showed a B8 concentration of 1.7 4- 0.7/tmoF1 (mean 4- 1 S.D.) and percent B6 of 2.2 4- 1.9 (mean total bilirubin 4- 1 S.D. = 118 4- 79/zmol/l). Although time consuming, this simple and precise method allows the measurement of B6 in laboratories without the need for specialized instruments.

Keywords: Conjugated bilirubin and unconjugated bilirubin; Delta bilirubin; Direct bilirubin; High performance liquid chromatography; Micro-column affinity chromatography

Abbreviations: I~, delta bilirubin; DBIL, direct bilirubin; TBIL, total bilirubin; HPLC, high performance liqlad chromatography; PB, phosphate buffer; PBT, phosphate buffered Triton X-100; BA, albumin and tmconjugated bilirubin; CB, caffeine-benzoate. * Corresponding author, Robarts Research Institute, University of Western Ontario, London, Ont. N6A 5K8, Canada. 0009-8981r95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0009-8981 (95)06097-W

38

J.R. Burnett et al./ Clinica Chiraica Acta 239 (1995) 37-46

1. Introduction

It is over 25 years since Kuenzle et al. [1,2] first described the existence of a fourth bilirubin fraction in serum. This fraction, named delta bilirubin (B~), was observed to be firmly bound to serum albumin. Little notice was taken of this finding until 1981 when Lauff et al. [3] developed a high performance reversed-phase liquid chromatography (HPLC) procedure for the measurement of B& This method separated bilirubin from icteric sera into four fractions, namely alpha or unconjugated bilirubin, beta or monoconjugated bilirubin, gamma or diconjugated bilirubin and B& Further characterization identified B6 as bilirubin linked covalently to albumin [4,5]. B6 does not undergo glomerular filtration and exhibits a half-life similar to albumin [6]. It can be formed in vitro by spontaneous non-enzymatic reaction of serum albumin with bilirubin glucuronides [7] and is a predominantly direct-reacting bilirubin with diazotized sulfanilic acid [4,5,8]. The clinical usefulness of the B6 fraction in the investigation of hyperbilirubinaemia has been reported by Weiss et al. [9] and in paediatric patients by Brett et al. [101. Various methods for separating B6 have been used including HPLC, immunoaffinity chromatography using Sepharose [11], anion exchange chromatography of the azodipyrrole derivatives [12], size exclusion chromatography [13], solvent extraction [14], enzymatic determination using bilirubin oxidase [15] and the commercially available Ektachem multilayer film process [16]. Simplified HPLC systems have been developed [17,18]; however, the equipment and technical expertise required may not be readily available in many laboratories. The triazinyl blue dye, Cibacron Blue 3G-A has become established as a group selective affinity ligand when directly coupled to a support matrix. Human serum albumin has been reported to bind to Cibacron Blue 3G-A and requires the entire Cibacron Blue molecule for most effective binding [19]. We present an alternative method based on affinity chromatography using Cibacron Blue to separate albumin bound bilirubin from free non-protein bound bilirubin for the estimation of B6 in serum.

2. Materials and methods

2.1. Apparatus Cibacron Blue 3G-A-Agarose type 3000-CL, C-1535 (Sigma Chemical Co., St. Louis, MO 63178, USA) was added to Polyprep chromatography columns (Bio-Rad Laboratories, Hercules, CA 94547, USA; Cat. no. 731-1550). A porous plastic disc (also available from Bio-Rad Laboratories; Cat. no. 920-7225, trimmed to size) was inserted to cap the affinity column. The final column height was 2.3 crn and the bed volume was 1 ml. The column was equilibrated in 50 mmoVl phosphate buffer (pH 5.5) containing 20 g/1 Triton X-100 (PBT). 2.2. Column fiactionation Before use, affinity colunms were primed using 5 ml PBT and then washed with 5 ml 50 mmoFl phosphate buffer, pH 5.5 (PB). Sample (100/~1) was then applied,

J.R. Burnett et al./ Clinica Chimica Acta 239 (1995) 37-46

39

follow~ by 200/~1 PB, 30 ml PBT, and 10 ml PB. Each was allowed to drain completely before the next solution was applied. B~ was eluted with 2 ml caffeinebenzoate (CB) reagent (56.0 g/l anhydrous sodium acetate, 56.0 g/l sodium benzoate, 1.0 g/1 disodium EDTA, 37.5 g/1 caffeine) which had been diluted 2-fold. 2.3. B8 quantitation

Bilirubin associated with the CB eluted fraction (BS) was quantitated by the method of .Iendrassik and Gr6f modified by Doumas et al. [20]. Three hundred microlitres of diazotized sulfanilic acid were added to fractions collected as described above and mixed. After exactly 10 min, 400/zl of alkaline tartrate (75.0 g/l NaOH, 320.0 g/l potassium sodium tartrate) were added. Absorbances were measured spectrophotometrically at 600 nm. A saline solution was treated as for samples and used as reagent blank. 2.4. Column re-equilibration

To re-equilibrate columns, a further 8 ml of CB reagent were applied. This was followed by a further 10 ml of PBT. At regular intervals the columns were washed with 10 ml urea-NaOH (480 g/l urea, 20 g/1 NaOH) and then re-equilibrated using PBT (2 x 30 ml). 2.5. Standardization and quality control

The affinity method was standardized using pooled icteric sera exhaustively dialyzed as described by Lauff et al. [5]. The total bilirubin (TBIL) value established for this preparation was 63/~mol/1. Two specimens prepared from pooled patients' sera we,re run with each batch as controls. 2.6. Preliminary investigations

After sample application (100-200 ~1), the affinity column was washed with a series of PB solutions at various pH values and containing various concentrations of Triton X-100. Six fractions were collected. Retained protein was then eluted with CB reagent. Total-protein was measured by biuret assay [21] and bilirubin by measuring absorbances of fractions at 420 mn to estimate recoveries for protein and bilirubin. Protein fractions were evaluated by agarose electrophoresis after 20-fold concentration. 2.7. Patient samples

Over a 3-month period, we determined both B6 by Cibacron Blue affinity chromatography and direct bilirubin (DBIL), in 297 serum samples from 95 adult patients with hyperbilirubinaemia. Samples from 35 of these patients also underwent B~ analysis by HPLC. In addition, Cibacron Blue affinity B6 and DBIL analyses were also determined on serum samples from 25 neonates less than 20 days old with hyperbilirubinaemia. Patients presenting with unconjugated hyperbilirubinaemia included neonates with physiological jaundice and patients with haemolysis or Gilbert's syndrome, whereas those presenting with conjugated hyperbilirubinaemia had a wide variety of hepatobiliary diseases.

40

J.R. Burnett et a l . / Clinica Chimica A c t a 239 (1995) 3 7 - 4 6

0

0

i .° 0

[--

0

~~

m

r,,)

8 ¢q

¢q

¢q

t,-q

i'q

¢',,I

o

o

e~

x_,

...7~

..,~_,x

o ..= 0

u3

t~

x

J.R. Burnett et al. / Clinica Chimica Acta 239 (1995J 37-46

41

3. Reselts

3.1. Recoveries of protein and bilirubin for affinity columns Two specimens, one a pool of icteric serum and one containing only human serum albumin (40 g/l) and unconjugated bilirnbin (200/~mol/1; No. B-4126, Lot 57F-0111, Sigma Chemical Company, St. Louis, MO 63178, USA) (BA) were used to assess recoveries. Results presented in Table 1 showed that recoveries for protein varied from 88.5 to 112.1%. As expected only small amounts of protein were recovered in the PBT fraction for BA (2.0-6.3%) which contained only added albumin, whereas 22.7-45.3% of protein was recovered from the PBT fraction of Control 2 which was serum-based. The recovery patterns for bilirubin showed that bilirnbin was largely recovered in the PBT fraction (> 75.9%) for BA. This was presumed to be added unconjugated bilirnbiLn (see Methods). On the other hand, 17.3-20.3% of bilirnbin was recovered in the CB fraction for Control 2 prepared from pooled adult serum. Varying the Triton X-100 concentration from 10 to 30 g/1 had little effect on the recovery of protein or bilirnbin from either fraction. Results do show a trend for increased protein recovery in the PBT fraction and decreased recovery in the CB fraction with increasing pH from 5.1 to 6.0; however, such pH effects were not evident in the recovery of bilirubin. A pH of 5.5 was chosen because the recovery of albumin from BA in the CB fraction was the highest while that from BA in the PBT fraction was the lowest.

3.2. Agarose electrophoresis Aliquots of fractions eluted in PBT and CB were concentrated 20-fold and elec1

300 / y

=

1.009x

- 5.49

,/

r = 0,959 O

n=35

E 200

•

// /

,

)

< i

2

!

.

I00

'

rm

o o 0

0 Delta

I)

,>

i

_

i

100

200

300

Bllirubin: HPLC ( u m o l / L )

Fig. I. Correlation of delta bilirubin by affinity chromatography with that by HPLC.

42

J.R. Burnett et aL I Clinica Chimica .4cta 239 (1995) 37-46

Table 2 Within-batch and between-batch precision of 1~ determinations Within-batch (n = 10)

Between-batch (n = 10)

Mean (1 S.D.),

Mean (! S.D.),

C.V. (%)

/~moFi Control i Control 2

C.V. (%)

/~mol/I

19.95 (0.63) 74.96 (1.71)

3.2 2.3

20.05 (0.78) 74.82 (2.28)

3.9 3.0

trophoresed on agarose gels. Whereas albumin was not detected in the PBT fraction, complete albumin recovery along with other serum proteins resulted following the CB wash (data not shown).

3.3. Validation of method by comparison with HPLC Samples from 35 hyperbilirubinaemic patients were analyzed by the Cibacron Blue affinity method and by the HPLC method of Lauff et al. [3]. Linear regression analysis gave the following equation: y (affinity) = 1.009x (HPLC) - 5.49. The correlation was acceptable (r = 0.959; Fig. 1). 3.4. Precision and linearity The reproducibility of the method was evaluated using two ~ooled serum

500

A

i

3007B

!

/i

i

400

n

0.994 297

--

~ r = 0,853 n = 297

lli~ i

-

,

~p~=~ 300

-~

;

• •

I

200

j

"

200

"~ 100

i

"

/

•

~ i."

• IIi

° g..,.

100

0

100 200 300 400 500 600 Total BIIIrubin (umol/L)

0

100 200 300 400 500 600 Total Billrubin (umol/L)

Fig. 2. Correlation of (A) direct bilirubin and (B) delta bilirubin (affinity chromatography) with total bilirubin in sera from adult patients.

J.R. Burnett et al. /Clinica Chimica Acta 239 (1995) 37-46

43

specimens (Controls 1 and 2) that were analyzed with each batch. The results of the within-batch and between-batch precision studies are summarized in Table 2. The method was linear to a B~ concentration of at least 200/~mol/1. 3.5. Measurement of B6 in serum of hyperbilirubinaemic adults A re,dew of liver function test results over a 3-month period identified 297 samples from 9:5 adult patients with hyperbilirubinaemia for B6 and DBIL analyses. DBIL and TBIL for this patient group correlated well (y [TBIL] = 1.238x [DBIL] + 13.6; r = 0.9!94). The relationship between B6 and total bilirubin was less well defined and characterized by much more scatter (y [TBIL] = 2.141x [B~] + 36.2; r = 0.853). These data are presented in Fig. 2. The poorer correlation between B~ and TBIL probably results from the dependence on albumin turnover for clearance of B6 from the sevarn [4,5]. 3.6. Measurement of B6 in serum of neonates Samples from 25 neonates less than 20 days old showed low B~ concentrations consistent with results using HPLC methods [14,16]. TBIL concentrations ranged from 31) to 281/zmol/l with the mean ± 1 S.D. for TBIL, B~ and percent Bdt being 118 ± 79/~mol/l, 1.7 ± 0.7/zmol/l and 2.2 4- 1.9%, respectively. TBIL concentrations for adults ranged from 22 to 557 #mol/l with the mean 4- 1 S.D. for TBIL, B6 and percent B6 being 92 4- 102 /zmol/1, 25 4- 22 #mol/1 and 31 4- 47 %, respectively. This contrast between the distribution of B~ against TBIL in neortates and adults is graphically presented in Figs. 2B and 3.

50 y

= O.O03x

+

1.37

r = 0.278 0

n =

E

25

>,

<

25

2

i

0

100

200

300

Total Bllirubln (umol/L)

Fig. 3. Correlation of delta bilirubin (affinity chromatography) with total bilirubin in sera from neonates.

44

J.R. Burnett et al./ Clinica Chimica Acta 239 (1995) 37-46

4. Discussion

The measurement of serum B~ has generally been limited to those laboratories with HPLC apparatus and expertise, or alternatively dry chemistry analysers, due to the lack of a suitable simple chemical technique. Our proposed method, based on the selective binding of human serum albumin by Cibacron Blue 3G-A, enables the isolation of albumin bound bilirubin which is subsequently eluted and quantitated. In contrast to alternative procedures, specialized instruments and technical expertise are not required and capital expenditure is minimal. Disadvantages of the method include lack of amenability to automation and the time consuming nature of the procedure due to the large volumes of PBT and PB required to remove all traces of bilirubin associated with, but not covalently bonded to, albumin. The flow rate could be increased by applying a small negative pressure across the columns. Advantages of the affinity procedure are, firstly, that albumin is tightly bound and associated bilirubin can be removed using large volumes of buffer without loss of bound albumin, secondly, the columns can be re-used at least 15 times and thirdly, re-equilibration is simple. Both conjugated and unconjugated bilirubin have high, though reversible, affinities for serum albumin, but it is known that the affinity of albumin for unconjugated bilirubin is probably higher than that for conjugated bilirubin [6]. We have therefore used a mixture of purified albumin and unconjugated bilirubin (BA) to guide our method development. Thus our method has been optimized by ensuring firstly, the complete recovery of albumin and secondly, the exhaustive dialysis of bilirubin from albumin held on the column. The low level of percent bilirubin found in the albumin fraction recovered from hyperbilirubinaemic neonatal specimens suggests that this has been achieved. Using the Cibacron Blue affinity method we have been able to show good correlation of B6 with the HPLC method of Lauff et al. [3] (r = 0.959). Our correlation is similar to others (Seligson et al. [12], r = 0.965: anion exchange vs. HPLC; Blanckaert et al. [14], r = 0.975: solvent extraction vs. difference between total bilirubin and HPLC determined conjugated and unconjugated bilirubins; Franzinni and Cattozzo [13], r = 0.926: molecular exclusion vs. Kodak Ektachem). Our analytical bias (slope = 1.009) is better than that of Seligson et al. (slope = 0.724) who used the same reference method. We have used a dialyzed serum for calibration because of the lengthy nature of the separation procedure and the known photosensitivity of bilirubin. The quantitation of eluted non-protein bound bilirubin is not possible because of the presence of large amounts of Triton X-100 and the large dilution; this potential alternative approach is therefore not feasible. In our opinion, standardization of the B6 calibrator using the TBIL method would also reduce any potential systematic bias between the assays for TBIL and B& However, although we have assumed our exhaustively dialyzed serum to be pure, HPLC analyses showed that this was only 89.5% pure. Re-calculation of our own results to reflect this impurity would cause our B6 results by the affinity method to be lowered by 11.7%.

J.R. Burnett et al./ Clinica Chimica Acta 239 (1995) 37-46

45

In practice, this difference is probably not clinically significant and as we required a procedure without the need for HPLC measurements, we have chosen not to apply this correction. For the record, the re-calculated linear regression analysis after correction gave the following equation: y [affinity] = 0.865x [ H P L C ] - 4.75, (r = 0.959). In recent years, several investigators have reported the usefulness of B6 measurements in the assessment of patients with liver disease. Weiss et al. [9] showed that B/~was not detected in appreciable amounts in serum from normal adults or patients with unconjugated hyperbilirubinaemia but was detectable in other types of hepatobiliary disease. Expressing serial measurements of B6 as a percentage of TBIL, an increase was seen following recovery from jaundice, reaching values in excess of 80%. On the other hand, a decreasing or low percentage of 1~ in patients with severe liver disease suggested a poor prognostic outlook for the patient [22]. Wu et al. [23] have recently demonstrated a role for the serial monitoring of B/~ following liver transplantation to aid in early diagnosis of rejection. We were able to demonstrate, using serial measurements of B6 in adult hyperbilirubinaemic patients, the changes in B6 that occur within the DBIL fraction as a result of its prolonged half-life relative to other direct reacting fractions (data not shown). Whether B/~ measurement will offer any significant additional clinical advantage over DBIL analysis alone remains to be established. In conclusion, we have developed a micro-column affinity method for B/~ based on Cibacron Blue which we have validated by comparison with an established HPLC method and by clinical studies. The method is simple and precise and the columns are easily re-equilibrated for repeat use. Referel~es [1] Kuenzle CC, Sommerhalder M, Ruttner JR, Maier C. Separation and quantitative estimation of four bilirubin fractions from serum and of three bilirubin fractions from bile. J Lab Clin Med 1966;67:282-293. [2] Kuenzle CC, Maier C, Ruttner JR. The nature of four bilirubin fractions from serum and of three bilirubin fractions from bile. J Lab Clin Meal 1966;67:294-306. [3] Lauff J J, Kasper ME, Ambrose RT. Separation of bilirubin species in serum and bile by high perfo[mance liquid chromatography. J Chromatogr 1981;226:391-402. [4] Wu TW, Lauff J J, Kasper ME, Ambrose RT. Delta bilirubin: preliminary physico-cheimcal characterization and its implications in bilirubin determination. J Clin Chem Clin Biochem 19131;19:881 (abstract). [5] Lauff J J, Kasper ME, Wu TW, Ambrose RT. Isolation and preliminary characterization of a fraction of bilirubin in serum that is firmly bound to protein. Clin Chem 1982;28:629-637. [6] Miwaca M, Fevery J, Blanckaert N. Analytical aspects and clinical interpretation of serum bilitrubins. Semin Liver Dis 1988;8:137-147. [7] Gautam A, Seligson H, Gordon ER, Seligson D, Boyer JL. Irreversible binding of conjugated biliirubin to albumin in cholestatic rats. J Clin Invest 1984;73:873-877. [8] De,ureas BT, Wu TW, Jendrzejczak B. Delta bilirubin: absorption spectra, molar absorptivity, and reactivity in the diazo reaction. Clin Chem 1987,33:769-774. [9] Weiss JS, Gautam A, Lauff JJ et al. The clinical importance of a protein-bound fraction of serum biliLrubin in patients with hyperbilirubinemia. N Engl J Med 1983;309:i47-150. [10] Brett EM, Hicks JM, Powers DM, Rand RN. Delta bilirubin in serum of pediatric patients: correlations with age and disease. Clin Chem 1984;30:1561-1564.

46

J.R. Burnett et al./ Clinica Chimica Acta 239 (1995) 37-46

[1 i] Blanckaert N, d'Argenio G. Presence of bilirubins covalently linked to albumin in serum of patients with cholestasis. Gastroenterology 1982;82:1222 (abstract). [12] Seligson D, Seligson H, Wu TW. An anion-exchange chromatographic method for measuring bilirubin covalently bound to albumin. Clin Chem 1985;31:!317-1321. [13] Franzini C, Cattozzo G, Spagliardi G. Micro-chromatographic method for measuring deltabilirubin in serum. Clin Chem 1987;33:1013 (abstract). [14] Blanckaert N, Servaes R, Leroy P. Measurement of bilirubin-protein conjugates in serum and application to human and rat sera. J Lab Clin Med 1986;108:77-87. [15] Kosaka A, Yamamoto C, Morishita Y, Nakane K. Enzymatic determination of bilirubin fractions in serum. Clin Biochem 1987;20:451-458. [16] Sundberg MW, Lauff JJ, Weiss JS et al. Estimation of unconjugated, conjugated, and "delta" bilirubin fractions in serum by use of two coated thin films. Clin Chem 1984;30:!314-1317. [17] Singh J, Bowers LD. Quantitative fractionation of serum bilirubin species by reversed-phase highperformance liquid chromatography. J Chromatogr 1986:380:321-330. [18] Adachi Y, Inufusa H, Yamashita M et al. Clinical application of serum bilirubin fractionation by simplified liquid chromatography. Clin Chem 1988;34:385-388. [19] Travis J, Bowen J, Tewksbury D, Johnson D, Pannell R. Isolation of albumin from whole blood plasma and fractionation of albumin-depleted plasma. Biochem J 1976;157:301-306. [20] Doumas BT, Kwok-Cheung PP, Perry BW et al. Candidate reference method for determination of total bilirubin in serum: development and validation. Clin Chem 1985;31:1779-1789. [21] Weichselbaum TE. An accurate and rapid method for the determination of proteins in small amounts of blood, serum and plasma. Am J Clin Pathol 1946;16:40-49. [22] Wu TW. Delta bilirubin: the fourth fraction of bile pigments in human serum. Isr J Chem 1983;23:241-247. [23] Wu TW, Levy GA, Yiu S et al. Delta and conjugated bilirubin as complementary markers of early rejection in liver-transplant recipients. Clin Chem 1990;36:9-14.