| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Clinical Biochemistry, Holbæk Hospital, 4300 Holbæk, Denmark.
2
Department of Analytical and Pharmaceutical Chemistry, Royal Danish School of Pharmacy, 2100 Copenhagen, Denmark.
3
Department of Endocrinology E and Clinical Chemistry, Frederiksberg Hospital, 2000 Frederiksberg, Denmark.
aAuthor for correspondence. Fax 45-59484409; e-mail chstho{at}vestamt.dk.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Three different potential reference techniques were investigated for adsorption of T4 to membrane materials, including any in vitro solid surfaces to which T4 might adsorb, and for efficient separation of the T4-binding proteins from the free hormone fraction. We compared ultrafiltration with different commercial ultrafiltration units, ultrafiltration by dialysis tubing, equilibrium dialysis, and ultracentrifugation. We measured the adsorption to membranes and materials with L-[125I]T4. Separation efficiency was determined by measuring the T4-binding protein albumin in the ultrafiltrate and the dialysate as well as in the supernatant from the ultracentrifugation with a double-antibody sandwich ELISA technique.
Results: We found a constant relationship between the amount of T4 adsorbed to the dialysis or ultrafiltration membranes/materials and the initial T4 concentration in HEPES buffer (protein-free medium). T4 was considerably less adsorbed from serum than from HEPES buffer (P <0.001). Serum T4 was less adsorbed upon ultracentrifugation than during dialysis and ultrafiltration (P <0.001). It was difficult to completely separate FT4 from the binding proteins by ultrafiltration and ultracentrifugation. Separation by ultrafiltration depended on the material used.
Conclusions: No investigated separation technique provides technically and theoretically correct separation of the free fraction of T4 from the protein-bound fraction. Equilibrium dialysis seems to be the least compromised.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The procedure for the reference method must combine separation of the free and protein-bound T4 fractions, without significant disturbance of the equilibrium, with subsequent highly accurate nonimmunologic quantification of the FT4 fraction. Established methods that comply fairly well with the above-mentioned requirements are based on equilibrium dialysis (1)(5)(6)(7) and ultrafiltration (8)(9)(10) followed by quantification of the free hormone by RIA in the dialysate/ultrafiltrate. Methods using nonimmunologic measurements of the FT4 fraction are not available. A possible reference technique for quantification of the T4 fractions could involve isotope dilution-liquid chromatography-mass spectrometry or isotope dilution-liquid chromatography-inductively coupled plasma mass spectrometry.
Even when methods that are related reference methodologies are used, significant differences among FT4 results are found. Higher FT4 concentrations have been found by the ultrafiltration technique than by equilibrium dialysis; consequently, no consensus has yet been achieved concerning the FT4 content in human serum (2)(11)(12).
The major technical difficulties in developing a reference method are as follows:
It has been shown that leakage of T4-binding proteins through dialysis tubing during ultrafiltration could cause spuriously increased results, which was confirmed by simultaneous measurements of the binding protein, albumin, in the ultrafiltrate (9). Cofiltration of binding proteins has also been examined for different commercial ultrafiltration devices (12). Only with the Ultracent-10, which is no longer in production, was it possible to omit the albumin assay to check the ultrafiltrate for cofiltration of binding protein (12). Thus, protein leakage during ultrafiltration or ultracentrifugation may induce relatively large errors. In a previous study, Nelson and Tomei (7) found it necessary to add rabbit IgG and gelatin to the dialysis buffer to prevent loss of T4 (up to 29%) attributable to adsorption to the dialysis cell components.
In the present study, three separation techniques involving initial physical separation of the free from the bound hormone were studied to find a method that fulfills the following conditions:
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
preparation of calibrators
Calibrators (controls, serum pools) were prepared in glassware cleaned for 10 min in 0.1 mol/L sodium hydroxide and then rinsed with destilled water. The disposable glass vials were cleaned just before use. The calibrators were kept at -20 °C in brown vials (T4 is light sensitive).
The stock solution containing 10 µmol/L L-thyroxine was prepared as follows: 38.9 mg of L-thyroxine {3-[4-(4-hydroxy-3,5-diiodophenoxy)-3,5-diiodophenyl]-L-alanine; cat. no. T-2376; Sigma} was dissolved in 2000 µL of methanol and 100 µL of 1 mol/L HCl. This solution was diluted with 0.05 mol/L sodium phosphate (pH 11.6) to a final volume of 5000 mL. HEPES buffer [0.01 mol/L (2.5 g/L), pH 7.4] was used for further dilution of the stock solution to give the calibrators.
preparation of albumin samples and human serum pools
Human albumin (30%; cat. no. A6909; Sigma) was diluted with 0.01 mol/L HEPES buffer (pH 7.4) to the final desired concentration. Human serum pools were prepared with fresh patient material just before use or with material frozen at -80 °C and thawed at room temperature just before use.
calibrator concentrations
The concentrations of the calibrator preparations were determined in T4 RIAs (Nichols Institute Diagnostics). The company guidelines were followed. All reagents were delivered with the assay package.
separation conditions
Sodium hydroxide-cleaned vials or disposable brown vials were used for the preparation of calibrators, controls, and serum pools. Each experiment was performed in duplicate. The amounts of initial material, ultrafiltrate, dialysate, and retentate were determined by weight. Disposable pipette tips were primarily flushed twice with sample material before use. Every lot of materials/devices was tested with "clean" (free from albumin and T4) 0.01 mol/L HEPES buffer (pH 7.4). The ultrafiltrate, dialysate, and retentate were collected in two fractions, both of which were analyzed for albumin and T4.
ultrafiltration by dialysis tubing
To study potential protein leakage as well as adsorption of T4, we used a modification of a previous method (8). Visking and Spectra/por® tubing (Table 1
) was prepared for ultrafiltration by rinsing the tubing with distilled water and then drying in a flow of air to make the tubing stiff. The tubing was bent in a U-shape and placed in 100 x 15 mm brown vials with the inflection point 1.5 cm above the bottom of the vials. Sample (3 mL) was pipetted into each tube. The tubing was centrifuged at 500g for 2 h for protein-free medium (T4 in 0.01 mol/L HEPES buffer, pH 7.4) and at 500g for 3 h for serum samples to obtain the same amount of ultrafiltrate.
|
ultrafiltration in devices
To examine protein leakage and T4 adsorption, we tested five different ultrafiltration devices. Two devices (Centricon® YM-3 and YM-10 and Omega Microsep® 1 and 3) were tested in two different nominal cutoffs. Specifications for the filters are presented in Table 1
. We used the ultrafiltration devices without precleaning the membranes. We filled the devices with the maximum amount of sample and closed them with a cap. The devices were centrifuged for 2 h at 500g for T4 in HEPES buffer and for 4–5 h at 500g for serum samples. The amount of ultrafiltrate varied and depended on the lot (material with the same specifications but produced at different time) and nominal cutoff of the device. The amounts of supernatant and ultrafiltrate were determined after ultrafiltration by weighing.
ultracentrifugation
The ultracentrifugation apparatus was an L5-65 ultracentrifuge from Beckman. Ultracentrifugation was at 141 000 and 230 000g. Two different centrifuge heads from Beckman were used, depending of the speed: an SW 28 swinging bucket (maximum rpm, 28 000), and a 45 Ti fixed angle rotor (maximum rpm, 40 000). The centrifuge tubing (38.5 mL of Ultra-Clear® or 94 mL of Polyallomer® tubing from Beckman) was filled with the maximum amount of sample and carefully closed. The samples were ultracentrifuged (24 h with the SW 28 rotor or 19 h with the 45 Ti rotor) at 20 °C. Each analysis consisted of six pieces of tubing. After centrifugation, the sample was pipetted very carefully as 5-mL fractions. The pipette tip was placed only a few millimeters below the sample surface in vials containing samples used for further analysis.
equilibrium dialysis
Equilibrium dialysis was performed using an assay from Nichols Institute Diagnostics. The company guidelines were followed. All reagents were used as delivered. We used 12 times more dialysis buffer than sample (2.4 mL of dialysis buffer and 0.2 mL of sample). The dialysis cell was covered with a layer of Parafilm® before it was placed in an incubator at 37 °C for 16–18 h. The amounts of retentate and dialysate after dialysis were determined by weighing.
albumin measurements
Albumin was measured with a double-antibody sandwich ELISA according to the recommendations given by DAKO. MaxiSorpTM ELISA plates were coated with rabbit anti-human albumin (cat. no. A0001; DAKO; 10 mg/L in 0.01 mol/L phosphate buffer, pH 7.2, containing 0.15 mol/L NaCl). The concentrations used for the calibration curve were 42–1667 pmol/L, made by diluting calibrator PROT-KAL9602 (13). The detector antibody was peroxidase-conjugated rabbit anti-human albumin (cat. no. P0356; DAKO) diluted 1:8000. For dilution of calibrators, samples, and detector antibody and washing of the plates, we used 0.01 mol/L phosphate buffer, pH 7.2, containing 0.5 mol/L NaCl and 1 mL/L Tween® 20. o-Phenylenediamine in citric acid-phosphate buffer, pH 5.2, was used as the chromogenic substrate.
adsorption of t4 to surfaces
We measured the adsorption of T4 to surfaces with L-[125I]T4 (specific activity 
100–116 Ci/mmol; NEX111; NEN®). The [125I]T4 was analyzed by HPLC to check radiochemical purity. Ultrafiltration devices, Visking and Spectra/por tubing, dialysis cells, and ultracentrifugation tubes were filled with sample material with added L-[125I]T4. The radioactivity was measured in the initial material and in the supernatants, retentates, ultrafiltrates, and materials to find the dispersion of T4. The same volume of sample was used in all vials for measurement of radioactivity. All materials (membranes, devices, tubes, and O-rings) were measured except for the dialysis cells, which could not be analyzed because of difficulties in destroying them and measuring the fragments in the gamma counter.
The recovery of [125I]T4 was checked in parallel experiments using direct RIA of T4 in supernatants and in ultrafiltrates (Nichols Institute Diagnostics).
measurements of t4 in ultrafiltrate by screening of different devices and visking and spectra/por tubing
The L-T4 concentrations in ultrafiltrates were measured with a T4 RIA (Nichols Institute Diagnostics). The company guidelines were followed. All reagents were used as delivered.
Statistical analysis for the data was performed using the Student t-test.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, the fraction of T4 adsorbed to the dialysis membranes was constant regardless of the initial T4 concentration (concentration range, 13–100 000 pmol/L). A comparison of the results in Fig. 1
with the results in Table 2
shows that the adsorption of T4 to the dialysis membranes was significantly less (P <0.001) in serum pools compared with HEPES buffer.
|
|
adsorption of t4 to ultrafiltration devices/dialysis tubing used as an ultrafiltration unit
Five different commercial devices and two different types of tubing (Table 1
) were screened for adsorption of T4 to the material. The Microcon®-3 and Centricon YM ultrafiltration membranes and the Visking and Spectra/por tubing had the lowest adsorption of T4. Membranes made of polysulfone or polyethersulfone showed a high percentage of T4 adsorbed compared with membranes made of cellulose. This could be attributable to ion binding between T4 and the sulfone groups in the membrane material (T4 is charged at pH 7.4, see below). The Microcon-3 was not suitable for ultrafiltration of serum samples because of the limited maximum sample volume. The Spectra/por tubing showed unacceptably high imprecision when the adsorption to the dialysis tubing was measured (results not shown).
Only the Centricon YM-3 and YM-10 ultrafiltration units (molecule weight cutoff, 3000 and 10 000, respectively) and the Visking dialysis tubing were suitable for ultrafiltration of serum samples because of the acceptable sample volume and the lower imprecision for measurement of the adsorption to the dialysis tubing. The relationship between molecular weight cutoff, adsorption, and T4 concentration was examined for the two ultrafiltration devices.
As shown in Fig. 1
, the fraction of T4 adsorbed to the ultrafiltration devices/dialysis tubing was constant at T4 concentrations of 13–100 000 pmol/L in HEPES buffer. Adsorption to the two ultrafiltration devices with different molecular weight cutoffs was similar, indicating that the adsorption was independent of molecular weight cutoff. The adsorption to the dialysis tubing appeared to be 10–15% higher than adsorption to the ultrafiltration devices, but this could be related to difficulties in emptying the tubes after ultrafiltration.
The relationship between T4 concentration in the ultrafiltrate and the amount of T4 bound to the filter part or dialysis tubing is shown in Figs. 1
and 2
. The total recovery was 94% ± 3% (n = 14) for each of the two ultrafiltration units and 91% ± 8% (n = 14) for the dialysis tubing. The fraction of T4 adsorbed to the membranes was constant at concentrations in the physiologic concentration range: 10 pmol/L (FT4) to 100 nmol/L (total T4).
|
The influence of the adsorption to membrane/material surfaces was evaluated for genuine materials (human serum pools). As shown in Table 2
, T4 adsorbed significantly less (P <0.001) to the two commercial ultrafiltration devices (0.63% ± 0.16% and 0.67% ± 0.19%) compared with the dialysis tubing (2.61% ± 0.36%). T4 in serum adsorbed significantly less than T4 in HEPES buffer (P <0.001).
When serum samples are subjected to ultrafiltration, the three binding proteins thyroxine-binding globulin (TBG), prealbumin (TTR), and albumin compete with the material/membrane in binding/adsorption of T4. The T4-binding proteins contain high-affinity binding sites (KTBG 1010, KTTR(1–2) 107/105, and Kalbumin(1) 105)
1
compared with the presumably low-affinity binding sites of the membrane/material. The percentage of T4 absorbed from serum onto the material was high compared with the free fraction usually found in normal plasma (0.02–0.03%).
As shown in Table 2
, adsorption to filters and ultrafiltration tubing on the retentate side probably had no influence on the determination of the FT4 concentration. Fig. 3
shows that the adsorption to vessel materials on the ultrafiltrate side had a minor effect on the FT4 determination.
|
adsorption of t4 to ultracentrifugation tubing
The adsorption of T4 to the ultracentrifugation tubing was lower in serum pools (0.21% ± 0.01% of the total T4 concentration before ultracentrifugation; Table 2
) compared with T4 in HEPES buffer (3.5% ± 1.5%; Fig. 1
). Thus, adsorption to the ultracentrifugation tubing did not significantly influence the determination of FT4 in serum samples. The total recovery (supernatant and ultracentrifugation tube) was 101% ± 2%.
separation of ft4 from the binding proteins by ultrafiltration
The albumin concentration in the ultrafiltrate was lot dependent (Fig. 4
), and no relationship between nominal molecular weight cutoff and the separation capacity of the tested ultrafiltration units (Centricon YM-3 and YM-10) was observed. For some lots, albumin concentrations >1667 pmol/L were found in the ultrafiltrate (Fig. 4
), and for one lot (Centricon YM-3; lot no. L8NM4783A) up to 18 nmol/L albumin was found in the ultrafiltrate.
|
For the Visking dialysis tubing, the albumin concentration in the ultrafiltrate was lower than that in the filtrate from the ultrafiltration units and <1667 pmol/L for all lot numbers.
separation of ft4 from the binding proteins by ultracentrifugation
Three serum samples with albumin concentrations 650 µmol/L were ultracentrifuged at 141 000g for 24 h or at 230 000g for 19 h. Even after ultracentrifugation at 230 000g for 19 h, the albumin concentration in the supernatant was still very high (>17 µmol/L). We found that ultracentrifugation under these conditions was not a qualified method for separation of FT4 from the bound fraction.
separation of ft4 from the binding proteins by dialysis
All the dialysates were found to be free from albumin. Thus, dialysis was the best choice of the methods compared for separation of FT4 from the bound fraction.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dilution is one of the drawbacks of the equilibrium dialysis-based methods. It may cause too low an estimate of FT4 in the presence of low-affinity binding protein inhibitors (6)(11). Attention must also be paid to the ionic composition and strength of the dialysis buffer. Cl-, H+, and HCO3- have been identified as important ions (15). It has been shown that changes in serum Cl-, H+, and HCO3- concentrations similar to those occurring in vivo alter or modify the protein binding of the thyroid hormones in serum (15). To a certain extent, the effects of in vivo changes in serum Cl-, H+, and HCO3- on free thyroid hormone concentrations tend to counteract each other. Therefore, the authors concluded that electrolyte and pH disturbances affect measurements of FT4 concentrations in serum by an equilibrium dialysis method to only a minor extent (15).
Another problem arises if T4 adsorbs to walls or membranes. Adsorption of hormone from serum (T4) implies "sequestration" of hormone from serum as a consequence of the adsorptive properties exhibited by, e.g., glass walls and dialysis membranes. This produces a net dissociation of bound hormone and establishes a new equilibrium throughout the system in which the free hormone concentration is reduced with respect to its original value. This concept can be extended to any process that produces selective hormone removal from serum, which causes both bound and free hormone concentrations in the new equilibrium to drop. This occurs, for example, if a serum sample is dialyzed. This process will cause sequestration of hormone from the sample into the dialysate. Nevertheless, the effect on the concentration of FT4 will generally be minor because of readjustments between free and bound pools, which occur immediately (16)(17).
In normal serum, the FT4 concentration is low (20 pmol/L) compared with the bound concentration (100 nmol/L). A 100-fold dilution causes depletion of the total bound hormone pool by 2%, implying a decrease of the ambient free hormone concentration in the diluted sample of the same order (16)(17).
We investigated the importance of adsorption in equilibrium dialysis. We found that loss by adsorption seems to cause, at most, a 2–3% (Table 2
) lower estimation of the FT4 concentration in equilibrium-based methods; this underestimation is likely less, because of adsorption of binding proteins, if the binding sites are not blocked or changed.
The advantage of ultrafiltration is that it can be performed with a minimum of dilution. Because the separation is not a equilibrium separation, the importance of loss by adsorption is uncertain. The adsorption did not exhibit any kind of saturation, and 13–34% of T4 in HEPES buffer (protein-free medium) was bound to the materials examined, probably to a pool of unspecific low-affinity binding sites on the membranes and the materials. It has been found that minimum adsorption of T4 on activated carbon occurs at pH 7–8. T4 is maximally charged at pH 7–8 according to the pKa values at 25 °C (pKa values: 
-COOH 2, -NH3+ 9, and 
-OH 7) (18). The adsorption of T4 to regenerated cellulose was substantially lower than the adsorption to cellulose ester at pH 7.4 (results not shown). The major difference between regenerated cellulose and cellulose ester is the hydrolyzed esters in regenerated cellulose, and the results seems to indicate that hydrophobic interactions are responsible for the adsorption of T4 to cellulose membranes (19)(20). Adsorption of T4 in serum to materials was substantially less than that of T4 in HEPES buffer: only 0.6–2% was bound to the different types of membranes. However, the percentage of serum T4 that adsorbed to material was high compared with the typical free fraction of T4 in normal plasma (0.02–0.03%). This indicates that during the filtration process, the surface of the membrane/material is coated with proteins and other serum ligands, including T4-binding proteins, because of the low binding of T4 in HEPES buffer, pH 7.4 (Fig. 1
). Therefore, it is difficult to find the contribution of adsorption to FT4 measurements.
Another problem arises from the fact that almost no proteins must leak during filtration, dialysis, and ultracentrifugation. This places high demands on the membrane materials and requires homogeneity between different lots to be used. Even small fractions of binding proteins in the ultrafiltrate, dialysate, or supernatant will induce an error in the estimation of FT4.
The three binding proteins are of similar size [albumin (66 kDa), TBG (54 kDa), and prealbumin (54 kDa)], and if it is assumed that:
 | (1) |
Eq. 1
shows that with cofiltration of all binding proteins, an albumin concentration in the ultrafiltrate (dialysate, supernatant) of 1667 pmol/L, a physiologically initial albumin concentration of 667 µmol/L, and values for total T4 of 100 nmol/L and for FT4 of 15 pmol/L, the FT4 result will be falsely increased by 2%. If only albumin passes through the membrane, the increase in FT4 will be 0.2%. However, the similarities of the three binding proteins, both physically and chemically [isoelectric point, Svedberg constant, and specific volume (21)], do not support this assumption. Weeke et al. (9) examined the correlation between T4 in the ultrafiltrate and albumin concentration in the same serum pool ultracentrifuged with increasing g force. They found that there was a linear correlation between albumin and T4 in the ultrafiltrate; they also found FT4 results that were falsely increased by 600%. The results obtained by Weeke et al. (9) strengthened the hypothesis that the other binding proteins (TBG and prealbumin) are cofiltered to the same extent and that 10% of the FT4 measurements were falsely high because the Visking tubing had been unable to retain proteins.
In a previous study testing different ultrafiltration devices and measuring albumin in the ultrafiltrate as a marker for thyroid-binding proteins, the limit of acceptance for albumin/proteins was set to <33 nmol/L (13). From Eq. 1
it can be seen that an albumin concentration in the ultrafiltrate of 33 nmol/L will give a falsely increased FT4 of 30% at physiologic conditions. In our view, such a high concentration of albumin (and the other binding proteins to the same extent) is not in accordance with suitable analytic performance goals. It has been recommended that bias in quantitative methods not exceed one-fourth of the group (within- plus between-subject) biologic variation expressed as CV (22). From the data of Browning et al. (23), Nelson and Wilcox (4) have calculated the acceptable bias for FT4 to be 4%. In our studies, we defined an albumin concentration <1667 pmol/L in the ultrafiltrate, dialysate, or supernatant as the maximum acceptable concentration and in reasonably good agreement with the analytic performance goals for acceptable bias for FT4 because it allows space for other sources that contribute to bias.
We found that separation of albumin (binding proteins) was very dependent on material type and lot in ultrafiltration. We were unable to separate FT4 from the bound fraction by ultracentrifugation. We found that dialysis was the best choice of the compared methods for separation of FT4 because all the measured dialysates were free from albumin.
An advantage of the ultracentrifugation process is that equilibrium separation does not take place around a membrane, which is the most important cause of loss by adsorption. It is therefore not very material dependent. However, ultracentrifugation of proteins can cause some disturbances of the equilibrium, and the equilibrium constant may change, e.g., the increased protein concentration at the bottom may cause steric hindrance of binding sites.
In conclusion, because of the different drawbacks and advantages, there is no perfect separation method that provides technically and theoretically correct separation of the free fraction of T4 from the protein-bound fraction. Equilibrium dialysis seems the least compromised method.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  B. Yue, A. L. Rockwood, T. Sandrock, S. L. La'ulu, M. M. Kushnir, and A. W. Meikle Free Thyroid Hormones in Serum by Direct Equilibrium Dialysis and Online Solid-Phase Extraction-Liquid Chromatography/Tandem Mass Spectrometry Clin. Chem., April 1, 2008; 54(4): 642 - 651. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. S. Fritz, R. B. Wilcox, and J. C. Nelson Quantifying Spurious Free T4 Results Attributable to Thyroxine-Binding Proteins in Serum Dialysates and Ultrafiltrates Clin. Chem., May 1, 2007; 53(5): 985 - 988. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Van Uytfanghe, D. Stockl, H A. Ross, and L. M. Thienpont Use of Frozen Sera for FT4 Standardization: Investigation by Equilibrium Dialysis Combined with Isotope Dilution-Mass Spectrometry and Immunoassay Clin. Chem., September 1, 2006; 52(9): 1817 - 1821. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Van Uytfanghe, D. Stockl, J. M. Kaufman, T. Fiers, H. A. Ross, A. P. De Leenheer, and L. M. Thienpont Evaluation of a Candidate Reference Measurement Procedure for Serum Free Testosterone Based on Ultrafiltration and Isotope Dilution-Gas Chromatography-Mass Spectrometry Clin. Chem., November 1, 2004; 50(11): 2101 - 2110. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, Visking tubing; 
, Centricon YM-10, filter part; +, equilibrium dialysis cell; 
, Centricon YM-3, filter part; 
, ultracentrifugation tubing.
