Specific immunoassay for protein dimers, trimers, and higher oligomers

Development and use of enzyme-linked immunosorbent assays (ELISA) for the detection of protein aggregates in interferon-alpha (IFN-alpha) formulations

PURPOSE

Protein aggregates are thought to be involved in the immunogenicity of recombinant proteins in humans. To probe human IFN-alpha formulations for the presence of soluble protein aggregates, enzyme-linked immunosorbent assays (ELISA) were developed.

METHODS

For the detection of IFN-alpha-IFN-alpha and HSA-IFN-alpha aggregates, sandwich ELISAs were developed using a monoclonal anti-IFN-alpha antibody as a capture antibody and the same anti-IFN-alpha antibody and an anti-human serum albumin (HSA) antibody (HRP-labeled), respectively.

RESULTS

Marketed freeze-dried, HSA-containing IFN-alpha-formulations tested in the ELISAs all contained IFN-alpha-IFN-alpha and/or HSA-IFN-alpha protein aggregates, although in varying amounts. These aggregates were predominantly IFN-alpha dimers and 1:1 conjugates of HSA with IFN-alpha. Test formulations revealed that aggregation of IFN-alpha was strongly affected by the presence of pharmaceutical excipients, pH of the formulation, lyophilisation procedure, and storage temperature and time.

CONCLUSIONS

The ELISAs are rapid, highly specific for aggregates in the presence of both IFN-alpha and HSA monomers and allow the direct detection of both types of aggregates in formulations in the nanogram range. The new assays will assist the monitoring of the aggregate-inducing processes during IFN-alpha formulation and storage in an early phase and the development of aggregate-free IFN-alpha formulations.

Refolding, isolation and characterization of crystallizable human interferon-alpha 8 expression in Saccharomyces cerevisiae

Human interferon-alpha 8 was expressed in Saccharomyces cerevisiae and found to accumulate intracellularly in an insoluble form. The protein could be solubilized and converted to a biologically active form with high yield by a denaturation-refolding procedure. The interferon-alpha 8 was further purified to apparent homogeneity by copper-chelate affinity chromatography and anion-exchange chromatography and fully characterized by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE), N-terminal sequence analysis, mass spectrometry, circular-dichroism (CD) spectroscopy and specific activity. Secondary-structure predictions from CD spectroscopy indicate that the molecule is correctly folded. Peptide mapping supported the correct sequence and the expected disulfide-bridge connectivity. The purified protein elutes on reversed-phase high-pressure liquid chromatography (RP-HPLC) as two peaks. Electrospray mass spectrometry and N-terminal sequence analysis of the minor component indicated the existence of an N-terminal acetyl group for the later eluting HPLC-component. In anti-viral assays, the two IFN forms were equally active. Hexagonal crystals of this interferon preparation could be obtained. On the basis of the electrophoretic mobility, HPLC profile, and biological activity assay, the crystalline material was judged to be identical to the uncrystallized interferon. Interferon in crystallized form was found to be stable for up to 24 months and, therefore, could be used for long-term storage, particularly for material intended for clinical use.

Human interferon-gamma quantified via a sensitive one-site monoclonal antibody in a sandwich ELISA. A PEG modification for measurement in serum

A new, specific and sensitive one-site ELISA for precise quantification of human interferon-gamma (HuIFN-gamma) at low levels in 50% human serum samples has been developed. The assay is based on the assumption that biologically active HuIFN-gamma is present exclusively as a dimer. Thus, in contrast to previous reports, the ELISA is based on a single monoclonal antibody (MAb) which is used in two ways: as "catching" antibody and as HRPO-labelled conjugate. The sensitivity could be improved five-fold by addition of (NH4)2SO4 to the conjugate solution; the lowest detectable amount of HuIFN-gamma is < 0.5 mu/ml. Non-specific interactions were not seen in interferon samples taken from cultures, or samples diluted in ordinary media or 1% BSA. However, > 30% of the (serum) samples gave non-specific false-positive results when the method was applied to samples containing 50-100% human serum from different donors. The false signals were related to the donors but could-at the expense of the sensitivity which was reduced to 1 mu/ml-be abolished by PEG treatment of the (donor) serum samples.

Generation and characterization of anti-idiotypic antibodies recognizing the interferon-alpha receptor: implications for ligand-receptor interactions

Monoclonal antibodies LI-1 and LI-8 against interferon-alpha A (IFN-alpha A) block IFN-alpha A activity and binding to its receptor, but they recognize distinct epitopes. Surprisingly, anti-idiotypic antibodies to both LI-1 and LI-8 have properties consistent with recognition of the receptor: anti-LI-1 and anti-LI-8 antibodies inhibit the binding of IFN-alpha A to its receptor. However, anti-LI-1 is an antagonist of IFN-alpha A, while anti-LI-8 is an agonist. Thus, at least some part of the epitopes on IFN-alpha A recognized by LI-1 and LI-8 are directly involved in receptor binding. Because these epitopes are spatially distinct, the implication is that the receptor binding site on IFN-alpha A must be extensive, or there are minimally two regions of IFN-alpha A involved in receptor interactions.

Monoclonal antibodies and enzyme immunoassays specific for human interferon (IFN) omega 1: evidence that IFN-omega 1 is a component of human leukocyte IFN

Four hybridoma cell lines secreting monoclonal IgG antibodies to human interferon (IFN) omega 1 (= IFN-alpha II 1) were developed, using spleen cells of mice immunized with IFN-omega 1 and/or a novel hybrid interferon, IFN-omega 1/alpha 2. All antibodies (OMG-2, -4, -5, and -7) neutralize the antiviral activity of IFN-omega 1 and show distinct patterns of reactivity with the hybrid proteins, IFN-omega 1/alpha 2 and IFN-alpha 2c/omega 1. However, none of the antibodies is able to neutralize human IFN-alpha, confirming earlier observations that IFN-omega 1 and IFN-alpha are antigenically unrelated. The epitope specificities of the antibodies were further characterized in direct and competitive enzyme immunoassays (ELISAs). All binary antibody combinations were tested for their suitability for a two-site ("sandwich") ELISA for IFN-omega 1, using horse radish peroxidase as the marker enzyme. A configuration employing OMG-2 for antigen capture and OMG-7 as the detector antibody resulted in the highest assay sensitivity (approximately 10 pg IFN-omega 1/ml) and was studied further. This one-step assay is highly specific for IFN-omega 1 and does not recognize human IFN-alpha, -beta, and -gamma, thus allowing for determination of IFN-omega 1 levels in natural mixtures of human IFNs. Using this ELISA, it was found that IFN-omega 1 is present in IFN preparations derived from virus-induced human peripheral blood leukocytes and may constitute as much as 15% of the total leukocyte IFN activity. IFN-omega 1 was also detected at somewhat lower levels in preparations of human "lymphoblastoid" IFN.

Stability of human interferon-beta 1: oligomeric human interferon-beta 1 is inactive but is reactivated by monomerization

Human interferon-beta 1 is extremely stable is a low ionic strength solution of pH 2 such as 10 mM HCl at 37 degrees C. However, the presence of 0.15 M NaCl led to a remarkable loss of antiviral activity. The molecular-sieve high-performance liquid chromatography revealed that, whereas completely active human interferon-beta 1 eluted as a 25 kDa species (monomeric form), the inactivated preparation eluted primarily as a 90 kDa species (oligomeric form). The specific activity (units per mg protein) of the oligomeric form was approx. 10% of that of the monomeric form. This observation shows that oligomeric human interferon-beta 1 is apparently in an inactive form. When the oligomeric eluate was resolved by polyacrylamide gel containing sodium dodecyl sulphate (SDS), it appeared to be monomeric under non-reducing conditions. Monomerization of the oligomeric human interferon-beta 1 by treatment with 1% SDS, fully regenerated its antiviral activity. These results suggest that the inactivation of the human interferon-beta 1 preparation was caused by its oligomerization via hydrophobic interactions without the formation of intermolecular disulphide bonds. These oligomers can be dissociated by SDS to restore biological activity.

Monoclonal antibodies as probes of the topological arrangement of the alpha subunits of Escherichia coli RNA polymerase

Three monoclonal anti-alpha antibodies were used to study the properties of the alpha subunit of Escherichia coli RNA polymerase. None of the monoclonal antibodies inhibited the d(A-T)n-directed synthesis of r(A-U)n. Reassembly of the RNA polymerase core was blocked by mAb 129C4 or mAb 126C6 while no effect was observed with mAb 124D1. The conversion of premature to mature core was partially inhibited by mAb 129C4 and almost totally inhibited by mAb 126C6. The data suggest that during the course of core assembly at least one of the alpha subunits undergoes conformational changes. The increase in affinity of mAb 126C6 for assembled alpha compared with free alpha also implies that alpha undergoes conformational changes during RNA polymerase assembly. Double antibody binding studies showed that the epitopes for mAb 124D1 and mAb 129C4 are available on only one of the alpha subunits in RNA polymerase. It would appear that the relevant domain on one of the alpha subunits in RNA polymerase is well exposed whereas this domain on the second alpha subunit is shielded by interaction with regions of the large beta and beta' subunits. The alpha domain in which the epitope for mAb 126C6 resides is not impeded by subunit interactions in the RNA polymerase. The data obtained also suggest that in the holoenzyme the sigma subunit may be positioned close to one of the alpha subunits, probably to the more exposed alpha. The alpha beta complex is the minimal stable subassembly since one of the alpha subunits dissociates from the alpha 2 beta complex following binding of any of the monoclonal antibodies studied.(ABSTRACT TRUNCATED AT 250 WORDS)

Escherichia coli-derived human interferon-gamma with Cys-Tyr-Cys at the N-terminus is partially N alpha-acylated

Purified preparations of recombinant human interferon-gamma (rIFN-gamma) with Cys-Tyr-Cys at the N-terminus ([ Cys-Tyr-Cys]IFN-gamma) derived from Escherichia coli gave two closely migrating bands on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and two peaks on reversed-phase high-performance liquid chromatography (rpHPLC). In contrast, rIFN-gamma without Cys-Tyr-Cys and rIFN-gamma in which both Cys-1 and Cys-3 were substituted with serine behaved as a single species on both SDS-PAGE and rpHPLC. These results suggest that the N-terminal portion of rIFN-gamma is heterogeneous. To elucidate the structure of the N-terminal portion, the N-terminal peptide preparation was obtained by binding rIFN-gamma to thiopropyl-Sepharose 6B gel with disulfide linkage followed by trypsin digestion and elution with 2-mercaptoethanol. The preparation gave four peaks (NT-1, NT-2, NT-3, and NT-4, in order of elution) on rpHPLC; all four were found to be Cys-1-Lys-9 by amino acid analysis after acid hydrolysis. Various analyses indicate that NT-1 is the intact nonapeptide, that NT-3 and NT-4 are N alpha-formyl and N alpha-acetyl forms of NT-1, respectively, and that NT-2 may be S-blocked at Cys-1. It is concluded that E. coli-derived [Cys-Tyr-Cys]IFN-gamma is partially N alpha-acylated. The data also suggest that N alpha-acylation does not affect the biological activity of [Cys-Tyr-Cys]IFN-gamma.

Normal human keratinocytes contain an interferon-like protein that may modulate their growth and differentiation

Epidermal growth and differentiation is a complex process which depends upon a balance between positive and negative growth signals, and in normal skin the majority of the cells in the germinative basal layer do not proliferate unless stimulated. Using the indirect immunofluorescent method, it can be demonstrated that purified polyclonal epidermis in cross sections of normal skin and to the basal layer of cultured keratinocyte colonies. Furthermore, extracts of keratinocyte cultures contain interferon bioactivity. With Western blot analysis, antibodies to interferon recognize a band of approximately 40 kD both in keratinocyte lanes and in recombinant interferon lanes that give in addition a band of approximately 20 kD. Addition of interferon to rapidly growing keratinocytes inhibits their growth by as much as 90% and promotes their terminal differentiation. The growth inhibitory effect of interferon is completely reversible. These data demonstrate that interferon or a closely related protein is present in human epidermis and suggest that this protein may act as a physiologic modulator of keratinocyte growth and differentiation.

Characterization of interferon-alpha binding sites on human cell lines

The binding sites for human interferon-alpha (IFN-alpha) have been characterized on human lymphoblastoid, melanoma, rhabdomyosarcoma, and cervical carcinoma cells. Crosslinking of iodinated-recombinant DNA-derived IFN-alpha-Con1, an analog of the known IFN-alpha subtypes, to the cell surface with disuccinimidyl suberate yielded four IFN-receptor complexes of 118, 138, 159, and 260 kD on all cell lines that specifically bind IFN-alpha. Since IFN-alpha exists in solution as monomers, dimers, and trimers, and the three lower molecular weight IFN-alpha-receptor complexes differ by the molecular weight of IFN-alpha (20 kD), this suggests that the human IFN-alpha receptor of 100 kD binds more than one molecule of IFN-alpha. The higher molecular weight complex of 260 kD may result from dimerization of the receptor. None of these complexes was observed in a rhabdomyosarcoma subclone that does not specifically bind IFN-alpha. Pretreatment of cells with trypsin abolished the formation of these complexes. Pretreatment of cells with neuraminidase did not reduce IFN-alpha binding, but increased the electrophoretic mobility of all four IFN-alpha-receptor complexes. Other glycosidases (i.e., mannosidase, beta-galactosidase, and endoglycosidase F) had no effects on IFN-alpha binding or mobility of complexes. Thus, although the IFN-alpha receptor is a glycoprotein, the glycosylated portion is apparently not part of the IFN-alpha-binding domain. The formation of IFN-alpha-receptor complexes is independent of the duration of incubation with IFN (from 5 min to 1 h at 15 degrees C).(ABSTRACT TRUNCATED AT 250 WORDS)

Large-scale chromatography of recombinant proteins

When recombinant proteins are expressed in bacterial cells and subsequently grown in fermentation tanks, there remains the problem of recovering the product in pure form. The empirical knowledge gained upon recovery of recombinant proteins indicates that a one-step purification process is very unlikely to succeed. However, combinations of modern techniques, such as immunoaffinity chromatography or immobilized-metal affinity chromatography, with classical techniques, such as ion-exchange chromatography, seem to be suitable for large-scale recovery of recombinant proteins.

Specificity of the immunoadsorbent used for large-scale recovery of interferon alpha-2a

After immunoaffinity chromatography, interferon alpha-2a synthesized in bacteria is not homogeneous. Beside oligomers and monomers, a fragment of molecular weight 15,000 was observed. Amino acid analysis and the determination of the amino terminal amino acid sequence of the fragment indicate that this polypeptide represents an interferon alpha-2a molecule lacking the 22 terminal amino acids. In order to define the epitope on the interferon alpha-2a molecule recognized by the immunoadsorbent, the binding to the immunoadsorbent of interferon alpha-2a fragments prepared by cyanogen bromide cleavage was studied. The results suggest that cyanogen bromide fragments Arg 22-Met 59 and Lys 112-Met 148 are recognized. Since the oligomers, the monomers and the fragment of molecular weight 15,000 of interferon alpha-2a share these sequences, the different forms are as expected co-purified on the immunoadsorbent.

Rapid dual-column chromatographic assay for recombinant leukocyte interferon alpha-2

A rapid dual-column chromatographic assay for determining recombinant leukocyte interferon alpha-2 in complex mixtures is described. The assay relies on the use of a high-performance monoclonal antibody affinity column connected in tandem with a reversed-phase column. The high specificity and selectivity of these columns permits the quantitation of subcomponent species, such as interferon oligomers that may be present in assay samples. The assay has a limit of sensitivity equal to 1 microgram/g over a range of 1-20 micrograms/g. The precision of the assay was estimated to be about 5%.

Cloned human interferons alpha: differential affinities for polyinosinic acid and relationship between molecular structure and species specificity

The HuIFN-alpha A and HuIFN-alpha D interferons, produced by two independent recombinant bacterial clones, have different affinities for polyinosinic acid (poly I). The monomeric form HuIFN-alpha A (FMM), but not the HuIFN-alpha D, binds to poly (I)-agarose and is protected by poly (I) from thermal inactivation. Other subtypes of HuIFN-alpha A including the monomer SMM and oligomers have no affinity for this polynucleotide. In addition, these interferons show different target cell preferences in agreement with our previous suggestion (23) that the polynucleotide binding domain may be responsible for species specificity. Two significant observations are 1) the fractions of HuIFN-alpha D and HuIFN-alpha A unbound on poly (I)-agarose show higher antiviral inducing activity on heterologous (MDBK) than on homologous (WISH) cells, whereas they induce about the same activity of 2'5' oligoadenylate synthetase in these two cell lines. These fractions are also active on L929 cells. 2) The bound fraction of HuIFN-alpha A induces almost the same antiviral and 2'5' oligoadenylate synthetase activities in MDBK and in WISH cells but neither activity in L929 cells.

Multiple molecular forms of interferon display different specific activities in the induction of the antiviral state and 2'5' oligoadenylate synthetase

Both Hu IFN-alpha A and Hu IFN-alpha D, produced by two independent recombinant bacterial clones, are mixtures of monomers, dimers and trimers. These forms, when assayed individually in heterologous MDBK cells, induced different degree of antiviral and 2'5' oligoadenylate synthetase (2'5' A synthetase) activities: the antiviral activity of the monomer is greater than that of the dimer and the trimer, whereas the activity of 2'5' A synthetase induction is lower with the monomer than with the dimer or the trimer. Similar differences are also observed on human cells. Compared to the mononeric form, the dimeric and the trimeric forms of Hu IFN-alpha A show higher antiviral inducing activity on heterologous MDBK cells than on homologous WISH cells, whereas the 2'5' A synthetase inducing activity in these two cell lines is about the same. Thus for the same antiviral activity, the trimer or the dimer compared to the monomer are much better inducers of the 2'5' A synthetase on human than on MDBK cells.

Normal human epidermis contains an interferon-like protein

Interferons have been postulated to participate in growth regulation of normal body tissues and are known to inhibit growth of human epidermal keratinocytes in vitro. Polyclonal antibodies to recombinant human interferon-alpha, purified by passage over an affinity column (Sepharose coupled to the recombinant interferon), used in the indirect immunofluorescent method specifically stained the proliferative (basal) compartment of human epidermis in histological cross-sections of normal skin and in cultured keratinocyte colonies. Extracts prepared from healthy nonvirally infected keratinocyte cultures contained interferon activity as determined by viral plaque inhibition assay. Using the Western blotting technique column-purified antibodies and antisera to recombinant human interferon-alpha recognized a band of approximately 40 kD when reacted with both extracted keratinocyte proteins and recombinant human interferon-alpha standards, that gave in addition a band of approximately 20 kD. The above findings suggest that interferon or a closely related protein is present in the proliferative compartment of normal epidermis in the absence of viral infection and therefore may serve as a physiological modulator of epidermal growth.

Chemical modifications of human and murine interferons

Methods that modify the carbohydrate and peptide portions of human and murine interferons have been described. Such work has helped considerably in our understanding of these fascinating substances. The most powerful of the methods relies on gene manipulation: it has already led, and will lead in the future, to further species with altered biological activities. Some of them may prove clinically advantageous, but one will have to be aware of the fact that some of these "unnatural" interferons could lead to immunological complications.

Analysis of different forms of recombinant human leukocyte interferons and synthetic fragments by high-performance liquid chromatography

Analytical high-performance liquid chromatography (HPLC) conditions are reported for the evaluation of recombinant human interferon monomers: recombinant human leukocyte interferon A (rIFN-alpha A), rIFN-alpha D, and the hybrid rIFN-alpha A1-62/D64-166. The two monomeric forms of rIFN-alpha A (slow-migrating monomer and fast-migrating monomer) were also resolved by HPLC. Conditions are reported for the HPLC separation of oligomers (dimers and trimers) of rIFN-alpha A. The synthesis and analytical HPLC of the carboxy-terminus fragment, corresponding to IFN-alpha D (140-166), and a series of analogues comprising the IFN-alpha A (105-125) region is reported. The syntheses were accomplished by the solid-phase peptide synthesis procedure and the products were purified by preparative HPLC.

Monoclonal antibodies as structural probes for oligomeric human interferon-gamma

Monoclonal antibodies (MAb) B1 and B3, specific for human interferon-gamma (IFN-gamma) failed to immunoprecipitate heat-inactivated human IFN-gamma in solution. However, both MAb retained some reactivity with denatured IFN-gamma immobilized on vinyl plates. The two MAb have been employed in a sensitive immunoradiometric assay (IRMA). In this IRMA one MAb was bound to polystyrene beads and used as immunoadsorbent. The second MAb, labeled with 125I, was used as the tracer to quantitate the amount of IFN-gamma bound to the immobilized MAb. Addition of unlabeled MAb B1 did not inhibit the binding of 125I-labeled MAb B3 (and vice versa), indicating that the two MAb react with two different and nonoverlapping epitopes. Yet, when the same MAb was used in IRMA as both immunoadsorbent and tracer, the amount of labeled MAb bound to a given concentration of natural or E. coli-derived recombinant human IFN-gamma was very similar as with two different MAb, indicating that a single IFN-gamma molecule must have two or more identical binding sites for each of the two MAb. These findings show that biologically active natural and recombinant human IFN-gamma exist in oligomeric form.