Human IgG Molecules Research Articles

N‐linked glycosylation is an important parameter for optimal selection of cell lines producing biopharmaceutical human IgG

We studied the variations in N-linked glycosylation of human IgG molecules derived from 105 different stable cell lines each expressing one of the six different antibodies. Antibody expression was based on glutamine synthetase selection technology in suspension growing CHO-K1SV cells. The glycans detected on the Fc fragment were mainly of the core-fucosylated complex type containing zero or one galactose and little to no sialic acid. The glycosylation was highly consistent for the same cell line when grown multiple times, indicating the robustness of the production and glycan analysis procedure. However, a twofold to threefold difference was observed in the level of galactosylation and/or non-core-fucosylation between the 105 different cell lines, suggesting clone-to-clone variation. These differences may change the Fc-mediated effector functions by such antibodies. Large variation was also observed in the oligomannose-5 glycan content, which, when present, may lead to undesired rapid clearance of the antibody in vivo. Statistically significant differences were noticed between the various glycan parameters for the six different antibodies, indicating that the variable domains and/or light chain isotype influence Fc glycosylation. The glycosylation altered when batch production in shaker was changed to fed-batch production in bioreactor, but was consistent again when the process was scaled from 400 to 5,000 L. Taken together, the observed clone-to-clone glycosylation variation but batch-to-batch consistency provides a rationale for selection of optimal production cell lines for large-scale manufacturing of biopharmaceutical human IgG.

N-Glycosylation as Novel Strategy to Improve Pharmacokinetic Properties of Bispecific Single-chain Diabodies

The therapeutic efficacy of recombinant antibodies such as single-chain Fv fragments and small bispecific or bifunctional molecules is often limited by rapid elimination from the circulation because of their small size. Here, we have investigated the effects of N-glycosylation on the activity and pharmacokinetics of a small bispecific single-chain diabody (scDb CEACD3) developed for the retargeting of cytotoxic T cells to CEA-expressing tumor cells. We could show that the introduction of N-glycosylation sequons into the flanking linker and a C-terminal extension results in the production of N-glycosylated molecules after expression in transfected HEK293 cells. N-Glycosylated scDb variants possessing 3, 6, or 9 N-glycosylation sites, respectively, retained antigen binding activity and bispecificity for target and effector cells as shown in a target cell-dependent IL-2 release assay, although activity was reduced approximately 3-5-fold compared with the unmodified scDb. All N-glycosylated scDb variants exhibited a prolonged circulation time compared with scDb, leading to a 2-3-fold increase of the area under curve (AUC). In comparison, conjugation of a branched 40-kDa PEG chain increased AUC by a factor of 10.6, while a chimeric anti-CEA IgG1 molecule had the longest circulation time with a 17-fold increase in AUC. Thus, N-glycosylation complements the repertoire of strategies to modulate pharmacokinetics of small recombinant antibody molecules by an approach that moderately prolongs circulation time.

Presence of IgG-CD4 Complexes in the Circulation

Many proteins of various origins are able to form complexes with Immunoglobulin G using reaction sites formed by residues of constant domains. Studies of IgG complexes of non-immune nature are important as biologically active proteins could escape from the circulation forming complexes with IgG as well as with other serum proteins whose concentration in serum is high. A quantitative characterization of circulating complexes in various diseases could give valuable information on the development of pathological processes. Immunoglobulins G are widely used for the treatment of a number of diseases and it is essential to understand what proteins are present in the preparations beside IgG. In the present study CD4 T-cell membrane glycoprotein was found in complexes with human IgG molecules isolated from donor sera as well as from sera of SLE patients via a sensitive quantitative dot-blot assay. According to immunoblotting experiments the CD4 part of the complexes had a molecular mass of about 50 kDa and is composed from all four extracellular domains. The CD4 content of the complexes varied among the studied human sera. There was no difference in IgG-CD4 complex concentration between the SLE patients and healthy controls. The data support the assumption that IgG molecules are able to act as scavengers and eliminate various proteins from the circulation including soluble CD4 protein.

Contrasting glycosylation profiles between Fab and Fc of a human IgG protein studied by electrospray ionization mass spectrometry

A conserved structural feature of human IgG molecules is the presence of an oligosaccharide moiety within the Fc region at Asn297. In addition, 15–20% of normal polyclonal IgG molecules bear N-linked oligosaccharides in the variable (V) regions of the light (L) and/or heavy (H) chains. Electrospray ionization mass spectrometry (ESI-MS) has been applied to the glycan analysis of two IgG1 myeloma proteins (Wid and Cri) after mild reduction and acidification. Heterogeneous ion peaks were observed for both the H and L chains of Wid in contrast to Cri whose L chain peak was homogeneous. Site-specific deglycosylation of the H and L chains of IgG Wid was achieved under native conditions with peptide- N-glycosidase F and endoglycosidase F2, respectively. The Fc glycoforms differed between the two proteins in that Cri-Fc bears diantennary complex-type glycans that are fully core-fucosylated and partially sialylated while Wid-Fc glycans are non-fucosylated, partially galactosylated and non-sialylated. In contrast to the Fc glycans, the L chain glycans of Wid were shown to be fucosylated, fully galactosylated and sialylated, indicating that the glycosylation machinery of the Wid-producing myeloma cells is intact. Thus, combination of the two endoglycosidases can provide a simple means of glycan analysis of both Fab and Fc by ESI-MS, which may contribute to the development of therapeutic IgG with customized glycan profiles.

Characterization of a novel modification on IgG2 light chain: Evidence for the presence of O-linked mannosylation

This paper describes the analysis of a novel modification identified on the light chain of a recombinant IgG2 antibody. This modification, a +162 Da adduct, suggestive of a single hexose addition, was observed by mass analysis of the reduced molecule. The modification was located on residue serine 66 of the light chain by investigation with LC–MS peptide mapping, mass spectrometry and N-terminal sequencing techniques. Location of the adduct on serine pointed the investigation toward O-linked glycosylation. Identification of the hexose residue was deduced from its elimination by action of α-mannosidase, providing evidence for the presence of an O-mannosylated light chain. This type of modification in the glycosylation profile of antibodies, to our knowledge, has not been reported for human IgG molecules.

Quantification of human immunoglobulin G immobilized on gold-coated silicon chip for biosensing applications

The most important aspect of biosensor development with high sensitivity is the oriented immobilization of antibodies on the solid substrate. Human IgG was immobilized on gold-coated silicon employing protein A, protein G and neutravidin immobilization procedures. The amount of human IgG immobilized was analyzed by 3, 3′, 4, 4′- tetramethyl benzidine (TMB) substrate assay and was maximum when protein A immobilization procedure was followed. Human IgG coated biosensing surface was regenerated by treatment with glycine-HCl buffer (50 mM, pH 2.2). Atomic force microscopy was used to analyze the distribution of biomolecules immobilized on gold-coated silicon. Our study indicates that human IgG molecules were uniformly bound to gold-coated silicon by protein A immobilization procedure.

A mutant human IgG molecule with only one C1q binding site can activate complement and induce lysis of target cells

There are potentially two binding sites for C1q on IgG, one on each C(H)2 domain of the gamma heavy chains, close to the lower hinge region. It is not clear whether the presence and involvement of both the C1q binding sites is necessary to induce the activation signal of human IgG. In order to clarify this issue, we made a hybrid mutant IgG1/IgG3 molecule where the IgG1 half of the molecule was made unable to activate complement through the introduction of a P329A mutation. The IgG3 half of the molecule was mutated to harbor a hinge region identical to that of IgG1, and for detection a peptide tag derived from p21ras was introduced into the FG loop of the C(H)1 domain. The hybrid IgG1P329A/IgG3h1-ras molecules were isolated by Protein A affinity chromatography and shown to activate complement and induce complement-mediated lysis at the same levels as wild-type IgG1 and IgG3h1-ras molecules. Thus, one C1q binding site per IgG is sufficient to induce activation. Wild-type human IgG molecules might also normally expose only one C1q binding site as already shown for interaction with FcgammaR, were IgG expose one binding site per molecule.

Preparation and optical scattering characterization of gold nanorods and their application to a dot-immunogold assay

We describe optical monitoring of the synthesis of gold nanorods (NRs) based on seed-mediated growth in the presence of the soft surfactant template cetyltrimethyilammonium bromide. To separate NRs from spheres and surfactants we fractionated samples in the density gradient of glycerol. The optical properties of NRs were characterized by extinction and differential light-scattering spectra (at 90 degrees, 450-800 nm) and by the depolarization light-scattering ratio, I(vh)/I(vv), measured at 90 degrees with a helium-neon laser. Theoretical spectra and the I(vh)/I(vv) ratios were calculated by the T-matrix method as applied to randomly oriented NRs, which were modeled by right-circular cylinders with semispherical ends. The simulated data were fitted to experimental observations by use of particle length and width as adjustable parameters, which were close to the data yielded by transmission electron microscopy. The sensitivity of the long-wavelength resonance of NRs to the dielectric surroundings was examined both experimentally and theoretically by comparison of the extinction spectra of NRs in water and in a 25% glycerol solution. Finally, we discuss the application of NR-protein A conjugates to a dot-immunogold assay with the example of biospecific staining of human IgG molecules adsorbed onto small membrane spots.

Immunoprecipitation for Rapid Detection of Macroprolactin in the Form of Prolactin–Immunoglobulin Complexes

The 3 major forms of serum prolactin (PRL), identifiable by gel-filtration chromatography (GFC), are monomeric PRL (23 kDa), big PRL (45–60 kDa), and big, big PRL or macroprolactin (150–170 kDa) (1). The common macroprolactin is PRL complexed with human IgG, but aggregates of PRL, some extensively glycosylated, may also occur (2)(3)(4)(5)(6). Macroprolactin has slower serum clearance than monomeric PRL and little or no apparent in vivo bioactivity, probably because it does not cross capillary walls (7)(8). Macroprolactin causes diagnostic confusion in evaluating hyperprolactinemia (9)(10). Polyethylene glycol (PEG) precipitation is a rapid screening method for macroprolactinemia (11)(12)(13)(14)(15)(16)(17), but PEG interference with PRL assays limits its general use. The method, moreover, shows nonspecific precipitation (up to 15%), necessitating use of a gray zone. We recently validated a screening method based on recognition of the IgG component of macroprolactin by anti-IgG-agarose (18); this method, however, had insufficient PRL assay sensitivity for use in moderate hyperprolactinemia (PRL <1000 mIU/L). A more rapid screening method based on PRL-IgG precipitation with protein A-Sepharose was described recently (19). We report a simple, rapid method, also based on precipitation of PRL-IgG complexes, with protein G-agarose (PGA) suspension and high IgG binding capacity (20 mg/mL of resin). The protein G polypeptide binds the Fc region of all subclasses of the human IgG molecule and also binds the IgG3 fraction (20). Results with this method were compared with those from GFC and PRL-PEG precipitation. For GFC, we used the fast FPLC system (Pharmacia) with a Superdex 200 HR10/300 prepacked column (Amersham Pharmacia) (11). Serum (100 μL) was applied and eluted with phosphate-buffered saline (PBS; 0.05 mol/L, pH 7.0) at a flow rate of …

Detection and quantification of the human IgG4 half‐molecule, HL, from unpurified cell‐culture supernatants

The presence and absence of inter-heavy-chain disulphide linkages contribute to the existence of the tetrameric (H(2)L(2)) and half (HL) human IgG molecules, respectively. Reduced effector response in the human IgG4 subclass presents an alternative therapeutic platform in a monoclonal-antibody (mAb) development program. During the initial cell-selection stage, titres of the recombinant human antibody present in crude cell-culture supernatants are determined by ELISA, a technique requiring nanogram quantities of mAb. In the case of an IgG4 antibody, this material is represented mainly by the combination of the tetrameric (H(2)L(2)) and dimeric (HL) forms of the antibody. The determination of concentrations or ratios of tetramer and dimer usually requires at least one chromatographic purification step, and thus frequently this is evaluated later in the mAb development process when the number of potential clones has been reduced. In the present paper we describe a Western-blot-based method that detects and quantifies IgG4 half-molecules, HL, from crude cell-culture supernatants without purification so that H(2)L(2)/HL ratios can be included as a part of early clonal evaluation along with the screening of mAb titres. This method was demonstrated (1) to have a linear HL detection range of 0.5-10 ng, (2) to require microlitre volumes of culture and (3) to react specifically with human IgG4 produced from hybridoma and Chinese-hamster ovary cell cultures. Moreover, this protocol is applicable to evaluate and monitor potential H(2)L(2)/HL variations as a result of changes during the process-development stage of a mAb development program.

Baboon immunoglobulin constant region heavy chains: identification of four IGHG genes.

The increasing use of nonhuman primate models in biomedical research and especially in vaccine development requires the characterization of their immunoglobulin genes and corresponding products. Therefore, we sequenced, cloned and characterized the four immunoglobulin gamma chain constant region genes ( IGHG) present in baboons. These four genes were designated IGHG1, IGHG2, IGHG3 and IGHG4 on the basis of sequence similarities with the four human genes encoding the IgG1, IgG2, IgG3 and IgG4 subclasses and the three known rhesus macaque IGHG genes. Specifically, the baboon IgG1, IgG2, IgG3 and IgG4 sequences exhibit 90.3%, 88.3%, 86.7% and 89.6% amino acid identity to their human counterpart. The percent of amino acid identity of baboon IgG1, IgG2 and IgG3 to the corresponding rhesus macaque sequences is 98.5, 93.1 and 94.4, respectively. Therefore, baboon and rhesus macaque IGHG genes are highly homologous to each other. The majority of differences existing between baboon and human sequences are clustered in the hinge region, with the upper hinge being the most diverse and containing several proline residues. Similar to rhesus macaques, the hinge regions of all baboon IGHG genes consist of a single exon, whereas in humans the IgG3 molecule is encoded by multiple exons. These results confirm the evolutionary instability of the hinge region and indicate that functional properties associated with the hinge regions of baboon and human IgG molecules might differ between the two species.

Differential Binding of Histidine-rich Glycoprotein (HRG) to Human IgG Subclasses and IgG Molecules Containing κ and λ Light Chains

In previous studies we showed that the plasma protein histidine-rich glycoprotein (HRG) binds strongly to pooled human IgG. In the present work myeloma proteins consisting of different human IgG subclasses were examined for their ability to interact with human HRG. Using an IAsys optical biosensor we found initially that IgG subclasses differ substantially in their affinity of interaction with HRG. However, the most striking finding was the observation that the kinetics of the HRG interaction was dramatically affected by whether the IgG subclasses contained the kappa or lambda light (L)-chains. Thus, the on-rate for the binding of HRG to the kappa L-chain containing IgG1 and IgG2 (IgG1kappa and IgG2kappa) was approximately 4- and approximately 10-fold faster than that for the binding of HRG to lambda L-chain containing IgG1 and IgG2 (IgG1lambda and IgG2lambda), respectively, with the dissociation constants (K(d)) in the range 3-5 nM and 112-189 nM for the kappa and lambda isoforms, respectively. In contrast, the on-rate for the binding of HRG to IgG3kappa and IgG4kappa was found to be 9- and 20-fold slower than that for the binding of HRG to IgG3lambda and IgG4lambda, respectively, with the K(d) in the range 147-268 nM and 96-109 nM for the kappa and lambda isoforms, respectively. The binding of HRG to immunoglobulins containing the kappa L-chain (particularly IgG1kappa) was generally potentiated in the presence of a physiological concentration (20 microM) of Zn(2+) (K(d) decreased to 0.60 +/- 0.01 for IgG1kappa), but Zn(2+) had no effect or slightly inhibited the binding of HRG to immobilized IgG subclasses possessing the lambda L-chain. Interestingly, HRG also bound differentially to Bence Jones (BJ) proteins containing kappa and lambda L-chains, with HRG having a 14-fold lower K(d) for BJkappa than for BJlambda when 20 microM Zn(2+) was present. HRG also bound to IgM (IgMkappa), but the affinity of this interaction (K(d) approximately 1.99 +/- 0.05 microM) was markedly lower than the interaction with IgG, and the affinity was actually decreased 4-fold in the presence of Zn(2+). The results demonstrate that both the heavy (H)- and L-chain type have a profound effect on the binding of HRG to different IgG subclasses and provide the first evidence of a functional difference between the kappa and lambda L-chains of immunoglobulins.

Recombinant human IgG molecules lacking Fcgamma receptor I binding and monocyte triggering activities.

Subclasses of human IgG have a range of activity levels with different effector systems but each triggers at least one mechanism of cell destruction. We are aiming to engineer non-destructive human IgG constant regions for therapeutic applications where depletion of cells bearing the target antigen is undesirable. The attributes required are a lack of killing via Fcgamma receptors (R) and complement but retention of neonatal FcR binding to maintain placental transport and the prolonged half-life of IgG. Eight variants of human IgG constant regions were made with anti-RhD and CD52 specificities. The mutations, in one or two key regions of the CH2 domain, were restricted to incorporation of motifs from other subclasses to minimize potential immunogenicity. IgG2 residues at positions 233 - 236, substituted into IgG1 and IgG4, reduced binding to FcgammaRI by 10(4)-fold and eliminated the human monocyte response to antibody-sensitized red blood cells, resulting in antibodies which blocked the functions of active antibodies. If glycine 236, which is deleted in IgG2, was restored to the IgG1 and IgG4 mutants, low levels of activity were observed. Introduction of the IgG4 residues at positions 327, 330 and 331 of IgG1 and IgG2 had no effect on FcgammaRI binding but caused a small decrease in monocyte triggering.

IgA class antibodies and flow cytometric cross-matching in renal transplantation.

The established method of pretransplant cross-matching does not detect IgA antibodies, and IgA antibodies have thus been ignored when assessing patients for transplantation. The aim of this study was to detect IgA allo- and autoreactive antibodies using flow cytometry and to correlate the results with transplant outcome. Pretransplant sera from 231 sequential renal recipients were tested for serum IgA levels and antibodies directed against the Fab portion of the human IgG molecule. Fifty-nine recipients with sufficient stored donor lymphocytes were also tested by flow cytometry for donor-specific alloantibodies of the IgA isotype. Graft survival was improved in recipients with higher IgA levels. High IgA anti-Fab levels led to a significantly higher 1-year graft survival (P<0.05). Graft survival was further enhanced where both serum IgA and IgA anti-Fab were raised (P<0.01). Although the mean IgA level tended to be higher for recipients with a positive IgA flow cytometric cross-match (FCXM), the IgA FCXM was not associated with increased IgA anti-Fab, suggesting that the IgA FCXM is detecting a different subset of IgA reactivity. Additionally, for primary grafts, a positive IgA FCXM was not associated with enhanced graft survival. Within the repertoire of IgA activity, there are two recognizable groups, the IgA anti-Fab specificity, which is significantly associated with enhanced graft survival, and that detected by the IgA FCXM, which surprisingly is more likely to be positive in less sensitized first grafts and is not associated with enhanced graft survival.

Properties of Whole Human IgG Molecules Produced by the Expression of Cloned Anti-DNA Antibody cDNA in Mammalian Cells

Antibodies to DNA are believed to play an important role in systemic lupus erythematosus (SLE). High affinity IgG antibodies which show marked specificity for double stranded DNA (dsDNA) are particularly closely linked to the occurrence and severity of tissue damage. Sequence analysis of mouse and human monoclonal antibodies has previously suggested that mutations in the complementarity determining regions (CDRs) play a major role in determining these binding properties. In many cases such mutations increase the overall number of basic residues in the CDRs. To further elucidate the role played by such mutations it is important to develop methods of expressing cloned autoantibody cDNA in the form of functional whole immunoglobulin molecules. We describe a system in which autoantibody VHand VLcDNA from monoclonal human anti-DNA antibodies, B3 and WRI176 were cloned into separate vectors which allowed their expression as whole heavy and whole light chains respectively. By cotransfecting mammalian cells with pairs of heavy and light chain vectors it was possible to produce whole IgG molecules from each of the four possible VH/VLcombinations. Only antibody produced by homologous VHand VLpairs bound DNA, suggesting that in these autoantibodies both chains are important in conferring this property.

Protein targeting in the analysis of learning and memory: a potential alternative to gene targeting.

Gene targeting using homologous recombination in embryonic stem (ES) cells offers unprecedented precision with which one may manipulate single genes and investigate the in vivo effects of defined mutations in the mouse. Geneticists argue that this technique abrogates the lack of highly specific pharmacological tools in the study of brain function and behavior. However, by now it has become clear that gene targeting has some limitations too. One problem is spatial and temporal specificity of the generated mutation, which may appear in multiple brain regions or even in other organs and may also be present throughout development, giving rise to complex, secondary phenotypical alterations. This may be a disadvantage in the functional analysis of a number of genes associated with learning and memory processes. For example, several proteins, including neurotrophins--cell-adhesion molecules--and protein kinases, that play a significant developmental role have recently been suggested to be also involved in neural and behavioral plasticity. Knocking out genes of such proteins may lead to developmental alterations or even embryonic lethality in the mouse, making it difficult to study their function in neural plasticity, learning, and memory. Therefore, alternative strategies to gene targeting may be needed. Here, we suggest a potentially useful in vivo strategy based on systemic application of immunoadhesins, genetically engineered fusion proteins possessing the Fc portion of the human IgG molecule and, for example, a binding domain of a receptor of interest. These proteins are stable in vivo and exhibit high binding specificity and affinity for the endogenous ligand of the receptor, but lack the ability to signal. Thus, if delivered to the brain, immunoadhesins may specifically block signalling of the receptor of interest. Using osmotic minipumps, the protein can be infused in a localized region of the brain for a specified period of time (days or weeks). Thus, the location and timing of delivery are controlled. Here, we present methodological details of this novel approach and argue that infusion of immunoadhesins will be useful for studying the role particular receptors play in behavioral and neural plasticity.

Enhancement of the sensitivity of surface plasmon resonance biosensor with colloidal gold labeling technique

In our experiment colloidal gold was used to enhance the signal obtained from the surface-plasmonresonance biosensor. Fab′ fragments of human IgG molecules were immobilized on the gold surface and reacted with the sheep-anti-human IgG(SAH-IgG) molecules in solution. It was found that when the colloidal gold labeling technique was introduced the sensitivity of surface plasmon resonance instruments increased by a factor of about 300.

Determination of anti-idiotype antibodies by surface plasmon resonance

A method based on surface plasmon resonance has been used to detect and to quantitate antibodies against other antibodies. This method has been applied to the detection of syngeneic antibodies. Concentrations as low as 0.3 μg/ml could be detected even when the antibodies were present in undiluted serum. Quantitation of both monoclonal and polyclonal antibodies could be performed. Anti-idiotype antibodies could be detected equally well against an IgM molecule derived directly from the B cell tumour or against a chimeric protein composed of the same idiotypic determinants embedded in a human IgG molecule. Surface plasmon resonance will be the method of choice for the detection of antibody responses in humans vaccinated against idiotypic determinants of their own tumour.

Direct demonstration of MuSK involvement in acetylcholine receptor clustering through identification of agonist ScFv.

MuSK is a tyrosine kinase localized to the postsynaptic surface of the neuromuscular junction. We have searched for modulators of MuSK function using a library of human single chain variable region antibodies (scFv) that can be displayed on M13 phage or expressed as soluble protein. A panel of 21 independent MuSK-specific scFv, identified in a screen for binding to MuSK-Fc immunoadhesin, were examined for ability to induce proliferation in a factor dependent cell line (Ba/F3) through a chimeric receptor, MuSK-Mpl. Four of the scFv induced a proliferative response, suggesting an ability to induce dimerization of MuSK. These scFv were also able to induce tyrosine phosphorylation of full-length MuSK and retained this ability when re-engineered to be expressed as authentic (and dimeric) human IgG molecules. Addition of agonist scFv to a cultured myotube cell line induced AChR clustering and tyrosine phosphorylation. These results provide direct evidence that MuSK activation is capable of triggering a key event in neuromuscular junction formation and further demonstrate that large libraries of phage-displayed scFv provide a robust method for generating highly specific agonist agents.

Cell-dislodging methods under serum-free conditions.

In this work, a BHK21 clone producing a fusion protein consisting of a recombinant human IgG molecule with a cytokine tail, growing in a protein-free medium, was used to test several alternatives to avoid the use of serum for trypsin inactivation, currently used in cell dislodging. These included (1) trypsin inactivated with soybean trypsin inhibitor (STI); (2) cell dissociation solution instead of trypsin; (3) dispase instead of trypsin; (4) trypsin inactivated with fetal calf serum (positive control); (5) non-inactivated trypsin (negative control). Use of a centrifugation step was also tested for each alternative. Results indicate that the best method regarding cell growth, viability and adherent fraction is to use trypsin inactivated with STI followed by a centrifugation step. For all methods tested, the utilization of a centrifugation step always led to improved results. The optimal proportion for total trypsin inactivation is 1:1 trypsin (0.2% w/v) to STI (1 mg ml-1). equivalent to 2 mg trypsin to 1 mg STI. No toxic effect was observed for STI at the concentrations used. Long-term subculturing with this new, alternative dislodging method did not affect cell growth, viability and productivity.