PEDS Advance Access originally published online on April 12, 2007
Protein Engineering Design and Selection 2007 20(5):235-241; doi:10.1093/protein/gzm016
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Chimeric protein for selective cell attachment onto cellulosic substrates
1 Graduate School of Biomedical Engineering 2 School of Biotechnology and Biomolecular Sciences, University of New South Wales, Australia
3 To whom correspondence should be addressed. E-mail: r.nordon{at}unsw.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We have developed a fusion protein (CBD-LG) incorporating a cellulose-binding domain and an antibody binding domain, protein LG, to provide an adaptor molecule for cell separation with regenerated cellulose hollow fiber arrays. A single hollow fiber cell adhesion assay utilizing a CD34+ cell line, KG1a, was used to investigate whether ligand affinity interactions were strong enough for cell attachment and separation. CBD-LG efficiently captured CD34+ cells labeled with the mouse IgG2a monoclonal antibody MHCD3400. However, it was not possible to bind CD34+ cells labeled with an IgG1 antibody (HPCA-2). The low affinity of HPCA-2 for LG was overcome by secondary antibodies: KG1a cells that were dual labeled with HPCA-2 followed by rat anti-mouse IgG1 adhered inside hollow fibers coated with CBD-LG. Alternatively, immobilized rabbit polyclonal anti-mouse IgG1 captured cells labeled with HPCA-2. The development of an adaptor molecule to display recombinant domains at the surface of hollow fibers will be an effective tool to investigate cellular ligand–receptor interactions, a necessary step in the development of hollow fiber bioreactors for manufacture of human cellular products.
Keywords: cell adhesion/cell separation/cellulose binding domain/cellulose hollow fibers/recombinant fusion proteins
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The manufacture of cells for transplantation in human therapy will require development of cost-effective, scalable and GMP compliant technologies for cell selection and culture. Monoclonal antibody based methods for large-scale cell separation have played an enabling role in the clinical evolution of blood stem cell graft manipulation to prevent adverse outcomes such as graft-versus-host disease (Butt et al., 2003
; Rodriguez-Luaces et al., 2004) or tumor recurrence (Burgess et al., 2001). The efficiency or cost of culture systems may require that target cells be enriched. For example, blood stem cell enrichment by CD34 cell selection from mobilized peripheral blood mononuclear cell concentrates has reduced the number of cells to be processed by two orders of magnitude. Ex vivo culture processes for gene transfer into hemopoietic stem cells (Kearns et al., 1997) or the production of blood progenitors to abrogate post-transplant cytopenias have utilized large-scale clinical systems for CD34 positive cell enrichment (Reiffers et al., 1999; Prince et al., 2004).
Several systems have been developed for pre-clinical and clinical-scale cell purification by monoclonal antibody. These are based on the plate panning technique (Morecki et al.,1990), fluidized bed affinity chromatography (Polouckova et al., 2001) or magnetic separation. Monodispersed magnetic beads (Ugelstad et al., 1993) or paramagnetic colloids (McNiece et al., 1997) are conjugated to monoclonal antibody which bind to and label cells that are then separated from unlabeled cells by retention within a magnetic field. A high gradient magnetic field is required to separate cells that are labeled with iron-dextran colloids (30–80 nm). For cell enrichment, antigen positive cells are recovered. Colloidal particles are not easily detached from cells, but are endocytosed and do not appear to influence cell function. There is a requirement for cell detachment from larger magnetic particles (DynabeadsTM: diameter 2.8 or 4.5 µm), with enzymatic treatments or competing peptides (Tseng-Law et al., 1999). DynabeadsTM (Isolex Magnetic Cell Selection SystemTM, Nexell Therapeutics, Inc.) or paramagnetic colloids (CliniMACSTM, Miltenyi Biotech) have been developed into large-scale clinical platforms for blood stem cell enrichment.
Hollow fiber affinity cell separation is a monoclonal antibody based cell separation process. Cells are bound directly or indirectly via surface epitopes by monoclonal antibody or secondary ligand that are immobilized onto the lumen side of hollow fibers. Deposited cells are fractionated on the basis of adhesion strength using the uniform shear field that is generated by flow through hollow fiber modules with well-defined header geometry (Nordon et al., 1996, 1998; Nordon and Schindhelm, 1997). A potential advantage of hollow fiber geometry is the use of a stationary solid phase adsorbent that can be regenerated. Hollow fiber devices are now used commonly for high-density cell culture (Yannelli, 1991), with the potential advantage to integrate cell selection and culture processes.
The coupling of proteins to hollow fibers by covalent chemistry can be problematic with low coupling yield, random orientation of antibody and possible alteration of the structural properties of the hollow fiber membrane by chemical cross-linking or degradation. Chimeric proteins that contain specific binding domains for the solid phase support provide a non-chemical method for orienting recombinant binding domains. Cellulose-binding domains (CBDs) are carbohydrate-binding modules that have moderately high and specific affinity for insoluble or soluble cellulosics. They have been used for applications involving the immobilization of enzymes, cytokines or other ligands. In this paper, we describe the design and application of a recombinant fusion protein for direct immobilization of antibodies and cells onto regenerated cellulose (Cuprophan) hollow fiber membranes. The chimeric protein for cell binding was a fusion between the type IIIa CBD isolated from Clostridium cellulovorans (Goldstein et al., 1993) and a hybrid antibody binding domain, protein LG (Kihlberg et al., 1992). The cell binding kinetics and immunoglobulin isotype specificity of the antibody binding domain were studied using a hollow fiber assay for ligand-mediated cell adhesion (Nordon et al., 2004).
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Strains and plasmids
XL10gold (Stratagene, USA) was used for all cloning work. BL21(DE3)pLysS was used for protein expression. pET23d+ and pET34b+ were from Novagen (Novagen/Calbiochem La Jolla, CA, USA). pLG (Kihlberg et al., 1996) was kindly donated by Dr Ulf Sjobring (University of Lund, Lund, Sweden).
PCR amplification was performed using Red Start Taq polymerase (Sigma, St Louis, MO, USA) and a geneamp 9600 thermal cycler (Perkin Elmer Boston, MA, USA). Unless stated otherwise all enzymes for DNA manipulations were from MBI Fermentas (Vilnius, Lithuania) and used according to the manufacturer's instruction. Gel extraction DNA purification was performed as per the manufacturer's instructions using the QIAquick gel extraction kit (Qiagen, Doncaster, Australia). TA cloning was performed using the pGEM-T Easy kit as per the manufacturer's instructions (Technical Manual, Promega, Madison, WI, USA). All other DNA manipulations were performed using standard methods.
The protein LG sequence was amplified using pLG as a template with the forward primer LG5KpnI (5' caggcggggggtaccaataaag 3') and reverse primer LG3SacI (5' caggtcgagtgagctcatttc 3') with introduced restriction enzyme sites shown in bold. The amplified product was cloned into pGEM-TEasy. Clones were sequenced to confirm the fidelity of the amplified product. The protein LG insert was removed using KpnI and SalI and subcloned into the same restriction enzyme sites in pET34b+ to give the plasmid pET34-LG. The vector pET34b+ contains a CBD fusion tag derived from C. cellulovorans CbpA scaffolding protein, a family IIIa CBD (Shoseyov et al., 1992). The fusion tag includes the minimal CBD domain (91 amino acids) and a further 67 a.a. of the CbpA protein as a linker domain.
The sequence for protein LG was amplified and TA cloned as above using the primers LG5BamHI2 (5' caggcggggatccaaaataaag 3') and LG3SacI (5' caggtcgagtgagctcatttc 3'). Protein LG was subcloned into the BamHI and SacI sites of pET23d+ to give pET23-LG.
Protein expression and purification
The CBD-LG and protein LG were expressed in 1 l of 2YT with selective antibiotics at 37°C for 2 h using 1 mM IPTG induction at OD600 nm = 0.8. Cells were collected (6000 g for 20 min at 4°C) and the pellet was frozen at –20°C. The soluble protein extract was isolated by resuspending the frozen cell pellet in 100 ml Cellytic lysis reagent (Sigma) + 10 µg/ml DNaseI +5 mM MgCl2 and incubating at room temperature for 15 min. The insoluble material was removed by centrifugation (12 000 g, 20 min, 4°C) and the soluble extract was filtered through a 0.22 µm filter ready for purification.
The soluble CBD-LG and protein LG were purified using affinity to IgG-sepharose FF (GE Healthcare, Castle Hill, Australia) according to the manufacturer's recommended procedure. The column was equilibrated in TBST buffer (20 mM Tris, 300 mM NaCl, 0.05% Tween-20 pH 7.6) followed by the loading of the sample. The column was subsequently washed with TBST and 5 mM ammonium acetate pH 5 prior to elution with 0.2 M acetic acid pH 2.4. The eluted fractions were immediately neutralized using 2 M Tris pH 8.4 followed by overnight dialysis against Dulbecco's Phosphate Buffered Saline (PBS). Protein concentrations were determined by Bicinchoninic Acid assay (BCA, Pierce, Rockford, IL, USA).
Binding of CBD-LG to regenerated cellulose
Semi-quantitative binding of the CBD-LG to regenerated cellulose was performed by overnight incubation of unpurified soluble lysate diluted in HEPES buffered saline (HBS, 20 mM HEPES buffer, pH 7.4, 150 mM NaCl) to a final concentration of 1.5, 0.75 and 0.5 mg/ml and stored at 4°C. Cellulose hollow fiber (Membrana GmbH, Wuppertal, Germany), snap frozen in a small quantity of liquid nitrogen, was ground to a fine powder by mortar and pestle. One 0.5 ml of protein solution was thoroughly mixed with 0.1 g ground hollow cellulose fiber and incubated at 4°C overnight. The suspension was centrifuged with collection of the supernatant (free fraction). The ground cellulose hollow fiber was then washed three times in 0.5 ml of washing buffer (20 mM Tris buffer, pH 7.4, 1 M NaCl, 0.05% v/v Tween-20) and the final wash collected for analysis (wash fraction). The soluble lysate, free and bound fractions were mixed with equal volumes of SDS dye (62.5 mM Tris–HCl, pH 6.8; 2% SDS, 10% glycerol, 2.5 mg/ml bromophenol blue) and heated at 95°C for 5 min. Bound CBD-LG was recovered from ground fiber by suspension in 0.5 ml SDS dye diluted in water (1 : 1). The fiber–SDS mixture was heated at 95°C for 15 min, followed by centrifugation and collection of the supernatant (bound fraction).
SDS-PAGE was performed under denaturing conditions on 4–12% NuPAGE gels (Invitrogen, Carlsbad, CA, USA) with morpholinoethanesulfonic acid (MES) SDS running buffer (50 mM 2-MES, 50 mM Tris, 1% w/v SDS, 1 mM EDTA, pH 7.3). See Blue 2 Plus molecular weight protein standards were used to determine the apparent molecular weights. Gels were stained for total protein using Gelcode® Blue Stain Reagent (Pierce, Rockford, IL, USA) as per the manufacturer's instructions. The gel photo was taken with a digital camera (KODAK EDAS120 system, Rochester, NY, USA). Densitometry measurements were made using Image J software.
KG1a cells were used as a model CD34 positive hemopoietic cell line (Koeffler et al., 1980). Five million KG1a cells in 100 µl of RPMI + 10% FCS were labeled at 4°C for 30 min using 50 µl of anti-CD34 mouse MAb, either IgG1(Becton-Dickenson, San Jose, CA, USA) or IgG2a isotype (MHCD3400, Caltag, Edward Keller, NSW, Australia). The cells were washed three times with 5 ml of PBS and 1% bovine serum albumin (BSA) before being resuspended in 0.5 ml of PBS and 1% BSA.
For labeling cells with secondary antibodies, the primary labeled cells were resuspended in 100 µl PBS and 1% BSA before adding 50 µl of the secondary antibody [rat anti-mouse IgG1 MAb (X56), Becton Dickenson or rabbit anti-mouse IgG1 polyclonal antibody, Zymed, San Francisco, USA] and incubating for 30 min at 4°C. The cells were washed three times in 5 ml PBS + 1% BSA and resuspended in 0.5 ml of PBS and 1% BSA.
Samples of the CBD-LG to be tested were injected at 100 µg/ml into the hollow fiber using a 1 ml syringe while the extra-capillary space (contained in the Petri dish) was filled with PBS to stop the fiber drying out. The sample was left to bind overnight at 4°C before washing with PBS + 1% BSA. Antibody was immobilized onto fiber via CBD-LG by injecting a 1 : 20 dilution of the antibody in PBS and incubating for 1 h at room temperature, followed by washing with PBS + 1% BSA.
The cell adhesion assay used is described in detail elsewhere (Nordon et al., 2004). Briefly, the assay device consists of a regenerated cellulose hollow fiber (inner diameter 200 µm) mounted at the base of a 35 mm polystyrene tissue culture dish and connected to stainless steel tube injection ports (23 gage) that pass through holes in the wall of the culture dish, and sealed with silicone elastomer. The cellulose fiber is transparent, and cells deposited on the interior of the fiber are viewed under an inverted microscope through a x10 objective lens. Cells were injected and recovered at various shear stresses using a simple injection and flow system. Images of the field of view were captured using a CCD video camera (Pulnix, Sunnyvale, CA, USA) and cells counted by image analysis (WIT image analysis software).
The single fiber module was primed on both the extra-capillary and intra-capillary space with PBS. The intra-capillary buffer was replaced with a mobile phase consisting of PBS containing 5 mg/ml BSA. A segment of fiber (2 mm) was positioned in the center of the microscopic field (x10 objective) and injected with a cell suspension (5 million cells per ml of mobile phase). Flow was stopped, and cells were allowed to deposit over 4 min. During this time, the first image was captured with counting of 
150 cells in the field of view. Flow was recommenced at 0.43 µl/s, which is equivalent to a shear stress of 5 dynes/cm2. A second image was captured after the cell detachment process was complete (60 s, Fig. 1). Cell attachment was expressed as the percentage of cells that remained bound. The hollow fiber membrane surface was regenerated by flushing out the remaining bound cells at high shear stress (>100 dynes/cm2) so the module could be reused for up to five adhesion measurements.
|
Analysis of binding domain interactions by cell adhesion
A CBD-LG coated module was used to assay adhesion for each of the labeled cell types described in Table I (5 runs per module). This was repeated with three other modules, but with a systematic change in the order of exposure to the different labeled cell types to balance for any membrane degradation effects. The percent bound for each cell type was expressed as the mean ± the standard error (n = 4). The unpaired Student's t-test was used to determine the significance of the difference in means.
|
The interaction of the LG domain with anti-CD34 mouse MAb with isotypes IgG1 or IgG2a designated HPCA-2 and MHCD3400, respectively, was measured by cell adhesion. Cells were labeled with antibodies before adherence on a fiber coated with CBD-LG (Table I, see cell types A and B). The binding of the LG-domain to rat anti-mouse IgG1 MAb (X56) and rabbit anti-mouse IgG1 polyclonal antibody was probed in a similar fashion by labeling cells with the primary MAb and the secondary Ab (Table I, labeled cell types C and D). The cell adhesion assay was also used to characterize the interactions between mouse IgG1 MAb and rat anti-mouse IgG1 MAb (X56) or rabbit anti-mouse IgG1 polyclonal antibody, as well as the interaction of the CD34 antigen and antiCD34 MAb (MHCD3400, Table II).
|
Surface plasmon resonance analysis of LG binding to antibody
Surface plasmon resonance (SPR) was performed on a Biacore 2000 system running Biacore 2000 control software v3.1.1. His6–tagged LG and CBD-LG extracts were immobilized onto a NTA chip (BR-1005-32, Biacore, Upsala, Sweden) via Ni2+/NTA chelation. Following extensive washing at 20 µl/min with eluent buffer (10 mM HEPES, 0.15 M NaCl, 0.05 mM EDTA, pH 7.4) the chip was loaded with 20 µl of 0.5 mM NiCl2 in eluent buffer. Following a 10 min wash with eluent buffer, channels 1 and 2 were loaded with 20 µl soluble extracts of His tagged LG and CBD-LG, respectively. The third channel was unmodified and served as a non-specific binding control. Binding of antibodies to channels 1–4 was analyzed at 20 µl/min by injecting 20 µl of undiluted MHCD3400, HPC2A MAbs or rabbit polyclonal anti-mouse IgG1 with the channels connected in series. The chip was regenerated using 10 mM HEPES, 0.15 M NaCl, 0.35 mM EDTA, pH 8.3 followed by injection of the nickel solution.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Cloning, expression and purification of protein LG and CBD-LG fusion
The protein LG sequenced was cloned into T7 RNA polymerase based expression plasmids pET23d and pET34b. These plasmids coded for protein LG with a c-terminal 6HIS tag (pET23-LG) and a fusion between the CBD from C. cellulovorans and protein LG again with a c-terminal 6HIS tag (pET34b-LG).
Soluble protein LG was expressed at high levels from the pET23-LG whereas the CBD-LG fusion protein was largely expressed as insoluble inclusion bodies when induced at 37°C for periods greater than 2 h. A significant proportion of soluble CBD-LG expression could be achieved using 2 h induction times. The soluble protein LG and CBD-LG were purified using the protein LG domains affinity for IgG.
Binding of CBD-LG to regenerated cellulose
A semi-quantitative study of binding equilibria was determined by overnight adsorption at 4°C. Figure 2 shows SDS-PAGE analysis of free and bound fractions. Three cell lysate concentrations (1.5, 0.75 and 0.5 mg/ml) were tested (lanes 1A–E, 2A–E and 3A–E, respectively). The molecular marker standards are shown in the A lanes. Densitometry was used to compare the intensity of bands on the SDS-PAGE gel and approximate protein concentrations. The intensity of the CBD-LG band in unpurified lysate stored overnight at 4°C without cellulose (control, B lanes) was directly proportional to the lysate concentrations as determined by BCA assay (r2 = 0.997, n = 3). A CBD-LG dimer, confirmed by western blot (data not shown), was present in lanes C and E. No protein was detected in the wash fraction (lane D). There was purification of CBD-LG in the bound fraction (lane 1E) with depletion of other proteins present in the lysate (see lanes 1B and C). The percentage of CBD-LG (including dimer) that bound to regenerated cellulose was inversely related to the total lysate concentration (26% at 1.5 mg/ml, 35% at 0.75 mg/ml and 46% at 0.50 mg/ml).
|
The interaction between CBD-LG and primary labeling mouse MAbs
Figure 3 illustrates the binding of KG1a cells labeled with primary or secondary antibody in fibers coated with CBD-LG. Duplicate experiments demonstrated that there was almost complete adhesion of KG1a cells labeled with the primary antibody MHCD3400 on fibers coated with CBD-LG (see bars B, 92.3 ± 2.4%, 96.4 ± 1.5%, P = 0.14). However, there was no adhesion if KG1a cells were labeled with the primary antibody HPCA-2 (see bars A, 3.4 ± 4.0%, 1.8 ± 1.3%, P = 0.72). There were very low levels of non-specific cell adhesion for unlabeled KG1a cells (bars E, 0.1 ± 0.7, 0.1 ± 0.7, P = 0.36).
|
Low affinity of HPCA-2 for LG is overcome by secondary labeling antibodies
Secondary antibodies that bind to mouse IgG1 were used to determine whether it was still possible to bind KG1a cells labeled with the primary antibody HPCA-2. Both rat anti-mouse IgG1 MAb and rabbit anti-mouse IgG1 polyclonal antibody were secondary labeling reagents that could promote adhesion of KG1a cells onto fibers coated with CBD-LG. KG1a cells labeled with rat anti-mouse MAb reagent had higher affinity compared with the polyclonal rabbit antibody (Fig. 3, bars C, 84.1 ± 7.3% versus 39.2 ± 9.4%, P = 0.005). Relevant negative controls demonstrating that primary antibody (HPCA-2) mediated binding of secondary antibody (rabbit anti-mouse IgG1 polyclonal antibody and rat anti-mouse IgG1 MAb) to KG1a cells was the absence of adhesion when primary antibody was omitted (bars D, 3.6 ± 2.6%, 1.8 ± 2.0%, P = 0.54). Therefore, it was possible to overcome the problem of low affinity of the primary MAb for LG by secondary labeling.
Adhesion by interaction between HPCA-2 and anti-mouse IgG1 antibodies
KG1a cells labeled with HPCA-2 did not adhere to CBD-LG coated fibers. This may be overcome by secondary adsorption of anti-mouse IgG1 antibodies on CBD-LG coated fibers. The ligands involved in the cell adhesive interaction would then be between HPCA-2 on KG1a cells and the rat or mouse anti-mouse IgG1 antibody immobilized on the hollow fiber membrane (Table II, see membranes I and II). Figure 4 shows that rabbit polyclonal anti-mouse IgG1 but not rat monoclonal anti-mouse IgG1 was able to adhere cells labeled with HPCA-2 (membranes I and II). The percentage of KG1a cells bound were 90 ± 6.3% (control 3.5 ± 4.0%, P = 2.0 x 10–5) versus 3.5 ± 2.4% (control 0%).
|
Adhesion by direct interaction between the CD34 antigen and MHCD3400
Finally, it was possible to show that the interaction between MHCD3400 and the CD34 antigen expressed on KG1a cells was strong enough for cell adhesion. MHCD3400 was immobilized onto CBD-LG coated hollow fibers and probed with KG1a cells. Both unlabeled and labeled KG1a cells (MHCD3400) bound to membrane III (Table II). Figure 4 (bar III) shows that there was almost complete cell adhesion for KG1a cells (96.0 ± 2.5%) or KG1a cells labeled with MHCD3400 (91.0 ± 4.3%).
SPR analysis of antibody binding by the LG domain
SPR was used to further characterize the binding properties of the LG domain. Figure 5 shows a sensogram that demonstrates the qualitative properties of the interaction of LG with mouse and rabbit antibodies. Channels 1 and 2 were injected separately with 20 µl of His tagged LG and CBD-LG, respectively (events 1 and 2). There was a higher mass of CBD-LG protein bound to the NTA chip in comparison to LG (3000 versus 650 RU). The level of non-specific binding to the NTA chip was 11.5 RU. Immobilization of these proteins via their His tag was reversible, since the baseline had a negative gradient (–11 and –40 RU/min, respectively, control –1.3 RU/min). At event 4, 20 µl of MHCD3400 was injected, and both LG and CBD-LG bound this mouse monoclonal antibody. The increase in mass bound was approximately +855 and +1733 RU, respectively (control +69 RU). Subsequent injection of HPCA-2 (mouse isotype IgG1) at event 5 did not significantly increase or decrease the amount of mass bound to LG or CBD-LG (+16 and –74 RU, respectively, control +99). There was a further increase in protein binding following injection of rabbit polyclonal anti-mouse IgG1 at event 6 (+711 versus +5046 RU, control +84).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Chemical cross-linking of immunoglobulins to solid phase supports may lead to multi-site attachment, random orientation and steric hindrance limiting specific activity (Wimalasena and Wilson, 1991
). CBD fusion proteins provide a non-covalent method to selectively immobilize and orient recombinant domains. The binding to various physical forms of cellulose will depend on the CBD subfamily. The CBD used in this application is a family subtype IIIa CBD (Tomme et al., 1998). Specific binding of the CBD by ground regenerated cellulose hollow fibers was demonstrated by SDS-PAGE analysis of free and bound fractions (Fig. 2). A significant percentage of the total CBD-LG protein bound to the ground fiber (26–46%) and binding specificity was demonstrated by depletion of non-CBD-LG bacterial lysate proteins in the bound fraction. Interestingly, a significant proportion of CBD-LG formed dimers in the presence of regenerated cellulose.
Three family IIIa CBDs have been extensively characterized. Crystallographic and modeling studies have been performed on CipA-CBD, from C. thermocellum (Tormo et al., 1996) and CipC-CBD from C. cellulolyticum (Shimon et al., 2000). Modeling and site-directed mutagenesis studies have been performed on CbpA-CBD from C. cellulovorans (Murashima et al., 2005). A fourth, CipA-CBD from C. josui has also been cloned and sequenced (Kakiuchi et al., 1998). Family IIIa CBDs are from scaffoldin proteins of cellulosomes and are responsible for anchoring the cellulosome complex onto cellulose. The molecular structure responsible for cellulose binding is a nine-stranded ß sandwich and jelly-roll topology with strands 1, 2, 3 and 7 forming a flat surface on one side that interacts with three adjacent cellulose chains. The opposite face has a shallow groove that binds a calcium ion (Tormo et al., 1996) which has been proposed to interact with the Pro/Thr-containing linker segments of the scaffoldin and other cellulosomal components (Shimon et al., 2000).
Cellulose fibers are composed of regenerated cellulose which is an insoluble form of primary non-crystalline cellulose (Boraston, 2005). The CBD IIIa family has been known to recognize and bind non-crystalline cellulose. According to the proposed cellulose interaction structure, the CBD stacks onto the first cellulose chain (at least six glucose molecules) via aromatic amino acids, and polar residues anchor the protein to the adjacent second and third chain.
Approximately 50% of CBD-LG was bound by ground cellulose hollow fiber when the total soluble lysate protein concentration was 0.5 mg/ml. If the purity of CBD-LG in the lysate was 10% (as determined by densitometry), then the molar concentration of CBD-LG would be 0.8 µM (66 000 g mol). The Kd and adsorption of CbpA-CBD protein to cellulosic substrates has been documented (Goldstein et al., 1993). The observed Kd was around 0.6–1.4 µM for the insoluble cellulose substrates tested. Our semi-quantitative study indicates that the Kd of the CBD domain of CBD-LG is close to those reported in the literature.
We calculate that 2–3 nmol of CBD-LG was bound by 1 g of ground cellulose hollow fiber. Goldstein et al. (1993) showed that the binding capacity of highly crystalline cellulose at saturating CBD concentration was 6.4 µmol/g whereas under the same conditions fibrous cellulose had a binding capacity of 0.2 µmol/g. We conclude that ground cellulose hollow fibres have lower binding capacity when compared to other forms of cellulose.
The binding of CbpA-CBD has been reported to be irreversible (Tomme et al., 1998). Despite moderately high affinities as quantified by the Langmuir adsorption model, no desorption can be observed when this equilibrium is disturbed by dilution of the soluble CBD fraction. Thus, the apparent Kd as determined by the Langmuir adsorption model is not indicative of the final binding strength, since binding is further strengthened by secondary interactions following initial binding. The secondary interactions that strengthen binding may include affinity between individual CBD family IIIa molecules as noted by Xu et al. (1995) which may explain why we observed the formation of a highly stable CBD-LG dimer by cellulose that was resistant to SDS-PAGE denaturation conditions. Thus, despite only moderately high affinity, the irreversible nature CbpA-CBD binding to cellulose makes it a robust domain for immobilization of secondary ligands for selective cell attachment.
CBD-protein A fusion proteins have been successfully used in a number of affinity purification systems based on cellulosic solid phases (Tomme et al., 1998; Shpigel et al., 2000). A significant number of monoclonal antibodies developed against human cell surface markers are mouse IgG1 isotype for which protein A and protein G have a low affinity. Therefore, we investigated the possibility that the hybrid molecule protein LG, which combines the four kappa-binding domains of protein L with two IgG Fc-binding domains of protein G, would provide a comprehensive range of antibody binding specificities for cell separation applications (Vola et al., 1995; Kihlberg et al., 1996).
The single hollow fiber cell adhesion assay provided a direct method to assess whether ligand interactions were strong enough for cell adhesion, a prerequisite for affinity cell separation. The mouse MAb binding specificity of protein LG was tested using an IgG1-
(HPCA-2) or IgG2a (MHCD3400) MAb against the CD34 antigen. Of the antibodies, CBD-LG efficiently captured KG1a cells labeled with MHCD3400, but not HPCA-2. It was also possible to demonstrate cell adhesion when MHCD3400 was immobilized onto the solid phase via CBD-LG (96.0 ± 2.5%). CBD-protein LGs lower affinity for HPCA-2 was confirmed by biosensor.
Kihlberg has shown that mouse IgG1 and IgG2a mouse MAbs have a variable and restricted pattern of binding to proteins L and G. All of these MAbs bound to protein LG, albeit with some variation in affinity (Kihlberg et al., 1992). In another study, Vola et al. (1995) measured the affinity of protein LG for various IgG1 or IgG2a mouse MAbs to be between 3–22 x 107 M–1 at neutral pH. The lower affinity interactions may not be sufficient for cell adhesion, and so direct cell binding to LG via primary mouse MAbs is not a universally applicable approach for cell selection.
We went on to study the use of secondary ligands, which could potentially have higher affinity interactions suitable for cell attachment. Anti-mouse IgG1 antibodies were bound to either the solid phase via CBD-LG or to cells that were labeled with primary mouse MAbs. The interaction between rat MAb and protein-LG resulted in cell adhesion (84.1 ± 7.3% bound), however, even though rat anti-mouse IgG could be used as a secondary cell label, the interaction was not strong enough for cell adhesion (3.5 ± 2.4%). As a secondary ligand, polyclonal rabbit anti-mouse IgG1 had converse properties. The interaction of the polyclonal antibody with protein LG was weaker than the interaction with the mouse MAb (90 ± 6.3% versus 39.2 ± 9.4% adhesion). The weaker interaction of a polyclonal antibody for protein LG may be the result of antibody heterogeneity where multiple Fc isotypes and kappa light chains with differing affinity are presented at the cell surface. Thus, immobilization of rabbit polyclonal anti-mouse IgG1 on CBD-LG coated fibers or secondary labeling of cells with rat anti-mouse IgG1 provide alternate approaches for cell separation if the affinity between primary antibodies and LG is not high enough for cell adhesion.
There are number of additional considerations for application of CBD-LG to separation of complex cellular mixes, such as bone marrow or peripheral blood mononuclear cells. The binding of many human immunoglobulins to LG is problematic since B-cell lineage cells express immunoglobulins at their surface. Therefore, it may be necessary to deplete B cell types before cell enrichment with MAbs. The recovery of cells from the solid phase can be assisted with enzymatic or competing peptides. Furthermore, the interaction of LG is pH dependent, and so it may be possible to enrich specific cell subsets using a combination of fluid shear and changes in pH.
A new methodology has been developed to investigate the mechanical properties of recombinant ligand interactions. This experimental approach is easily generalized to include the study of receptor–ligand interactions that occur at surfaces and are responsible for directing mammalian cell differentiation and growth. The single hollow fiber assay will be an effective tool to screen and optimize biomimetic surfaces for hollow fiber bioreactor systems.
|
Footnotes |
|---|
Edited by Patrick Stayton
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Plasmids containing protein LG was kindly provided by Dr Ulf Sjobring (Department of Laboratory Medicine, Section for Microbiology, Immunology and Glycobiology, Lund University).
This work was supported by an Australian Research Council (ARC) SPIRT grant in collaboration with Becton Dickenson immunocytometry systems and a grant from the Energy Research Development Corporation of Australia (ERDC, Australia).
S.J.C. was supported by an Australian Postgraduate Award (Industry) provided by the ARC.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Rodriguez-Luaces M., et al. Eur. J. Haematol. (2004) 73:162–168.[CrossRef][ISI][Medline]
Boraston A.B. Biochem. J. (2005) 385(Pt 2):479–484.[CrossRef][ISI][Medline]
Burgess J., et al. Cytotherapy (2001) 3:285–294.[CrossRef][ISI][Medline]
Butt N.M., McGinnity N., Clark R.E. Leuk. Lymphoma (2003) 44:1509–1513.[ISI][Medline]
Goldstein M.A., et al. J. Bacteriol. (1993) 175:5762–5768.
Kakiuchi M., et al. J. Bacteriol. (1998) 180:4303–4308.
Kearns K., et al. Hum. Gene Ther. (1997) 8:301–311.[ISI][Medline]
Kihlberg B.M., et al. J. Biol. Chem. (1992) 267:25583–25588.
Kihlberg B.M., et al. Eur. J. Biochem. (1996) 240:556–563.[ISI][Medline]
Koeffler H.P., et al. Blood (1980) 56:265–273.
McNiece I., et al. J. Hematother. (1997) 6:5–11.[ISI][Medline]
Morecki S., et al. J. Biol. Response. Mod. (1990) 9:463–474.[ISI][Medline]
Murashima K., Kosugi A., Doi R.H. J. Bacteriol. (2005) 187:7146–7149.
Nordon R.E., Schindhelm K. Artif. Organs (1997) 21:107–115.[ISI][Medline]
Nordon R.E., et al. Cytometry (1996) 24:340–347.[CrossRef][ISI][Medline]
Nordon R.E., et al. Cell Separation Device (1998) USA: Unisearch Ltd.
Nordon R.E., Shu A., Camacho F., Milthorpe B.K. Cytometry (2004) 57A:39–44.[Medline]
Polouckova A., et al. Neoplasma (2001) 48:374–381.[ISI][Medline]
Prince H.M., et al. Br. J. Haematol. (2004) 126:536–545.[CrossRef][ISI][Medline]
Reiffers J., et al. Lancet (1999) 354:1092–1093.[CrossRef][ISI][Medline]
Ugelstad J., et al. Blood Purif. (1993) 11:349–369.[ISI][Medline]
Shimon L.J., et al. Acta Crystallogr. D Biol. Crystallogr. (2000) 56(Pt 12):1560–1568.[CrossRef][Medline]
Shoseyov O., et al. Proc. Natl Acad. Sci. USA (1992) 89:3483–3487.
Shpigel E., et al. Biotechnol. Appl. Biochem. (2000) 31(Pt 3):197–203.[CrossRef][ISI]
Tomme P., et al. J. Chromatogr. B Biomed. Sci. Appl. (1998) 715:283–296.[CrossRef][Medline]
Tormo J., et al. EMBO J. (1996) 15:5739–5751.[ISI][Medline]
Tseng-Law J., et al. Exp. Hematol. (1999) 27:936–945.[CrossRef][ISI][Medline]
Vola R., et al. J. Chromatogr. B Biomed. Appl. (1995) 668:209–218.[CrossRef][ISI][Medline]
Wimalasena R.L., Wilson G.S. J. Chromatogr. (1991) 572:85–102.[CrossRef][ISI][Medline]
Xu G.Y., et al. Biochemistry (1995) 34:6993–7009.[CrossRef][Medline]
Yannelli J.R. J. Immunol. Methods (1991) 139:1–16.[CrossRef][ISI][Medline]
Received September 1, 2006; revised January 25, 2007; accepted February 28, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||