PEDS Advance Access originally published online on November 24, 2008
Protein Engineering Design and Selection 2009 22(2):59-66; doi:10.1093/protein/gzn071
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Aggregation-resistant VHs selected by in vitro evolution tend to have disulfide-bonded loops and acidic isoelectric points*
1 Institute for Biological Sciences, National Research Council of Canada, Ottawa, ON, Canada K1A 0R6 2Department of Environmental Biology, Ontario Agricultural College, University of Guelph, Guelph, ON, Canada N1G 2W1 3Depeartment of Biochemistry, Microbiology and Immunology, University of Ottawa, Ottawa, ON, Canada K1 N 6N5
5 To whom correspondence should be addressed. E-mail: jamshid.tanha{at}nrc-cnrc.gc.ca
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
When panned with a transient heat denaturation approach against target enzymes, a human VH (antibody heavy chain variable domain) phage display library yielded VHs with composite characteristics of binding, non-aggregation and reversible thermal unfolding. Moreover, selection was characterized by enrichment for VHs with (i) an even number of disulfide forming Cys residues in complementarity-determining region (CDR) 1 and CDR3 and (ii) acidic isoelectric points. This parallels naturally occurring camelid and shark single-domain antibodies (sdAbs) which are also characterized by (i) solubility and reversible unfolding, (ii) a high occurrence of disulfide forming Cys in their CDRs, particularly, in CDR1 and CDR3 and (iii) acidic VHs as inferred here by a pI distribution analysis, reported here, of pools of human and camelid VH and VHH (camelid heavy chain antibody VH) sequences. Our results, reinforced by previous observations by others, suggest that protein acidification may yet be another mechanism nature has devised to create functional sdAbs and that this concept along with the inclusion of inter-CDR disulfide linkages may be applied to human VH domains/libraries for non-aggregation optimization. In addition, calculation of theoretical pIs of VHs selected by panning may be used for rapid and precise identification of non-aggregating VHs.
Keywords: disulfide-bonded loops/isoelectric point/non-aggregating VH/phage display library and panning/single-domain antibody
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The main incentive for using human VHs as opposed to their ‘naturally occurring’ sdAb counterparts, camelid VHHs and shark VNARs (Ig new antigen receptor variable domains), is their expected lower immunogenicity in immunotherapy (Holt et al., 2003
; Holliger and Hudson, 2005). Synthetic VH libraries, which serve as the source of binders (VHs), contain a significant percentage of aggregating VHs (VHs tend to aggregate) and, when subjected to conventional panning, yield aggregating VHs. Thus, a number of methods have been developed for effective isolation of non-aggregating VHs from synthetic libraries (Davies and Riechmann, 1994; Tanha et al., 2001; Tanaka et al., 2003; Jespers et al., 2004a; To et al., 2005; Tanha et al., 2006; Christ et al., 2007; Tanaka et al., 2007). An interesting approach (Jespers et al., 2004a) for selecting soluble, binding VHs involves the use of a phage display library and sequential steps of subjecting the library to (i) heating (80°C for 10 min) to denature phage-displayed VHs, (ii) cooling and (iii) exposure to target antigen in the binding stage of panning. While VHs with reversible unfolding characteristics regain their binding during the cooling step and are subsequently selected during the binding step, those with irreversible denaturation characteristics, which include insoluble VHs, are lost to aggregation and eliminated. This selection method leads to the isolation of non-aggregating VHs with reversible thermal unfolding properties. The approach requires multivalent display of VHs on the phage surface, since it was demonstrated that it was effective with phage vector-based display libraries but not in a monovalent display format bestowed with phagemid vector-based systems (Winter et al., 1994; Bradbury and Marks, 2004).
Here, we describe the construction of a human VH phagemid library with a disulfide bond-constrained CDR3 loop, the adaptation of the above-mentioned transient heat denaturation approach (Jespers et al., 2004a) to panning the library and the isolation of several VHs characterized by non-aggregation and reversible thermal unfolding properties. We find that the non-aggregating VHs tend to have disulfide-bonded CDRs and acidic pIs. Consistently, by a pI distribution analysis of pools of human and camelid VH and VHH sequences, we reveal that the majority of VHs are basic while the majority of VHHs are acidic. The implications and applications of our findings are discussed.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Expression, purification and size exclusion chromatography analysis of VHs
VH genes were sub-cloned into pSJF2H vector for soluble expression in Escherichia coli strain TG1 (Arbabi-Ghahroudi et al., 2008) using the primers HVHP430Bam and HVHP430Bbs (Supplementary data are available at PEDS online, Table SI). [The pSJF2H vector is identical to pSJF2 (Tanha et al., 2003) except that it results in the expression of proteins with a His6 tag instead of a His5 tag.) The VHs with amber codons replaced with Glu codons were constructed by splice overlap extension and polymerase chain reaction (PCR) (Ho et al., 1989; Aiyar et al., 1996) using pSJF2H vectors containing VH genes as templates. Size exclusion chromatography of the purified VHs was performed as described (To et al., 2005). Percent monomer was obtained by integrating the area under the monomeric and multimeric peaks of size exclusion chromatograms and adjusted for the amounts of VHs that were aggregated on top of the column before application (Jespers et al., 2004a).
Thermal refolding efficiency measurements
Thermal refolding efficiencies of VHs at concentrations of 0.5 and 5 µM were determined as described (To et al., 2005) except that 600 resonance units (RUs) of protein A (Sigma, Oakville, ON, Canada) or ovalbumin (Sigma) as a reference protein were immobilized. Refolding efficiencies were calculated from the concentration assay of active VH bound at steady state.
Theoretical pI distribution analysis of camelid and human VHH and VH sequences
The pIs of the VHs/VHHs were determined using the software Laser gene v6.0 (DNASTAR, Inc. Madison, WI, USA). There is minor disagreement between pI values obtained here (higher by 2%) and those reported elsewhere (Jespers et al., 2004a). The Camelus dromedarius germline VHH and VH segments are referenced in Nguyen et al. (Nguyen et al., 2000) and the germline human VH segments are deposited in V BASE (http://vbase.mrc-cpe.cam.ac.uk/). The numbers of rearranged domains included in the analysis were 141 L. glama VHHs (Harmsen et al., 2000; Tanha et al., 2002), 495 C. dromedarius VHHs (NCBI database, accession Nos AB091838
[GenBank]
–AB092333
[GenBank]
) and 356 human VHs (NCBI database, accession Nos Z80363
[GenBank]
–Z80770) (Brezinschek et al., 1997). Forty-seven human VHs with stop codons or ambiguous amino acids were excluded from the analysis. The missing N-terminal amino acids of rearranged human VHs were replaced with their corresponding germline residues (the germline designation of the human VHs were reported in the database and for the few unassigned ones the germline sequence assignment was performed using DNAPLOT software Version 2.0.1 and V BASE version 1.0 (http://vbase.dnaplot.de/cgi-bin/vbase/vsearch.pl). The last nine residues were missing but this should not lower the pI values of the human VHs; in the instance of a JH2 gene usage there would be a pI increase (see V BASE).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Construction of a synthetic phagemid-based VH phage display library with constrained CDR3 loop
The objective of this study was to construct and pan a synthetic human VH library in a phagemid format, introducing a feature expected to make the library a good source of enzyme inhibitors and to investigate it in terms of yielding non-aggregating VH binders. Previously, we identified a non-aggregating human VH, i.e. HVHP430 (Supplementary Fig. S1(a)), which has Cys residues in CDR3 at positions 99 and 100d (To et al., 2005). By alkylation reactions/mass spectrometry, we verified that the two CDR3 Cys residues formed an intra-CDR3 disulfide linkage (Supplementary data are available at PEDS online, Fig. S2). Therefore, in constructing our VH library on the HVHP430 scaffold, we did not randomize these Cys residues in order to maintain formation of disulfide-constrained CDR3 loops and in hope of increasing the proportion of enzyme inhibiting VHs in the library. [It has been shown that the disulfide bond-constrained CDR3 loops of a VHH and a VNAR that inhibit lysozyme protrude into the enzyme active site (Desmyter et al., 1996; Stanfield et al., 2004).] The remaining 15 CDR3 positions, position 94 and eight hypervariable loop 1 (H1)/CDR1 positions were randomized (Supplementary data are available at PEDS online, Fig. S1(a)). CDR2 was left untouched as it has been shown to be involved in protein A binding (Randen et al., 1993; Bond et al., 2003). [It is noteworthy that CDR2-deficient VNARs (Stanfield et al., 2004) and camelid VHHs utilizing only CDR1 and CDR3 (Decanniere et al., 1999) or only CDR3 (Desmyter et al., 2001) for antigen recognition have nanomolar affinities.] The library, which has a diversity of 2 x 109, was constructed according to the scheme shown in Supplementary data available at PEDS online, Fig. S1(b) using a phagemid vector (Supplementary data are available at PEDS online, Fig. S3) and characterized as described in Supplementary data available at PEDS online, Fig. S1.
Panning in a monovalent display format
In our first selection attempt, we panned the VH phage library in a monovalent display format (O'Connell et al., 2002). Our initial aim was to explore the quality of the library as a source of enzyme inhibitors. We, thus, panned (four rounds) against three enzymes: 
-amylase, lysozyme and carbonic anhydrase. Sequencing of 80 clones from various rounds showed that over 40% of the VHs had an amber stop codon (TAG), almost exclusively at position 32 in CDR1. (The unpanned library had 53% TAG-containing VHs.)
Following panning, 10–20 round 4 phage clones were tested for binding to their target antigens by enzyme-linked immunosorbent assay (ELISA); 6/20, 10/10 and 19/20 were positive for binding to lysozyme, -amylase and carbonic anhydrase, respectively. Twelve VHs (three -amylase binders, four lysozyme binders and five carbonic anyhdrase binders) were sub-cloned into an expression vector and expressed in 1 l cultures. The VHs were subjected to Superdex 75 gel filtration chromatography to evaluate their aggregation states. Aggregation tendencies of VHs were assessed quantitatively as the percentage of VH existing as monomer. All the VHs showed strong aggregation tendencies with the monomer percentage ranging from as low as 10 to 84% at best (Fig. 1A and C). Additionally, several VHs precipitated at 4ºC, not long after purification. Dissociation constants of the VHs, determined by performing surface plasmon resonance (SPR) on purified monomeric fractions, ranged from 10 µM to 40 nM (Supplementary data are available at PEDS online, Fig. S4(a)).
|
100% monomer) and non-monomeric (display significant aggregation) and were compared using chi-squared (Panning in a multivalent display format with heat denaturation
The panning results demonstrated that selection based solely on binding was not sufficient to yield non-aggregating binders, not to mention functional enzyme inhibitors. We, thus, shifted our focus to identifying a panning strategy which would select for non-aggregating binders. We adapted the heat denaturation approach which had been shown to efficiently yield non-aggregating binders from VH phage display libraries (Jespers et al., 2004a). The multivalency requirement of the approach was fulfilled by using hyperphage (M13KO7
pIII) (Rondot et al., 2001) for superinfection. We performed three rounds of panning against -amylase, incorporating a heat denaturation step before each panning. A non-treatment panning was carried out in parallel. Following three rounds of panning, for each condition 20 clones were tested by phage ELISA and all were found to bind to -amylase (data not shown). All 40 clones were subjected to DNA sequencing, revealing no common clones between the treatment and non-treatment VHs. Again, all VHs had amber stop codons, predominantly in CDR1 (position 32) but also in CDR3. In contrast to non-treatment panning, which like panning in the monovalent display format did not yield repeating clones, panning under heat denaturation yielded VHs that occurred more than once (Table I, huVHAm302, huVHAm309, huVHAm316), a pattern typically seen with enrichment for binders. We, therefore, continued only with panning under conditions of heat denaturation. Twenty-seven ELISA-positive clones from rounds 4 were sequenced and eight out of the nine identified VHs, had amber stop codons (Table I). Biased enrichment for binders (scFvs) with amber codons has been observed with other synthetic libraries (Yan et al., 2004; Marcus et al., 2006a; Marcus et al., 2006b). Reduced expression of VHs with amber codons in E.coli TG1 compared to those without should confer a growth advantage to phage displaying such VHs (possibly due to reduced toxicity), leading to their preferential selection. We observed that almost 70% of the isolated clones (Table I) have an A or a G immediately following the amber codons which, in contrast to T or C, leads to good suppression of the amber codon (Bossi, 1983; Miller and Albertini, 1983). Thus, balanced VH expression, low enough to reduce toxicity but high enough to survive selection, appears to form the basis of the selection process.
|
Interestingly, after the third round of panning, we observed significant enrichment for VHs which had acidic pIs (
2 and t-tests, P < 0.01) and/or an even number of Cys residues in their CDRs (2 test, P < 0.001), with one CDR1 Cys and one or three Cys residues in CDR3 (Table I: see VHs with double-underlined Cys residues). In some clones, an inter CDR1–CDR3 disulfide bridge appears to be formed by selection for a third Cys in CDR3 or mutation of one of the two non-randomized CDR3 Cys residues. Others (Jespers et al., 2004a) have reported that panning under heat denaturation conditions predominantly enriched for VHs with acidic pIs. Strong selection for the above two properties is further underlined by the fact that only 7% of the 103 VHs from the unpanned library had acidic pIs (8.4 ± 1.0, mean ± SD; 2 test, P < 0.00001; t-test, P < 0.0001) and none had paired Cys residues in CDR1 and CDR3 (2 test, P < 0.0001). (In the absence of data for panning without heat treatment (4th round) one cannot conclusively claim that the selection for the aforementioned properties is due to the heat treatment.) Characterization of VHs selected by heat denaturation panning
We randomly selected nine VHs and sub-cloned them for further analysis (Table I). However, all except one (huVHAm431) had an amber stop codon which would impede their expression even in an amber suppressing strain such as TG1 which was to be used as the expression host (in TG1, the amber stop codon is sometimes read as an amino acid but mostly as a stop codon). Of the eight TAG-containing VHs, only one, huVHAm302, was expressed, although poorly. We found, by mass spectrometric sequencing of huVHAm302, that the amber codon codes for Glu in E.coli TG1 (Hoogenboom et al., 1991; Baek et al., 2002) (Supplementary data are available at PEDS online, Fig. S5) and not for Gln as reported by others (Marcus et al., 2006a; Soltes et al., 2003). We, thus, recloned eight of the nine aforementioned VHs, substituting the amber codon with a Glu codon. Size exclusion chromatography of the VHs showed a significant improvement in solubility compared to those selected in a monovalent display format. VHs isolated by panning with heat denaturation in a multivalent display format have median and mean monomer contents of 78 and 85%, respectively, compared to 73 and 62% for those isolated in a monovalent display format (t-test, P < 0.05), with four in the former group (huVHAm304, huVHAm309, huVHAm416, huVHAm428) being completely monomeric (Fig. 1B and C; Table I). Interestingly, of the four non-aggregating VHs, three are acidic (huVHAm304, pI 5.3; huVHAm416, pI 5.8; huVHAm428, pI 5.8), whereas only one is basic (huVHAm309, pI 8.2) (Fig. 1C, inset; Table I). The remaining five VHs were basic or near neutral. Of the four aggregating VHs with the lowest monomer proportion, three (huVHAm302, huVHAm315, huVHAm427) had pIs around neutral pH (7.3, 7.0, 6.4). Previously, it has been observed that out of seven human serum albumin-specific VHs, one (obtained with the heat denaturation approach) with an acidic pI (5.7) showed reversible folding upon heat denaturation whereas the other six (obtained without the heat step) with higher pIs (7.4 ± 1.2, mean ± SD) did not (Jespers et al., 2004a). Also, of six aggregation-resistant protein A binding VHs, four had acidic pIs (4.3–4.7), whereas two had neutral (7.0) and basic (8.0) pIs. A highly refoldable and non-aggregating lysozyme-specific VH, HEL4 (Jespers et al., 2004b), also had an acidic pI, 4.7. In agreement with the above findings, none of the aggregating VHs isolated in the present study with the monovalent display format (Fig. 1) had acidic pIs (9.1 ± 0.3, mean ± SD). Interestingly, all the non-aggregating, acidic VHs obtained previously (Jespers et al., 2004a; Jespers et al., 2004b) and in this study have theoretical pIs of less than 6.
We further tested five VHs for the presence of intra- and inter-CDR disulfide linkages by alkylation reaction/mass spectrometry experiments. As shown in Supplementary data available at PEDS online, Figs S2 and S5 and summarized in Table I, all CDR cysteines are engaged in disulfide linkages. Thus, huVHAm302, huVHAm304 and huVHAm309 have intra-CDR3 disulfide linkages, huVHAm428 has a CDR1–CDR3 disulfide linkage and huVHAm416 has both the intra- and inter-CDR disulfide linkages. SDS–PAGE analyses of the five aggregating VHs (huVHAm302, huVHAm315, huVHAm316, huVHAm427 and huVHAm431) revealed dimer species on non-reducing gels but not on reducing gels for four of the VHs, indicating the existence of inter-domain disulfide linkages in these VHs (Fig. 2A; Table I). Thus, for these VHs, the non-canonical Cys residues may contribute to their aggregation.
|
The four non-aggregating VHs were analyzed for their reversible thermal unfolding properties by comparing equilibrium dissociation constants for the binding of native (KD) and heat-treated/cooled (KDref) VHs to protein A (To et al., 2005
). The ratio of KD to KDref, defined as the thermal refolding efficiency (TRE), gives a measure of the degree to which the VHs refold to their native state following thermal denaturation. Denaturation and measurement of TREs were performed at two different VH concentrations: 0.5 and 5 µM (Fig. 2B; Table I). Only huVHAm304 showed a concentration-dependent decrease in refolding from 99% TRE at 0.5 µM to 87% at 5 µM. Aggregation, which is accelerated at higher protein concentrations, is most likely the cause of this decrease. However, for all VHs, the TRE values were still very high at 5 µM ranging from 87 to 97%. The VH with the highest TRE, huVHAm416, has one more non-canonical disulfide linkage than the other three (see above), underlining the importance of non-canonical disulfide linkages in sdAb solubility and stability.
All nine VHs, bound to their target antigen, -amylase, in ELISA experiments (Supplementary data are available at PEDS online, Fig. S4(b)), and for huVHAm416 a KD of 4 µM were obtained by SPR (Supplementary data are available at PEDS online, Fig. S4(c)). The VHs also bound to protein A (Supplementary data are available at PEDS online, Fig. S4(b); Table I), a property which can be exploited for VH purification and detection in diagnostic tests, immunoblotting and immunocytochemistry. Moreover, one of the five VHs tested (huVHAm302), which has an intra-CDR3 disulfide linkage, was an enzyme inhibitor with an IC50 of 1.7 µM (Supplementary data are available at PEDS online, Fig. S6). However, it remains to be seen if the inhibition is mediated by the penetration of an intra-CDR3 disulfide linkage dependent protruding CDR3 loop into the -amylase active site.
Theoretical pI distribution analysis of camelid and human VHH and VH sequences
We carried out a theoretical pI distribution analysis of rearranged and germline VHH and VH sequences from camelids and humans (Fig. 3). In the case of rearranged VHHs, although Lama glama VHH1 and VHH2 subfamily members are mostly basic, the L. glama VHH3 subfamily and C. dromedarius sequences, which are more closely related to each other, are mostly acidic. The combined pool of camelid VHHs (646 sequences) is also mostly acidic (Fig. 3F). Many of the L. glama sequence entries are missing the first few framework region 1 (FR1) amino acids, which often has an acidic amino acid at position 1. With an acidic residue included in FR1, the proportion of acidic VHHs would increase to 79% for VHH3 and 53% for the combined camelid VHHs (versus 43% for basic VHHs). In the case of germline clones (C. dromedaries), while the VH pool, which would be comprised primarily of aggregating VHs, consists mainly of VH segments with basic pIs, the opposite is true of the VHH pool, which is predominately populated with VHH segments with acidic pIs. Consistent with this is the fact that the overwhelming majority of human rearranged and germline VH sequences have basic pIs. Thus, based on the biophysical and statistical data accumulated so far on human and camelid VHs/VHHs in this study and by others (Jespers et al., 2004a; Jespers et al., 2004b), it is possible that the high abundance of acidic VHHs in the camelid sdAb repertoire is not a random occurrence but rather the result of nature arriving at a solution to generate soluble and stable sdAbs by in vivo evolution.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Here, we successfully extended a previously described transient heat denaturation approach (Jespers et al., 2004a
) to a phagemid display library and isolated several VHs which were characterized by non-aggregation and reversible thermal unfolding properties. The selection for VHs with solubility and reversible thermal unfolding overlapped with selection for VHs with inter-CDR1–CDR3 disulfide linkages and/or with acidic pIs. Such selection overlap has parallels in nature. Naturally occurring camelid VHH and shark VNAR sdAbs are characterized, on the one hand, by solubility and reversible thermal unfolding (van der Linden et al., 1999; Perez et al., 2001; Dooley et al., 2003; Goldman et al., 2006; Ladenson et al., 2006; Tanha et al., 2006; Liu et al., 2007), and, on the other hand, by non-canonical Cys pairs in their CDRs and FRs (Muyldermans et al., 1994; Vu et al., 1997; Harmsen et al., 2000), forming disulfide linkages between CDR3 and CDR1, CDR2, FR2 or FR4 (Desmyter et al., 1996; Vu et al., 1997; Stanfield et al., 2004; Streltsov et al., 2005; Stanfield et al., 2007) (see also NCBI database: Of the 495 C. dromedarius VHHs in the NCBI database [accession numbers AB091838–AB092333], 44% have Cys residues in CDR3 that can potentially form disulfide linkages with Cys residues in CDR1 [81%] or CDR2 and FR2 [19%]). This salient feature, which is more pronounced in the case of shark VNAR domains is under strong selection at both the germline and somatic levels (Muyldermans et al., 1994; Vu et al., 1997; Harmsen et al., 2000; Nguyen et al., 2000; Diaz et al., 2002), underscoring its importance in sdAb stability and solubility. A conceptually similar example of evolutionary convergence was observed when libraries based on the fibronectin type III domain were subjected to in vitro evolutionary selection (Lipovsek et al., 2007). We found that, as in the case with non-canonical disulfide linkages, there is strong selection for acidic VHHs both at the somatic and germline levels. Taken together, the results described here and elsewhere (Jespers et al., 2004a; Jespers et al., 2004b) suggest that protein acidification may be yet another mechanism devised by nature to create functional single-domain antibodies. However, it is premature to state that convergent evolution leads to the selection of acidic pIs as the data in Fig. 3 can be interpreted in a number of ways and the number of clones isolated by the panning experiments are too few to provide enough statistical significance to establish a clear trend between non-aggregation and acidity. Net charge alone (Jespers et al., 2004a) does not explain why the VHs with solubility and reversible thermal unfolding have predominantly acidic pIs and not basic pIs conferring comparable but opposite net charge. Perhaps, a net positive-charge is conducive to folding intermediates and/or native folds which are more prone to aggregation. Any mechanistic explanation at this time would be mere speculation and requires further detailed studies.
Non-aggregating VHs from the present library (described here and unpublished results) have KDs ranging from a few µM to 100 nM. In a number of therapeutic applications which involve avid binding of VHs to cell surface antigens, VHs in this affinity range should be suitable, even desirable. Examples include applications involving (i) bispecific T cell engaging antibodies (Dreier et al., 2002; Bargou et al., 2008), (ii) internalizing antibodies that bind to cell surface receptors which require self-cross linking for internalization (Becerril et al., 1999) and (iii) dual-specific antibodies which bind to both epitopes on the surface of diseased cells but only to one of the two epitopes on normal cells (patent application WO/2004/003019). Lower affinities in the lower end of KD range seen here are also an advantage in terms of tumor penetration (Adams et al., 2001). For many other applications, however, the VHs obtained from the present library need to be affinity matured into the low nM–pM range to render them useful.
Our enrichment results and published data on camelid VHH and shark VNAR sdAbs (see above) strongly suggest that non-intra-CDR3 (inter-CDRs and inter-CDR-FR) non-cananical disulfide linkages are better suited for imparting non-aggregation and stability to VHs than the intra-CDR3 bridges. Being pinned together by an extra, non-canonical, disulfide linkage which stretches from CDR1 to CDR3, such VHs presumably refold to their native structure more efficiently, and aggregate less, than those with an intra-CDR3 disulfide linkage during the refolding step of the panning procedure and are, therefore, preferentially selected during the binding step. In one study, it was shown that an inter-CDR1–CDR3 disulfide linkage increased the stability of a human VH (Davies and Riechmann, 1996). Thus, including non-canonical inter-CDR disulfide linkages as a part of library design should be a beneficial option for imparting solubility and stability on the VHs.
The fact that non-aggregating VHs tend to be acidic implies that, following panning, non-aggregating VHs may be distinguished from aggregating ones based on pI values calculated from amino acid sequences, saving significant time and effort which would otherwise be dedicated to sub-cloning, expression, purification and biophysical characterization of a large number of VHs in order to identify the non-aggregating subset. Moreover, one should be able to increase the proportion of non-aggregating VHs in the library by using an acidic scaffold—assuming that it is the global pI effect as opposed to the local pI effect in the CDRs that confer the aggregation resistance- and/or biasing randomization towards acidic residues and/or against basic ones. Such improvement should be very significant considering that in the present case the library yielded a good number of non-aggregating acidic binders despite the paucity of acidic VHs in the library. As for our selection approach, it combines the capability of selecting for desired biophysical properties offered by phage vector systems with the advantages offered by phagemid-based systems, such as easier construction of large libraries and the selection for high affinity binders.
|
Footnotes |
|---|
* This is National Research Council of Canada Publication 42530.
4 Present address: Canadian Border Services Agency, Ottawa, ON, Canada K2E 6T7.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
We would like to acknowledge Jacek Stupak and Jianjun Li for MW determination of VHs by mass spectrometry, Sonia Leclerc for DNA sequencing and Tom Devecseri for preparing publication quality figures. We are indebted to John Kelly for his helpful suggestions on mass spectrometric analysis of huVHAm302, Pejman Hanifi-Moghaddam for performing data statistical analyses and Tassnim Moradipour for retrieving and organizing camel VHH sequences from databases. Requests for pMED1 vector should be addressed to Mehdi Arbabi-Ghahroudi.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Adams G.P., Schier R., McCall A.M., Simmons H.H., Horak E.M., Alpaugh R.K., Marks J.D., Weiner L.M. Cancer Res. (2001) 61:4750–4755.
Aiyar A., Xiang Y., Leis J. Methods Mol. Biol. (1996) 57:177–191.[Medline]
Arbabi-Ghahroudi M., Tanha J., MacKenzie R. Methods Mol. Biol. (2009) In press.
Baek H., Suk K.H., Kim Y.H., Cha S. Nucleic Acids Res. (2002) 30:e18.
Bargou R., et al. Science (2008) 321:974–977.
Becerril B., Poul M.A., Marks J.D. Biochem. Biophys. Res. Commun. (1999) 255:386–393.[CrossRef][Web of Science][Medline]
Bond C.J., Marsters J.C., Sidhu S.S. J. Mol. Biol. (2003) 332:643–655.[CrossRef][Web of Science][Medline]
Bossi L. J. Mol. Biol. (1983) 164:73–87.[CrossRef][Web of Science][Medline]
Bradbury A.R., Marks J.D. J. Immunol. Methods (2004) 290:29–49.[CrossRef][Web of Science][Medline]
Brezinschek H.P., Foster S.J., Brezinschek R.I., Dorner T., Domiati-Saad R., Lipsky P.E. J. Clin. Invest. (1997) 99:2488–2501.[Web of Science][Medline]
Christ D., Famm K., Winter G. Protein Eng. Des. Sel. (2007) 20:413–416.
Davies J., Riechmann L. FEBS Lett. (1994) 339:285–290.[CrossRef][Web of Science][Medline]
Davies J., Riechmann L. Protein Eng. (1996) 9:531–537.
Decanniere K., Desmyter A., Lauwereys M., Arbabi-Ghahroudi M., Muyldermans S., Wyns L. Structure (1999) 7:361–370.[Medline]
Desmyter A., Decanniere K., Muyldermans S., Wyns L. J. Biol. Chem. (2001) 276:26285–26290.
Desmyter A., Transue T.R., Arbabi-Ghahroudi M., Thi M.H., Poortmans F., Hamers R., Muyldermans S., Wyns L. Nat. Struct. Biol. (1996) 3:803–811.[CrossRef][Web of Science][Medline]
Diaz M., Stanfield R.L., Greenberg A.S., Flajnik M.F. Immunogenetics (2002) 54:501–512.[CrossRef][Web of Science][Medline]
Dooley H., Flajnik M.F., Porter A.J. Mol. Immunol. (2003) 40:25–33.[CrossRef][Web of Science][Medline]
Dreier T., Lorenczewski G., Brandl C., Hoffmann P., Syring U., Hanakam F., Kufer P., Riethmuller G., Bargou R., Baeuerle P.A. Int. J. Cancer (2002) 100:690–697.[CrossRef][Web of Science][Medline]
Goldman E.R., Anderson G.P., Liu J.L., Delehanty J.B., Sherwood L.J., Osborn L.E., Cummins L.B., Hayhurst A. Anal. Chem. (2006) 78:8245–8255.[Medline]
Harmsen M.M., Ruuls R.C., Nijman I.J., Niewold T.A., Frenken L.G.J., de Geus B. Mol. Immunol. (2000) 37:579–590.[CrossRef][Web of Science][Medline]
Ho S.N., Hunt H.D., Horton R.M., Pullen J.K., Pease L.R. Gene (1989) 77:51–59.[CrossRef][Web of Science][Medline]
Holliger P., Hudson P.J. Nat. Biotechnol. (2005) 23:1126–1136.[CrossRef][Web of Science][Medline]
Holt L.J., Herring C., Jespers L.S., Woolven B.P., Tomlinson I.M. Trends Biotechnol. (2003) 21:484–490.[CrossRef][Web of Science][Medline]
Hoogenboom H.R., Griffiths A.D., Johnson K.S., Chiswell D.J., Hudson P., Winter G. Nucleic Acids Res. (1991) 19:4133–4137.
Jespers L., Schon O., Famm K., Winter G. Nat. Biotechnol. (2004) a 22:1161–1165.[CrossRef][Web of Science][Medline]
Jespers L., Schon O., James L.C., Veprintsev D., Winter G. J. Mol. Biol. (2004) b 337:893–903.[CrossRef][Web of Science][Medline]
Ladenson R.C., Crimmins D.L., Landt Y., Ladenson J.H. Anal. Chem. (2006) 78:4501–4508.[Medline]
Lipovsek D., Lippow S.M., Hackel B.J., Gregson M.W., Cheng P., Kapila A., Wittrup K.D. J. Mol. Biol. (2007) 368:1024–1041.[CrossRef][Medline]
Liu J.L., Anderson G.P., Delehanty J.B., Baumann R., Hayhurst A., Goldman E.R. Mol. Immunol. (2007) 44:1775–1783.[CrossRef][Web of Science][Medline]
Marcus W.D., Lindsay S.M., Sierks M.R. Biotechnol. Prog. (2006) a 22:919–922.[Medline]
Marcus W.D., Wang H., Lohr D., Sierks M.R., Lindsay S.M. Biochem. Biophys. Res. Commun. (2006) b 342:1123–1129.[CrossRef][Web of Science][Medline]
Miller J.H., Albertini A.M. J. Mol. Biol. (1983) 164:59–71.[CrossRef][Web of Science][Medline]
Muyldermans S., Atarhouch T., Saldanha J., Barbosa J.A., Hamers R. Protein Eng. (1994) 7:1129–1135.
Nguyen V.K., Hamers R., Wyns L., Muyldermans S. EMBO J. (2000) 19:921–930.[CrossRef][Web of Science][Medline]
O'Connell D., Becerril B., Roy-Burman A., Daws M., Marks J.D. J. Mol. Biol. (2002) 321:49–56.[CrossRef][Web of Science][Medline]
Perez J.M.J., Renisio J.G., Prompers J.J., van Platerink C.J., Cambillau C., Darbon H., Frenken L.G.J. Biochemistry (2001) 40:74–83.[CrossRef][Web of Science][Medline]
Randen I., Potter K.N., Li Y., Thompson K.M., Pascual V., Forre O., Natvig J.B., Capra J.D. Eur. J. Immunol. (1993) 23:2682–2686.[Web of Science][Medline]
Rondot S., Koch J., Breitling F., Dubel S. Nat. Biotechnol. (2001) 19:75–78.[CrossRef][Web of Science][Medline]
Soltes G., Barker H., Marmai K., Pun E., Yuen A., Wiersma E.J. J. Immunol. Methods (2003) 274:233–244.[CrossRef][Web of Science][Medline]
Stanfield R.L., Dooley H., Flajnik M.F., Wilson I.A. Science (2004) 305:1770–1773.
Stanfield R.L., Dooley H., Verdino P., Flajnik M.F., Wilson I.A. J. Mol. Biol. (2007) 367:358–372.[CrossRef][Web of Science][Medline]
Streltsov V.A., Carmichael J.A., Nuttall S.D. Protein Sci. (2005) 14:2901–2909.[CrossRef][Web of Science][Medline]
Tanaka T., Lobato M.N., Rabbitts T.H. J. Mol. Biol. (2003) 331:1109–1120.[CrossRef][Web of Science][Medline]
Tanaka T., Williams R.L., Rabbitts T.H. EMBO J. (2007) 26:3250–3259.[CrossRef][Web of Science][Medline]
Tanha J., Dubuc G., Hirama T., Narang S.A., MacKenzie C.R. J. Immunol. Methods (2002) 263:97–109.[CrossRef][Web of Science][Medline]
Tanha J., Muruganandam A., Stanimirovic D. Methods Mol. Med. (2003) 89:435–449.[Medline]
Tanha J., Nguyen T.D., Ng A., Ryan S., Ni F., MacKenzie R. Protein Eng. Des. Sel. (2006) 19:503–509.
Tanha J., Xu P., Chen Z.G., Ni F., Kaplan H., Narang S.A., MacKenzie C.R. J. Biol. Chem. (2001) 276:24774–24780.
To R., Hirama T., Arbabi-Ghahroudi M., MacKenzie R., Wang P., Xu P., Ni F., Tanha J. J. Biol. Chem. (2005) 280:41395–41403.
van der Linden R.H., Frenken L.G., de Geus B., Harmsen M.M., Ruuls R.C., Stok W., de Ron L., Wilson S., Davis P., Verrips C.T. Biochim. Biophys. Acta (1999) 1431:37–46.[CrossRef][Medline]
Vu K.B., Arbabi-Ghahroudi M., Wyns L., Muyldermans S. Mol. Immunol. (1997) 34:1121–1131.[CrossRef][Web of Science][Medline]
Winter G., Griffiths A.D., Hawkins R.E., Hoogenboom H.R. Annu. Rev. Immunol. (1994) 12:433–455.[Web of Science][Medline]
Yan J.P., Ko J.H., Qi Y.P. Thromb. Res. (2004) 114:205–211.[CrossRef][Web of Science][Medline]
Received August 7, 2008; revised October 28, 2008; accepted October 31, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||