This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.
Agricola
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.

 Previous Article  |  Next Article 

Applied and Environmental Microbiology, January 2005, p. 442-450, Vol. 71, No. 1
0099-2240/05/$08.00+0     doi:10.1128/AEM.71.1.442-450.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.

Isolation of Llama Antibody Fragments for Prevention of Dandruff by Phage Display in Shampoo

Edward Dolk,1 Marcel van der Vaart,2 David Lutje Hulsik,1 Gert Vriend,3 Hans de Haard,2,{dagger} Silvia Spinelli,4 Christian Cambillau,4 Leon Frenken,2 and Theo Verrips1,2*

Department of Molecular and Cellular Biology, University of Utrecht, Utrecht,1 Unilever Research and Development Vlaardingen, Vlaardingen,2 CMBI, KUN, Nijmegen, The Netherlands,3 Architecture et Fonction des Macromolécules Biologiques, UMR 6098 CNRS and Universités Aix-Marseille I & II, Marseille, France4

Received 4 February 2004/ Accepted 9 August 2004


arrow
ABSTRACT
 
As part of research exploring the feasibility of using antibody fragments to inhibit the growth of organisms implicated in dandruff, we isolated antibody fragments that bind to a cell surface protein of Malassezia furfur in the presence of shampoo. We found that phage display of llama single-domain antibody fragments (VHHs) can be extended to very harsh conditions, such as the presence of shampoo containing nonionic and anionic surfactants. We selected several VHHs that bind to the cell wall protein Malf1 of M. furfur, a fungus implicated in causing dandruff. In addition to high stability in the presence of shampoo, these VHHs are also stable under other denaturing conditions, such as high urea concentrations. Many of the stable VHHs were found to contain arginine at position 44. Replacement of the native amino acid at position 44 with arginine in the most stable VHH that lacked this arginine resulted in a dramatic further increase in the stability. The combination of the unique properties of VHHs together with applied phage display and protein engineering is a powerful method for obtaining highly stable VHHs that can be used in a wide range of applications.


arrow
INTRODUCTION
 
In the human scalp there is a very complex balance among many microorganisms. One of these organisms, Malassezia furfur, is an anthropophilic fungus that belongs to the physiological skin flora. Increased turnover of M. furfur has been implicated in the formation of dandruff (6, 11, 38, 42) and other skin diseases, such as psoriasis (1). Significantly, more M. furfur was found on the scalps of people with dandruff than on the scalps of people without dandruff (34, 39). Reductions in the numbers of M. furfur cells on the scalps of dandruff sufferers resulted in a decrease in the severity of the dandruff (32).

To date, the treatment and/or prevention of dandruff has involved the use of chemical antifungal compounds in shampoos (23), compounds such as ketoconazole (33), selenium sulfide (6), cyclopyrox olamine, piroctone olamine, zinc pyrithione, and sulfur-containing substances (38).

Here we describe a novel approach for preventing the formation of dandruff by inhibition of M. furfur with antibodies. For successful use of antibodies in consumer goods they must meet certain requirements regarding cost of production, specificity, affinity, and especially stability under application conditions.

Camelid heavy-chain antibodies have been shown to have great potential in many biotechnological applications (9, 13, 25, 43) because of their unique characteristics involving production, folding, and stability (12, 30). They lack light chains, and therefore the variable domain of the heavy chain (VHH) is the single binding domain (14). The simple, one-domain structure of these VHHs give them unique characteristics, but they have properties with regard to specificity and affinity that are similar to the properties of conventional antibodies (41).

Furthermore, the extralong protruding third binding loop (CDR3) of VHHs is considered an advantage for efficient inhibition of enzymes and small organisms (7, 8, 9, 20). Therefore, VHHs are good candidates for antibody-mediated delivery of antidandruff agents or even direct neutralization of M. furfur.

The objective of this study was to select a series of VHHs that are functional in application conditions and could be used in shampoo for the prevention of dandruff. Phage display is a method commonly used for selection of specific antibody fragments under laboratory conditions (17).

Below we describe selection of VHHs specific for the M. furfur cell surface protein (Malf1) via phage display with the high concentrations of nonionic and anionic surfactants present in shampoos, such as Andrelon and Organics. We also show the importance of a basic amino acid at position 44 for the stability of these antibodies.


arrow
MATERIALS AND METHODS
 
Strains and growth media.
Escherichia coli strain TG1 [F' traD36 lacIq {Delta}(lacZ)M15 proA+B+/supE {Delta}(hsdM-mcrB)5 (rk– mk– McrB–) thi {Delta}(lac-proAB)] was used for phage and plasmid production. This strain was grown in 2TY medium (1.6% tryptone, 1% yeast extract, 86 mM NaCl) with 100 µg of ampicillin per ml and/or 2% (vol/vol) glucose when appropriate.

VHHs were produced in Saccharomyces cerevisiae VWK18 gal1 (36) as described previously (40). Briefly, individual S. cerevisiae colonies were transferred to test tubes containing selective minimal medium (0.7% yeast nitrogen base, 2% glucose) and were grown for 24 h at 30°C. Subsequently, the cultures were diluted 10-fold in YP medium (1% yeast extract, 2% Bacto Peptone [Difco], 2% glucose) supplemented with 2% galactose for induction. After 48 h of induction with galactose, the cells were harvested by centrifugation. Antibody fragments were purified with a 5-ml protein A column (Hi-Trap; Pharmacia) or with TALON (Clontech) by using the His tag according to the manufacturer's protocol.

The M. furfur extract was obtained from John Whitley of the Karolinska Hospital in Stockholm, Sweden. Preparation of this extract has been described by Zargari et al. (44). Recombinant Malf1 (rMalf1) was produced from E. coli JM109-DE3 that was grown overnight at 37°C in 2TY medium and then diluted 100-fold in 2TY medium and grown to an optical cell density at 600 nm (OD600) of 0.6. Protein production was induced with 1 mM (final concentration) isopropyl-ß-D-thiogalactoside (IPTG) (Roche Diagnostics), and production was continued for 2 h at 37°C. Cells were harvested, and rMalf1 was purified from inclusion bodies by using standard protocols (37).

Induction of a humoral immune response in llama.
A llama was immunized subcutaneously and intramuscularly with an extract of M. furfur and with rMalf1 in an oil emulsion (1:9 [vol/vol] antigen in phosphate-buffered saline [PBS]-Specol) (2). Immunizations were performed by using the following time schedule: the second immunization was performed 3 weeks after the first injection, and the third immunization was performed 2 weeks after the second immunization. In each immunization round 0.75 to 1.5 ml of a water-in-oil emulsion containing 1 mg of M. furfur extract or 200 µg of rMalf1 protein was injected. The immune response was monitored by titration of serum samples by an enzyme-linked immunosorbent assay (ELISA) with rMalf1 immobilized on Nunc Maxisorb plates (the coat solution contained 5 µg of rMalf1 per ml diluted in PBS [100 µl/well]). Subsequently, wells were blocked with 4% Marvel in PBS (2 mM NaH2PO4, 13 mM Na2HPO4, 150 mM NaCl; pH 7.4) (200 µl/well) and incubated with serum dilutions in 0.15% Tween 20-PBS. The bound llama antibodies were detected with polyclonal rabbit anti-llama immunoglobulin G (R906, ID-DLO) in 0.15% Tween 20—PBS (100 µl/well) (obtained by immunizing rabbits with llama immunoglobulins purified with ProtA and ProtG columns) and swine anti-rabbit immunoglobulins (Dako) conjugated to horseradish peroxidase (in 0.15% Tween 20—PBS; 100 µl/well). Finally, the peroxidase enzyme activity was determined with tetramethylbenzidine and urea peroxide as the substrates. After termination of the reaction by addition of 50 µl of 1 M H2SO4, the optical density at 450 nm was measured.

Isolation of VHH fragments.
A blood sample (about 200 ml) was taken from an immunized llama. An enriched lymphocyte population was obtained by centrifugation on a Ficoll (Pharmacia) discontinuous gradient. Total RNA was isolated from these cells by acid guanidium thiocyanate extraction (5). After first-strand cDNA synthesis by using Moloney murine leukemia virus reverse transcriptase (Gibco-BRL) and random oligonucleotide primers (Pharmacia), DNA fragments encoding VHH fragments and part of the long or short hinge region were amplified by PCR by using the following specific primers: VH-2B (AGGTSMARCTGCAGSAGTCWGG, where S is C or G, M is A or C, R is A or G, and W is A or T), Lam-07 (AACAGTTAAGCTTCCGCTTGCGGCCGCGGAGCTGGGGTCTTCGCTGTGGTGCG) (short hinge), and Lam-08 (AACAGTTAAGCTTCCGCTTGCGGCCGCTGGTTGTGGTTTTGGTGTCTTGGGTT) (long hinge). The DNA fragments generated by PCR were digested with PstI (coinciding with codons 4 and 5 of the VHH domain, encoding amino acids L and Q; site underlined in the sequence of primer VH-2B) and HindIII (site underlined in primers Lam-07 and Lam-08). The digested PCR products, which were between 300 and 400 bp long (encoding the complete VHH domain but lacking the first and last three codons), were separated by electrophoresis on agarose gels and, after purification from the gel slices with a Qiaex-II extraction kit (QIAGEN), cloned (PstI/HindIII) in the phage display vector pUR4676. This vector was derived from pHEN1 (16) by insertion of the lacI gene and construction of a His6 tag behind the Myc tag.

Selection of VHHs binding to Malf1.
Phage display libraries were created from RNA isolated from two different bleeds. One library was created from blood taken after three immunizations with the M. furfur extract, and the other library was created from blood obtained after six immunizations with the recombinant Malf1 protein. Each library was composed of two sublibraries, one containing DNA fragments derived from short-hinged antibody fragments and one derived from long-hinged antibody fragments (26). After phage production from each of the four sublibraries (M. furfur extract long hinged, M. furfur extract short hinged, rMalf1 long hinged, and rMalf1 short hinged), phages were isolated and pooled.

A display library with 107 clones, 75% of which were estimated to contain a complete VHH-encoding insert (results not shown), was constructed in phagemid vector pUR4676. Phage particles expressing VHH fragments were prepared by infection of E. coli TG1 cells harboring the phagemid with helper phage VCS-M13 (Pharmacia). Free VHHs were removed by precipitation of phage from the culture supernatant with polyethylene glycol 6000, thereby avoiding competition for binding to antigen between phage-bound and free VHH domains. One hundred microliters of phage in 2% Marvel was incubated for 1 h in a Maxisorp tube with immobilized rMalf1. After 20 washes with 200 µl of PBS, the phage particles complexed to rMalf1 via their exposed VHHs were eluted with a pH shock (100 µl of 0.1 M triethylamine, pH 10.0). This phage population, which was enriched for Malf1 binders, was rescued by infection of E. coli host cells (21). Individual E. coli clones were grown in wells of microtiter plates, and the production of VHHs was induced by addition of IPTG (0.1 mM). Culture supernatants containing free VHHs were tested in ELISA experiments for rMalf1 binding by using the Myc tag for detection. OD450 values of >0.2 (twice the background value) were considered positive for binding to Malf1.

Selection of VHHs binding to Malf1 via phage display under application conditions.
The shampoo stability of Malf1 in the ELISA was tested by incubating plates with various concentrations of shampoo 1 (10 to 40% Andrelon) and shampoo 2 (2 to 20% Organics). After washing, all of the ELISA plates were still immunoreactive.

VHH selection under application conditions was performed in the presence of 2 to 40% (vol/vol) shampoo 1 and in the presence of 2 to 20% (vol/vol) shampoo 2. The chemical compositions of these shampoos are available in the supplemental material at http://www.cmbi.kun.nl/articles_ext/. Dilutions of shampoo were made in PBS with 0.1% (wt/vol) Marvel and 0.05% (vol/vol) Tween 20. Selection was performed as described above. Individual E. coli clones were grown, and the production of VHH was induced by addition of IPTG (0.1 mM). Culture supernatants containing free VHHs were tested in ELISA experiments with 10% shampoo 1 or 2% shampoo 2 for binding to rMalf1 by using the Myc tag for detection. OD450 values of >0.2 were considered positive for binding to Malf1.

ELISA.
Maxisorp plates were coated overnight with 10 µg of rMalf1 in 100 µl per well at 4°C. The plates were blocked with 200 µl of 4% Marvel in PBS. Equal amounts of purified VHH (0.33 to 3.3 µM) were incubated for 1 h at room temperature in 2% Marvel-0.05% Tween 20 in PBS or in the desired concentration of shampoo 1, shampoo 2, urea, or guanidine HCl diluted in PBS-0.5% Marvel-0.05% Tween 20 (100 µl/well). After the plates were washed with PBS—0.05% Tween 20, they were incubated with polyclonal rabbit anti-llama immunoglobulin G and swine anti-rabbit immunoglobulins (Dako) conjugated to horseradish peroxidase (in 0.05% Tween 20—PBS; 100 µl/well). Finally, the peroxidase enzyme activity was determined with tetramethylbenzidine and urea peroxide as substrates. The optical density at 450 nm was measured after termination of the reaction by addition of 50 µl of 1 M H2SO4. All experiments were performed at least in duplicate, and typical results are shown below.

Construction of the mutants.
Mutants of VHH-A7 and VHH-D12 were created with PCR by splicing by overlap extension (15). Briefly, complementary primers with nucleotides substituted in order to generate Arg44, Lys44, or Glu44 were used in a two-step PCR to generate DNA fragments containing the desired mutations. The PCR products were digested with restriction enzymes and cloned into the original vector. The correctness of the mutated genes was confirmed by sequence analysis. All numbers and complementary determining region (CDR) designations were assigned as described by Kabat et al. (18).

Surface plasma resonance.
The binding affinities of the VHHs were determined by using a Biacore 3000 (Biacore AB, Uppsala, Sweden) in combination with a CM5 sensor chip (Biacore). Approximately 2,500 response units of rMalf1 was immobilized on different flow cells, and different concentrations of VHHs (5, 25, 50, 250, and 500 nM) were injected at a flow rate of 30 µl/min. Association was measured for 3 min, and dissociation was measured for 15 min. Regeneration was achieved by washing with 10 mM HCl for 3 min. Kd values were calculated by using the BIAevaluation software.

Crystallization conditions.
VHH-OE7 was purified from crude yeast culture broth by affinity chromatography by using a 5-ml protein A-Sepharose FF column (Pharmacia). Crystals of VHH-OE7 were obtained in 10% polyethylene glycol 6000-100 mM HEPES (pH 7.3)-5% 2-methyl-2,4-pentane diol (MPD) by using the vapor diffusion method. One drop was obtained from 1 µl of protein at a concentration of 11 to 15 mg/ml and 1 ml of reservoir solution. The VHH-OE7 crystals belonged to trigonal space group P3221 with the following unit cell dimensions: a = b = 71.40 Å and c = 74.84 Å. They contained one molecule per asymmetric unit giving (assuming a molecular mass of 12,500 Da) a Vm = 4.4 Å3/Da, which corresponded to a solvent content of 72% (22). Crystals were cryoprotected with 7% MPD and flash frozen at 100 K.

Data collection and processing.
Data at a 2.2-Å resolution were collected on the ID14-EH2 beamline (ESRF, Grenoble, France) by using a charge-coupled device detector ({lambda} = 0.9326 Å). The data were indexed and integrated by using Denzo (28) and were scaled by using SCALA, and structure factor amplitudes were calculated by using TRUNCATE (4). The data processing statistics to a resolution of 2.2 Å are shown in Table S1 in the supplemental material (http://www.cmbi.kun.nl/articles_ext/).

Phasing, model building, and refinement.
The crystal structure of the llama VHH-OE7 fragment (PDB 1SJX) was solved by the molecular replacement method with the AmoRe program (27) by using the llama HC-V fragment (PDB code 1HCV) as the search model. A solution giving a correlation coefficient of 0.45 and R factors of 40% was obtained, and the electron density map obtained was readily interpretable.

Preliminary refinement was performed with Crystallography and NMR System software (3) and the high-resolution (2.2-Å) data set by using bulk solvent correction and standard CNS protocols. Cycles of unrestrained, maximum-likelihood refinement and total least square refinement with REFMAC (4) were alternated with manual inspection of the model by using the program Turbo-Frodo (35). Final refinement data are summarized in Table S1 in the supplemental material (http://www.cmbi.kun.nl/articles_ext/).

The complete model has an Rwork of 19% and an Rfree of 20.6% (see Table S1 in supplemental material [http://www.cmbi.kun.nl/articles_ext/]). The stereochemistry was analyzed with Procheck (19), which indicated that 95% of the residues were in the most favorable region and 5% were in the additionally allowed region.

A comparison with other VHHs was performed by using VHH-R2 (anti-RR6; PDB 1QD0), VHH-R9 (anti-RR6; PDB 1SJV), Car24 (PDB 1U0Q), Amy-B07 (antiamylase; PDB 1KXT), Amy-D08 (antiamylase; PDB 1KXQ), and Amy-D10 (antiamylase; PDB 1KXV).


arrow
RESULTS
 
Initial selection of VHHs that can bind to Malf1.
A VHH phage display library with 107 clones was created from the blood of a llama immunized with Malf1. The results of two rounds of phage display with this library obtained by using Malf1 for selection and PBS buffer in the binding step are shown in Table 1. The second round consisted of three parallel selections performed with the output of experiment C of the first step.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Phage titers, enrichment, and percentages of binders for the two rounds of selection for Malf1 binding in PBSa

Nearly 200 monoclonal VHHs were expressed, and the Malf1 binding of these fragments was tested in ELISA experiments. The percentages of VHHs capable of binding to Malf1 were determined (Table 1). Ten VHHs were selected and tested for Malf1 binding under application conditions. Figure 1A shows that all of the VHHs selected could bind to Malf1 in PBS. Figures 1B and C show that only one VHH could bind to Malf1 in shampoo.



View larger version (25K):
[in this window]
[in a new window]
  		FIG. 1. ELISA experiments to determine the Malf1 binding capacity of 10 randomly selected VHHs after two rounds of selection in PBS. (A) Binding in PBS; (B) binding in 10% shampoo 1; (C) binding in 2% shampoo 2. Four dilutions of the periplasmic fraction of bacteria producing VHHs (see Materials and Methods for the procedures used for preparation) were tested for binding. VHHs were measured.

Phage display in shampoo.
In order to isolate VHHs that are able to bind to Malf1 in shampoo, two more rounds of selection were performed in the presence of shampoo. The first round was started with the output of experiment C (Table 1).

Monoclonal VHHs from these selections were expressed, and the Malf1 binding of these VHHs was tested in the respective shampoos (Table 2). Ten positive VHHs were picked from experiment F2 (2% shampoo 2), and 10 positive VHHs were picked from experiment G3 (10% shampoo 1). These 20 VHHs were tested for binding to Malf1 in PBS and in the shampoo used for selection (Fig. 2). Eight of the 10 shampoo 1-selected VHHs and 8 of the 10 shampoo 2-selected VHHs bound to Malf1 equally well in the respective shampoos (Fig. 2C and D). In both cases one VHH bound significantly worse, and one VHH bound significantly better. All but one shampoo-selected VHH could bind to Malf1 in PBS (Fig. 2A and B), indicating that the shampoo-selected VHHs did not require shampoo for binding.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Phage titers, enrichment, and percentages of binders for the two rounds of selection for Malf1 binding in shampooa



View larger version (29K):
[in this window]
[in a new window]
  		FIG. 2. ELISA experiments to determine the Malf1 binding capacity of 10 randomly selected VHHs after two rounds of selection in shampoo 1 or shampoo 2. (A) Binding of shampoo 2-selected VHHs in PBS; (B) binding of shampoo 1-selected VHHs in PBS; (C) binding of shampoo 1-selected VHHs in 10% shampoo 1; (D) binding of shampoo 2-selected VHHs in 2% shampoo 2. Four dilutions of the periplasmic fraction of bacteria producing VHHs (see Materials and Methods for the procedures used for preparation) were tested for binding.

Only one of the PBS-selected VHHs could bind to Malf1 in shampoo. To further explore the difference in binding in shampoo, the Malf1 binding of the 10 PBS-selected clones in different concentrations of the two shampoos and the Malf1 binding of the 20 shampoo-selected clones in different concentrations of the respective shampoos were examined. It was found that all shampoo 1-selected VHHs bound to Malf1 in up to 40% shampoo 1 and that all shampoo 2-selected VHHs bound to Malf1 in up to 5% shampoo 2. The one PBS-selected VHH that could bind to Malf1 in shampoo did so as well as the shampoo-selected VHHs except that it bound at higher concentrations of shampoo 1 (data not shown).

For unknown reasons, it was easier to select for VHHs that could bind to Malf1 in high concentrations shampoo 1 than it was to select for VHHs that could bind to Malf1 in high concentrations of shampoo 2. We therefore concentrated on shampoo 1 during the remainder of our research. This choice was mainly based on the fact that high concentrations (i.e., 10 to 25%) of shampoo are used in application conditions.

Crystal structure.
The crystal structure of a shampoo-selected VHH (VHH-OE7; PDB 1SJX) was determined in order to investigate whether the structure of the shampoo-selected VHHs was different from that of VHHs selected under physiological conditions.

The VHH-OE7 structure is similar to that of other VHHs, except for the three complementary determining regions (Fig. 3A). In particular, loop 39-45 has a very well conserved conformation (Fig. 3B). The precise epitope of VHH-OE7 is unknown. In the crystal structure, however, two crevices were identified close to the CDRs. The first crevice is located on top of the VHH (Fig. 4A), between CDR1 and CDR2, and contains a serendipitous MPD molecule from the crystallization liquor. The second crevice, located on the VHH side between CDR1 and CDR3 and close to the beginning of CDR2 (Fig. 4B), also contains an MPD molecule. Above these two crevices, two stretches of residues formed by CDR1 and CDR3 protrude from the VHH body. These stretches are positively charged (Fig. 4C), since they contain Arg 30, Arg 97, and Arg 99, as well as His 98 (Fig. 4C).



View larger version (29K):
[in this window]
[in a new window]
  		FIG. 3. Conformation of VHH-OE7 compared to the conformations of other VHHs (VHH-R2, VHH-R9, Car24, Amy-B07, Amy-D08, and Amy-D10). (A) Overall fold of the framework. The CDR area has been removed for clarity. (B) Comparison of the conformation of loop 38-46 of VHH-OE7 with the conformation of loop 38-46 of VHH-R2, VHH-R9, Car24, Amy-B07, Amy-D08, and Amy-D10. The conserved Arg38 and Arg45 residues are indicated, as is position 44.



View larger version (38K):
[in this window]
[in a new window]
  		FIG. 4. Molecular surface of VHH-OE7, with the three CDRs indicated by red, green, and blue. (A) Top view of the combining area with a MPD molecule bound (see explanation in the text). (B) Side view of the VHH with the second MPD molecule bound. (C) Top view of the VHH with the positively charged patch formed by CDR1 and CDR3 (blue).

The conformation of the 38-45 loop is very well conserved, irrespective of the residues at position 44. Gln44 points toward the solvent in VHH-OE7, as well as all the side chains in the other VHHs. This Gln is also present in VHH-R9, but in VHH-R2, Car24, Amy-B07, and Amy-D08 there is a Glu residue at position 44 and in Amy-D10 the amino acid at position 44 is Gly. With the exception of Amy-B07, which has a Cys residue forming a disulfide bridge with its CDR3 (8), there is an arginine at position 45. None of these substitutions, however, influences the loop or the framework conformations.

Sequence analysis.
The shampoo 1-selected VHH-D12 bound to Malf1 significantly better at high shampoo concentrations than the 29 other VHHs tested, but the binding of this VHH to Malf1 in PBS was similar to that of all other VHHs. This suggests that VHH-D12 has one or more mutations that are beneficial for Malf1 binding in shampoo. Furthermore, PBS-selected VHH-A7 was the only VHH selected in PBS that was able to bind to Malf1 in shampoo. This suggests that VHH-A7 also has one or more mutations beneficial for Malf1 binding in shampoo. The sequences of these VHHs, six PBS-selected Malf1-binding VHHs, and 11 shampoo-selected Malf1-binding VHHs were determined (Fig. 5).



View larger version (46K):
[in this window]
[in a new window]
  		FIG. 5. Sequences of VHHs selected in the presence of PBS or in the presence of shampoo. S1 to S9 are VHHs that bind to Malf1 well in shampoo 1, and S10 and S11 bind well in shampoo 2. P1 to P6 bind to Malf1 well in PBS. VHH-D12 is the best shampoo-selected Malf1 binder in shampoo, and VHH-A7 is the best PBS-selected Malf1 binder in shampoo. CDRs and framework regions are indicated. The five positions where VHH-D12 differs from many shampoo-selected VHHs are enclosed in boxes. The three positions where VHH-A7 differs from VHH-D12 are indicated by asterisks. A phylogenetic tree of the 19 sequences is available in the supplemental material at http://www.cmbi.kun.nl/articles_ext/.

Not surprisingly, the variability among the PBS-selected sequences was considerably higher than the variability among the shampoo-selected sequences, especially in the CDRs (Fig. 5). The CDRs of the shampoo-selected VHHs were on average more hydrophilic than the CDRs of the PBS-selected VHHs. This suggests that the hydrophobic tails of the soap-like shampoo components might shield interactions of antibodies that recognize more hydrophobic epitopes.

The sequence comparison revealed five positions where VHH-D12 differs from a significant number of the other shampoo-selected VHHs (Fig. 5). Three of these positions are located in the antigen binding loops (CDRs), which make them poor candidates for further analysis in mutation studies, because it is likely that these mutations also (negatively) affect the binding affinity. Two of these positions, R44 and K75, are located in the framework. In 684 VHH sequences, determined in unrelated experiments with llama antibody fragments, arginine occurs at position 44 only seven times, whereas lysine was observed at position 75 in 528 VHHs. This suggests that R44 might be (partially) responsible for the good Malf1 binding in shampoo.

To study this effect further, the residues at position 44 in VHH-D12 and VHH-A7 were exchanged. Figure 6 shows the Malf1 binding of wild-type VHH-A7 and VHH-D12 and the VHH-A7-Q44R and VHH-D12-R44Q mutants. There was a clear correlation between the presence of arginine at position 44 and Malf1 binding. However, this effect was much stronger in shampoo than in PBS.



View larger version (15K):
[in this window]
[in a new window]
  		FIG. 6. ELISA experiments for determining Malf1 binding of four different concentrations of the wild-type VHHs VHH-D12 and VHH-A7 and the mutants VHH-D12-R44Q and VHH-A7-Q44R in the presence of PBS (A) and in the presence of 10% shampoo 1 (B).

To investigate whether the stability caused by R44 was shampoo specific or more general, the Malf1 binding of the four VHHs shown if Fig. 6 was tested in the presence of guanidine HCl and urea (Fig. 7). In both denaturants the Malf1 binding was better for the R44 variants than for the Q44 variants. With the exception of VHH-A7-Q44R, which bound Malf1 in 3.5 M urea, binding was not observed at guanidine HCl concentrations greater than 2 M or at urea concentrations greater than 2.5 M.



View larger version (13K):
[in this window]
[in a new window]
  		FIG. 7. ELISA experiments to determine Malf1 binding of the wild-type VHHs VHH-D12 and VHH-A7 and the mutants VHH-D12-R44Q and VHH-A7-Q44R in the presence of increasing concentrations of guanidine HCl (range, 0.5 to 3.5 M) (A) and in the presence of increasing concentrations of urea (range, 0.5 to 3.5 M) (B).

To exclude the possibility that differences in affinity for Malf1 caused the enhanced binding of the VHHs containing R44, binding affinities were determined by performing surface plasma resonance experiments. VHH-D12 and VHH-D12-R44Q showed identical affinities for Malf1 (75 and 73 nM, respectively). This indicates that differences in affinity alone cannot explain the change in binding capacity in shampoo. However, the affinities of VHH-A7 and VHH-A7-Q44R (150 and 15 nM, respectively) indicate that there might be a role for position 44 in binding to Malf1.

In order to shed light on the enhanced Malf1 binding of VHH variants with arginine at position 44, we also produced the variants VHH-A7-Q44K and VHH-D12-R44K. Arginine and lysine both are positively charged residues, so these variants could indicate if the enhanced Malf1 binding in shampoo is caused by the introduction of a positive charge or by specific characteristics of arginine. Lysine and arginine are structurally comparable; however, arginine has a relatively rigid side chain, whereas lysine has a relatively flexible side chain. At pH 11, arginine is expected to be positively charged, whereas lysine is expected to be only partially charged at this pH.

Figure 8 shows that the Malf1 binding of the Q44, K44, and R44 VHH variants was significantly reduced at pH 11 in both PBS and shampoo 1. This was not unexpected, as proteins tend to be less stable at this pH. At pH 11 the R44 variants still bound to Malf1 very well in 10% shampoo 1, while the Malf1 binding of the K44 variants was reduced by a factor of about 2. In PBS, on the other hand, the reductions in the binding of Malf1 were about the same for all four variants. This experiment strongly suggests that the enhanced Malf1 binding in shampoo was caused by the introduction of a positive charge at position 44.



View larger version (18K):
[in this window]
[in a new window]
  		FIG. 8. ELISA experiments to determine Malf1 binding of the wild-type VHHs VHH-A7 and VHH-D12 and the mutants VHH-A7-Q44K and VHH-D12-R44K in the presence of PBS at pH 7 (A), in the presence of shampoo at pH 7 (B), in the presence of PBS at pH 11 (C), and in the presence of shampoo at pH 11 (D).


arrow
DISCUSSION
 
Phage display was successfully performed for isolation of Malf1-specific VHH antibody fragments. Although many VHHs showed binding to Malf1 under physiological conditions, most of them did not bind to Malf1 in the presence of shampoo. As Malf1 binding is also strongly reduced in the presence of denaturants such as urea or guanidine HCl, it seems likely that stability of the VHH is the main determinant for Malf1 binding.

We used phage display in shampoo to select a series of VHHs that are functional in shampoo. It has been observed many times that mutations that protect a protein against one type of external stress also protect the protein against other forms of external stress (10, 29). Indeed, the VHH variants that were selected for stability in shampoo also showed enhanced resilience in the presence of urea and guanidine HCl.

It was possible to perform phage display in the presence of high concentrations of shampoo to select VHHs that can function well in shampoo. It was surprising that this procedure worked under such harsh conditions without any major modifications to the standard phage display protocols. The growing interest in industry for antibody-based applications has generated a demand for rapid selection and isolation of antibody fragments which have high affinity and are stable in the environment in which they should perform their function. Antibodies are often used as therapeutics and should be functional in the gastrointestinal tract with proteases and at low pH (31). Addition of proteases and low pH to the selection procedure could increase stability under gastrointestinal conditions.

In this study we showed that selection pressure can improve VHH performance in a predictable direction. This could eradicate time-consuming improvement of antibody fragments by mutagenesis (24).

Phage display helped us detect a series of VHHs with enhanced binding properties. The variability within the sequence families provides a good impression of the possible motifs involved in protein stability and binding. Multiple-sequence alignment analyses of VHH families can reveal potent sites for directed mutagenesis experiments that can either explain the observations or further improve the VHH variants obtained.

From the analysis of the sequences of 12 shampoo-selected VHHs and seven PBS-selected VHHs, we concluded that an arginine at position 44 is important for VHH functioning in shampoo. This was confirmed by mutagenesis studies. Introduction of a lysine at position 44 had equally beneficial effects, indicating that the positive charge is more important than arginine-specific characteristics. This is somewhat surprising in light of the observation that lysine was never observed at position 44 in the VHH variants analyzed in this study and was observed only four times in 684 VHH variants isolated in previous selection studies that had no relation to shampoo. Thus, even though selection by phage display can quickly produce good VHH variants, classical mutagenesis work can still add more information.

There was no significant difference in affinity for Malf1 between VHH-D12 and VHH-D12-R44Q. Although there was a significant difference in affinity for Malf1 between VHH-A7 and VHH-A7-Q44R, it is unlikely that this was the sole reason for the increased binding observed. This conclusion was supported by investigation of the crystal structure of a shampoo-selected VHH. Loop 38-45 is very well conserved in all VHHs, including VHH-OE7, which makes it unlikely that mutation at position 44 is involved in Malf1 binding. However, it is still possible that mutations at this position have an influence on the position and flexibility of the CDR loops and thereby affect binding.

The influence of the Gln->Arg mutation is most probably an electrostatic effect. This conclusion is supported by the fact that a lysine residue induces the same stabilization effect. Indeed, one should keep in mind that the harsh medium in which the llama antibodies are raised contains a high concentration of negatively charged molecules. This fact might explain the charged apex CDR1 and CDR3 of VHH-OE7, as well as the positive effect of the framework mutation. The fact that this residue is barely observed as a naturally occurring residue indicates that there was selection of a rare subpopulation of VHHs, induced by the harsh medium conditions.

The final goal of our project is the production of VHHs that can be added to shampoo to reduce dandruff. We produced VHHs that can bind to Malf1 under realistic shampoo conditions. Whether this binding also leads to inhibition of growth of M. furfur on the scalp is still to be proven. Even if this were not the case, the VHHs can still be used to deliver antidandruff agents to the surface of M. furfur.


arrow
ACKNOWLEDGMENTS
 
We thank John Whitley for providing the M. furfur extract, Nathalie van Egmond for affinity measurements, John Chapman for critical reading of the manuscript, and Janice Perez for fruitful discussions.


arrow
FOOTNOTES
 
* Corresponding author. Mailing address: Department of Molecular and Cellular Biology, University of Utrecht, Padualaan 8, 3584 CH Utrecht, The Netherlands. Phone: 31-30-2533421. Fax: 31-30-2513655. E-mail: c.t.verrips{at}bio.uu.nl. Back

{dagger} Present address: Ablynx NV, B-9052 Ghent, Belgium. Back


arrow
REFERENCES
 
    1
  1. Baroni, A., I. Paoletti, E. Ruocco, M. Agozzino, M. A. Tufano, and G. Donnarumma. 2004. Possible role of Malassezia furfur in psoriasis: modulation of TGF-beta1, integrin, and HSP70 expression in human keratinocytes and in the skin of psoriasis-affected patients. J. Cutan. Pathol. 31:35-42.[CrossRef][Medline]
    2. 2
  2. Bokhout, B. A., A. T. Bianchi, P. J. van der Heijden, J. W. Scholten, and W. Stok. 1986. The influence of a water-in-oil emulsion on humoral immunity. Comp. Immunol. Microbiol. Infect. Dis. 9:161-168.[CrossRef][Medline]
    3. 3
  3. Brunger, A. T., P. D. Adams, G. M. Clore, W. L. DeLane, P. Gros, R. W. Grosse-Kustleve, J. S. Jiang, J. Kuszewski, M. Nilges, N. S. Panun, R. J. Read, L. M. Rice, T. Simonson, and G. L. Warren. 1998. Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. Sect. D 54:905-921.[CrossRef][Medline]
    4. 4
  4. CCP4: Collaborative Computational Project No. 4. 1994. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. Sect. D 50:760-763.[CrossRef][Medline]
    5. 5
  5. Chomczynski, P., and N. Sacchi. 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162:156-159.[Medline]
    6. 6
  6. Danby, F. W., W. S. Maddin, L. J. Margesson, and D. Rosenthal. 1993. A randomized, double-blind, placebo-controlled trial of ketoconazole 2% shampoo versus selenium sulfide 2.5% shampoo in the treatment of moderate to severe dandruff. J. Am. Acad. Dermatol. 29:1008-1012.[Medline]
    7. 7
  7. Dekker, S., W. Toussaint, G. Panayotou, T. de Wit, P. Visser, F. Grosveld, and D. Drabek. 2003. Intracellularly expressed single-domain antibody against p15 matrix protein prevents the production of porcine retroviruses. J. Virol. 77:12132-12139.[Abstract/Free Full Text]
    8. 8
  8. Desmyter, A., K. Decanniere, S. Muyldermans, and L. Wyns. 2001. Antigen specificity and high affinity binding provided by one single loop of a camel single-domain antibody. J. Biol. Chem. 276:26285-26290.[Abstract/Free Full Text]
    9. 9
  9. Dumoulin, M., A. M. Last, A. Desmyter, K. Decanniere, D. Canet, G. Larsson, A. Spencer, D. B. Archer, J. Sasse, S. Muyldermans, L. Wyns, C. Redfield, A. Matagne, C. V. Robinson, and C. M. Dobson. 2003. A camelid antibody fragment inhibits the formation of amyloid fibrils by human lysozyme. Nature 424:783-788.[CrossRef][Medline]
    10. 10
  10. Eijsink, V. G., B. W. Dijkstra, G. Vriend, J. R. van der Zee, O. R. Veltman, B. van der Vinne, B. van den Burg, S. Kempe, and G. Venema. 1992. The effect of cavity-filling mutations on the thermostability of Bacillus stearothermophilus neutral protease. Protein Eng. 5:421-426.[Abstract/Free Full Text]
    11. 11
  11. Faergemann, J. 1985. Lipophilic yeasts in skin-disease. Semin. Dermatol. 4:173-184.
    12. 12
  12. Frenken, L. G., J. G. Hessing, C. A. Van den Hondel, and C. T. Verrips. 1998. Recent advances in the large-scale production of antibody fragments using lower eukaryotic microorganisms. Res. Immunol. 149:589-599.[CrossRef][Medline]
    13. 13
  13. Frenken, L. G., R. H. van der Linden, P. W. Hermans, J. W. Bos, R. C. Ruuls, B. de Geus, and C. T. Verrips. 2000. Isolation of antigen specific llama VHH antibody fragments and their high level secretion by Saccharomyces cerevisiae. J. Biotechnol. 78:11-21.[CrossRef][Medline]
    14. 14
  14. Hamers-Casterman, C., T. Atarhouch, S. Muyldermans, G. Robinson, C. Hamers, E. B. Songa, N. Bendahman, and R. Hamers. 1993. Naturally occurring antibodies devoid of light chains. Nature 363:446-448.[CrossRef][Medline]
    15. 15
  15. Ho, S. N., H. D. Hunt, R. M. Horton, J. K. Pullen, and L. R. Pease. 1989. Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene 77:51-59.[CrossRef][Medline]
    16. 16
  16. Hoogenboom, H. R., A. D. Griffiths, K. S. Johnson, D. J. Chiswell, P. Hudson, and G. Winter. 1991. Multi-subunit proteins on the surface of filamentous phage: methodologies for displaying antibody (Fab) heavy and light chains. Nucleic Acids Res. 19:4133-4137.[Abstract/Free Full Text]
    17. 17
  17. Hoogenboom, H. R., A. P. de Bruine, S. E. Hufton, R. M. Hoet, J. W. Arends, and R. C. Roovers. 1998. Antibody phage display technology and its applications. Immunotechnology 4:1-20.[CrossRef][Medline]
    18. 18
  18. Kabat, E. A., T. T. Wu, H. M. Perry, K. S. Gottesman, and C. Foeller. 1991. Sequences of proteins of immunological interest, 5th ed. Publication no. 91-3242. National Institutes of Health, Public Health Service, U.S. Department of Health and Human Services, Bethesda, Md.
    19. 19
  19. Laskowski, R. A., M. W. MacArthur, D. S. Moss, and J. M. Thornton. 1993. PROCHECK: a program to check stereochemical quality of protein structures. J. Appl. Crystallogr. 26:283-291.[CrossRef]
    20. 20
  20. Ledeboer, A. M., S. Bezemer, J. J. de Hiaard, I. M. Schaffers, C. T. Verrips, C. van Vliet, E. M. Dusterhoft, P. Zoon, S. Moineau, and L. G. Frenken. 2002. Preventing phage lysis of Lactococcus lactis in cheese production using a neutralizing heavy-chain antibody fragment from llama. J. Dairy Sci. 85:1376-1382.[Abstract]
    21. 21
  21. Marks, J. D., H. R. Hoogenboom, T. P. Bonnert, J. McCafferty, A. D. Griffiths, and G. Winter. 1991. By-passing immunization. Human antibodies from V-gene libraries displayed on phage. J. Mol. Biol. 222:581-597.
    22. 22
  22. Matthews, B. W. 1968. Solvent content of protein crystals. J. Mol. Biol. 33:491-497.[Medline]
    23. 23
  23. Mayser, P., H. Argembeaux, and F. Rippke. 2003. The hair strand test—a new method for testing antifungal effects of antidandruff preparations. J. Cosmet. Sci. 54:263-270.[Medline]
    24. 24
  24. McDonagh, C. F., K. S. Beam, G. J. Wu, J. H. Chen, D. F. Chace, P. D. Senter, and J. A. Francisco. 2003. Improved yield and stability of L49-sFv-beta-lactamase, a single-chain antibody fusion protein for anticancer prodrug activation, by protein engineering. Bioconjug. Chem. 14:860-869.[CrossRef][Medline]
    25. 25
  25. Muyldermans, S. 2001. Single domain camel antibodies: current status. J. Biotechnol. 74:277-302.[CrossRef][Medline]
    26. 26
  26. Muyldermans, S., C. Cambillau, and L. Wyns. 2001. Recognition of antigens by single-domain antibody fragments: the superfluous luxury of paired domains. Trends Biochem. Sci. 26:230-235.[CrossRef][Medline]
    27. 27
  27. Navaza, J. 1992. AMORE, p. 87-90. In E. J. Dodson, S. Gover, and W. Wolf (ed.), Proceedings of Daresbury Study Weekend on Molecular Replacement. SERC Daresbury Laboratory, Warington, United Kingdom.
    28. 28
  28. Otwinowski, Z., and W. Minor. 1997. Processing of X-ray diffraction data collected on oscillation mode. Methods Enzymol. 276:307-326.
    29. 29
  29. Parsell, D. A., and R. T. Sauer. 1989. The structural stability of a protein is an important determinant of its proteolytic susceptibility in Escherichia coli. J. Biol. Chem. 264:7590-7595.[Abstract/Free Full Text]
    30. 30
  30. Perez, J. M., J. G. Renisio, J. J. Prompers, C. J. van Platerink, C. Cambillau, H. Darbon, and L. G. Frenken. 2001. Thermal unfolding of a llama antibody fragment: a two-state reversible process. Biochemistry 40:74-83.[CrossRef][Medline]
    31. 31
  31. Petschow, B. W., and R. D. Talbott. 1994. Reduction in virus-neutralizing activity of a bovine colostrum immunoglobulin concentrate by gastric acid and digestive enzymes. J. Pediatr. Gastroenterol. Nutr. 19:228-235.[Medline]
    32. 32
  32. Pierard-Franchimont, C., J. E. Arrese, G. Durupt, G. Ries, G. Cauwenbergh, and G. E. Pierard. 1998. Correlation between Malassezia spp. load and dandruff severity. J. Mycol. Med. 8:83-86.
    33. 33
  33. Pierard-Franchimont, C., G. E. Pierard, J. E. Arrese, and P. De Doncker. 2001. Effect of ketoconazole 1% and 2% shampoos on severe dandruff and seborrhoeic dermatitis: clinical, squamometric and mycological assessments. Dermatology 202:171-176.[CrossRef][Medline]
    34. 34
  34. Priestley, G. C., and J. A. Savin. 1976. The microbiology of dandruff. Br. J. Dermatol. 94:469-471.[CrossRef][Medline]
    35. 35
  35. Roussel, A., and C. Cambillau. 1991. The Turbo-Frodo graphics package, p. 81. In Silicon Graphics geometry partners directory. Silicon Graphics, Mountain View, Calif.
    36. 36
  36. Sagt, C. M., W. H. Muller, J. Boonstra, A. J. Verkleij, and C. T. Verrips. 1998. Impaired secretion of a hydrophobic cutinase by Saccharomyces cerevisiae correlates with an increased association with immunoglobulin heavy-chain binding protein (BiP). Appl. Environ. Microbiol. 64:316-324.[Abstract/Free Full Text]
    37. 37
  37. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
    38. 38
  38. Schmidt, A. 1997. Malassezia furfur: a fungus belonging to the physiological skin flora and its relevance in skin disorders. Cutis 59:21-24.[Medline]
    39. 39
  39. Shuster, S. 1984. The aetiology of dandruff and the mode of action of therapeutic agents. Br. J. Dermatol. 111:235-242.[Medline]
    40. 40
  40. Thomassen, Y. E., W. Meijer, L. Sierkstra, and C. T. Verrips. 2002. Large-scale production of VHH antibody fragments by Saccharomyces cerevisiae. Enzyme Microb. Technol. 30:273-278.[CrossRef]
    41. 41
  41. Van der Linden, R. H., L. G. Frenken, B. de Geus, M. M. Harmsen, R. C. Ruuls, W. Stok, L. de Ron, S. Wilson, P. Davis, and C. T. Verrips. 1999. Comparison of physical chemical properties of llama VHH antibody fragments and mouse monoclonal antibodies. Biochim. Biophys. Acta 1431:37-46.[CrossRef][Medline]
    42. 42
  42. Van der Wyk, R., and K. E. Hechemy. 1967. A Comparison of the bacterial and yeast flora of the human scalp and their effect upon dandruff production. J. Soc. Cosmet. Chem. 18:629-639.
    43. 43
  43. Verheesen, P., M. R. ten Haaft, N. Lindner, C. T. Verrips, and J. J. de Haard. 2003. Beneficial properties of single-domain antibody fragments for application in immunoaffinity purification and immuno-perfusion chromatography. Biochim. Biophys. Acta 1624:21-28.[Medline]
    44. 44
  44. Zargari, A., B. Harfast, S. Johansson, S. G. Johansson, and A. Scheynius. 1994. Identification of allergen components of the opportunistic yeast Pityrosporum orbiculare by monoclonal antibodies. Allergy 49:50-56.[Medline]

Applied and Environmental Microbiology, January 2005, p. 442-450, Vol. 71, No. 1
0099-2240/05/$08.00+0     doi:10.1128/AEM.71.1.442-450.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.



This article has been cited by other articles:

  • Forsman, A., Beirnaert, E., Aasa-Chapman, M. M. I., Hoorelbeke, B., Hijazi, K., Koh, W., Tack, V., Szynol, A., Kelly, C., McKnight, A., Verrips, T., Haard, H. d., Weiss, R. A. (2008). Llama Antibody Fragments with Cross-Subtype Human Immunodeficiency Virus Type 1 (HIV-1)-Neutralizing Properties and High Affinity for HIV-1 gp120. J. Virol. 82: 12069-12081 [Abstract] [Full Text]  
  • Hagihara, Y., Mine, S., Uegaki, K. (2007). Stabilization of an Immunoglobulin Fold Domain by an Engineered Disulfide Bond at the Buried Hydrophobic Region. J. Biol. Chem. 282: 36489-36495 [Abstract] [Full Text]  
  • El Khattabi, M., Adams, H., Heezius, E., Hermans, P., Detmers, F., Maassen, B., van der Ley, P., Tommassen, J., Verrips, T., Stam, J. (2006). Llama Single-Chain Antibody That Blocks Lipopolysaccharide Binding and Signaling: Prospects for Therapeutic Applications. CVI 13: 1079-1086 [Abstract] [Full Text]  
  • Regan-Klapisz, E., Sorokina, I., Voortman, J., de Keizer, P., Roovers, R. C., Verheesen, P., Urbe, S., Fallon, L., Fon, E. A., Verkleij, A., Benmerah, A., van Bergen en Henegouwen, P. M. P. (2005). Ubiquilin recruits Eps15 into ubiquitin-rich cytoplasmic aggregates via a UIM-UBL interaction. J. Cell Sci. 118: 4437-4450 [Abstract] [Full Text]  

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.
Agricola
Right arrow Articles by Dolk, E.
Right arrow Articles by Verrips, T.

Copyright © 2009 by the American Society for Microbiology.   For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»