Use of Phage Antibodies to Distinguish Closely Related Species of Protozoan Parasites

Acanthamoeba are typically identified in the laboratory using culture and microscopic observation. In this paper we describe the isolation and specificity of antibody fragments that can be used for the identification of Acanthamoeba. A phage library expressing a large repertoire (approx. 5×109) of antibody fragments was used to generate two libraries one enriched for bacteriophage that exhibit genus specific binding and the other containing bacteriophage that bind specifically to pathogenic Acanthamoeba. Individual clones were isolated on the basis of binding by ELISA, and then flowcytometry and immunofluorescence were used for further characterisation. Four monoclonal antibodies were isolated, specific for Acanthamoeba at the generic level with clone HPPG6 exhibiting the highest level of binding. Furthermore clone HPPG55 was specific for pathogenic species of Acanthamoeba.


Introduction
Methods used for the detection of parasites have undergone a rapid change over the past ten years; however morphology is still the single most useful diagnostic character. Species determination of non-culturable or morphologically similar organisms relies on the use of PCR based technology or monoclonal antibodies (mAb), the latter typically being used in either an ELISA format or for fluorescence microscopy. Both methodologies are used in laboratories for the detection and identification of parasites such as Plasmodium, Leishmania and Cryptosporidium, but each has some disadvantages. mAb often show cross reactivity within a genus or, in extreme cases, between different genus groups (as seen with cryptosporidium) [8]; PCR is more reliable although the cost is prohibitively high when small numbers of assays are performed annually. For those organisms that are difficult to culture or are nonculturable in vitro, identification of phenotypic characteristics such as drug resistance or pathogenicity are difficult using current technology (see Freeman "Images and antibodies -the potential of bacteriophage antibody display to clinical microbiology"). A good example of this problem is seen with the identification of Acanthamoeba infections in man.
Acanthamoeba is a free-living opportunistic protozoan parasite of man, which can cause a fatal meningoencephalitis disease, but is more commonly associated with eye infections (Acanthamoeba keratitis). The number of cases annually in the UK of Acanthamoeba keratitis is approximately 150. Acanthamoeba keratitis is typically diagnosed after viral and bacterial causes have been eliminated [1,5] and, as a result, there is a significant delay before appropriate treatment is administered. Because of the severity of Acanthamoeba keratitis a significant loss of visual acuity is common and in many cases total loss of sight in the eye occurs [4,6]. Although commonly isolated, only a limited number of Acanthamoeba are pathogenic thus, the development of a rapid, simple detection method for such strains is important [2,17]. Current methods of detection involve culture and microscopic identification [16]. These methods are time consuming, laborious and open to error. This paper outlines our approach for generating antibody fragments that can be used to differentiate pathogenic Acanthamoeba either by immunofluorescence microscopy or flow cytometry. Table 1 Acanthamoeba species used in this study. The pathogenicity of the strains was determined by observing cytopathic effects and is taken from Khan et al. [10].

Cell cultures
Pathogenic and non-pathogenic Acanthamoeba species were either obtained from CCAP (Culture Collection of Algae and Protozoa) or from Dr. S. Kilvington (Leicester PHL). All Acanthamoeba spp. were maintained as described by Khan et al. [10]. The Acanthamoeba species used in this study are shown in Table 1. For all experiments, Acanthamoeba were harvested by centrifugation at 800 × g for 8 min and resuspended in PAS (Page's amoeba saline) [10]. Centrifugation was repeated a further three times. Cells used for isolating antibody fragments, ELISA and FACS analysis were fixed in 50% (v/v) methanol in PAS. After fixation, cells were washed three times with PAS as described above. Organisms used for immunofluorescence microscopy were harvested and placed directly onto slides prior to fixation.

Scanning Electron Microscopy (SEM)
Cells were prepared for SEM using the following protocol: coverslips were cleaned by soaking in 100% ethanol, wiped with lint free paper and air dried for at least 1 h. A drop of 0.1% (w/v) poly-l-lysine was placed on each coverslip, dried overnight, and then rinsed in distilled water. Acanthamoeba in PAS were placed on the coverslip and left for 5-10 min to allow cells to adhere. The coverslips were washed with 2.5% (v/v) gluteraldehyde at 4 • C, and left overnight. The following day, they were washed again with ice cold PAS and left overnight. Samples were finally washed with distilled water at room temperature and left for 30 min. Coverslips were then dehydrated in a series of acetone and water mixtures from 30% acetone in water to 100% acetone. Samples were finally heated at 36 • C, mounted, and observed under SEM.

Bacteriophage display library
The semi-synthetic ScFv library no. 1 (Nissim Library) was obtained from G. Winter (Centre for Protein Engineering, Medical Research Council Centre, Cambridge, United Kingdom) [14].

Preparation of bacteriophage particles
For bacteriophage preparation, the library stock or individual phage clones were added to a culture of Escherichia coli (TG1) grown in 2 × TY (0.8% (w/v) NaCl, 1.6% (w/v) tryptone, 0.5% (w/v) yeast extract) supplemented with 100 µg/ml ampicillin and 1% (w/v) glucose. The culture was then incubated at 37 • C until the absorbance at 600 nm was between 0.4-0.5. VCS-M13 helper phage was added to this culture and incubated for a further 30 min at 37 • C without shaking. Cells were then centrifuged at 1500 × g for 10 min. and the pellet resuspended in 2 × TY containing 100 µg/ml ampicillin and incubated with shaking at 30 • C overnight. The overnight culture was centrifuged at 10,800 × g for 10 min. and the pellet resuspended in 1/5 volume of 20% (w/v) polyethylene glycol 6000 in 2.5 M NaCl for 1 h at 4 • C. Phage particles were precipitated by centrifugation at 10,800 × g for 30 min. then resuspended in 10 ml H 2 O containing 20% (w/v) PEG/NaCl and incubated for 20 min. at 4 • C. This solution was centrifuged at 3,300 × g for 30 min. and the pellet containing phage particles was then resuspended in 2 ml PAS / 15% (v/v) glycerol. Lastly, the supernatant, containing the phage was filtered through a 0.45 µm filter and then stored at −80 • C prior to use. The bacteriophage titre was determined by serial dilution of infected E. coli plated out on TYE (1.5%(w/v) Bacto-agar 0.8% (w/v) NaCl, 1% (w/v) tryptone and 5% (w/v) yeast extract) supplemented with 25 µg/ml kanamycin.

Use of phage antibody display library
To isolate Acanthamoeba specific antibody fragments or those that can differentiate pathogenic isolates, an aliquot of the bacteriophage library (approx. 2 × 10 11 bacteriophage), was added to a suspension of whole fixed cells (2 × 10 6 ) which had been blocked by incubation in MPAS [2% (w/v) dried milk powder, 1% (w/v) bovine serum albumin in PAS] at 37 • C for 1 h. prior to use. The mixture of phage and cells was incubated at 20 • C with gentle shaking for 1 h. then centrifuged at 800 × g for 5 min. The pellet was resuspended in 50 ml of 0.1% (w/v) BSA in PAS then centrifuged again; this process was performed 10 times in total to remove unbound phage. Citric acid (76 mM) was then added to the pellet and the suspension incubated at 20 • C with shaking to remove bound phage from cells, the pH of the mixture was then adjusted to 7.0 by adding Tris-HCl (1M) pH 7.4. This procedure was termed a panning round. The phage selected by this procedure were than amplified as described previously (see "Preparation of Bacteriophage Particles" above) except that bacteria infected with bacteriophage were spread onto TYE in a bio-assay dish and incubated at 30 • C overnight. Overnight grown E. coli containing bacteriophage were scraped from the plate with 2 × TY containing 15% (v/v) glycerol and the suspension (library stock) was kept at −70 • C until required.
The protozoan, Hartmanella sp. was used for subtractive (negative) panning in the genus specific protocol; a pool of non-pathogenic species was used as the negative for the isolation of pathogen specific bacteriophage. Briefly, these cells were treated and blocked as described for target cells. The cells were then resuspended in MPAS and 2 × 10 11 phage (from the phage library isolated after one round of panning against the target organism) were added and incubated at 20 • C with gentle shaking for 1 h. Cells were centrifuged and  supernatant containing unbound phage was incubated with a second aliquot of blocked cells. These were incubated with gentle shaking at 20 • C for 30 min. This procedure was repeated a further three times. Finally the supernatant containing unbound bacteriophage was incubated with fixed and blocked target cells and a positive panning round was performed. Two further rounds of positive panning were then performed. The complete panning procedure is shown in Fig. 1.

PCR
PCR was performed to demonstrate the diversity of phage clones isolated after each panning round [13,14]. PCR fragments were amplified directly from bacterial colonies using CDR-FOR (5 CAG GGT ACC TTG GCC CCA 3 ) and CDR-BACK (5 GTG TAT TAC TGT GCA AGA 3 ) primers and were separated on 5% (w/v) MetaPhor agarose gels (FMC Bioproducts, Flowgen).

ELISA
Single bacterial colonies were picked directly from the bioassay dish and inoculated into 2 × TY supplemented with 100 µg/ml ampicillin +1%(w/v) glucose and grown in 96 well round bottomed plates overnight at 37 • C. Glycerol stocks of the overnight incubations were made by adding 15% (w/v) glycerol and stored at −80 • C until required. To rescue bacteriophage from glycerol stock cultures, 5-10 µl were transferred to a second round bottomed plate containing 2 × TY and incubated at 37 • C for 1 h. VCS-M13 helper bacteriophage (1 × 10 9 ) were added to each well and in-  cubated for a further 30 min. at 37 • C without shaking, and then for 1 h with shaking. The plate was centrifuged at 400 × g for 10 min, the supernatant aspirated, and the pellet resuspended in 2 × TY supplemented with 50 µg/ml kanamycin and incubated overnight at 30 • C with shaking (phage plate). The relevant fixed and blocked target cells (2 × 10 5 ) were added to the wells of a conical bottomed plate (Assay plates) and centrifuged at 200 × g for 5 min. The supernatant was carefully removed and cells were washed with PAS. Phage plates grown overnight were centrifuged at 400 × g for 10 min and supernatant was used as a source of the bacteriophage. Bacteriophage particles were added to the assay plates and these were then incubated with shaking at 20 • C for 1 h and then washed with PAS. Sheep anti M13-Horseradish peroxidase conjugate (Pharmacia, diluted 1:500 in blocking buffer) was added and incubated with shaking at 20 • C for 1 h. Cells were washed twice with PAS before 2,2azinobis(3-ethylbenzthiazoline sulfonic acid) (ABTS) substrate (Vector Laboratories) was added, and cells were transferred to a flat bottomed plate. The plate was then incubated at 20 • C for 30 min. in the dark before the supernatant was transferred to a flat-bottomed plate and the absorbance at 405-609 nm determined.

Flow cytometry
For flow cytometry, 10 6 fixed and blocked target cells were mixed with 10 12 phage from the appropriate clone and this was incubated at 4 • C for 1 h. Cells were washed three times with 0.25% (w/v) bovine serum albumin, 0.04% (w/v) Sodium azide in PAS (PAA) and resuspended in sheep anti-M13 antibody (10 µg/ml) and incubated at 4 • C for 1 h. Cells were washed twice with PAA and resuspended in 50 µl of fluorescein isothiocyanate-conjugated donkey anti-sheep immunogobulin G (IgG) (10 µg/ml, Vector Laboratories) and incubated at 4 • C for 1 h. Cells were washed twice as above then resuspended in PAA and analysed by flow cytometry. Fluorescent activity was measured on a FACSCalibur and Cellquest software (Becton Dickinson) using an excitation wavelength of 488 nm.

Scanning electron microscopy
Scanning electron microscopy showed the presence of characteristic acanthopodia (Fig. 2). The nonpathogenic strains of Acanthamoeba tested have < 20 acanthopodia while the pathogenic Acanthamoeba exhibited > 100 acanthopodia per cell (Table 1). However, one species of non-pathogenic Acanthamoeba (A. royreba), had > 100 acanthopodia per cell. Table 4 The level and specificity of binding of bacteriophage clones to a range of pathogenic and non-pathogenic Acanthamoeba

ELISA
Over 400 clones were analysed by ELISA from each of the two libraries in an attempt to isolate genus and pathogenic strain specificities. From the genus specific library, 10 clones with the highest binding to A. palestinensis (OD A 405 -A 690 > 0.8) were selected initially. All showed high binding (> 1.0 absorbance units) to most or all of the Acanthamoeba spp. (Table 2). PCR analysis showed diversity within these clones (data not shown). Clone HPPG6 exhibited a high level of binding to all Acanthamoeba species, with no binding to a range of other cell types (Table 3). From the second "pathogenic" library, 8 clones were selected on the basis of ELISA reactivity using pooled pathogenic, and non-pathogenic Acanthamoeba. Only one clone (HPPG55) exhibited a useful binding profile (Table 4).

Flow cytometry
Flow cytometry showed strong binding of these antibody fragments against Acanthamoeba cells with no binding against other amoeba, bacterial strains or lymphocytes. Mean fluorescence for clone HPPG6 with A. palestinensis was 200 as compared with 48 for cells incubated with the irrelevant bacteriophage. No increase in fluorescence was seen with Hartmanella spp. (Fig. 3(c), (d)). This demonstrated the specificity of antibody fragments with Acanthamoeba spp. Clone HPPG55, isolated against pathogenic Acanthamoeba spp. showed mean channel fluorescence of 150 with A. castellanii as compared with 56 for cells incubated with irrelevant bacteriophage. No change in fluorescence was observed with A. astronyxis (Fig. 3(a), (b)).

Discussion
In this study we have shown that antibody fragments showing a high level of specificity for Acanthamoeba spp. can be isolated from a naive phage display library. This represents the first time that phage antibody display technology has been used in the development of a useful diagnostic antibody in microbiology. We were also able to isolate bacteriophage that can distinguish between pathogenic and non-pathogenic isolates cultured under various conditions.
Rabbit polyclonal antiserum has been used previously for detecting Acanthamoeba in corneal scrapings and in tissues of suspected meningoencephalitis, however these reagents showed considerable cross reactivity with other cell types and their sensitivity has been questioned [7]. Current methods for "rapid" identification of amoeba involve staining with Giemsa, calcofluor white, methylene blue or acridine orange. Accurate diagnosis and interpretation of results requires highly trained personnel together with a strong clinical suspicion of amoebic infection. A few groups have used the polymerase chain reaction as a rapid detection method for Acanthamoeba but its clinical use, specificity and sensitivity has not been tested [4]. With Acanthamoeba spp. becoming more important as causative agents for eye keratitis, due to increased contact lens use and meningoencephalitis in immunocompromised people, the need for a simple and rapid method for the specific detection is obvious.
To our knowledge the antibodies described above are the only demonstration of such an application on protozoan parasites, previous work has focussed on mammalian systems, e.g. tumours such as malignant melanoma [3]. Phage antibody display technique is a very useful and powerful tool for generating antibody fragments against a diverse array of molecules, many of  which can potentially be used in new diagnostic assays and perhaps chemotherapy.
Interestingly, using EM we also show that the number of acanthapodia is a marker for pathogenically, however our observation of significant numbers of acanthapodia on the non-pathogen A. royreba suggest that the relationship requires further verification using a larger number of species. Also, as acanthapodia are difficult to visualise using light microscopy, their use in routine identification is limited.

Ongoing work
We are currently expanding our work to look at two other protozoan parasites, Giardia intestinalis and Cryptosporidium parvum. Our work on Giardia is discussed in another part of this supplement (Hough et al. "Isolation of anti-Giardia antibodies from a bacteriophage display library"). In addition, we are using bacteriophage to identify phenotypic characteristics that can differentiate the human infective and animal infective genotypes of C. parvum.