Foot and mouth disease (FMD) is the most contagious disease of mammals and has a great potential for causing severe economic loss in susceptible cloven-hoofed animals. There are seven serotypes of FMD virus (FMDV), namely, O, A, C, SAT 1, SAT 2, SAT 3 and Asia 1. Infection with one serotype does not confer immunity against another. FMD cannot be differentiated clinically from other vesicular diseases, such as swine vesicular disease, vesicular stomatitis and vesicular exanthema. Laboratory diagnosis of any suspected FMD case is therefore a matter of urgency.
Typical cases of FMD are characterized by a vesicular condition of the feet, buccal mucosa and, in females, the mammary glands. Clinical signs can vary from mild to severe, and fatalities may occur, especially in young animals. In some species the infection may be subclinical, e.g. African buffalo (Syncerus caffer). The preferred tissue for diagnosis is epithelium from unruptured or freshly ruptured vesicles or vesicular fluid. Where collecting this is not possible, blood and/or oesophageal–pharyngeal (OP) fluid samples taken by probang cup in ruminants or throat swabs from pigs provide an alternative source of virus. Myocardial tissue or blood can be submitted from fatal cases, but vesicles are again preferable if present.
Diagnosis of FMD is by virus isolation or by the demonstration of FMD viral antigen or nucleic acid in samples of tissue or fluid. Detection of virus-specific antibody can also be used for diagnosis, and antibodies to viral nonstructural proteins (NSPs) can be used as indicators of infection, irrespective of vaccination status.
1. Identification of the agent
A range of sample types, including epithelium, OP samples and serum, may be examined by virus isolation or RT-PCR. By contrast, ELISA is suited to the examination of epithelial suspensions, vesicular fluids or cell culture supernatants, but are insufficiently sensitive for the direct examination of OP samples or serum.
a) Virus isolation
The epithelium sample should be taken from the PBS/glycerol, blotted dry on absorbent paper to reduce the glycerol content, which is toxic for cell cultures, and weighed. A suspension should be prepared by grinding the sample in sterile sand in a sterile pestle and mortar with a small volume of tissue culture medium and antibiotics. Further medium should be added until a final volume of nine times that of the epithelial sample has been added, giving a 10% suspension. This is clarified on a bench centrifuge at 2000 g for 10 minutes. Once clarified, such suspensions of field samples suspected to contain FMDV are inoculated onto cell cultures. Sensitive cell culture systems include primary bovine (calf) thyroid cells and primary pig, calf or lamb kidney cells. Established cell lines, such as BHK-21 (baby hamster kidney) and IBRS-2 cells, may also be used but are generally less sensitive than primary cells for detecting low amounts of infectivity. The sensitivity of any cells used should be tested with standard preparations of FMDV. The use of IB-RS-2 cells aids the differentiation of swine vesicular disease virus (SVDV) from FMDV (as SVDV will only grow in cells of porcine origin) and is often essential for the isolation of porcinophilic strains, such as O Cathay. The cell cultures should be examined for cytopathic effect (CPE) for 48 hours. If no CPE is detected, the cells should be frozen and thawed, used to inoculate fresh cultures and examined for CPE for another 48 hours.
b) Immunological methods
• Enzyme-linked immunosorbent assay
The preferred procedure for the detection of FMD viral antigen and identification of viral serotype is the ELISA. This is an indirect sandwich test in which different rows in multiwell plates are coated with rabbit antisera to each of the seven serotypes of FMDV. These are the ‘capture’ sera. Test sample suspensions are added to each of the rows, and appropriate controls are also included. Guinea-pig antisera to each of the serotypes of FMDV are added next, followed by rabbit anti-guinea-pig serum conjugated to an enzyme. Extensive washing is carried out
between each stage to remove unbound reagents. A colour reaction on the addition of enzyme substrate and chromogen indicates a positive reaction. With strong positive reactions, this will be evident to the naked eye, but results can also be read spectrophotometrically at an appropriate wavelength. In this case, an absorbance reading greater than 0.1 above background indicates a positive reaction; the serotype of FMDV can also be identified. Values close to 0.1 should be confirmed by retesting or by amplification of the antigen by tissue culture passage and testing the supernatant once a CPE has developed. A suitable protocol is given below. Other protocols are available with slightly different formats and interpretation criteria.
Depending on the species affected and the geographical origin of samples, it may be appropriate to simultaneously test for SVDV or vesicular stomatitis virus (VSV). Ideally a complete differential diagnosis should be undertaken in all vesicular conditions.
Rabbit antiserum to the 146S antigen of each of the seven serotypes of FMDV (plus SVDV or VSV if required) is used as a trapping antibody at a predetermined optimal concentration in carbonate/bicarbonate buffer, pH 9.6.
Control antigens are prepared from selected strains of each of the seven types of FMDV (plus SVDV or VSV if appropriate) grown on monolayer cultures of BHK-21 cells (IB-RS-2 cells for SVD or VSV). The unpurified supernatants are used and pretitrated on ELISA plates. The final dilution chosen is that which gives an absorbance at the top of the linear region of the titration curve (optical density approximately 2.0), so that the five-fold dilutions of the control antigens used in the test give two additional lower optical density readings from which the titration curve can be derived. PBS containing 0.05% Tween 20 and phenol red indicator is used as a diluent (PBST).
Guinea-pig antisera prepared by inoculating guinea-pigs with 146S antigen of one of the seven serotypes of FMDV (plus SVDV if required) and preblocked with normal bovine serum (NBS) is used as the detecting antibody. Predetermined optimal concentrations are prepared in PBS containing 0.05% Tween 20, and 5%
dried, nonfat skimmed milk (PBSTM).
Rabbit (or sheep) anti-guinea-pig immunoglobulin conjugated to horseradish peroxidase and preblocked with NBS is used at a predetermined optimum concentration in PBSTM. As an alternative to guinea-pig or rabbit antisera, suitable monoclonal antibodies (MAbs) can be used coated to the ELISA plates as capture antibody or peroxidase-conjugated as detecting antibody.
• Test procedure
i.) ELISA plates are coated with 50 μl/well rabbit antiviral sera in 0.05 M carbonate/bicarbonate buffer, pH 9.6. Rows A to H receive, respectively, antisera to serotypes O, A, C, SAT 1, SAT 2, SAT 3, Asia 1 and SVDV or VSV (optional).
ii.) Leave overnight at 4°C in a stationary position or place on an orbital shaker set at 100–120 revolutions per minute in a 37°C incubator for 1 hour.
iii.) Prepare test sample suspension (10% original sample suspension or undiluted clarified cell culture supernatant fluid).
iv.) The ELISA plates are washed five times in PBS.
v.) On each plate, load wells of columns 4, 8 and 12 with 50 μl PBST. Additionally, add 50 μl of PBST to wells 1, 2 and 3 of rows A to H on plate 1. To well 1 of row A of plate 1 add 12.5 μl of control antigen type O, to well 1 of row B add 12.5 μl of control antigen type A; continue in this manner for control antigen of types C, SAT 1, SAT 2, SAT 3, Asia 1 and SVDV or VS (if appropriate) in order to well 1, rows C to H. Mix diluent in well 1 of rows A to H and transfer 12.5 μl from well 1 to 2 (rows A to H), mix and transfer 12.5 μl from well 2 to 3, mix and discard 12.5 μl from well 3 (rows A to H) (this gives a fivefold dilution series of each control antigen). It is only necessary to change pipette tips on the micropipette between antigens. The remainder of the plate can be loaded with the test sample(s). Add 50 μl of sample one to wells 5, 6 and 7 of rows A to H, the second sample is placed similarly in columns 9, 10 and 11, rows A to H.
If more than two samples are to be tested at the same time, the other ELISA plates should be used as follows:
Dispense 50 μl of the PBST to the wells (rows A to H) of columns 4, 8 and 12 (buffer control columns). Note that the control antigens are not required on these plates. These test samples may be added in 50 μl volumes in rows A to H to columns 1, 2, 3; 5, 6, 7; 9, 10, 11, respectively.
vi.) Cover with lids and place on an orbital shaker at 37°C for 1 hour.
vii.) Wash the plates by flooding with PBS – wash three times as before and empty residual wash fluid. Blot the plates dry.
viii.) Transfer 50 μl volumes of each guinea-pig serum dilution to each plate well in the appropriate order, e.g. rows A to H receive, respectively, antisera to serotypes O, A, C, SAT 1, SAT 2, SAT 3, Asia 1 and SVDV or VSV (optional).
ix.) Cover the plates with lids and replace on the orbital shaker. Incubate at 37°C for 1 hour.
x.) The plates are washed again three times, and 50 μl of rabbit anti-guinea-pig immunoglobulin conjugated to horseradish peroxidase is added to each well. The plates are incubated at 37°C for 1 hour on a rotary shaker.
xi.) The plates are washed again three times, and 50 μl of substrate solution, containing 0.05% % H2O2 plus orthophenylene diamine or a suitable alternative chromogen, is added to each well.
xii.) The reaction is stopped after 15 minutes by the addition of 50 μl of 1.25 M sulphuric acid. The plates are read at 492 nm on a spectrophotometer linked to a computer.
c) Nucleic acid recognition methods
RT-PCR can be used to amplify genome fragments of FMDV in diagnostic materials including epithelium, milk, serum and OP samples. RT combined with real-time PCR has a sensitivity comparable to that of virus isolation and automated procedures enhance sample throughput. Serotyping primers have also been developed.
• Agarose gel-based RT-PCR assay
The procedure used at the OIE Reference Laboratory at Pirbright is described. The RT-PCR assay consists of the three successive procedures of (i) extraction of template RNA from the test or control sample followed by (ii) RT of the extracted RNA, (iii) PCR amplification of the RT product and (iv) detection of the PCR products by agarose gel electrophoresis.
• Test procedure
i.) Add 200 μl of test sample to 1 ml of TRIzol® Reagent in a sterile tube. Store at –70°C until required for RNA extraction.
ii.) Transfer 1 ml of the solution from i) into a fresh, sterile tube containing 200 μl of chloroform. Vortex mix for about 10–15 seconds and leave at room temperature for 3 minutes.
iii.) Centrifuge for 15 minutes at 20,000 g.
iv.) Transfer 500 μl of the aqueous phase into a fresh, sterile tube containing 1 μl of glycogen (20 mg/ml) and add 500 μl of iso-propyl-alcohol (propan-2-ol). Vortex mix for a few seconds.
v.) Leave at room temperature for 10 minutes then centrifuge for 10 minutes at 20,000 g.
vi.) Discard the supernatant fluid from each tube and add 1 ml of 70% ethanol. Vortex mix for a few seconds.
vii.) Centrifuge for 10 minutes at 20,000 g.
viii.) Carefully remove the supernatant fluid from each tube taking care not to dislodge or lose any pellet at the bottom of the tube.
ix.) Air dry each tube at room temperature for 2–3 minutes.
x.) Resuspend each pellet by adding 20 μl of nuclease-free water to the tube.
xi.) Keep the extracted RNA samples on ice if the RT step is about to be performed. Otherwise store at –70°C.
xii.) For each sample to be assayed, add 2 μl of random hexamers (20 μg/ml) and 5 μl of nuclease-free water into a sterile 0.5 ml microcentrifuge tube. It is recommended to prepare the dilution in bulk for the total number of samples to be assayed but allowing for one extra sample.
xiii.) Add 5 μl of RNA from the extraction procedure described above to give a volume of 12 μl in each tube. Mix by gently pipetting up and down.
xiv.) Incubate at 70°C for 5 minutes.
xv.) Cool at room temperature for 10 minutes.
xvi.) During the 10-minute incubation period, prepare the RT reaction mixture described below for each sample. Prepare the reaction mixture in bulk in a sterile 1.5 ml microcentrifuge tube for the number of samples to be assayed plus one extra sample. First strand buffer, 5× conc. (4 μl); bovine serum albumin (acetylated), 1 mg/ml (2 μl); dNTPs, 10 mM mixture each of dATP, dCTP, dGTP, dTTP (1 μl); DTT, 1 M (0.2 μl); Moloney Murine Reverse Transcriptase, 200 U/μl (1 μl).
xvii.) Add 8 μl reaction mix to the 12 μl of random primer/RNA mix. Mix by gently pipetting.
xviii.) Incubate at 37°C for 45 minutes.
xix.) Keep the RT products on ice if the PCR amplification step is about to be performed, otherwise store at –20°C.
xx.) Prepare the PCR mix described below for each sample. It is recommended to prepare the mix in bulk for the number of samples to be tested plus one extra sample.Nuclease-free water (35 μl); PCR reaction buffer, 10× conc (5 μl); MgCl2, 50 mM (1.5 μl); dNTPs, 10 mM mixture each of dATP, dCTP, dGTP, dTTP (1 μl); primer 1, 10 pmol/μl (1 μl); primer 2, 10 pmol/μl (1 μl); Taq Polymerase, 5 units/μl (0.5 μl).
xxi.) Add 45 μl of PCR reaction mix to a well of a PCR plate or to a microcentifuge tube for each sample to be assayed followed by 5 μl of the RT product to give a final reaction volume of 50 μl.
xxii.) Spin the plate or tubes for 1 minute in a suitable centrifuge to mix the contents of each well.
xxiii.) Place the plate in a thermal cycler for PCR amplification and run the following programme:
94°C for 5 minutes: 1 cycle;
94°C for 1 minute, 55°C for 1 minute, 72°C for 2 minutes: 30 cycles;
72°C for 7 minutes: 1 cycle.
xxiv.) Mix a 20 μl aliquot of each PCR reaction product with 4 μl of staining solution and load onto a 1.5% agarose gel. After electrophoresis a positive result is indicated by the presence of a 328 bp band corresponding to FMDV sequence in the 5’ untranslated region of the genome.
i.) Nuclease-free water, TRIzol® Reagent, chloroform, glycogen, iso-propyl-alcohol (propan-2-ol), ethanol, random hexanucleotide primers, First strand buffer, BSA (acetylated), dNTPs, DTT, Moloney Murine Reverse Transcriptase, PCR reaction buffer (10×), MgCl2 and Taq Polymerase are commercially available.
ii.) Primers at a concentration of 10 pmol/μl:
Primer 1 sequence 5’-GCCTG-GTCTT-TCCAG-GTCT-3’ (positive strand);
Primer 2 sequence 5’-CCAGT-CCCCT-TCTCA-GATC-3’ (negative strand).
2. Serological tests
Serological tests for FMD are performed in support of four main purposes namely: 1) to certify individual animals prior to import or export (i.e. for trade); 2) to confirm suspected cases of FMD; 3) to substantiate absence of infection; 4) to demonstrate the efficacy of vaccination. For substantiating freedom from infection, different approaches are required according to whether the population has been vaccinated or not and if vaccination has been used, whether this has been applied as an emergency application or as part of an ongoing programme of vaccination. Different tests and different interpretations of test results will be appropriate according to the abovementioned purposes and the validation of the selected procedure must take account of the purpose. For example, test cut-offs may be set at a different threshold for herd-based serosurveillance than is appropriate for certifying freedom from infection for individual animals for the purposes of international trade.
Serological tests for FMD are of two types; those that detect antibodies to viral structural proteins (SP) and those that detect antibodies to viral nonstructural proteins (NSPs).
a) Virus neutralisation test (a prescribed test for international trade)
The quantitative VN microtest for FMD antibody is performed with IB-RS-2, BHK-21, lamb or pig kidney cells in flat-bottomed tissue-culture grade microtitre plates.
Stock virus is grown in cell monolayers and stored at –20°C after the addition of 50% glycerol. (Virus has been found to be stable under these conditions for at least 1 year.) The sera are inactivated at 56°C for 30 minutes before testing. The control standard serum is 21-day convalescent or post-vaccination serum. A suitable medium is Eagle’s complete medium/LYH (Hank’s balanced salt solution with yeast lactalbumin hydrolysate) with hepes buffer and antibiotics.
The test is an equal volume test in 50 μl amounts.
• Test procedure
i.) Starting from a 1/4 dilution, sera are diluted in a twofold, dilution series across the plate, using at least two rows of wells per serum, preferably four rows, and a volume of 50 μl.
ii.) Previously titrated virus is added; each 50 μl unit volume of virus suspension should contain about 100 TCID50 (50% tissue culture infective dose) within an accepted range (e.g. 32–320 TCID50).
iii.) Controls include a standard antiserum of known titre, a cell control, a medium control, and a virus titration used to calculate the actual virus titre used in the test.
iv.) Incubate at 37°C for 1 hour with the plates covered.
v.) A cell suspension at 106 cells/ml is made up in medium containing 10% bovine serum (specific antibody negative) for cell growth. A volume of 50 μl of cell suspension is added to each well.
vi.) Plates are sealed with pressure-sensitive tape and incubated at 37°C for 2–3 days. Alternatively, the plates may be covered with loosely fitting lids and incubated in an atmosphere of 3–5% carbon dioxide at 37°C for 2–3 days.
vii.) Microscope readings may be feasible after 48 hours. The plates are finally fixed and stained routinely on the third day. Fixation is effected with 10% formol/saline for 30 minutes. For staining, the plates are immersed in 0.05% methylene blue in 10% formalin for 30 minutes. An alternative fixative/stain solution is naphthalene blue black solution (0.4% [w/v] naphthalene blue black, 8% [w/v] citric acid in saline). The plates are rinsed in tap water.
viii.) Positive wells (where the virus has been neutralised and the cells remain intact) are seen to contain blue-stained cells sheets; the negative wells (where virus has not been neutralised) are empty. Titres are expressed as the final dilution of serum present in the serum/virus mixture where 50% of wells are protected. The test is considered to be valid when the amount of virus used per well is in the range log10 1.5–2.5 TCID50, and the positive standard serum is within twofold of its expected titre.
ix.) Interpretation of tests can vary between laboratories in regard to the negative/positive cut-off threshold. Laboratories should establish their own criteria by reference to standard reagents that can be obtained from the OIE Reference Laboratory at Pirbright. In general, a titre of 1/45 or more of the final serum dilution in the serum/virus mixture is regarded as positive. A titre of less than 1/16 is considered to be negative. For certification of individual animals for the purposes of international trade, titres of 1/16 to 1/32 are considered to be doubtful, and further serum samples may be requested for testing; results are considered to be positive if the second sample has a titre of 1/16 or greater. For the purposes of herd-based serosurveillance as part of a statistically valid serological survey, a cut-off of 1/45 may be appropriate. Cut-off titres for evaluating immunological protection afforded by vaccination have to be established from experience of potency test results with the relevant vaccine and target species.
b) Liquid-phase blocking enzyme-linked immunosorbent assay (a prescribed test for international trade)
Antigens are prepared from selected strains of FMDV grown on monolayers of BHK-21 cells. The unpurified supernatants are used and pretitrated in a twofold dilution series but without serum. The final dilution chosen is that which, after addition of an equal volume of diluent (see below), gives an absorbance on the upper part of the linear region of the titration curve (optical density approximately 1.5). PBS containing 0.05% Tween 20 and phenol red indicator is used as a diluent (PBST). The other reagents used in the test are the same as those in the solid-phase blocking ELISA. An example of the test procedure is described below. Temperature and incubation times can vary depending on the protocol.
• Test procedure
i.) ELISA plates are coated with 50 μl/well rabbit antiserum homologous to the antigen being used and left overnight in a humid chamber at room temperature.
ii.) The ELISA plates are washed three times with PBS.
iii.) In U-bottomed multiwell plates (carrier plates) 50 μl of a duplicate, twofold series of each test serum is prepared, starting at 1/8. To each well, 50 μl of a constant dose of viral antigen that is homologous to the rabbit antisera used to coat the plates is added and the mixtures are left overnight at 4°C, or incubated at 37°C for 1 hour. The addition of the antigen increases the final serum dilution to 1/16.
iv.) Then 50 μl of serum/antigen mixtures is transferred from the carrier plates to the rabbit-serum coated ELISA plates and the plates are incubated at 37°C for 1 hour on a rotary shaker.
v.) After washing, 50 μl of guinea-pig antiserum homologous to the viral antigen used in the previous step
vi.) (iv) (preblocked with NBS and diluted in PBST containing 5% skimmed milk powder) is added to each well. The plates are then incubated at 37°C for 1 hour on a rotary shaker.
vii.) The plates are washed and 50 μl of rabbit anti-guinea-pig immunoglobulin conjugated to horseradish peroxidase (preblocked with NBS and diluted in PBST containing 5% skimmed milk powder) is added to each well. The plates are incubated at 37°C for 1 hour on a rotary shaker.
viii.) The plates are washed again three times and 50 μl of substrate solution, containing 0.05% H2O2 plus orthophenylene diamine or a suitable alternative chromogen, is added to each well.
ix.) The reaction is stopped after 15 minutes by the addition of 50 μl of 1 M sulphuric acid. The plates are read at 492 nm on a spectrophotometer linked to a computer.
x.) Controls: A minimum of four wells each of strong positive, weak positive and negative bovine reference sera at a final dilution of 1/32 should be included on each plate together with an equivalent number of reaction (antigen) control wells containing antigen in diluent alone without serum. For end-point titration tests, duplicate twofold dilution series of positive and negative homologous bovine reference sera should be included on at least one plate of every run.
xi.) Interpretation of the results: Antibody titres are expressed as the 50% end-point titre, i.e. the dilution at which the reaction of the test sera results in an optical density equal to 50% inhibition of the median optical density of the reaction (antigen) control wells. The median is calculated as the mean of two mid-values of the reaction control wells, eliminating from the calculation the highest and lowest values (alternatively, the mean value can be used after setting suitable tolerance limits to control for inter-well variation). In general sera with titres greater than or equal to 1/90 are considered to be positive. A titre of less than 1/40 is considered to be negative. For certification of individual animals for the purposes of international trade, titres of greater than 1/40, but less than 1/90 are considered to be doubtful, and further serum samples may be requested for testing; results are considered to be positive if the second sample has a titre of 1/40 or greater. For the purposes of herd-based serosurveillance as part of a statistically valid serological survey, a cut-off of 1/90 may be appropriate. Cut-off titres for evaluating immunological protection afforded by vaccination have to be established from experience of potency test results with the relevant vaccine and target species.
c) Nonstructural protein (NSP) antibody tests
Antibody to expressed recombinant FMDV NSPs (e.g. 3A, 3B, 2B, 2C, 3ABC) can be measured by different ELISA formats or immunoblotting. These ELISAs either use purified antigens absorbed directly to microplates or use PAbs or MAbs to trap specific antigens from semi-purified preparations.
• Indirect enzyme-linked immunosorbent assay
• Test procedure
i.) Microplates are coated overnight at 4°C with 1 μg/ml of the fusion antigen 3ABC in carbonate/ bicarbonate buffer, pH 9.6 (100 μl per well). Antigen 3ABC was expressed and purified as indicated for the EITB (enzyme-linked immunoelectrotransfer blot) tests.
ii.) The plates are washed six times with PBS, pH 7.2, supplemented with 0.05% Tween 20 (PBST).
iii.) Test sera (100 μl per well) are added in a 1/20 dilution in blocking buffer consisting of PBS, 0.05% Tween 20, 5% nonfat dry milk, 10% equine sera and 0.1% Escherichia coli lysate. Each plate includes a set of strong and weak positive and negative controls calibrated against the International Standard Sera described below.
iv.) The plates are incubated for 30 minutes at 37°C and washed six times in PBST.
v.) Horseradish-peroxidase-conjugated rabbit anti-species IgG is diluted optimally in the blocking buffer, added at 100 μl per well and the plates are incubated for 30 minutes at 37°C.
vi.) After six washings, each well is filled with 100 μl of 3’3’, 5’5’-tetramethylbenzidine plus 0.004% (w/v) H2O2 in phosphate/citrate buffer, pH 5.5.
vii.) The reaction is stopped after 15 minutes of incubation at room temperature by adding 100 μl of 0.5 M H2SO4. Absorbance is read at 450 nm and at 620 nm for background correction.
viii.) Interpreting the results: Test results are expressed as per cent positivity relative to the strong positive control [(optical density of test or control wells/optical density of strong positive control) × 100] or alternatively as a test to control (T/C) index relative to a cut-off (i.e. threshold positive) control. Profiling the NSP antibody reactivity levels in herds along with age/vaccination stratification aids interpretation of herd infection status in vaccinated populations. Test cut-off values, with or without suspicious zones, need to be determined with consideration to the purpose of testing and the intended target population. Inconclusive results may be followed up using confirmatory tests, retesting with EITB or a second NSP ELISA (taking account of the conditional dependence of the two tests). The overall test system sensitivity and specificity must be taken into account when designing the serosurveillance programme. Although not a prescribed test for trade, NSP ELISAs may be a valuable adjunct in circumstances where the serotype or subtype of virus in the originating country is not known.
Manual of Diagnostic Tests and Vaccines for Terrestrial Animals 2013. Chapter 2.1.5. Foot and mouth disease. http://www.oie.int/fileadmin/Home/eng/Health_standards/tahm/2.01.05_FMD.pdf