IVD Technology Magazine
IVDT Article Index

Originally published March, 1998

Development of monoclonal antibodies and a rapid assay for tuberculosis

Richard T. Root, Abu F. Raisur Rahman, Minal Ashtekar, and Salman H. Siddiqi

An IVD is urgently needed to help combat the worldwide TB pandemic. But to be really useful in underdeveloped countries, diagnostic tools must be inexpensive, easy to use, and easy to interpret.

Since 1978, the increasing incidence of tuberculosis (TB) infection has reawakened awareness of this disease in developed nations.^1—3 Thought to be the principal killer of the human race since antiquity, TB is even today considered "probably the largest single infectious cause of human mortality."⁴ Informed estimates of current infection and mortality rates are truly terrifying.^5,6 Although the majority of TB infections are in Third World countries, the ease of worldwide travel ensures that, for the foreseeable future, the disease will also be a problem for developed nations.

Mycobacterium tuberculosis, an ancient and powerful foe of humankind. Photo by Alfred Pasieka/Science Photo Library

Diagnosis of TB is particularly challenging. Because patients may present with a wide variety of symptoms, a major obstacle to diagnosing TB is the difficulty of distinguishing it from other respiratory complaints. In developed countries, where TB has been nearly eradicated for many years, the current generation of physicians lacks the experience needed to quickly differentiate the disease from other respiratory problems.^7,8

TB is usually diagnosed by culturing processed sputum samples to obtain visible growth, then confirming the presence of Mycobacterium tuberculosis (Mtb) through further culturing on selective media and biochemical testing. Completion of this lengthy procedure requires four to eight weeks. Although a new generation of nucleic acid—based diagnostics promises to reduce the time required for a confirmed diagnosis, the high costs associated with such tests have so far limited their acceptance by health authorities.⁹

In underdeveloped countries, where the vast majority of TB cases are located, diagnosis is made still more difficult by the chronic scarcity of medical clinics, trained personnel, and funding for public health operations. The standard method of using microscopy and an acid-fast staining procedure suffers from a number of problems and is of limited use in very rural areas.¹⁰ Amplified DNA probe methods are also not practical because they require sophisticated instrumentation operated by trained personnel.

To be really useful in underdeveloped countries, a diagnostic tool for TB must be inexpensive, easy to use, and easy to interpret. In this article, the authors describe their progress toward developing a rapid, manual immunoassay using monoclonal antibodies for diagnosing TB infection. Although this research was directed toward creating a culture confirmation test, an inexpensive direct TB assay would be of equal or greater utility for Third World countries.

Antibody Development and Production

Mycobacterium tuberculosis is a notoriously difficult organism to handle, requiring biohazard level III facilities and a program of immune status surveillance for laboratory personnel. Common problems include the typically slow growth of the organism and the fact that clumping and cord formation during culturing prevent standard methods of enumeration from yielding consistent results. When cultured, Mtb secretes a variety of proteins into the growth media, and produces many other proteins that are localized within the bacterial cell or cell wall. Antibodies to most of these proteins are known to lack specificity and to cross-react with other mycobacterial species. Successful development of monoclonal antibodies specific to Mtb therefore requires that particular attention be given to the strategy used for screening the fusions.

The authors determined to screen first for specificity using the mycobacteria most likely to be encountered during primary culture isolation from a patient suspected of having tuberculosis. This was accomplished using a panel of mycobacteria other than tuberculosis (MOTT) to screen for monoclonal antibodies that bind only to Mtb (see Table I). An important factor in interpreting the results of this screening was the probability that multiple clones might be growing in any of the tested wells. Fused cells were therefore plated at an extremely low density, resulting in growth in no more than about 30% of the wells. Although this process required extra work in handling the microwell plates, it paid off by ensuring that desirable clones would not be lost because of the presence of undesirable cross-reacting clones in the same well.

ELISA Assay Format. In a level III biohazard laboratory, mycobacteria were grown to late log phase in a defined nonprotein-containing medium (Proskauer Beck), then subjected to heat killing in a 95°C water bath for 90 minutes. All preparations were tested for nonviability by subculturing on Bactec 12B liquid medium (Becton Dickinson Microbiology Systems, Sparks, MD). The killed bacteria were then collected by centrifugation, resuspended in distilled water with 0.02% thimerosal, and stored refrigerated until nonviability was confirmed—typically four weeks. The cells were then sonicated at 4°C by multiple pulses until a dispersion was obtained. The bacteria were quantitated by measuring their optical density at 600 nm.

For assay use, the cells were diluted to an OD^₆₀₀ of 0.125 in 100 mm sodium carbonate buffer, pH 9.5, pipetted into 96-well microplates, and dried at 37°C in an incubator with circulated air. The plates were then blocked with 100 µl of a goat gamma globulin and bovine serum albumin (BSA) solution. These plates could be used immediately or air-dried and stored desiccated at 4°C until needed. Spent broth from Mtb grown in the defined medium was also collected and treated, and was used to detect monoclonal antibodies against secreted mycobacterial antigens.

For use in the screening assays, hybridoma culture supernatant was diluted in a medium containing 10% fetal calf serum; 0.05 ml of this solution was pipetted into a coated well and then incubated for one hour at 37°C. The plates were then washed with phosphate-buffered saline and probed with horseradish peroxidase (HRP)—labeled goat antimouse IgG, A & M antibody for one hour, followed by washing. Tetramethylbenzidine substrate was pipetted into the wells and allowed to develop for 15 to 30 minutes. Development was halted by the addition of 2 M phosphoric acid and the results were read in a microplate reader at 450 nm.

Immunizations. Mice were immunized with a mixture of heat-killed cells and extracts of a defined culture broth in which Mtb H37Rv had been grown. Because H37Rv is a highly virulent strain of Mtb, only verified nonviable bacteria were used. To stimulate antibody production, an emulsion of antigen in Freund's incomplete adjuvant was employed. The mice received multiple injections intraperitoneally over the course of a year, with a final injection in the hind footpads.

In a separate series of experiments, the authors also used synthetic peptides of known Mtb-specific protein sequences linked to keyhole limpet hemocyanin (KLH). Mouse immune responses to these peptides were periodically assayed from blood samples using ELISA methods and corresponding immunogens. To eliminate interference by carrier protein, BSA was substituted for KLH for these assays.

Fusions. Once good immune responses had been obtained from the mice, hybridomas were produced via standard PEG fusion methods. Murine myeloma P3X63Ag8U.1 cells were used as the fusion partner. Spleen cells as well as popliteal and inguinal lymph nodes were used for the fusions. In all, eight fusions were carried out over a three-year period. From these fusions, the researchers plated 10,752 culture wells, screened 2216 wells, and selected 193 clones.

The separate experiment using peptide fusions resulted in monoclonal antibodies that proved capable of binding peptide-BSA conjugates very well. However, they did not bind significantly to either mycobacteria or their proteins, and this line of work was discontinued.

Figure 1. Flowchart of the hybridoma selection process for Mtb, to be used in the ColorPAC TB assay. The list of organisms screened is provided in Tables I—III.

Screening of the 193 hybridoma clones was a multitiered process (see Figure 1). First, the clones were tested for antibodies that would bind to either Mtb cells or to proteins adsorbed to plastic plates. Clones that tested positive in that assay were expanded, with aliquots frozen in liquid nitrogen to ensure against loss, while subcloning via limiting dilution was performed.

Once subcloning was complete, screening was continued using the seven other heat-killed mycobacterial species in the MOTT panel as well as Mtb H37Rv as a positive control (see Table I). Clones that displayed the greatest specificity toward Mtb were expanded for further testing using an extended MOTT panel and a panel of respiratory pathogens. The extended MOTT panel consisted of 41 strains of 14 other mycobacterial species (see Table II). The panel of respiratory pathogens included 16 fungal and bacterial species (other than mycobacteria) that are commonly found as contaminants in primary screening cultures for tuberculosis (see Table III).

Table III. Panel of respiratory pathogens used for screening antibody produced for the ColorPAC TB assay.

The hybridoma that demonstrated the best specificity toward Mtb was designated Mtb 5. Poisson analysis comparing the number of wells with growth for that fusion (710) with the total number of plated wells (1920) indicated a 5.4% probability that this cell culture could contain more than one clone. Therefore, Mtb 5 was twice subjected to subcloning by limiting dilution. To assess the stability of antibody production, levels in the supernatant from all subclones were normalized for growth and compared. The ELISA results of all subclones from the second round of cloning were within a 10% coefficient of variance.

The subclone selected for further production was Mtb 5.3.5, which showed the highest normalized level of antibody. The Mtb 5.3.5 antibody is an IgG1,k isotype. Further investigation of its specificity revealed that the epitope it recognizes is sensitive to pronase or trypsin treatment of the heat-killed bacterium, but not to periodate or neuraminidase treatment—a result suggestive of a protein epitope.

When TB antigens were probed with Mtb 5.3.5, after electrophoresis under denaturing and reducing conditions, Western blotting confirmed that the epitope was present on only one protein of approximately 65 kDa. This protein has been very well described in the literature, and is present in low amounts in the bacterial spent media and in large amounts in the bacteria itself. It is known to be a heat-shock protein with approximately 50% homology to a wide range of different organisms, but also containing species-specific epitopes.^11—13

Antibody Production and Purification. Several lots of the monoclonal antibody were produced via ascites in BALB/cj mice. In subsequent testing, however, these lots exhibited much greater cross-reactivity than had been observed during the screening phase. Since earlier batches of antibody produced in vitro were confirmed to possess the high specificity found during screening, it was suspected that the method of production caused variable cross-reactions.

This hypothesis was examined by comparing the specificity of new lots of the Mtb 5.3.5 antibody produced by both in vivo (mouse ascites) and in vitro methods. The results clearly showed a loss of specificity for the ascites-produced antibody, presumably because of variations among the normal immune responses of the host animals (see sidebar, page 42). Similar observations have been described by other researchers.¹⁴

Although the amounts of antibody produced by both the mouse ascites and artificial bioreactor methods were comparable—in the range of 1—5 mg/ml—in other respects the two methods differed greatly. The cross-reactivity of ascites-produced antibody was found to vary with each lot. Moreover, the ascites antibody required clotting, centrifugation, delipidation, high-speed centrifugation, and prefiltering before it could be purified. The prefiltering step proved to be extremely problematic, because the ascites-produced antibody tended to clog relatively large 5-µm filters rapidly even when it was diluted into phosphate-buffered saline. The result was a significant loss of the total amount of antibody produced via ascites.

By contrast, the bioreactor-produced antibody exhibited a consistent lot-to-lot specificity. It also required considerably less treatment prior to purification; centrifugation to remove cell mass followed by adjustment of pH to 7.8 was all that was required. Protein G chromatography was used for all purifications.

After comparing the two methods of production, the researchers determined that all future production of the Mtb 5.3.5 antibody would be carried out by use of a hollow-fiber bioreactor system.

Antibody specificity: Ascites vs. bioreactor production

To investigate the specificity of M. tuberculosis 5.3.5, the antibody selected for the ColorPAC assay, several lots of monoclonal antibody were produced by means of ascites in BALB/cj mice. These monoclonal antibodies were tested for cross-reactivity to mycobacteria other than tuberculosis (MOTT) via a dry-plate technique (see accompanying article).

The tests revealed patterns of cross-reactivity that differed for each batch of ascites-produced antibody. In some cases, antibody purified from mouse ascites showed greater cross-reactivity to nontubercular bacteria—such as M. kansasii 711, M. avium 33 and 61, M. terrae 1504 and 1505, and M. gastri 1305—than to Mtb cells. In other cases, ascites-produced antibody showed much less cross-reactivity. Such variations were not unexpected, since other researchers have reported that the normal immune responses of mouse hosts can greatly influence the specificity of ascites-produced antibody. In light of these test results, however, the researchers determined that it would be extremely difficult to support the ColorPAC assay by means of ascites-produced antibody, which would have required very expensive affinity purifications of the reagent.

As an alternative, the researchers explored the specificity and cross-reactivity of Mtb 5.3.5 antibody produced by means of a hollow-fiber bioreactor. Like its ascites-produced cousin, the bioreactor-produced antibody also exhibited very high binding capacity for all members of the TB complex, including M. microti, M. africanum, and M. bovis (not shown). However, the specificity and lot-to-lot consistency of the bioreactor-produced antibody was much better.

The bioreactor used for production of the antibody was the AcuMouse system (Cellex Biosciences, Minneapolis), which is now available as the TechnoMouse system (Integra Biosystems, Lowell, MA). The bioreactor portion of the system consists of group of hollow fibers encased within silicon membranes. In operation, a mixture of air and carbon dioxide was circulated outside the silicon membranes, through which it was able to effectively perfuse into the extracapillary space. A growth medium, no longer needed to deliver oxygen, was circulated once through the hollow fibers at about 1 ml/min, and then into a waste container. Glucose uptake and antibody yields suggested that an efficient metabolic state was maintained within the bioreactor.

This system was capable of maintaining antibody production for long periods of time; it was not unusual to carry on a production run for six months. Although the instrument could operate up to five cartridges at once, the production capacity of each cartridge was limited to small-scale runs of less than 100 mg/month. This self-contained benchtop instrument was very useful for producing R&D quantities of antibody from many different hybridoma clones.

Rapid Manual Assay for Tuberculosis Culture Confirmation

The ColorPAC TB assay (Becton Dickinson Microbiology Systems) consists of a microporous nylon membrane to trap the bacteria, under which are layers of absorbent paper (see Figure 2). The device features a positive control by means of a circular blot of goat antimouse antibody placed in the exact center of the area exposed to the test sample. The sample to be tested is applied via a focusing cup with a triangular orifice that limits application to an area immediately surrounding the control dot. The whole assembly is encased in a hard plastic shell. Reagents consist of wash buffers, an antibody conjugated to alkaline phosphatase, and a two-component enzyme substrate (NBT and BCIP).

Figure 2. The ColorPAC TB assay. Photo Courtesy Richard Root

The antibody selected for use to confirm TB was conjugated to enzyme via standard procedures. Subsequent testing with heat-killed mycobacteria demonstrated that when the Mtb 5.3.5—alkaline phosphatase conjugate was substituted into an existing ColorPAC product, it worked very well. In use, appearance of a purple triangle would indicate a positive result; appearance of the circular control dot would indicate only a negative result; and absence of the control dot would indicate a need to repeat the test.

As the assay was finally designed, operation began with the removal of colonies from Lowenstein-Jensen (L-J) slants via a sterile loop. The removed cells were suspended in distilled water with 0.02% Tween 80 to a density of MacFarland 0.5 to 1.0, then heated in a water bath at 95°C for 15 to 30 minutes. Using the focusing cup, 0.5 ml of the killed suspension was then applied to the device, the cup was removed, and the membrane was blocked and washed with a protein-containing buffer. Several drops of the conjugated antibody were then applied directly from the package vial and allowed to react for two minutes at room temperature. Buffer was then applied to wash away unbound conjugate. Finally, substrate solutions were applied and development allowed to proceed for five minutes prior to reading the result.

Third World Alternative. Because the production costs of the ColorPAC TB kit were judged to be too high for successful introduction in underdeveloped countries, the researchers also designed and produced a small number of the devices in a low-cost version. The materials used in this version included a cardboard slide, absorbent paper, adhesive tape, and a small square of nylon membrane. The focusing cup was replaced by a bit of white tape with a triangular hole placed over the membrane. To further reduce costs, it was determined that the reagents could also be packaged in larger amounts.

On the whole, the low-cost device functioned as well as ColorPAC, though with a somewhat lower capacity for absorbing liquid. The manufacturing cost of the device was estimated to be about $0.25, including labor. This device was used only for demonstrating the feasibility of a low-cost test, and was not included in the preclinical trials.

Preliminary Testing. The device was subjected to preliminary testing using a variety of Mtb strains (100 cultures), MOTT strains (200 cultures), and many other bacterial cultures. With one exception, all of the test results were negative for all non-TB cultures and positive for all Mtb-complex cultures.

The exception was seen in about 10% of the cultures of M. kansasii, an organism that is known to be difficult to antigenically resolve from Mtb in clinical samples. The researchers subsequently found that several of the test-positive streak plates from M. kansasii cultures actually showed two different colony types. While Mtb colonies produce a colorless streak, M. kansasii produces a yellow pigment when exposed to light. The test-positive streak plates showed both types of colonies, and it is possible that the colorless type was Mtb. This cross-reactivity increased with an increase of culture incubation time beyond the first appearance of growth, and the reaction also became more intense for Mtb.

The cross-reactivity of the assay with M. kansasii was not considered an invalidating effect. Although M. kansasii is rare outside of a few African nations, infection with the organism results in a disease very similar to tuberculosis. Moreover, consultations revealed that while M. kansasii is somewhat less sensitive to the antibiotics used to treat tuberculosis, it is treated identically. Considering the similarities between the two organisms, the researchers did not feel that assay cross-reactivity was detrimental in this case.

Although the presence of prescribed drugs might have had an effect on the results of the ColorPAC TB assay, the preliminary tests that were conducted suggest that no such influences are likely. In all the testing that was done, none of the results were affected by the presence of chemotherapeutic levels of antimycobacterial drugs such as streptomycin, rifampin, ethambutol, or isoniazid, or of other antimicrobials such as penicillin, cefloxitin, or cephalothin.

Since the ColorPAC assay was originally envisioned to be a culture confirmation test, preliminary testing also investigated the effects of using various media for sample culturing. Mtb cells taken from L-J slants consistently provided the best results. Less reliable test results were obtained when the cultures were grown in liquid media such as Bactec 12B, Septichek AFB, or the mycobacteria growth indicator tube (all from Becton Dickinson Microbiology Systems). It was observed that prolonged incubation beyond the positivity threshold of these media was required in order to obtain a reliably positive assay result.

Processed Sputum Testing. The device was also submitted to testing using processed sputum samples, without the intermediate and time-consuming step of growing a culture. Sputa were processed at a clinical laboratory by the N-acetyl L-cysteine-NaOH method, which reduces contaminants and liquefies the sample. This was centrifuged and the pellet resuspended in a small amount of buffer, a portion of which was delivered the same day for testing.

The results were initially encouraging: sputa containing Mtb (confirmed by acid-fast staining) gave positive results in the assay, suggesting that the test might be sensitive enough for such a use. However, further testing revealed problems with clogging of the membranes by both processed TB-positive and unprocessed sputum specimens seeded with Mtb. With limited time in the development calendar, the researchers were unable to find a reliable method of pretreating sputum to eliminate the clogging problems before the device went into preclinical trials. As a result, only the culture confirmation format of the assay was involved in preclinical testing.

Limited Preclinical Testing. Preclinical testing of ColorPAC TB was carried out at three sites. All patient samples were subjected to culturing and identified either by conventional biochemical reactions or by use of the Accuprobe molecular diagnostic assay (Gen-Probe, San Diego). No patient data were collected on HIV coinfection.

Cultures were grown on L-J slants, 7H10/7H11 Middlebrook agar, or Septichek (BBL, Cockeysville, MD) media. As soon as bacterial growth was evident, a suspension (0.5—1.0 MacFarland turbidity) was made in PBS buffer containing Tween 20. This suspension was tested using the ColorPAC TB assay. All together, the device was tested on more than 300 Mtb and MOTT cultures.

Analysis of the test results showed sensitivity of about 92% and specificity greater than 95% (contractual provisions preclude the inclusion of actual test data in this article). To obtain such high numbers, however, it is necessary to exclude those tests for which results were uninterpretable because there was poor flow or no flow of specimen through the membrane (7.8% of all specimens). Further work is needed to resolve this problem.

The results were very encouraging, especially in the areas of ease of use, reduced labor, and time savings. After initial growth in culture, testing took less than one hour, compared to another one to two weeks' incubation required for biochemical identification of Mtb.

Testing revealed two areas of concern: clogging of the membrane, which sometimes affected the test outcome; and the medium used for culturing patient samples. This test worked well only if growth occurred on solid media; performance was not satisfactory if the sample was cultured in liquid media.

Conclusion

When used in conjunction with standard culture methods, a rapid, manual immunoassay based on monoclonal antibodies can successfully diagnose specific mycobacterial infections. The need for such rapid and inexpensive diagnostics is clearly rising. Beyond Mtb, diagnostics are also needed for M. avium complex, M. kansasii, M. haemophilum, and M. xenopi, because of the association of such infections in HIV-infected persons.¹⁵

The work described above demonstrates the feasibility of developing an antibody and rapid manual assay for the diagnosis of tuberculosis. Along the way a number of key lessons were learned, some of which also provide opportunities for future development of this and other such assays.

• To prevent a loss of specificity, researchers should consider producing antibodies by means of in vitro (hollow-fiber bioreactor) technology.

• Initial culturing can affect test results; this assay performed well with cultures grown on solid media such as L-J slants, but not with samples cultured using liquid media.

• Membrane clogging remains an obstacle for future assays to overcome; doing so may also open the door to direct testing of patient samples.

• Low-cost versions of this and other assays appear to be feasible, and could open new markets in Third World countries.

• Cross-reactivity should not always be considered detrimental to an assay; this assay cross-reacted with M. kansasii in about 10% of cultures.

Careful planning of screening strategies and production methods can result in cost-effective monoclonal antibodies with the requisite specificities. In the near future, it may be possible to successfully develop a diagnostic that is capable of direct detection of the infection in patient material without the need for preliminary culturing. Such a tool would be especially appropriate to the conditions that exist in the developing countries of the world.

References

1. Ellner JJ, Hinman AR, Dooley SW, et al., "Tuberculosis Symposium: Emerging Problems and Promise," J Infect Dis, 168(3):537—551, 1993.

2. Grange JM, "Mycobacterial Diseases," Current Topics in Inf Series, 1:1—2, 1980.

3. Rieder HL, "Tuberculosis among American Indians of the Contiguous United States," Public Health Reports, 104(6):653—657, 1989.

4. Murray CJL, Stylbo K, and Rouillon A, "Tuberculosis in Developing Countries: Burden, Intervention and Cost," Bull Int Union against Tuberc Lung Dis, 65:6—24, 1990.

5. Starke JR, "Tuberculosis in Children," Curr Opin Pediatr, 7(3):268—277, 1995.

6. Fox JL, "TB: A Grim Disease of Numbers," ASM News, 56(7):363—365, 1990.

7. Wiseman KC, "Tuberculosis: An Old Disease with a New Face," Amer Neph Nurs J, 22(6): 541—554, quiz 555—556, 1995.

8. Fry DE, "The Reemergence of Mycobacterial Infections," Arch Surg, 131(1):14—17, 1996.

9. Scheck A, "Cost Resistance Slows Adoption of Nucleic Acid TB Tests," IVD Technol, 3(5):21—22, 1997.

10. Sommers HM, and Good RM, "Mycobacterium," in Manual of Clinical Microbiology, 4th ed, Lennette EH, Balows A, Hausler WJ, Jr, et al. (eds), Washington, DC, American Society for Microbiology, pp 216—248, 1985.

11. Young DB, "The Immune Response to Mycobacterial Heat Shock Proteins," Autoimmunity, 7:237—244, 1990.

12. Shinnick TM, Vodkin MH, and Williams JC, "The Mycobacterium Tuberculosis 65 Kilodalton Antigen Is a Heat Shock Protein which Corresponds to Common Antigen and to the Escherichia coli GroEl Protein," Infect Immun, 56(2): 446—451, 1988.

13. Kaufmann SHE, "Heat Shock Proteins and the Immune Response," Immunology Today, 11(4):129—136, 1990.

14. Ivanyi J, Morris JA, and Keen M, "Studies with Monoclonal Antibodies to Mycobacteria," Monoclonal Antibodies against Bacteria, 1:63, 1985.

15. Small PM, and Jacobson MA, "Human Immunodeficiency Virus and Mycobacterial Infections," in Tuberculosis, Schlossberg (ed), Berlin, Springer, pp 265—275, 1994.

Richard T. Root is senior project leader and head of the antibody technology laboratory at Bard Diagnostic Sciences, Inc. (Redmond, WA), and a member of the IVD Technology editorial advisory board; Abu F. Raisur Rahman was formerly manager of rare reagents at Becton Dickinson Microbiology Systems (Sparks, MD); Salman H. Siddiqi is a research fellow and Minal Ashtekar was formerly a research scientist at Becton Dickinson Diagnostic Instruments Systems (Sparks, MD).