In Vivo Adaptation of the Wayne Model of Latent Tuberculosis

Abstract

Cultures of Mycobacterium tuberculosis grown under oxygen depletion conditions enter into a state of nonreplicating persistence that may reflect a physiologically latent state. When these cultures were harvested and injected intranasally into mice, no bacteria could be recovered from the lungs for about 3 weeks, but after that evidence of regrowth was observed. Preimmunization of mice with a panel of selected vaccine candidates slowed or prevented this event. This simple model has potential for identifying vaccines targeting latent tuberculosis.

It has been estimated that as much as one-third of the global population has been exposed to Mycobacterium tuberculosis and harbors this organism in some form of dormant or latent infection (2, 3, 11). This represents a vast reservoir of individuals who may suffer reactivation disease at some point in the future.

There is no satisfactory cost-effective animal model of latent tuberculosis. Much of the emphasis to date has been on the chronic disease state in the lungs that develops in mice after low-dose aerosol exposure (18). Data have accumulated, however, to indicate that this remains a dynamic system both in terms of bacterial metabolism and susceptibility to chemotherapy and in terms of the host response, which maintains an activated profile (6, 8, 15). Using a similar model, another approach has been to produce bacterial mutants that fail to survive upon entering the chronic phase of the disease, which has supported the hypothesis that the bacilli change their respiration to an anaerobic mode (12, 13). Finally, studies of residual bacteria left after chemotherapy (the so-called “Cornell model”) have provided some information, but the method is unreliable and requires large numbers of mice; there is also the concern that it is actually a model of bacterial persistence rather than of true latency (9, 18).

In this study, we took the tactic that rather than rely on the mouse host response to generate latent bacilli, we would harvest the bacilli from in vitro cultures in which oxygen had been depleted. In this “Wayne model,” surviving bacteria are thought to be under conditions similar to those believed to occur in latent disease (22-24). Having established that these bacteria could slowly regrow in the mouse lung, we then applied this model to determine if certain vaccine candidates could slow or prevent the establishment of lung disease. Although contrived, this simple model may be useful in identifying vaccines or drugs specifically targeting latent tuberculosis.

Female specific-pathogen-free C57BL/6 mice aged 6 to 8 weeks were purchased from Jackson Laboratories, Bar Harbor, ME. Mice were housed in a biosafety level-3 facility and maintained with sterile bedding, water, and mouse chow. Specific-pathogen-free status was verified by testing sentinel mice housed within the colony.

M. tuberculosis strain H37Rv bacteria were grown to mid-log phase in Proskauer-Beck liquid medium supplemented with 0.05% Tween 80 (Sigma-Aldrich). Aliquots were frozen at −80°C until needed. Bacterial CFU was determined by plating serial dilutions of individual whole-lung homogenates on Middlebrook 7H11 agar (Difco, Detroit, MI) supplemented with oleic acid-albumin-dextrose-catalase (Becton Dickinson, Sparks, MD). Colonies were counted after 21 days of incubation at 37°C.

In keeping with the Wayne model (22-24), 10-ml cultures of H37Rv were started from 1:10 dilutions of frozen H37Rv stock and incubated at 37°C with fast stirring. Once bacteria had reached mid-log phase (optical density at 600 nm, 0.4 to 0.6), the culture was diluted 100-fold in Dubos medium and transferred to screw-top tubes (150 by 20 mm). The cultures were incubated with screw tops tightly closed at 37°C with slow stirring for 28 days. Control tubes contained methylene blue (1.5 μg/ml) as a visual indicator of oxygen depletion. The blue dye fades and finally disappears as oxygen is depleted. At day 28, a sample of the culture was carefully removed from the middle portion of the tube by pipette to avoid disturbing the pellicle on the surface or the sediment on the bottom. Cultures were centrifuged at 3,000 rpm and 4°C for 10 min. The pellet was vortexed briefly to resuspend it and was diluted with sterile phosphate-buffered saline and 0.05% Tween 80 to obtain optical density (600 nm) readings of approximately 0.6.

For challenge, mice were anesthetized by intraperitoneal injection of 0.08 ml xylazine (2 mg/ml) and ketamine (10 mg/ml) in phosphate-buffered saline. Bacteria were harvested as described above, using midtube sampling to avoid bacteria in the pellicle or at the bottom of the tube, which can be highly clumped. Harvested bacteria were washed and resuspended as described above to prepare single-cell suspensions. Using a micropipette, 10³ CFU of M. tuberculosis in 20 μl of sterile saline was gently placed beneath one nostril of each mouse. To verify bacterial uptake, several mice were euthanized by CO₂ inhalation 1 day after infection. The intranasal route was used instead of the aerosol method due to preliminary data indicating a considerable drop in cell viability immediately after aerosol administration, possibly due to the high force of the nebulization procedure. This effect was not seen if intranasal inoculation was given.

A panel of vaccine candidates was chosen based on identification by proteomic analysis of bacteria grown under low-oxygen conditions (J. T. Belisle, unpublished data). By using H37Rv genomic DNA as a template, the genes encoding Rv2031 (HspX), Rv3841 (BfrB), and Rv3810 (Erp) were amplified by PCR and then cloned into the pET23b expression vector by using restriction enzymes NdeI and XhoI. Identity was confirmed by restriction digestion and verified by DNA sequencing. The recombinant plasmids were transformed into the expression host BL21(DE3)pLysS (Rv2031), BL21 Star(DE3)pLysS (Rv3810), or BL21(DE3)pLysE (Rv3810). The transformants were induced with 0.5 mM isopropyl-β-d-thiogalactoside (IPTG), and the recombinant His-tagged proteins were purified by passage on nickel-nitrilotriacetic acid His bind resin (Novagen, Darmstadt, Germany). The purified protein Rv3810 was refolded by dialysis in a urea gradient decreasing from 8 M to zero. The endotoxin levels were <5 ng of lipopolysaccharide/mg yield of purified protein. Purified early secretory antigenic target 6 (ESAT-6) (Rv3875) was provided under the NIH “Tuberculosis Vaccine Testing and Research Materials” contract at Colorado State University. Vaccines were delivered in an adjuvant formulation of MPL (monophosphoryl lipid A) and DDA (dimethyldioctadecyl ammonium bromide). Stable emulsified MPL (MPL-SE) was kindly provided by Corixa Ribi. DDA was prepared by dissolving 50 mg DDA powder (Sigma-Aldrich) in 10 ml sterile distilled water. This was heated to 80°C with gentle stirring for 10 min and then allowed to cool to room temperature before use.

We should note, though it is not the specific topic of this paper, that these proteins have been tested in our laboratory in prophylactic studies. ESAT-6 routinely gives strong protection (a 0.5- to 1-log-unit reduction in bacterial numbers), whereas the other three proteins had minimal activity (reductions in bacterial numbers, 0.31 log unit with Rv2031, 0.23 log unit with Rv3841, and 0.37 log unit with Rv3810).

Animals were immunized subcutaneously in the scruff of the neck by using a 26-gauge needle. Vaccines consisted of 10 μg antigen, 25 μl MPL-SE (1 mg/ml), and 50 μl DDA. Vaccinations were given three times, 3 weeks apart. As controls, one group of animals was immunized with adjuvant alone, whereas another group was given a single subcutaneous injection of Mycobacterium bovis BCG Pasteur at a dose of 10⁶ CFU/mouse.

For histology, lung caudal lobes were preserved in 10% formalin and subsequently embedded in paraffin, sectioned, and stained with hematoxylin and eosin or by acid-fast staining.

As shown in Fig. 1, freshly thawed bacterial cultures established infection in the lungs and began to grow progressively, as anticipated. Oxygen-deprived bacilli could not initially be detected in the lungs, but after 20 to 30 days, colonies could be recovered from all mice and the infection showed evidence of growing at a similar rate. We noted that the variance of the data points was far higher in this model than in our experience using the aerosol model.

FIG. 1.

Growth of freshly thawed regular cultures (squares) and “Wayne cultures” (triangles) adjusted for similar viabilities in the lungs of mice after intranasal inoculation. Data shown are means for four to five mice plus standard errors. Because of the high variance, the day 30 values were marginally significant (P = 0.04) whereas the day 62 counts were not significant.

Having established that bacteria could readapt to the lung conditions and actively cause productive infection, we examined the effect of vaccinating mice prior to infection with either one of three mycobacterial proteins thought to be associated with latent or dormant infections, ESAT-6, or the BCG vaccine. Mice were then again infected via the nares, with an adjusted inoculum (10-fold increase) to increase the day 30 bacterial load in the controls. Thirty days after infection, we determined how many mice had recoverable bacteria, and for mice with bacteria, we determined the bacterial load (Table 1). Compared with the adjuvant control group, all test groups showed reductions in the lung bacterial load, with the Rv3810 and BCG groups showing reductions of 1.05 and 1.66 log₁₀ CFU, respectively. However, because of the high variance, the result with Rv3810 was not significant. In contrast, Rv2031 gave a half-log unit of protection, which was statistically significant. One mouse immunized with ESAT-6 had a bacterial load of 4.95 log₁₀ CFU, but more importantly, the other four mice in the group had no detectable colonies. Forty percent of mice in the Rv3810 group and 33% in the Rv3841 group also were devoid of colonies, whereas samples from all mice vaccinated with Rv2031 grew bacterial colonies.

TABLE 1.

Proportions of mice with recoverable colonies and bacterial loads

Vaccine	% of mice with recoverable colonies	Log10 CFU + SEMa
Adjuvant only	80	5.58 ± 0.16
BCG	75	3.92 ± 0.19*
Rv2031	100	5.00 ± 0.07**
Rv3841	67	5.28 ± 0.15
Rv3810	60	4.53 ± 0.97
ESAT-6	20	4.95

^a*, P = 0.002 compared to negative control; **, P < 0.001 compared to negative control.

Histologic examination of the lungs showed severe granulomatous inflammation in the lungs of mice inoculated with adjuvant alone (Fig. 2), including areas of developing necrosis. In mice prevaccinated with Rv3841 (BfrB) or Rv3810 (Erp), granulomatous lesions were also evident but appeared relatively small and densely packed with lymphocytes (lymphocytic). Despite modest protection, multiple lesions were also seen in the lungs of mice inoculated with Rv2031 (HspX), although here lesions were associated with much higher numbers of acid-fast bacteria than for the other groups. In contrast, small areas of granulomatous inflammation were seen around airways in the lungs of mice given BCG or in those of the single ESAT-vaccinated mouse showing bacterial growth, but most of the lung tissues were clear. For the four ESAT-vaccinated animals in which no bacilli were detected, the lung sections were all completely clear.

FIG. 2.

Histopathological appearance of representative lung sections taken from vaccinated mice showing evidence of bacterial regrowth. (A) Severe granulomatous inflammation with evident areas of necrosis in mice injected with adjuvant only. (B) Tight lymphocytic mass around a vessel in the single ESAT-vaccinated mouse showing regrowth. (C) Larger lymphocytic granuloma in mice vaccinated with BfrB. Similar patterns were seen in the Erp-vaccinated group (not shown). (D) Lesion in BCG-vaccinated mice. Note the similarity to panel B. (E and F) In contrast, mice vaccinated with HspX showed much larger, less organized granulomas (E), often associated with evident acid-fast bacilli (F). For panels A through E, hematoxylin-and-eosin staining was used; magnification, ×200. For panel F, acid-fast bacillus staining was used.

A large percentage of M. tuberculosis bacteria gradually deprived of oxygen in stationary cultures in the so-called Wayne model appear to survive but do so by considerably modifying their metabolic activity (1, 4, 5, 17, 19-21, 24). As a result, their susceptibility to certain drugs changes; they remain susceptible to rifampin and become susceptible to metronidazole but resistant to isoniazid (10). In the current study, we found that they are rather fragile at this point and appeared to be damaged by the high initial air pressure of aerosolization but could be introduced into mice by the more gentle intranasal route without much loss of viability. However, while control bacteria grew well in the lungs after inoculation, the oxygen-deprived bacilli took about 20 days before any evidence of resuscitation.

We then investigated whether immunization of mice prior to infection with the oxygen-deprived bacteria would slow or prevent the subsequent recovery of the bacteria. To address that question, we selected a small panel of mycobacterial antigens identified by proteomic analysis that are preferentially produced by bacilli grown under adverse conditions including oxygen depletion, as well as the immunogenic protein ESAT-6 and BCG as controls. Two effects were observed: a reduction in bacterial load and reductions in the number of mice showing evidence of any regrowth. In the latter case, ESAT-6 was highly effective.

Histologic analysis also revealed interesting and rather unusual results. Although considerable regrowth was seen in mice vaccinated with Rv2031 (HspX), Rv3841 (BfrB), or Rv3810 (Erp), significant lung damage was seen only in the HspX group, with considerable numbers of acid-fast bacteria detectable, whereas in the latter two groups the lesions were highly lymphocytic, suggesting that a considerable protective response had been generated. In the single mouse in the ESAT group showing regrowth, as well as in the BCG group, lesions were unusual in that they were small and highly lymphocytic and were restricted to areas of inflammation around vessels, possibly involving the lymphatic vessels. This might suggest rapid recognition of the presence of the infection, preventing deeper penetration of the lung tissues.

These results allow us to hypothesize that the active vaccines identified here may have promise as therapeutic vaccines for latent tuberculosis and the prevention of reactivation disease. It is possible that vaccines completely preventing the reappearance of the inoculum may be generating protective T cells recognizing antigens produced by bacteria in a stable state of latency or antigens signaling the initiation of active growth. Other antigens might be produced later in this process, so that T-cell recognition would slow growth but not initially prevent it. At this point it is hard to tell, but either mechanism would be of benefit.

We have labeled this model the “Wayne in vivo” model for the reason that we have driven the cultured bacteria into a state of “nonreplicating persistence” (NRP). This appears to allow bacilli to survive in a dormant or latent state for which several adaptations have been described (20). These are metabolic, with the organism switching to nitrate respiration, duplication of the bacterial chromosome, and cell wall structural changes (thickening). Exposure to oxygen is the primary trigger for resuscitation (20).

A major genetic control of this process is the dormancy survival regulator (DosR), and expression of the 48 genes in this regulon is a dominant factor in the response to oxygen limitation. These events are known to occur in the establishment of the NRP state. As more information accumulates, however, the role of DosR is becoming broader, relating both to virulence and to adaptations to low oxygen tension (7, 14, 25). This, of course, goes well beyond simple “latency,” and indeed that term is used loosely here, given our previous skepticism (16).

What is of interest to us from a practical perspective is not so much the generation of the NRP state, which, as discussed above, is becoming well understood, but rather what is happening in the mouse for the first 20 days or so after inoculation with these bacilli. At this time there is no technology to address this question that would not change the physiological status of the bacilli; hence, right now, we can only hypothesize that the bacteria are in a state of NRP from which at least some of them resuscitate and begin to grow.

This rather contrived model obviously needs further analysis to see if it truly mimics a state of latency and if it could be used to answer the interesting questions posed above. From a practical viewpoint, it could provide a cost-effective method of assessing the efficacy of therapeutic tuberculosis vaccines. Moreover, by infecting mice first and then giving a short course of drug therapy, the model could potentially also be used to identify new drug candidates for the treatment of latent tuberculosis.