Abstract

Ziehl-Neelsen (ZN) staining is the key technique for diagnosis of mycobacterial infections; however, a high percentage of patients exhibit positive signs of tuberculosis, as indicated by pathology, culture of mycobacteria, and polymerase chain-reaction analysis, and yet show negative results on ZN staining. In this report we present evidence that such ZN-negative specimens represent Mycobacterium tuberculosis bacilli in a dormant state with distinct cell-wall alterations: the classical cell-wall composition–dependent ZN staining of M. tuberculosis in lung sections gradually discontinued with persistence of infection, both in mice and in human patients; in contrast, detection of mycobacteria by cell-wall composition–independent staining using a polyclonal anti–M. bovis Bacille-Calmette-Guérin serum continued with persistence of infection. These findings have important implications for diagnosis, as well as for both chemotherapy and development of vaccine strategies.`

Materials and methodsC57BL/6 mice were bred at the animal facilities of the Max-Planck Institute for Infection Biology, at the Bundesinstitut für Risikobewertung in Berlin, under specific pathogen–free conditions. Sex- and age-matched mice were used for all experiments, which were conducted under conventional housing conditions. The part of the study investigating human specimens was approved by the Ethics Committee of the University Hospital Charité, Berlin

Mycobacterium tuberculosis strain H37Rv was grown in Middlebrook 7H9 broth (Difco) containing 0.05% Tween 80 supplemented with albumin-dextrose complex and was stored in aliquots at −70°C. Mice were aerosol infected by means of a Middlebrook Airborne Infection Apparatus (Middlebrook) adjusted to administer ∼10–100 cfu of M. tuberculosis/lung. For determination of bacterial colony-forming units, organ homogenates were diluted in PBS containing 0.05% Tween 80. Dilutions were plated on Middlebrook 7H11 agar plates supplemented with oleic acid/albumin-dextrose complex (Difco) and were incubated at 37°C for 3–4 weeks

Lungs were fixed in 4% paraformaldehyde overnight and were embedded in paraffin. Tissue sections (5 μm thick) were cut. For cold Ziehl-Neelsen (ZN) staining and rhodamine/auramine staining, color-reaction materials from Difco were used according to the manufacturer’s instructions. In brief, tissue sections were dewaxed, stained with carbolfuchsin (Bacto TB Carbolfuchsin KF; Difco) for 4 min, and, after being washed, decolorized with HCl (Bacto TB Decolorizer) until the stain was completely dissolved. Counterstaining was performed with brilliant green (Bacto TB Brilliant Green K) for 30 s. Quantification of bacilli that were positive for ZN staining (ZN+) followed the guidelines for microbiological diagnostics procedures for detection and quantification of mycobacteria [1]. Tissue sections were analyzed by 4 investigators independently, in a blinded fashion. The number of ZN+ rods was considered to be the number of bacilli per 400 fields of view at 1000× magnification. In order not to miss ZN+ bacilli in latently infected mouse or human tissue specimens, ⩾50 sections (in 5-μm increments of thickness) of representative organs were ZN stained, and every fifth section was stained with a polyclonal rabbit anti–M. bovis Bacille-Calmette-Guérin serum (pAbBCG) (Dako). Serial sections also were stained with pAbBCG. Rabbit serum was detected by goat anti-rabbit monoclonal antibody labeled with alkaline phosphatase (AP) (Dianova) for 50 min. After being washed with Tris-buffered saline, sections were incubated with a third antibody, donkey anti-goat monoclonal antibody labeled with AP (Dianova). Binding was visualized by means of naphthol ASBI phosphate (Sigma) and New Fuchsin (Merck) as substrate for the AP. Endogenous AP activity was blocked by levamisole (Sigma). All color reactions were performed for 15 min at room temperature in the dark. Sections were counterstained with hemalaun (Merck)

ResultsTo analyze whether ZN staining of M. tuberculosis depends on a distinct, cell wall–defined state of the mycobacteria in vivo, infected specimens were analyzed, in a comparative way, by the following 2 detection techniques: (1) conventional, cell-wall composition (CWC)–dependent ZN staining [2] and (2) CWC–independent immunohistochemistry using pAbBCG [3], verified by determination of viable bacterial counts (colony-forming units). C57BL/6 mice were aerosol infected with 100 cfu of M. tuberculosis H37Rv. As shown in figure 1, beginning 4 weeks after infection, when bacterial titers reached levels above the detection limit of 104 organisms/g of tissue, ZN staining of the lung sections revealed typical pink stains of rod-shaped M. tuberculosis bacilli; in contrast, 10 weeks after infection, CWC-dependent ZN stains were gradually lost, and they were barely detectable 30–49 weeks after infection (figure 1AD). This phenomenon was quantitatively verified by analysis of 400 field views of tissue sections, which were examined by several investigators, in a blinded fashion (figure 1O)

Figure 1

Loss of Ziehl-Neelsen (ZN) staining during persistence of Mycobacterium tuberculosis in mouse lung. Mice were aerosol infected with M. tuberculosis strain H37Rv. At different time points after infection, ZN staining of lung sections was used to analyze for acid-fast bacilli (A–D) mycobacterial titers were determined on the basis of lung homogenates (E–H; each graph shows the mean ± SE for 4 mice), and distribution of mycobacterial antigens in lung sections was analyzed by immunohistochemistry using a polyclonal rabbit anti–M. bovis Bacille-Calmette-Guérin serum (pAbBCG) (I–L). Results of control stainings, applying either pAbBCG onto uninfected lung sections (M) or secondary (2°) staining reagents onto M. tuberculosis–infected lung sections (N), remained negative. Quantification of ZN-positive mycobacteria on tissue sections was performed by 4 investigators in blinded form (O). Original magnifications, 800× (for A–D and I–N) and 40× (for insets in I–N), respectively. All panels show representative results of 5 similar experiments. n.d., not detectable

Figure 1

Loss of Ziehl-Neelsen (ZN) staining during persistence of Mycobacterium tuberculosis in mouse lung. Mice were aerosol infected with M. tuberculosis strain H37Rv. At different time points after infection, ZN staining of lung sections was used to analyze for acid-fast bacilli (A–D) mycobacterial titers were determined on the basis of lung homogenates (E–H; each graph shows the mean ± SE for 4 mice), and distribution of mycobacterial antigens in lung sections was analyzed by immunohistochemistry using a polyclonal rabbit anti–M. bovis Bacille-Calmette-Guérin serum (pAbBCG) (I–L). Results of control stainings, applying either pAbBCG onto uninfected lung sections (M) or secondary (2°) staining reagents onto M. tuberculosis–infected lung sections (N), remained negative. Quantification of ZN-positive mycobacteria on tissue sections was performed by 4 investigators in blinded form (O). Original magnifications, 800× (for A–D and I–N) and 40× (for insets in I–N), respectively. All panels show representative results of 5 similar experiments. n.d., not detectable

Our findings were corroborated by staining with rhodamine/auramine, another staining method that is commonly used to detect mycobacteria on the basis of their acid fastness and that therefore, like ZN staining, is CWC dependent. Consistent with the results of ZN staining, which have been described above, staining with rhodamine/auramine detected hardly any mycobacteria in tissue sections 49 weeks after infection (data not shown). Despite the dramatic reduction in acid-fast staining in lung sections of mice infected for >30 weeks, mycobacterial titers in the lung remained constant throughout the >70-week observation period (figure 1EH). To ensure that M. tuberculosis bacilli were still present in the sample sections investigated, serial sections were immunohistochemically stained with pAbBCG. Because of its polyclonal nature, pAbBCG recognizes multiple determinants of mycobacteria, including particular cell-wall components. Such recognition, however, is independent of CWC—that is, it is independent of both the relative amount of each particular component and these components’ relative structural connections within the cell wall. Staining with pAbBCG gave positive results at all times, even in those sections that were negative for ZN staining (ZN) (figure 1IL). The infected tissue was broadly stained, but only a few or no rod-shaped bacilli were detectable. Staining with pAbBCG is not expected to reveal rod-shaped staining predominantly. Because of the polyclonal nature of the serum, shed mycobacterial material also is expected to be stained by pAbBCG. In addition, thus far it is not clear whether M. tuberculosis bacilli persist in rod shape or in any other form. Our observations indicate that viable M. tuberculosis (1) were present at constant levels in the lung during the acute and persistent phase of infection, (2) were dispersed throughout the infected lung, and (3), as the duration of persistence of infection continued, lost CWC-dependent ZN staining

Virtually identical results were obtained in samples from patients with tuberculosis (TB). We analyzed lung sections from patients with acute TB, from patients with reactivated TB, and from patients with TB diagnosed circumstantially, the latter group representing patients with persistent latent TB (for definitions of latency and persistence, see [4]). As summarized in table 1, M. tuberculosis in tissue sections from the patients with either acute or reactivated TB (i.e., patients 1, 4, 6, 10, and 12) were positive both for CWC-dependent ZN staining and for staining with CWC-independent pAbBCG; in contrast, M. tuberculosis in tissue sections from the patients with persistent latent TB (i.e., patients 2, 3, 7–9, and 11) were positive for staining with CWC-independent pAbBCG but remained ZN. Figure 2 shows representative tissue sections from the patients with either latent or active TB (patient identification numbers correspond to those in table 1). As for the mouse specimens, staining with pAbBCG revealed only a few or no rod-shaped bacilli (figure 2, arrows in middle 4 panels), but broad staining of infected tissue was detectable. We conclude that, both in latent experimental TB and in patients with TB, loss of CWC-dependent ZN staining represents a characteristic attribute of dormant M. tuberculosis and can best be explained by mycobacterial cell-wall alterations abolishing ZN staining—that is, loss of acid fastness during persistence of infection

Table 1

Tuberculosis (TB) status of patients investigated by use of Ziehl-Neelsen (ZN) staining and staining with a polyclonal rabbit anti–Mycobacterium bovis Bacille-Calmette-Guérin serum (pAbBCG)

Table 1

Tuberculosis (TB) status of patients investigated by use of Ziehl-Neelsen (ZN) staining and staining with a polyclonal rabbit anti–Mycobacterium bovis Bacille-Calmette-Guérin serum (pAbBCG)

Figure 2

Ziehl-Neelsen (ZN) staining (top 4 panels) and staining with a polyclonal rabbit anti–M. bovis Bacille-Calmette-Guérin serum (pAbBCG) (middle 4 panels) in tissue sections from human lungs (patients 2 and 12), lymph nodes (patient 6), and muscle (patient 11). Clinical diagnosis was confirmed both by polymerase chain reaction and histopathology analysis and by microbiological diagnosis (see table 1). Results of control stainings applying secondary tissue sections remained negative (bottom 4 panels). Original magnification, 800×

Figure 2

Ziehl-Neelsen (ZN) staining (top 4 panels) and staining with a polyclonal rabbit anti–M. bovis Bacille-Calmette-Guérin serum (pAbBCG) (middle 4 panels) in tissue sections from human lungs (patients 2 and 12), lymph nodes (patient 6), and muscle (patient 11). Clinical diagnosis was confirmed both by polymerase chain reaction and histopathology analysis and by microbiological diagnosis (see table 1). Results of control stainings applying secondary tissue sections remained negative (bottom 4 panels). Original magnification, 800×

DiscussionWe have presented direct in vivo evidence for a distinct dormant state of M. tuberculosis a state that is characterized by cell-wall alterations during persistence of infection. The mycobacteria were analyzed directly in tissue sections from infected lungs, without either removal or in vitro cultivation. Both CWC-dependent ZN staining and CWC-dependent staining with rhodamine/auramine were lost during persistence of infection; only CWC-independent staining with pAbBCG was maintained. Bacterial titers remained constant throughout the observation period. Hence, M. tuberculosis did not retreat from the tissue but entered a dormant state during persistence of infection, a state that is characterized by a loss of CWC-dependent staining

Our observations can best be explained by mycobacterial cell-wall alterations. Such alterations could be caused by loss or reorganization of cell-wall components such that acid fastness—that is, retention of the ZN stain carbolfuchsin in the bacterial soma—is lost. Alternatively, reorganization of the cell wall could prevent access of the ZN stain to the bacterial soma. Whether this is due to changes in CWC or to reorganization of the cell-wall structure is a question that remains technically difficult to address. Ample evidence exists for a unique physiological state of M. tuberculosis during dormancy in vitro in the presence of oxygen deprivation [5]. However, in vivo proof to support this notion is difficult to obtain [6]. In vitro investigations during the early 1900s suggested several phenotypically different forms of viable M. tuberculosis in vitro and in vivo—the rod-shaped, ZN+ form and a series of ZN forms called “ultra fine” or “L-form” [7], “ghost” [8], “mycococcus” [9] or “Much’s granule” [10, 11]; many could be transformed from ZN+ forms into ZN forms, and vice versa, depending on the nutrient supply, yet neither of these forms was correlated with persistence of M. tuberculosis in vivo

Detailed knowledge of the dormant state of M. tuberculosis is crucial to the effort to combat this global threat. Here we have described a distinct state of latency-related, dormant M. tuberculosis in vivo. This finding has implications for diagnosis, chemotherapy, and development of vaccine strategies. First, the use of ZN staining as a major diagnostic tool has to be reconsidered [12]: our observations may explain ZN results in the diagnosis of tuberculosis; if a specimen remains ZN and, at the same time, results of pAbBCG staining are positive, then the ZN result may indicate a dormant stage of the M. tuberculosis infection, rather than absence of infection. It is tempting to speculate that ZN granulomatous pathologies of unclear etiology may be attributable—at least in part—to persistence of mycobacteria [13, 14]

Second, a different physiological state [15] and/or cell-wall organization during persistence of infection provides a clue to why chemotherapy susceptibility differs between dormant and active M. tuberculosis bacilli [16]. Chemotherapy may be refined case by case, on the basis of the physiological state of the infecting mycobacteria. Refinements in diagnosis—as well as refinements in chemotherapy—might have a direct effect on the World Health Organization’s recommended TB Control Strategy, the Directly Observed Treatment, Short (DOTS) course (summarized at http://www.who.int/gtb/publications/whatisdots/summary.html), which currently is the most effective strategy available for controlling the TB epidemic by coordination of diagnosis, chemotherapy, and the control of both treatment and treatment success

Third, our findings strongly support the notion that expression patterns of M. tuberculosis vary between different stages of the infection, implying that each subunit vaccine currently under development should be composed of either an antigen expressed throughout the infection or a cocktail of antigens expressed during ⩾1 per stage of infection [17,18]

Our results emphasize that M. tuberculosis infection should be investigated at different stages, to distinguish between the acute and persistent phases of infection. Most clinical cases may represent latent infection that, thus far, has not been considered by experimental in vivo research and has been missed by conventional ZN staining. Our observations may contribute to the understanding of persistent M. tuberculosis by considering cell-wall reorganization as an attribute of dormancy

Acknowledgments

We thank J. Zerrahn and U. Schaible for helpful discussions and careful reading of the manuscript

References

1
Nolte
FS
Metchock
B
Murray
PR
Baron
EJ
Pfaller
MA
Tenover
FC
Yolken
RH
Mycobacterium
Manual of clinical microbiology
 (pg. 
400
-
37
)
2
Bishop
PJ
Neumann
G
The history of the Ziehl-Neelsen stain
Tubercle
 , vol. 
51
 (pg. 
196
-
206
)
3
Wiley
EL
Mulhollan
TJ
Beck
B
Tyndall
JA
Freeman
RG
Polyclonal antibodies raised against bacillus Calmette-Guerin, Mycobacterium duvalii and Mycobacterium paratuberculosis used to detect mycobacteria in tissue with the use of immunohistochemical techniques
Am J Clin Pathol
 , vol. 
94
 (pg. 
307
-
12
)
4
Ulrichs
T
Kaufmann
SH
Mycobacterial persistence and immunity
Front Biosci
 , vol. 
7
 (pg. 
d458
-
69
)
5
Wayne
LG
Sohaskey
CD
Nonreplicating persistence of Mycobacterium tuberculosis
Annu Rev Microbiol
 , vol. 
55
 (pg. 
139
-
63
)
6
Orme
IM
The latent tuberculosis bacillus (I’ll let you know if I ever meet one)
Int J Tuberc Lung Dis
 , vol. 
5
 (pg. 
589
-
93
)
7
Khomenko
AG
The variability of Mycobacterium tuberculosis in patients with cavitary pulmonary tuberculosis in the course of chemotherapy
Tubercle
 , vol. 
68
 (pg. 
243
-
53
)
8
Canetti
G
The tubercle bacillus
New York: Springer Publishing
 (pg. 
1
-
55
)
9
Csillag
A
The mycococcus form of mycobacteria
J Gen Microbiol
 , vol. 
34
 (pg. 
341
-
52
)
10
Much
H
Über die granuläre, nach Ziehl nicht färbbare Form des Tuberkulosevirus
Beitr Klin Tuberk
 , vol. 
8
 (pg. 
85
-
99
)
11
Much
H
Die nach Ziehl nicht darstellbaren Formen des Tuberkelbacillus
Berl Klin Wochenschr
 , vol. 
45
 (pg. 
691
-
4
)
12
Behr
MA
Warren
SA
Salamon
H
, et al.  . 
Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli
Lancet
 , vol. 
353
 (pg. 
444
-
9
)
13
Bocart
D
Lecossier
D
De Lassence
A
Valeyre
D
Battesti
JP
A search for mycobacterial DNA in granulomatous tissues from patients with sarcoidosis using the polymerase chain reaction
Am Rev Respir Dis
 , vol. 
145
 (pg. 
1142
-
8
)
14
Saboor
SA
Johnson
NM
McFadden
J
Detection of mycobacterial DNA in sarcoidosis and tuberculosis with polymerase chain reaction
Lancet
 , vol. 
339
 (pg. 
1012
-
5
)
15
Mitchison
DA
Basic mechanisms of chemotherapy
Chest
 , vol. 
76
 (pg. 
771
-
81
)
16
McKinney
JD
In vivo veritas: the search for TB drug targets goes live
Nat Med
 , vol. 
6
 (pg. 
1330
-
3
)
17
Collins
HL
Kaufmann
SH
Prospects for better tuberculosis vaccines
Lancet Infect Dis
 , vol. 
1
 (pg. 
21
-
8
)
18
Kaufmann
SH
How can immunology contribute to the control of tuberculosis?
Nat Rev Immunol
 , vol. 
1
 (pg. 
20
-
30
)
Financial support: Alexander von Humboldt-Foundation (support to P.S.)
P.S. and T.U. contributed equally to this work
Present affiliation: Institut für Medizinische Mikrobiologie und Hygiene, University of Freiburg, Freiburg, Germany