Download Virulence Mechanisms in Tuberculosis

Virulence Mechanisms in Tuberculosis
Priscille BRODIN
Institut Pasteur Korea, Inserm Avenir, Seoul, Korea
I will discuss what is currently known about the virulence mechanisms in tuberculosis.
I.
Definition of Mycobacterium Tuberculosis Virulence
1. The Etiological Agent
Mycobacterium tuberculosis is caused by a mycobacteria belonging to the mycobacterium
genus, which comprises over 17,000 different strains. M. tuberculosis agent can be stained
with a specific method, the Ziehl Neelsen method, due to the rich composition of its cell-wall
in mycolic acids and glycolipids.
Working with mycobacteria is considered rather difficult, due to their slow replication time approximately one day. In addition, mycobacteria are highly infectious, so researchers are
obliged to work in Class III biosafety laboratories.
2. Disease Process
When a patient is infected with M. tuberculosis, the bacilli reach the alveoli and infect the
alveolar macrophages in the lungs. At this stage, it is considered an intracellular bacillus,
since it is located within the host cell. The next stage is the latency stage, and in
approximately 10% of the cases, the tuberculosis disease is developed.
A critical feature of the latency stage is granuloma formation, with the presence of giant cells.
This leads to caseous necrosis and the release of bacilli, which can then be transmitted to new
patients and infect them. At this stage, the bacilli are referred to as extracellular bacilli, and
their condition is anaerobic. Then, is some cases, calcification occurs, while in others,
dissemination takes place, with miliary tuberculosis or meningitis.
3. Defining Virulence
I believe virulence is slightly different than resistance. I would define virulence as the ability
of a microbe to cause disease in a host. It can also be defined as the degree of pathogenicity
of a microbe. Thus, virulence is always related to a host body or cell.
Virulence is caused by several factors that are encoded by the bacteria. There are then several
different ways by which the bacteria can cause the disease. First, the virulence factors can
induce cell adhesion to the host cell. Second, they can increase the colonisation of the host
body and the persistence. Third, they can favour invasion into host cells. Fourth, the bacteria
can express inhibitors of the immune response. Fifth, some bacteria can express toxins.
It has often been noted that there are no classical bacterial virulence factors for M.
tuberculosis. That is, no toxin has been identified, and there is no pathogenicity island in the
genome. Thus, although there are numerous putative virulence factors, we must admit that
our knowledge is still limited.
4. Models for Measuring Virulence
Nevertheless, TB virulence can be quantified, with the help of two models: the animal model
and the cellular model. The animal model utilises mice, guinea pigs, and non-human
primates. Typically, the animals are infected either via the aerosol route, or intranasally, like
the natural infection. The animals usually receive 100 CFU ((colony forming unit, or number
of living bacteria) at day zero. Then, typically, following a month, there is a 3 log increase of
the CFU. Then, the number of bacteria remains constant for two or three months.
a. The Mouse Model
There are typically two target organs in the mouse model - the lungs and the spleen.
In parallel to examining the CFU, we also examine the lesion, to determine the severity of the
strain. The parameters associated with virulence in the mouse model are colonization,
persistence, and pathogenicity. These depend on the route of entry and the infecting dose, and
we also examine morbidity and mortality. One the host side, it is possible to examine
immunity.
b. The Cellular Model
The second model used to quantify virulence is the cellular model. Mycobacteria can infect
the macrophages cells and dendritic cells, and can also enter the pneumocytes and adipocytes
cells.
The first feature we examine is invasion survival replication. Today, it is possible to follow
the growth and spread of mycobacteria within the cells using automated fluorescent
microscopy. We are thus able to quantify the bacterial burden, and obtain a precise measure
of the invasion parameters, survival and multiplication inside the cells. This method can
currently be applied to several hundreds of strains.
A second feature we examine inside the cells is trafficking. The bacterium has the ability to
block phagosome maturation. One of the critical features of mycobacteria is that they reside
in endosomes that cannot be acidified into lysosomes. While we still do not understand the
factors responsible for this phenomena, we estimate that it helps the bacteria survive and
multiply within the cells.
Examining cellular models also allows to study the cytokines release and profile.
For example, it is possible to examine the release of TNFα, IL-2, IL-12, and IFNγ.
II. Features of M. Tuberculosis
The features of M. tuberculosis can be examined using a phylogenetic tree, based on 16S
rRNA, hsp65, sodA, and rpoB sequences. In mycobacterial terminology, bacteria are divided
into slow-growing and fast-growing types. M. tuberculosis, which is a slow-growing bacteria,
belongs to the tuberculosis complex. It is striking that the Identity in nucleotidic sequence of
the species in this complex is greater than 99.9%.
Other slow-growing mycobacterium bacteria include the M. marinum and M. ulcerans. The
latter are causing the Buruli ulcer, which is an emerging disease caused by mycobacteria, and
is found mostly in Africa.
1. The M. Tuberculosis Complex
The M. tuberculosis complex consists of six members. One is of course M. tuberculosis,
which is the main agent in human TB. The other five are Mycobacterium africanum, bovis,
canettii, pinnipedii and microti. The tropism of the different species varies. They each come
from isolates, from some of which vaccines were developed. The most popular strain used in
the laboratory is H37Rv, from which the H37Ra vaccine is derived, similarly to the way BCG
has been obtained from passage of M. bovis.
Currently, significant work is being conducted on the W-Beijing strain family, and
particularly on the GC1237 isolate. M. bovis has the largest tropism, and can infect most
mammals. It used to cause TB in humans, but more rarely, and all the BCG strains were
derived from it. M. pinnipedii was shown to cause disease in Seals in Australia.
I would like to briefly discuss M. microti. It was isolated in England in the 1930s among the
population of small rodents called voles, which exhibited a disease similar to tuberculosis.
More recently, human patients with active tuberculosis were diagnosed as infected with the
M. microti. M. microti OV254 is a naturally attenuated strain, and its derivatives have been
used in Czechoslovakia as a vaccine in the 1960s, instead of BCG.
In conclusion, since the different isolates of M. tuberculosis differ in tropism and virulence, it
follows that these strains must encode virulence factors.
2. Mycobacterium Specific Features: Cell Envelope
One of the specific features of the M. tuberculosis is its cell envelope. On the membrane,
there is arabinoglactan on top of the peptidoglycan, and then mycolic acid. There are also
some Phenolic Glycolipid (PGL), and Lipoarabinomannan (LAM). It was recently shown
that PGL seems to be reproduced by the Beijing strain, and can account for the increased
virulence of the strain. LAM is an immunomodulator.
3. The TB Genome
The first sequence of the TB genome, the H37Rv strain, was obtained 10 years ago. It is
comprised of one circular chromosome, of 4.4 Mb, which contains about 4,000 genes.
No extrachromosomic DNA has been found in this strain.
The M. tuberculosis genome includes all the genes that enable the bacteria to survive in
aerobic, microaerophilic, and anaerobic conditions. In comparison to other bacteria, it has a
complex regulatory potential. It includes 13 sigma factors, 11 two-component systems, and
histidine kinase factors. It was also striking to discover 11 Serine Threonine Protein Kinases
in the TB genome. These kinases are one of the preferred targets of the pharmaceutical
industry when looking for new drugs.
The TB genome also includes a large number of genes involved in lipid metabolism, which is
required for the metabolism of the cell wall. It also includes some degradative enzymes, such
as 20 cytochrome P- 450, as well as many insertion elements, such as IS6110, which are the
basis of genotyping.
Another striking discovery was the presence of large gene families. There are about 160
members of the PE/PPE family in the genome, but we are not sure yet what their function is.
Other families include mce, mmpL, and ESAT-6. Again, no toxin has been found, nor any
obvious pathogenicity island, and merely a few classical virulence genes.
4. Summary
To summarise, virulence in tuberculosis is different from that of classical bacteria. Although
the major virulence factors remain to be identified, robust cellular and animal models allow to
measure the virulence. According to clinical data, the M. tuberculosis strains and the clinical
isolates exhibit significant virulence differences, so much can still be learned. Finally, there
are many putative candidates arising from biochemical and genome global analyses.
III. Approaches for Identifying TB Virulence Mechanisms
For the past ten years, many laboratories have attempted to identify virulence mechanisms.
Their approaches can be roughly classified into four major groups.
1. Strain Comparison: Comparative Genomics
The first approach is comparative genomics. This method is to compare the genome content
of virulent and non-virulent strains. This task requires an available sequence, typically of the
virulent kind. We use the M. tuberculosis H37Rv sequence. Then, we use an attenuated or
vaccine strain that is extremely similar genetically, such as M. bovis BCG or M. microti
OV254. We also require genomics, genetic, and DNA based tools, such as Bacterial Artificial
Chromosomes (BAC) libraries, microarrays and bioinformatics.
Let us examine the differences between the genomes.
First, we consider the
Single Nucleotide Polymorphism (SNP), which is a single nucleotide that is mutated from one
strain to another. 2,000 mutations were found between the two TB sequenced strains, H37Rv
and CDC1551. Most of these mutations have no significant effect.
A feature that is specific for TB is the regions of difference (RD regions). These regions
correspond to a piece of DNA that has been lost from one strain to the other. The regions are
numbered from RD1 to RD26. Then, the opposite phenomenon has been discovered: a
complementary piece of DNA that was present in the attenuated strain, but absent in the
virulent strains. These regions are called RvD regions. A third region, TbD, designates DNA
that is absent in the tuberculosis strains. A famous example is TbD1, which is a piece of
DNA that is absent in most of the strains, but still found in Indian strains.
Once we have identified the genes that have been lost during attenuation, our goal was to
discover what these genes encode. To achieve that goal, we required functional genomics.
However, first, it was necessary to extend the study to multiple strains, in order to select the
relevant regions for virulence. For that purpose, we have used internally controlled PCR
amplification. We used two set of primers. The first set was an internal primer, which
amplified a product within the RD region. The second set was a flanking primer, which
provides a product merely when the region was lost. The aim of the PCR amplification was
to determine, in many strains, whether a certain region is present or absent.
We also used focused macro-arrays. These are basically membranes that have been spotted
by different probes, corresponding to different genes from the TB genome, particularly from
the RD regions. The technique is used to study a couple of hundreds of strains.
I will now briefly describe how we demonstrated that RD1 was absent in M. microti. The
RD1 region is displayed, and it encodes two secreted proteins, CFP10 and ESAT-6. ESAT-6,
which is the basis of the new diagnostic method QuantiFERON, is a strong T-cell antigen.
However, we still do not know its function at the cellular level.
We obtained genomic DNA from different species of the tuberculosis complex. Then, we
transferred the DNA into a membrane, and hybridised it with a probe that was generated with
an ESAT-6 primer. We have obtained a product for the M. bovis, M. canettii and Rv stains,
thereby demonstrating that ESAT-6 was indeed present in this genome. On the other hand,
we did not obtain a product with BCG or with an entire set of M. microti isolates.
As a confirmation, we used a RD1 probe, which also provided evidence of the RD1 division.
We have demonstrated that the RD1 division was not the same in M. microti and BCG.
Further, the findings might suggest that the RDI region could play an important role in
virulence. Since, as I have noted, M. microti and BCG are the only two members of the
tuberculosis complex that are non-virulent in humans.
We conducted this study with all the members of the TB complex, to see what DNA regions
have been lost by each species. The largest number of lost regions occurred in BCG and M.
microti. Our goal was to find the region that has been lost in all the attenuated strains, but is
still present in all the virulent ones. We discovered only one such region - RD1. RD1 is
absent from the BCG and M. microti strains, but present in the human infectious strains.
In order to determine whether the RD1 region is involved in virulence, we over-expressed the
region within the BCG. For that purpose, we used the integrative cosmid tool, which allows
the expression of a large DNA fragment inside the genome. We called the strain obtained
BCG:RD1; that is, a BCG strain that expresses the entire RD1 region. Strikingly, we
observed a difference in the morphology of the colony. We conducted the same experience
with M. microti, to ensure that it was not a unique feature of BCG, and we observed the same
phenomenon.
Then, we infected mice with the BCG and BCG:RD1 strains, and examined the growth of the
CFU in the lungs. We discovered that the BCG strain did not continue to grow, while the
BCG:RD1 did. In addition, we found some lesion present in the BCG:RD1 stain, as well as a
clear expression of the antigen ESAT-6.
Since the increased bacterial burden and lesion formation is a correlate of virulence, we
concluded that RD1 is a virulence factor. Furthermore, in a study conducted by W.R
JACOBS et al, the RD1 strain has been removed from the RD region of the Rv strains. The
researchers have found a significant decrease in the number of bacteria. The same decrease
has been achieved by removing the ESAT6. It was therefore rapidly confirmed that the major
virulence factor in the RD1 region is ESAT6.
However, it is important to remember that when RD1 is added to the BCG strain, the result is
not TB, and when it is deleted from the TB strains, the result is not BCG. Thus, RD1 is
merely one of the virulence factors.
2. The Genetic Mutant Approach
The first step of the genetic mutant approach is to formulate an hypothesis. Then, we create a
mutant of the wild type strain, through a gene knock-out process. The deletion is performed
either through allelic exchange or transposon insertion. The gene or the operon is not
transcribed, and the protein is not synthesised. We thus obtain a genetic KO-mutant.
Let us examine the isocitrate lyase (icl) example. Our initial observation was that
isocitrate lyase activity increases when M. tuberculosis infects macrophages. Thus, it was
reasonable to assume that it plays is an important role in macrophage replication. We thus
isolated the wild type strain sequence from the genome analysis, and obtain a knock-out
mutant, Δicl, in which an antibiotic marker replaced the icl gene.
Then, we infected mice either with the wild type strain or the knock-out mutant. We observed
that the growth of the two strains was identical during the first three weeks. However, then,
the number of bacteria in the lungs of the icl mutant has decreased, while the number bacteria
in the wild type strain remained unchanged.
A survival study revealed that while 100% of the mutant mice survived until the end of the 24
week experiment period, all the mice infected with the wild type strains died within 16 weeks.
These results led us to the conclusion that the icl gene is involved in persistence in mice,
which is a virulence parameter.
However, we wanted to confirm that the deleted gene is indeed responsible for the effect
observed, and that it is not due to a side effect. We have therefore decided to add back the
gene that we have omitted to the KO mutant. This has been achieved through insertion.
Since the virulence has been restored when the gene was restored, we were provided with
definite proof that the gene studied is indeed a virulence factor.
3. The Large Mutant Pool Approach
Several laboratories have adopted the large mutant pool approach. It utilises a transposon
mutant library, which is a pool of different genetically modified bacteria. Each bacteria
mutant has one gene that has been inactivated through a transposon insertion. Thus, we
obtain a mixture of bacteria that have different inactivated genes. Considering that the TB
genome includes 4,000 genes, In a 100,000-member library, there are approximately 25
mutants for each gene.
A study conducted by RUBIN et al examined genes that are essential for growth inside the
mice. The researchers thus injected the pool of mutants into the mice, waited for a few
weeks, extracted the target organs, and then recovered the bacteria.
They then decided to compare the diversity of the output and input pools. If the genes that
played an important role in growth inside the mice have indeed been deleted, they would not
be found in the output pool. Using DNA PCR, labelling, and hybridisation to a genomic
microarray, they managed to identify and compare the diversity of the bacteria responsible for
growth in-vivo. The same experiment was conducted in-vitro, and the authors compiled a list
of 194 genes that are necessary for growth in-vivo.
The same approach was adopted in several studies focusing on macrophages over the past few
years. While these studies do provide insight into virulence factors, they still require
validation with individual mutants, as well as modelisation of molecular mechanisms.
4. Global Gene Expression Analysis
The global gene expression analysis approach utilises regular tuberculosis or BCG strains
instead of mutants. The element that is being modified is the model. While it is still possible
to use mice, it is also possible to use in-vitro models, which are believed to mimic the in-vivo
conditions of the persistent bacteria. Thus, it is possible to compare the bacteria state in
anaerobic and aerobic conditions, when the bacteria are in nutrient starvation, or in different
persistence models.
For that purpose, it is possible to use postgenomics techniques, such as Transcriptomics,
Proteomics and Lipidomics. These techniques enable us to search for the genes or proteins
that are over-expressed in one condition, in order to compare them with another condition.
Once again, we managed to obtain a list of genes that we believe are encoding putative
virulence factors. In some studies, the analyses of the bacteria and the host factors have been
conducted in parallel. These findings also require confirmation through genetic mutant
studies.
IV. M. Tuberculosis Virulence Factors
I would like to provide a slightly more detailed description of the tuberculosis virulence
factors, which I roughly classify into four categories.
1. Secreted Factors
The first category is secreted factors, which are products that are exported outside the
bacteria. They include culture filtrate proteins, such as HspX, Esat6/CFP-10, and 19-kD.
These proteins have been studied thoroughly, due to their potential for vaccine and diagnostic
applications. More recently, the protein Kinase G was shown to be involved in phagosome
maturation arrest.
2. Cell Surface Components
The cell surface components include proteins involved in synthesis of the cell wall. Among
them, I would like to emphasise the LAM, which has multiple functions. First, it is an
immunomudulator. In addition, it has been shown that LAM is able to inhibit the release of
calcium.
3. Enzymes Involved in Metabolism
The third category includes enzymes involved in general or cellular metabolism. Some of
these enzymes are involved in lipid and fatty acid metabolism, which allows bacteria to grow
on fatty acids. It was also discovered that when introducing knock-out mutants to the amino
acid and purine biosynthesis pathway, then the bacteria has a different phenotype in virulence.
A similar phenomenon was observed when disrupting the genes involved in metal uptake, as
well as factors involved in anaerobic respiration and oxidative stress proteins.
4. Transcriptional Regulators
The final category of virulence factors includes the transcriptional regulators. Following the
systematic knock-out of all the sigma factors, four have been found to account for change in
virulence in the mouse model. The PhoP/PhoR response regulatory is also an important
virulence factor, and a knock-out mutant of PhoP/PhoR is considered a promising vaccine
candidate.
5. Summary
To summarise this section, numerous focused and global approaches have been undertaken to
identify M. tuberculosis virulence factors. We have obtained long lists of genes, which I
believe require further confirmation and investigation. Finally, the genes encoding virulence
factors belong to different functional classes: secreted factors, cell surface components,
metabolism enzymes, and transcriptional regulators
I.
Conclusions for Future Research
I would like to offer several recommendations for future research.
first, I believe it is necessary to increase the comparison of whole genome sequences from
clinical isolates. Since it is easier to sequence a genome today, we should be able to learn
more about the clinical isolates.
Concerning the pooled screens, I believe it would be useful to investigate in different genetic
backgrounds, such as an immuno-deficient background, or different animal models.
When we study mutants in the animal model, it would be useful to be able to turn the gene
expression ‘on’ or ‘off’. Therefore, inducible promoters that can be turned on in-vivo would
be extremely helpful for determining in which part of the replication process of the bacteria
the gene plays an important role in the animal model.
Finally, we must remember that tuberculosis is a human infection. For that reason, we could
greatly benefit from a global transcriptional and proteomics analysis of human infections.
I.
Question and Answer Session
Ying ZHANG
What is the size of the RD1 deletion in the microtic strain vs. the BCG strain? Do you know
what causes the deletion?
Pricille BRODIN
The deletions are roughly the same size. However, in the microtic strain it is shifted from
Rv3864 to Rv3876, whereas in the BCG1 it is shifted from Rv3871 to Rv3879.
We do not know what causes the deletion; it is simply a loss of DNA that is being cut within
the gene. When we compared the microti OV strains that were isolated in the 1930s with the
microti isolates that were found in human TB patients, and we found exactly the same
deletion. However, some microti isolates, which have been isolated in 1995 do not have the
RD1 region, and are still considered as agents responsible for TB in humans. In this case,
RD1 is not the virulent factor.
Steffen STENGER
What do you know about the immunology of the Mycobacterium microti, compared to other
mycobacterium in the M. tuberculosis complex? It seems peculiar that organisms that are
99.9% identical on the genetic level would be so different in terms of the clinical disease.
Pricille BRODIN
We know that M. microti grows even slower than TB or BCG. Thus, it takes four weeks to
obtain a colony on a plate, and eight weeks to obtain a clinical strain. OV254 is the model
strain with which we have conducted all the investigations on the RD1. Generally, we did not
find any difference between OV254 and BCG, besides the fact that the CFU grew somewhat
slower. However, when we infected a mouse with the clinical isolate of microti, the bacteria
grew significantly faster than on the plate, and killed the mouse.