Please refer to the following very good study and their Methods

If you have further questions, consider contacting the authors of this study

[email protected] or [email protected].

Proc Natl Acad Sci U S A. 2015 Jan 5. pii: 201416951

To assess the bactericidal and sterilizing activity of the first-line regimen, we infected 223 female BALB/c mice by aerosol with M. tuberculosis strain H37Rv, achieving a mean implantation of 2.62 (SD 0.44) log 10 cfus in the lungs, with the spleens being culture-negative the day after infection.

Our objective was to establish a heavy infection with high cfu counts in the lungs and the spleens; as such, our original protocol was to initiate treatment 6 wk after infection. However, by 4 wk after infection, 12 mice had already died and all remaining mice were sick; therefore, we began treatment at 4 wk postinfection.

An additional 10 mice died during the first 4 d of treatment. To compensate for this loss of mice, we reduced the numbers of mice killed on day 0 (the day of treatment initiation) as well as at some of the relapse assessment time points (detailed in Materials and Methods).

Thanks a lot. 

The answer to this question depends very much on what you want to test. The paper cited above established a very rapid fatal infection, but it used a very high dose to obtain this. This is good if for example you want to study dissemination, or to test a drug regimen against rapidly growing bacteria. It would not be ideal for testing a vaccine, or looking at more physiological function.

The initial kinetics of MTB growth in mice are very dose dependent. If you outline what question you are asking, we could give a more precise answer.

Thanks a lot for your answer and Interest.  We wanted to optimize s/c infection model in mice and study biomarkers of TB infection . We have used a dose of about 2×106 (power 6) and found considerable load of mycobacteria in lung at around 30-32 days. The same study was repeated using I.V route along with S/c route and we found bacterial colonization as early as 21 days. 

Since there is lot of variable evidences in literature I wanted to just confirm whether mycobacteria can colonize in lungs around the mentioned timepoints using s/c route of infection with mentioned dose ??

They certainly can. In the past, I tried different routes (sc., iv., nasal and oral, etc) and used doses of 10e3 to 10e6. I did whole-organ liquid culture for individual mice and by sc infection with high dose, bacteria started to be detectable (although in low numbers, and not in all mice) as soon as a week post-infection, with rapid growth from there. We used PCR to measure immune responses and in the spleen noted elevated IFN as soon as 4 days post infection.

Some things to note, though. Using a high dose like this and infection sc. or iv. gives quite different kinetics and a different immune response from low dose or pulmonary infection. With sc. infection (and even more with iv.) you get rapid colonisation of the spleen and lung (it typically is detectable first in the spleen and then shortly after in the lung, but in some mice I saw it on the same day). Also, in these mice colonisation of the spleen was substantial and prolonged - in contrast, with lower doses and pulmonary infection, dissemination to the spleen happened much later than the lung and was typically transient - bacterial load started to decrease after a couple of months.

Responses with high dose had a typically Th1 pattern, but were also characterised by high levels of IL-10 TNF-alpha and IL-6. In these mice, the IFN-gamma response declined or flattened out as the bacterial load increased. This is different from the pattern we saw with lower-dose or pulmonary infection where the Th1 response was much more prominent and lasted longer. High dose infections in mice may look more like the responses we see in progressive TB patients, who are typically already mildly immunocompromised when they are seen in the clinic, and who also have high ratios of IL-10 and TNF-a compared to healthy latently-infected individuals

Dear Sir, I thank you for your expert comments and for sharing your valuable experiences. It will surely add to our knowledge.

Actually, I forgot to add one important point! You asked about the best time to measure CFU.

Based on our own experiments in BALB/cJ and C57BL/6 mice, I would expect the inital peak of bacterial growth around 21-28 days. The kinetics of the high dose sc infection showed a rapid growth in the first 3 weeks, folowed by a small decline. This corresponded with the IFN-gamma response peaking. After the decline (which lasts 1-3 weeks), bacterial load started to slowly increase again and this corresponded with a fall in IFN-gamma levels. From this point the mice started to lose weight and had to be euthanised after a few more weeks.

Lower dose infections follow the same pattern with regard to CFU but somewhat delayed initially and at a much lowerpeak CFU load. They also display a much longer period of plateau (some months) after the initial decline, where CFU remain more or less stable before starting to rise. Mice infected with a low dose will remain alive for montsh although they also start to show growth retardation after a couple of months.

Thanks a lot. Actually I also observed similar trend with bacterial load at 30 days post s/c infection with high dose. Most importantly our rationale was to optimize time points for antigen and Antibody titres with that of Interferon response in spleenocytes and histopathological changes in lung.  

Can such studies be used to determine surrogate endpoints of protection?  Since low titres against MTB antigens and low antigen response post vaccination be used to determine protective efficacy of new vaccine?

Not sure if this adds anything, since it was used to study treatment response and not protection: there are two papers that compare in vivo Tb models used for drug testing, both come from U Colorado:

http://www.ncbi.nlm.nih.gov/pubmed/22940006

http://www.ncbi.nlm.nih.gov/pubmed/21135176

Maybe that' s a question more to Dr Doherty, and comes from my work in TB treatment research where we are increasingly recognizing the differences between mouse and human TB and problems in predicting human data from the mouse: since histopathology and prognosis in the mouse seems to be different from human TB, would you think a more "human-like" murine TB model is needed for biomarker research?

Just be carefull about the inoculum and the strain used and its virulence.

We have used live Imaging after in vivo infection upon tail injection, using 10(4) we have detected a baciili in lung after 17 days. By aerosolization it is more faster using less bacterial inoculum concentration.

@Norbert.  It's true that a mouse is not a human, and mouse TB does not always match our ideal of human TB, so we always have to be careful extrapolating. But that said, we now have a great deal of experience comparing immune responses in humans and mice and in general, they are far more alike than they are different.

I think with any animal model, the key is identifying what, exactly, you are measuring and then check to see if your model accurately reflects that aspect. So for example, known species of mice are a poor model for low-grade chronic infection. They don't seem to spontaneously develop anything that shows the same features of latent TB that we can see in patients (long term control of infection, without obvious symptoms). But monkeys do. The same is true of some features of severe disease like liquefaction of large granullomas, and this probably reflects the real physiological differences of these species.

On the other hand, the type of immune response developed in response to low-dose infection in mice, or to vaccination, matches almost precisely that seen in humans. The small - but not significant - degree of protection over BCG seen in the recent MVA-85A trial in South Africa is an almost exact replay in humans of the results in mice, and the immune responses were of precisely the same type as well.

So small animal models can be very useful in TB - you just need to ensure that the aspect of the disease you want to model is in fact accurately represented. In general, given that the immune responses to acute infection that we can measure are very similar between humans and mice, I think that using mice to search for biomarkers makes sense - but I would want to test anything we found in mice rigourously in humans. This is, in fact, exactly how we found the markers that were developed into the Quantiferon and T.Spot.TB tests.