Rv2623 regulates mycobacterial growth in vivo 
Although USP family proteins are expressed by many bacterial pathogens [7],[8], to date, there has only been one in vivo study, which showed that a Salmonella USP promotes virulence in mice [17].
The observation that Rv2623 modulates mycobacterial growth in vitro prompted us to examine the effect of this USP on the in vivo kinetics of M. tuberculosis infection.
Low dose aerosol infection of outbred Hartley guinea pigs with approximately30 CFU revealed a clear growth advantage of the Deltarv2623 mutant strain relative to wildtype.
As early as 20 days post-infection, the number of M. tuberculosis bacilli present in the lungs of Deltarv2623-infected guinea pigs was approximately10-fold higher (p<0.05) than those infected with wildtype Erdman, and continued to rise, attaining a 15-fold (p<0.001) difference by 60 days post-infection (Figure 3A).
Guinea pigs are able to control the growth of Erdman bacilli following the onset of adaptive immunity at approximately3 weeks post-infection, as evident by the relatively stable pulmonary bacterial burden beyond the 3 week time point, yet levels of Deltarv2623 bacilli continued to increase at a reduced but steady rate resulting in a rapidly progressing infection.
Moreover, Deltarv2623-infected guinea pigs were moribund at 60 days post-infection, while those challenged with wildtype Erdman remained relatively healthy, providing further evidence that the mutant strain is hypervirulent in this model.
Finally, complementation with a single integrated copy of rv2623 expressed from a constitutive mycobacterial promoter (Deltarv2623 attB::Phsp60Rv2623) abrogated the growth advantage of the deletion mutant (Figure 3A).
Also consistent with the fulminate disease progression displayed by Deltarv2623-infected guinea pigs are the more severe pathological changes observed as early as 20 days post-infection in the lungs of these animals, as assessed by histopathological studies, including the semi-quantitative Total Lung Score analysis (Figure 3B and Protocol S1).
Overall, the progression of pulmonic lesions was accelerated in Deltarv2623-infected animals compared to those infected with wildtype Erdman, accompanied by more extensive necrosis and widespread fibrosis.
This increase in lung pathology was also largely reversed in animals infected with the complemented Deltarv2623 attB::Phsp60Rv2623 strain (Figure 3B and C).
Results of the complementation experiments were further validated using a complemented strain Deltarv2623 attB::Prv2623Rv2623, whose expression of the wildtype universal stress protein is driven by the native rv2623 promoter [18] (Figure 3D and E).
In contrast to the result of the guinea pig study, we observed no difference in the kinetics of infection between C57BL/6 mice infected with wildtype M. tuberculosis, Deltarv2623, or the attB::Phsp60 Rv2623 complemented strain in a low dose aerogenic model [19], as assessed by lung bacterial burden (Figure 4A).
However, the mouse is a relatively resistant host to M. tuberculosis, particularly in strains such as C57BL/6 [20],[21].
In fact, evidence exists that M. tuberculosis triggers an immune response in mice that is in excess of that required for controlling the infection [22],[23].
Thus, the hypervirulence phenotype of Deltarv2623 observed in the susceptible guinea pig model could have been masked in the C57BL/6 mice.
Consequently, we examined the virulence of Deltarv2623 in the relatively susceptible C3H/HeJ mouse strain [24].
Indeed, the Deltarv2623 mutant was markedly more virulent relative to wildtype Erdman M. tuberculosis following aerogenic infection, as assessed by the mean survival time of C3H/HeJ mice infected with these strains (62 and 25.5 days post infection for Erdman- and Deltarv2623-infected mice, respectively, p=0.0014; Figure 4B).
In agreement with the survival data, quantification of tissue bacterial burden revealed a growth advantage for the Rv2623-deficient mutant relative to wildtype M. tuberculosis Erdman (Figure 4C).
Manifestation of this hypervirulence phenotype is apparent as early as 3 weeks post-infection, with the lung bacterial burden of mice infected with Deltarv2623 M. tuberculosis approximately100 fold higher than that in the wildtype-infected animals.
As in the guinea pig studies, results of complementation experiments involving the reintroduction of a single copy of wildtype rv2623 into Deltarv2623 M.tuberculosis reverses the hypervirulence (Figure 4C) exhibited in the C3H/HeJ model, thus indicating that the observed growth phenotype of the tubercle bacillus deficient for the universal stress protein is rv2623-specific.
Finally, survival of Deltarv2623-infected mice was also significantly reduced in another susceptible mouse strain, C3HeB/FeJ (Figure S1).
Together, the animal studies provide strong evidence that Rv2623 regulates the growth of M. tuberculosis in vivo: in the absence of Rv2623, the tubercle bacillus fails to establish a chronic persistent infection, exhibiting a hypervirulent phenotype.
