Reactivation of latent tuberculosis through modulation of resuscitation promoting factors by diabetes

The evidence of an association between diabetes and latent tuberculosis infection (LTBI) remains limited and inconsistent. Thus, the study aims to delineate the role of diabetes in activation of latent tuberculosis infection. Murine model of latent tuberculosis and diabetes was developed, bacillary load and gene expression of resuscitation promoting factors (rpfA-E) along with histopathological changes in the lungs and spleen were studied. Treatment for LTBI [Rifampicin (RIF) + Isoniazid (INH)] was also given to latently infected mice with or without diabetes for 4 weeks. Diabetes was found to activate latent tuberculosis as the colony forming unit (CFU) counts were observed to be > 104 in lungs and spleen. The gene expression of hspX was downregulated and that of rpfB and rpfD was observed to be upregulated in latently infected mice with diabetes compared to those without diabetes. However, no significant reduction in the CFU counts was observed after 4 weeks of treatment with RIF and INH. Diabetes helps in the progression of LTBI to active disease mainly through altered expression of resuscitation promoting factors rpfB and rpfD, which can serve as important targets to reduce the shared burden of tuberculosis and diabetes.

Diabetes affected the granuloma formation in the lungs. At week 9, ill formed granulomas were observed as represented by collections of epithelioid histiocytes along with the inflammation and haemorrhage in the lung tissues of animals from latent tuberculosis only group ( Fig. 2A). In the lung tissues of animals from latent tuberculosis with diabetes group, peri-vascular and peri-bronchiolar inflammation was observed in lungs as indicated by the infiltration of lymphocytes with no granulomas (Fig. 2B). Similarly, in the lung tissues of animals from latent tuberculosis with immunosuppression group mild peri-vascular and peri-bronchiolar inflammation was observed but no granulomas were identified (Fig. 2C). At week 13, a collection of pale histiocytes admixed with surrounding lymphocytes and increased inflammation was observed in the lung tissues of Group-I animals (Fig. 2D). However, the lungs of latently infected mice with diabetes showed very dense peri-vascular and peri-bronchial inflammation. Ruptured air spaces were also seen in the lung tissue sections (Fig. 2E). In www.nature.com/scientificreports/ the lung tissue of Group-III animals, lymphocytic infiltration was observed (Fig. 2F). The level of significance between Group-I, II and III for granuloma formation, peribronchiolitis and perivasculitis at week 9 and 13 is given in Table 1. However, no changes were found in the spleen in all groups at week 9 and 13 of diabetes induction.
Reactivation of M. tuberculosis by resuscitation promoting factors. The expression of hspX gene was significantly downregulated in latent tuberculosis with immunosuppression group as compared to latent tuberculosis only group which signifies the activation from latency at week 9 (Fig. 3A). However, the expression of all the resuscitation promoting factor (rpf) genes was found to be significantly downregulated in latent tuberculosis with immunosuppression group (Group-III) as compared to latent tuberculosis only (Group-I) and latent tuberculosis with diabetes group (Group-II) at week 9 ( Fig. 3B-F). Moreover, no significant change was observed in the expression of any of the genes mentioned above in latent tuberculosis with diabetes group as compared to latent tuberculosis only group after 2 weeks of diabetes development i.e. week 9, indicating that Group-II is behaving similar to Group-I ( Fig. 3A-F). However, at week 13, a significant decrease was observed in the expression of hspX gene in latent tuberculosis with diabetes group (Group-II) as well as in latent tuberculosis with immunosuppression group (Group-III) as compared to latent tuberculosis only group (Group-I).
No significant difference was observed in the expression of hspX between Group-II and Group-III (Fig. 4A).

Figure 2.
Histological examination of lung morphology to assess the presence of granulomas. After diabetes development, Hematoxylin & Eosin staining was performed on lung tissue of mice from Group-I, II and III at week 9 and 13 of streptozotocin treatment. Lung section of (A) Group-I mice (B) Group-II mice (C) Group-III mice after 9 weeks; Lung section of (D) Group-I mice (E) Group-II mice (F) Group-III mice after 13 weeks of streptozotocin treatment. Arrows indicate granulomas (A,D) and inflammation (B,C,E,F). Images were taken at 20× magnification. Group-I: latent tuberculosis only, Group-II: latent tuberculosis with diabetes and Group-III: latent tuberculosis with immunosuppression.  www.nature.com/scientificreports/ A significant increase was observed in the expression of rpfB gene i.e. approximately 2.3-fold, and rpfD gene, i.e. 2.8-fold in latently infected animals having diabetes as compared to animals having latent tuberculosis only (Fig. 4C,E). In latent tuberculosis with immunosuppression group also, around 1.5-fold increase was observed in the expression of rpfB gene as compared to latent tuberculosis only group but the expression was significantly decreased when compared to Group-II (Fig. 4C). The expression of rpfA gene was found to be upregulated in Group-III as compared to Group-II and the expression of rpfC gene remained same in all the groups whereas the gene expression of rpfE was found to be significantly downregulated in both Group-II and Group-III as compared to Group-I ( Fig. 4B,D,F). It was observed that at earlier time point (week 9) Group-II behaved like Group-I but with increase in time duration (week 13), Group-II behaved similar to Group-III which indicates that the duration of hyperglycemia plays a vital role in modulating the expression of different resuscitation and dormancy associated genes. These findings suggest that diabetes promotes resuscitation of latent bacilli through rpfB and rpfD and thus favours its activation.
Diabetes had no effect on treatment outcome. INH + RIF therapy had no effect on the bacillary loads in lungs of latently infected animals with or without diabetes. Although a significant decrease in log 10 CFU/mL in spleen was observed after therapy in both latent tuberculosis only treated with INH + RIF (Group-Ib) and latent tuberculosis with diabetes treated with INH + RIF (Group-IIb) as compared to latent tuberculosis only, untreated (Group-Ia) and latent tuberculosis with diabetes, untreated (Group-IIa) respectively (Fig. 5A,B). After 4 weeks of treatment, granulomas were observed in lung tissues of Group-Ia (Mean score-4.66) and Group-Ib animals (Mean score-1.33) (Fig. 6A,B). In Group-IIa inflammation (Mean score-4.33) was observed but no granulomas were seen whereas in Group-IIb, histiocytic collection was seen and granuloma formation was observed (Mean score-0.33) (Fig. 6C,D). However, the changes were non-significant between untreated and www.nature.com/scientificreports/ treated groups. Moreover, a significant decrease in the expression of hspX gene was observed in Group-Ib as compared to Group-Ia but no significant change was observed in the expression between Group-IIa and Group-IIb (Fig. 7A). The expression of rpfB and rpfD was found to be significantly downregulated in Group-IIb as compared to Group-IIa but no change was observed in its expression between Group-Ia and Ib (Fig. 7E). The expression of rpfE was found to be significantly downregulated in latent tuberculosis treated group as compared to latent tuberculosis untreated group and expression was significantly upregulated in latent tuberculosis with diabetes treated group as compared to latent tuberculosis with diabetes untreated group (Fig. 7F). However, no significant changes were observed in the expression of rpfA and rpfC between untreated and treated groups (Fig. 7B,D). The results indicate that after treatment, observations of the latent tuberculosis with diabetes group  www.nature.com/scientificreports/ (Group IIb) were not very different from only latent tuberculosis group. Thus, it was concluded that longer duration therapy i.e. more than 4 weeks is required to evaluate the treatment outcome.

Discussion
Diabetes acts as a predisposing risk factor for tuberculosis making it necessary to delineate its role in LTBI activation. This prompted us to develop an animal model of latent TB with diabetes so as to thoroughly investigate the underlying factors that may lead to the reactivation of latent tuberculosis under diabetic conditions. The conversion of LTBI to active TB was strongly supported by CFU counts which were found to be more than 10 4 . Moreover, a study by Dutta et al have shown similar results in which a model of latent tuberculosis was developed by immunization with rBCG and an increase in CFU counts was seen after treating the mice with anti-TNF-α mAb 18 . Similarly, increased CFU counts were seen after administration of aminoguanidine, a nitric oxide synthase inhibitor, resulting in immune suppression in latent tuberculosis mice model 19 . TNF-α neutralization has also been shown to reactivate chronic persistent tuberculosis in a low dose mice model of latent tuberculosis 20 . Abnormal regulation of inflammatory mechanisms in diabetes is also a characteristic feature that can cause exaggerated inflammatory responses in the lung, resulting in impaired lung function 21 . In the present study also, diabetes was found to be associated with compromised lung morphology with severe inflammation as indicated by lymphocytic infiltration and no granulomas. Our findings were supported by a study wherein granulomas became disorganised and inflammation was increased in the lungs of latently infected mice after treatment with anti-TNF-α mAb 18 . Another study on zebrafish infected with latent M. marinum also reported loss of granuloma www.nature.com/scientificreports/ integrity upon dexamethasone treatment 22 . Similar results were observed after blocking TNF-α in chronically infected C57BL/6 mice along with the dissolution of granulomatous tissues and increased lung inflammation 23 . Alpha crystallin is believed to sustain the tubercle bacilli during latent or dormant phase of infection, thus indicating HspX as a potential marker of latent tuberculosis infection 24,25 . The gene expression of hspX was downregulated in latently infected mice induced with diabetes as compared to mice infected with latent tuberculosis only. This finding further confirmed the active state of tubercle bacilli after diabetes development. Increased expression of HspX in the latent state of tubercle bacilli and its reversal to normal levels during the exponential growth has also been reported earlier 25 . Other studies also showed that the levels of anti-Acr IgG were lower in active TB as compared to latent TB thus supporting the present observation 26,27 . Similarly, a study by Muttucumaru et al strongly supported our findings and reported that alpha-crystallin was induced in the non-replicative state of M. tuberculosis as compared to active culture 28 .
Since TB reactivation has an inevitable link with resuscitation promoting factors, it was mandatory to study the expression of various rpf genes under diabetic conditions. The expression of rpfB and rpfD genes was significantly upregulated in latent tuberculosis and diabetes group in comparison to only latent tuberculosis group. These results strongly suggest that diabetes promotes resuscitation to latent tubercle bacilli and helps in its activation through resuscitation promoting factors B & D. Goldman et al have also shown the importance of Rpf in reactivation of chronic TB. RpfAB double knockout mutants exhibited a deficiency in reactivation in C57BL/6 mice upon immune suppression induced by nitric oxide synthase inhibitor (aminoguanidine) and by CD 4+ T-cell depletion 29 . A delay in reactivation of bacterial growth upon immune suppression in the mouse model of chronic tuberculosis after deletion of rpfB gene has been shown 30 . Rpf triple and double mutants were observed to be defective in their ability to re-grow after immunosuppression induced by administration of aminoguanidine and anti-TNF-α antibodies 31 . A significant differential attenuation in virulence has been observed in the above stated triple mutants in a mouse model of chronic TB 32 . Kapoor et al reported that the dormant M. tuberculosis within the granuloma were resuscitated when treated with anti-TNFα-mAb resulting in increased expression of rpf genes i.e. rpfA, rpfB and rpfC 33 . Recently an Rpf-interacting protein A (RipA) was identified which interacts explicitly with RpfB and RpfE and efficiently promotes hydrolysis of peptidoglycan during reactivation and hence is responsible for the release of bacteria from dormancy 34 .
Since diabetes is associated with poor tuberculosis treatment outcomes, the effect of diabetes on LTBI treatment was studied. Latent TB therapy (INH + RIF) did not affect the bacillary load between untreated and treated LTBI with diabetes group. The results were supported by a clinical study wherein tuberculosis patients with diabetes showed poor TB treatment outcomes 35 . A meta-analysis study showed that tuberculosis treatment failure was higher in tuberculosis patients with diabetes 36 . In the present study, no significant difference was observed in the bacillary load between latent tuberculosis only untreated and treated groups, which could be because of the short treatment period. A study by Zhang et al showed that the therapy for LTBI (Rifapentine + Pyrazinamide and Rifapentine + Isoniazid + Pyrazinamide) was ineffective for less than 6-8 weeks 37 . Other studies have also shown that it takes 10-12 weeks for the standard regimen RIF + INH + PZA to cure active TB mice where the bacterial burden is high and 20-24 weeks to prevent relapse 38,39 . Thus, longer duration therapy (> 4-5 weeks) is required to eliminate minimal bacillary load. The gene expression of hspX was upregulated and the expression of rpfB and rpfD was downregulated in latent tuberculosis with diabetes treated group as compared to latent tuberculosis with diabetes untreated group. These results may suggest that upon treatment, M. tuberculosis may revert to the latent state which could be the reason for relapse of tuberculosis. However, it needs further exploration as therapy was given for only four weeks. Apart from the short duration of therapy, another limitation of the present study is that a single high dose of streptozotocin was administered to the animals to induce diabetes which is more reflective of type 1 diabetes and not of type 2 diabetes. The reason for using a high dose of streptozotocin is that animals were immunized with BCG which is thought to play a role in protection against diabetes. Several studies have shown that multiple low doses of streptozotocin did not induce diabetes in mice previously immunized with BCG 40,41 .
In conclusion, diabetes leads to disruption of granuloma formation and ultimately leads to the activation of latent tuberculosis. The major factors involved in this activation or resuscitation process are the resuscitation promoting factors rpfB and rpfD which can be targeted to contain the tuberculosis infection and reduce the shared burden of tuberculosis and diabetes co-pathology.

Materials and methods
Animal procurement and ethics statement. All  www.nature.com/scientificreports/ as described by Ventura-Sobrevilla et al. 44 . A flow chart of the distribution of animals into different groups with experimental endpoints is given in Fig. 8.

Induction of immunosuppression in latently infected mice.
Group-II animals were induced with diabetes and Group-III animals were treated with dexamethasone to induce immunosuppression after six weeks of infection with M. tuberculosis H37Rv. Group-III animals were used as a positive control for active tuberculosis. Three doses of dexamethasone was administered to Group-III animals at a dose of 6 mg/kg body weight subcutaneously 2 days apart 46 .
Treatment therapy for latent tuberculosis. After the induction of diabetes, Group-I (latent tuberculosis only) and Group-II (latent tuberculosis with diabetes) animals were further subdivided into four subgroups i.e. Group-Ia: latent tuberculosis untreated, Group-Ib: latent tuberculosis treated with isoniazid (INH) and rifampicin (RIF), Group-IIa: latent tuberculosis with diabetes, untreated, Group-IIb: latent tuberculosis with diabetes, treated with INH + RIF. 10 animals from both Group-I and Group-II each were treated with isoniazid (INH) and rifampicin (RIF). At week 9 of diabetes induction, Group-Ib and Group-IIb animals were administered with INH (10 mg/kg body weight) and RIF (10 mg/kg body weight) orally through gavage, 5 days per week for 4 weeks. After 4 weeks animals (n = 8) were sacrificed from all groups, their lungs and spleens were harvested and processed for CFU enumeration, histopathology and gene analysis.
Histopathological studies. Aseptically removed lungs and spleen were transferred to 10% buffered formalin and further processed for paraffin embedding and sectioning 47 . Paraffin sections were deparaffinized and stained by standard haematoxylin and eosin (H&E) stain and acid-fast bacilli (AFB) stain. According to Valdez et al., a scoring system [0 = within normal limits (no change); 1 = minimal changes; 2 = mild changes; 3 = moderate changes; 4 = marked changes; 5 = very severe changes] involving examination of lungs for peribronchiolitis, perivasculitis and granuloma formation was used to give a total score for lungs from each mouse in all groups 48 .