Uganda ranks 16th among the world's 22 countries with the highest tuberculosis burden in the world [1]. The country had more than 132,000 TB cases in 2007, with an estimated incidence rate of 330 per 100,000 people with greater Mbarara contributing about 26% of all TB cases. This region has been heavily affected by the TB/HIV epidemic. In 2005, the case notification for Mbarara was 175/100,000 people compared to the 147/100,000 people for Uganda [2]. In 2008 the TB/HIV co-infection rate for Mbarara was 65% [3]

To date, there are very limited data available pertaining strains circulating in Mbarara yet evidence indicates that M. tuberculosis' ability to spread varies from strain to strain and those different strains have different geographical and/or host specificities [4, 5]. The presence of Human immunodeficiency virus (HIV) has caused an increase in Mycobacterium tuberculosis complex (MTC) infection [6] and rapid progression of the infection [7] and is also known to increase MTC transmission rates at the community level, further threatening the health and survival of HIV sero-negative individuals as well [8].

Since the discovery of DNA polymorphisms in M. tuberculosis, molecular typing of strains has become an invaluable tool for the study of epidemiology of TB. Some of the applications include; predicting transmission rates and identifying dominant strains associated with outbreak [9] severe disease [10] and drug resistance.

Comparative-genomics approaches greatly enhanced our understanding of the mechanisms of insertion and deletion of DNA and the resulting distribution of variable regions around the genomes of tubercle bacilli [11–13]. There are 20 variable regions of which 14 regions of difference (RD1 to RD14) were found to be absent from Bacillus Calmette-Guérin (BCG) Pasteur relative to M. tuberculosis H37Rv [13–15]. Six regions, H37Rv-related deletions (RvD1 to RvD5 and M. tuberculosis specific deletion 1 (TbD1) are absent from the M. tuberculosis H37Rv genome relative to other members of the M. tuberculosis complex. Based on the presence or absence of the TbD1 region, M. tuberculosis strains can be divided into "ancestral" and "modern" types. The Beijing, Haarlem, and African strains responsible for major epidemics are modern types [11, 16].

Clustered regularly interspaced short palindromic repeats (CRISPRs) are repetitive structures in bacteria and archaea composed of exact repeat sequences 24 to 48 bases long separated by unique spacers of similar length [17, 18]. The CRISPR sequences appear to be among the most rapidly evolving elements in the genome, to the point that closely related species and strains, sometimes more than 99% identical at the DNA level, differ in their CRISPR composition [19, 20]. Direct Repeat loci (DR) are members of the CRISPR [21]. The Direct Repeat locus consists of alternating identical DRs and variable spacers can be assessed using the spoligotyping fingerprinting methodology. Variability in the direct repeat locus of M. tuberculosis [22] most likely occurs by one of three mechanisms--homologous recombination between neighboring or distant direct variable repeats, IS-mediated transposition, and DNA replication slippage [22].

Drug resistance among mycobacteria is a threat to the treatment of tuberculosis globally and threatens to reverse the gains made so far in the fight against tuberculosis. Several competing technologies have been proposed for rapid detection of drug resistant tuberculosis. Some commercial assays are currently available including INNO-LiPA Rif.TB (Innogenetics N.V, Ghent, Belgium) and GenoType^® MTBDR (HAIN Lifesciences GmbH, Nehren, Germany) [23]. The new version of the latter assay (GenoType^® MTBDRplus), targeting the rpoB gene associated with the resistance to rifampicin (RIF) and both genes (katG and inhA) commonly associated with the resistance to isoniazid (INH) has been evaluated mainly on cultures and clinical specimens in various low incidence settings, demonstrating excellent specificity and good concordance with phenotypic drug susceptibility test (DST) results [24, 25].

The current study was carried out in the Mbarara region of South-Western Uganda between May 2007 and April 2008 in order to know the clones of M. tuberculosis and their drug resistance patterns.

Deletion analysis using 16SrRNAS, RD4, RD 9, RD12 and TbD1 revealed that all the strains investigated were M. tuberculosis. This finding is comparable to that of the study by Byarugaba et al 2009 [34] which found no M. bovis among the 75 MTC isolates from the mainly cattle rearing communities of greater Mbarara. This could be a result of improved animal husbandry in Mbarara, or due to improved practices of the people such as boiling of milk before consumption. On the other hand, it may be due to the fact all of our samples were pulmonary sputum, which might have limited the chances of culturing M. bovis.

The diversity of the M. tuberculosis found in the present study (63.2%) is high compared to 16.6% described from a study in central Uganda [35], which may indicate geographical differences in the epidemiology of TB in Uganda. Another hypothesis could be because of differences in the population density. Central Uganda has the major cities of the country and therefore has a high transmission rate, hence limited diversity. Mbarara is a big district, sparsely populated because of the main economic activity: cattle keeping. In such an area, it is possible to have less transmission and a highly diverse bacterial population. This is not easily determined using spoligotyping alone and would therefore require more discriminatory techniques such as MIRU and restriction fragment length polymorphism (RFLP) to confirm this.

In this study 69 (87%) of the spoligopatterns could not be typed on the basis of the existing SpolBD4 database. This indicates the current absence of knowledge on the genetic diversity of M. tuberculosis strains in from this region. This therefore calls for more clinical epidemiological studies in different regions of not only Uganda but Africa at large so as to clearly understand the genetic diversity of the TB epidemic on the continent.

Majority (59.2%) of the strains in our sample were of the Uganda genotype, a finding which is in keeping with those of earlier studies in Uganda [36–38] as opposed to findings from the surrounding East African countries[39, 40]. For example a study in Kenya [39] found only eight (11%) of 73 isolates to be of the Uganda family while in northern Tanzania [40] only four (3%) of 130 strains were T2-Uganda. Collectively these findings are in agreement with those of other studies showing a tendency for local genotypes to form a greater proportion of the circulating strains in different parts of Africa [41–44]. These results indicate that each of the major lineages of M. tuberculosis have evolved to most efficiently transmit within an original human population. These results are further in agreement with other findings where it was noted that different strains of M. tuberculosis have adapted to specific human populations, and that such local strains are more likely to transmit compared to others [4, 45].

The relative frequencies of major M. tuberculosis spoligotype families were in range with the overall frequencies in Uganda and other East African countries [35, 36, 39, 40, 46]. The largely predominant Uganda family identified in more than 59% of the strains in our study is ubiquitous in this country. Both its high degree of dissemination and its preponderance among the new (shared as well as orphan) patterns are manifestations of the current adaptive evolution of the Uganda genotype in this setting. Other significant spoligotypes in our study were LAM (7.2%), CAS (8%), EAI (5.6%), Cameroon (4%) and Ghana (2.4%). A study by Asiimwe et al 2008 from Kampala showed proportions of CAS1-Kili (3.5%), LAM9 (2.6%), CAS1-Delhi (2.6%), LAM3/S (1.7%), CAS1 (1.7%), and LAM11-ZWE (1.5%). In comparison to other studies in the region, the CAS, LAM and EAI families were reported at 37%, 22% and 17% respectively of a total of 147 isolates in a study in Dar es Salaam, Tanzania [46]; while in northern Tanzania, the most predominant families were CAS-Kili (30%), LAM11-ZWE (14.6%), EAI (6.2%), Beijing (5.4%), and CAS1-Delhi, T1 and LAM9 at 3.8% [40]. In Kenya, on the other hand, 35.6% of 73 isolates were of the CAS family, while 11% were LAM [39]. These studies show more success of the CAS, LAM and EAI families in the neighboring East African countries, while in Central and Western Uganda, the Uganda family of strains predominates.

We found resistance to isoniazid and rifampicin to be 3.2% and 5.6% respectively, while MDR was 1.6% (2/125). Our result show differences compared with findings that were obtained in the last National anti-tuberculosis drug resistance survey in Uganda of 1996-97 that indicated a primary resistance to isoniazid of 6.7%, that to rifampicin at 0.8%, and MDR of 0.5% [47]. More recently, a study in peri-urban Kampala showed resistance to isoniazid of 8.1%, rifampicin resistance of 4.4% and MDR was found to be 4.4% [35]. These differences may probably be due to sampling strategy employed in each study and the numbers involved. While the National survey randomly sampled districts in Uganda, the peri-urban study in Kampala looked at a single division known to be the second most TB burdened in the city, while the current study sampled patients from various villages of a rural district in western Uganda.

Studies from neighbouring East African countries show varied results. In a study in Rwanda, resistance to isoniazid was found at 6.2%, that to rifampicin was 3.9% with all rifampicin resistant isolates being multidrug-resistant [48]. In northern Tanzania, on the other hand, a study of 111 isolates showed that 9.9% were resistant to isoniazid, 2.7% to rifampicin, while MDR was 2.7%[40]. Generally, the drug resistance rates in the current study are fairly within the range of those found in previous studies both in-country and around the region. However there is evidence of an increase in the MDR rate in Uganda in the last two studies compared to the first National survey albeit on smaller samples. Although a number of patients were not tested for HIV and could be dually infected, two thirds of those tested for were infected by both HIV and TB, a common trend in sub Saharan Africa [49].

Despite the high prevalence of the Uganda family there was no significant association with anti-TB drug resistance (P-value = 0.076) This finding in agreement with the finding by Asiimwe et al [35] who found that there was no significant difference in the resistance pattern of the predominant T 2 Uganda genotype versus the Non T family. This therefore implies that T2 Uganda might not be the driving force of anti-tuberculosis drug resistance in this community Mutations in codon 315 of the katG gene were the only ones found in our study. These mutations are found in the vast majority of isoniazid-resistant isolates [50, 51]. However, the frequency of codon 315 mutations in isoniazid-resistant isolates in other populations has been reported to range from 35% to 97% [52]. Mutations in codon 315 do not significantly decrease the peroxidase activity of the katG gene product, but do decrease its ability to activate isoniazid[53]. These features allow the mutants to maintain the peroxidase activity required for virulence, and to resist killing by isoniazid. Such isolates often display resistance to only lower levels of isoniazid, and resistance to higher levels appears to correlate with loss of catalase activity or acquisition of mutations in multiple genes implicated in isoniazid resistance, e.g., inhA or ahpC[54, 55].

The most frequent mutation in the rpoB codons in our study was 516 (43%). As opposed to mutation 531 which is the most commonly observed mutation in rifampicin-resistant isolates in many parts of the world, e.g., in Brazil (54%) [56], the USA (35%) [57], India (38.7%) [58], Germany (65%) [59] and Australia (52%) [60].

Differences in the frequencies of mutations in the 516, 526 and 531 codons of the rpoB gene among isolates of the Beijing, Haarlem and LAM families may primarily reflect differences in mutational frequencies or the relative fitness of the mutations in the strain families, as opposed to a possible sampling bias caused by extensive transmission of individual MDR strains.

Overall, the data indicate that the frequencies of individual mutations in the genes associated with rifampicin and isoniazid resistance vary among isolates belonging to different genotype families. Such variation may influence the performance of molecular diagnostic tests designed to detect mutations associated with drug resistance in M. tuberculosis isolates. This emphasises the importance of validating the performance of a diagnostic test in the population being tested. The biological significance of the predominance of certain mutations in particular genotype families remains to be determined.

This study provides an insight into the M. tuberculosis strains circulating in the rural district of Mbarara, South Western Uganda. We have shown that the majority of the Mycobacteria in Mbarara are modern M. tuberculosis with a wide diversity of spoligotypes and a predominance of the Uganda genotype. Additionally, strain types in our sample were not associated with drug resistance.