An Improved Fluorescent Amplified Fragment Length Polymorphism Method for Typing Mycobacterium tuberculosis

We have evaluated fluorescent amplified fragment length polymorphism(FAFLP) fingerprinting as a complementary technique for genotypingMycobacterium tuberculosis, which may aid in the elucidationof the transmission dynamics of tuberculosis. Earlier FAFLPstudies (1, 2, 3, 5) broadly employed EcoRI and MseI restrictionenzymes, which are known to have a low cleavage frequency forGC genomes of >65 mol% (4). By contrast, BamHI and MspI restrictionendonucleases were used in this study because they have a highercleavage frequency (as judged by in silico calculations) forthe M. tuberculosis genome and do not show polymorphisms withinIS6110/IS986.

Study 1.

Top

Study 1.

FAFLP was performed on 39 IS6110 restriction fragment lengthpolymorphism (RFLP) genotyped strains. The FAFLP profiles ofthese isolates generated an average of 74 amplified fragmentsbetween 50 bp and 400 bp, and 24 of these bands were polymorphic.Cluster analysis differentiated between two main groups. Thefirst cluster consisted of seven strains with multiple copiesof the IS6110 element, and they lacked the 118-bp and 243-bpfragments but shared 201-bp and 286-bp polymorphic bands. Thesecond cluster was comprised of strains with only a single copyof the IS6110 elements, which were poorly differentiated byRFLP, and all possessed the 200-bp and 286-bp polymorphism bands.These single-copy strains were further grouped into four subgroups,differentiated by 229-bp, 243-bp, 288-bp, 316-bp, and 372-bppolymorphic fragments.

Study 2.

Top

Study 2.

Study 2 consisted of 44 M. tuberculosis strains, all of whichwere previously characterized by IS6110 RFLP. Analysis of theseFAFLP profiles produced two main clusters (Fig. 1). The bandingpatterns of the 10 isolates belonging to the first cluster correlatedwith those of the single IS6110 copy isolates because of thepresence of the 200-bp and 286-bp polymorphisms. However, thesestrains lacked the 243-bp polymorphism, which was prominentin the single IS6110 copy strains of study 1. The second clusterconsisted of multiple IS6110 strains, and they were dividedinto three subgroups, as shown in Fig. 1.

View larger version (65K):
[in this window]
[in a new window]
FIG. 1. Comparison of the FAFLP dendrogam and the IS6110 RFLP patterns of all 44 isolates. Clustering employed the unpaired-group method with arithmetic averages by use of a multivariate statistical package.

Epidemiological and medical records of the source patients ofthese study 2 isolates revealed that the patients were residentsof London, United Kingdom, and were of either Somalian or Caucasianethnic origin. The IS6110 RFLP data that were obtained for comparisonpurposes (6) show significant genetic heterogeneity within strainsof both Caucasian and Somalian origins. A direct comparisonof the FAFLP and the IS6110 RFLP data for the 42 strains revealedthat only isolates 99/1154 and 00/10121 of Caucasian originclustered at 100% genetic similarity for the two typing techniques.However, visual inspections of the IS6110 profiles indicatethat samples of the same FAFLP cluster have significantly similarthough not identical RFLP banding patterns. This suggests asignificant degree of congruence between the two typing techniques(Fig. 1).

Is this FAFLP method dominated by the IS6110/IS986 genetic marker?

Top

Is this faflp method...

As there was some shared congruence between FAFLP and IS6110RFLP patterns, an obvious question that needed addressing waswhether the FAFLP using BamH1 and MspI was reliant on IS6110,since these restriction enzymes have restriction sites withinthe IS6110 element.

Although sequences flanking the IS6110 elements can be detectedas FAFLP, experiments utilizing selective nucleotides complementaryto sequences immediately around BamHI and MspI sites withinthe IS6110 element detected none of the observed polymorphisms(Fig. 2). This confirmed that BamHI plus T and MspI plus 0 FAFLPwas not influenced by the IS6110 element and the flanking sequences,which has important implications for the complementation ofthis approach to IS6110 RFLP.

View larger version (21K):
[in this window]
[in a new window]
FIG. 2. FAFLP profiles from BamHI plus T (B+T)/MspI plus 0 (M+0) and BamHI plus T/MspI plus GGCG (M+GGCG) for M. tuberculosis isolate 5079, demonstrating that the observed polymorphisms (dashed boxes) are not generated by the IS6110 element. The boxed numbers under the peaks are fragment sizes in base pairs, and the vertical scales at right are fragment intensities.

Conclusions.

Top

Conclusions.

The differentiation of M. tuberculosis isolates containing onlya single copy of IS6110 confirms the potential of FAFLP as ahighly discriminatory fingerprinting technique which is complementaryto IS6110 RFLP. Furthermore, BamHI and MspI FAFLP requires onlytwo primers, compared to eight primers for EcoRI and MseI multiplexingdone by Goulding et al. (3), which has resource implications.Since only one data set was generated for each isolate usingBamHI plus T and MspI plus 0 FAFLP, our approach involves lessdata analysis and as such we believe it will significantly enhancesubsequent interlaboratory data comparison.

In conclusion, the results presented in this study stronglysuggest that in conjunction with IS6110 RFLP, FAFLP using BamHIplus T and MspI plus 0 could significantly improve strain identificationof M. tuberculosis and in turn the control of tuberculosis.

ACKNOWLEDGMENTS

We are very grateful to Leighton Pritchard for help with AFLPin silico calculations. Y.K. thanks the Overseas Research StudentsAwards Scheme (ORS) for his studentship, and R.G. is indebtedto the United Kingdom BBSRC for financial support.

REFERENCES

Top