In E. histolytica
there have been no reports comparing the genotypes of amebae identified in stool with those in the liver abscess pus of the same patients. However, in Leishmania
, it was demonstrated that isolates from mucosal and cutaneous lesions of the same patients were genetically distinct and those authors suggested that a subpopulation of the parasite had reached the mucosal tissue from the original cutaneous lesion site 
. We therefore were interested in investigating whether a similar phenomenon could be detected when comparing invasive and intestinal E. histolytica
. In order to detect the genetic differences we used 6 highly polymorphic loci in E. histolytica
. These loci are not likely to be directly involved in virulence, but they may be acting as surrogate markers and be physically linked to loci that are having a direct effect on the outcome of infection. Using this genotyping method, we have previously shown that the range of E. histolytica
genotypes in liver abscess samples is different from that detected in intestinal samples in the Bangladeshi population 
. One limitation of our previous genotype comparisons was that they were conducted on samples from 3 separate clinical groups: asymptomatic, diarrhea/dysentery, and liver abscess; the first two groups provided stool samples, while the liver abscess patients gave only pus samples.
In 2004, while developing the genotyping methodology, we received two DNA samples from Prof. Egbert Tannich (Bernhard Nocht Institute, Hamburg) – one extracted from liver abscess pus from a woman who had never travelled outside of Europe and the other extracted from a stool sample provided by her husband. Typing showed the gel electrophoresis patterns for PCR products of the two samples to be distinct. This result inspired the present study, in which 18 pairs of E. histolytica
DNA samples derived from stool and liver abscess pus of the same patient were collected. The samples were analysed by PCR genotype and sequencing to determine whether differences could be detected between amebae that had metastasized to liver tissue and those that remained in the intestine. The results we found were dramatic: the intestinal amebae were distinct from those in the liver abscess in all patients investigated, from Bangladesh, Italy and the USA. This observation based on sequencing of tRNA-linked STR loci was also supported by nested PCR of the widely studied polymorphic serine-rich E. histolytica
protein (SREHP) gene. This gene encodes an immunodominant surface antigen containing tandem repeats of related dodeca- and octa-peptides 
. Analysis of SREHP in three of the sample pairs from Bangladesh revealed that the intestinal and liver abscess amebae were indeed different from each other at this locus also (data not shown).
We believe that there are three possible explanations for these findings. Firstly, ALA may occur many months after the initial infection. Due to the time elapsed the original intestinal infection may have been cleared 
and the amebae in our stool samples could be the result of a second infection with a distinct genotype. Genetic differences would then be expected due to the extensive diversity that exists even in geographically localized populations 
. While a second infection is a reasonable possibility in an endemic country like Bangladesh, and could indeed be true in some of our cases, it seems extremely unlikely to be a realistic explanation for the cases from Italy and the USA where E. histolytica
infection is rare.
Secondly, the original intestinal infection may have contained multiple genotypes, of which only one migrated to the liver via the bloodstream to cause the liver abscess (i.e. a “genetic bottleneck”). There are two different situations that could lead to multiple genotypes. The first is a ‘double infection’ scenario, where 2 or more unrelated genotypes are present, either through separate infections or exposure to a mixed source. The second scenario is a ‘heterogeneous infection’ of 2 or more related genotypes, most likely recently derived from a single source. The latter is more in line with what we see in the USA and Italian cases, where the stool and ALA pus genotypes were very similar - matching at 5 out of 6 loci sequenced. It would be unexpected to find infection with two such closely related genotypes by chance, given the diversity that exists in nature 
. In the Bangladeshi cases paired samples often differed at more loci than they shared, which is more in line with the double infection scenario. In both scenarios, however, only a minor genotype must have reached the liver.
Finally, there may be DNA reorganisation or recombination events taking place when the amebae migrate from the intestine to the liver. The STR loci at which the differences are being detected exist in multiple copies within a larger tandem array 
. Such chromosomal structures are generally less stable than single-copy sequences 
. The increase/decrease in STR number observed suggests a recombination event and tRNA genes are known to be ‘hotspots’ for recombination 
. Although only one locus was detected as undergoing change in the Italian and USA paired samples, it should be remembered that only 6 unlinked STR loci were studied and there are 25 distinct tRNA gene arrays in the E. histolytica
genome, each of which can have from 1 to 5 STRs. Even though only one locus was observed to change, recombination may well be much more extensive and widespread in the genome. It is also likely that in some future sample pairs no differences will be detected using these loci but that changes are nevertheless occurring elsewhere in the genome.
Interpreting the observations of ‘liver abscess’ sequences in stool DNA samples and vice versa is not straightforward. They could indicate the presence of a heterogeneous infection and the difference would then be the result of a genetic bottleneck through which a minor population of cells passed that happened to have indels involving STRs or single bases. It then becomes a problem to explain why in every case only a minor variant ended up in the liver, while the major genotype remained in the intestine. Alternatively, the observations could be the result of expansion of a variant array unit to replace the predominant version present in the intestinal cell population, implying radical recombination/reorganization of the array. Distinguishing between the two alternatives depends on whether all the cells in the intestinal infection would produce the same sequence trace or whether the relative proportions of the two sequence variants differs among cells. It is impossible to know which is the case.
Obtaining material for further studies of this type is not straightforward. Reports on the proportion of amebic liver abscess patients who harbor concomitant intestinal E. histolytica
infections vary, with values ranging from less than 10% 
to as high as 70% of patients 
. Additionally, in most countries abscess drainage is not performed unless imminent rupture is anticipated, restricting the availability of ALA pus for study although to the benefit of the patient. However, in a study by Haque et al 
, lectin antigens were detected in the serum of 32 out of 98 ALA patients in Bangladesh and, in a more recent study by Parija and Khairnar 
, E. histolytica
DNA was detected in the urine of 21 out of 53 ALA patients in India. These indicate that it might be possible to amplify E. histolytica
DNA from the blood and/or urine of the ALA patients, removing the need for pus samples. It would then be possible to investigate intestinal and invasive ameba genotypes in many more ALA patients.
In order to determine if the differences between intestinal and liver isolates of E. histolytica
in the same patient represent a genetic bottleneck effect or mutation will require additional experimentation. The ability to produce intestinal infections in mice using cloned isolates of E. histolytica 
may allow the testing of our hypotheses. Nevertheless, our observations to date give an insight into why in our previous work the ALA genotypes differed from those in the intestine 
and reinforce the likelihood that the ameba genotype is a significant player in the outcome of infection with E. histolytica.