Abstract:
Nile perch (Lates niloticus) is one of the most important fish species in the Lake Victoria
region due to its market potential and health benefits. Despite such potentials, reports on
detention of exported fillet due to Salmonella have been reported, yet not much studied. A
cross-sectional study design was employed to investigate the microbial safety of Nile
perch and its salted sun-dried products, as well as their contamination with Salmonella
spp. and Escherichia coli producing extended-spectrum beta-lactamases (ESBLs). The
specific objectives of the study were to: i) Establish the prevalence and diversity of
Salmonella serotypes in water and Nile perch from Lake Victoria, Tanzania, ii) Evaluate
genetic diversity, antimicrobial resistance and pathogenic potential of Salmonella serovars
isolated from water and Nile perch, iii) Determine the prevalence of ESBL-producing E.
coli in water and Nile perch from Lake Victoria, Tanzania, iv) Establish genomic
characteristics of ESBL-producing E. coli isolates from water and Nile perch; and, v)
Assess microbial quality of frozen Nile perch and salted sun-dried products sold in
domestic and regional markets.
Collected samples included Nile perch (from fishing grounds, landing sites, and domestic
fish markets), lake water, swabs from surfaces of facilities used for fish transportation and
salted sun-dried Nile perch products. Standard methods were used for analysis of bacteria,
moisture content (MC) and water activity (A w ) in samples. Antimicrobial resistance in
Salmonella spp. was determined by minimum inhibitory concentration (MIC) and for
ESBL-producing E. coli, standard disc diffusion method was used. ESBL-producing E.
coli were screened on MacConkey agar supplemented with 2 μg/mL cefotaxime and were
confirmed by polymerase chain reaction (PCR). Fourteen Salmonella and 11 ESBL-
producing E. coli isolates were selected for whole genome sequencing (WGS) using Illumina MiSeq. The genomic characterisation and phylogenetic relatedness analysis of
the isolates were established using WGS.
The study revealed 12 different Salmonella serovars, commonly S. enterica subsp.
salamae 42:r:- and S. waycross. The prevalence of S. enterica subsp. salamae 42:r:- was
10% (6/60) in Nile perch from fishing grounds, 1.7% (1/60) from landing sites and 1.7%
(1/60) from the markets. The magnitude of contamination in Nile perch from landing sites
was significantly lower compared to other sources (P<0.05). The prevalence of S.
waycross in the Nile perch from fishing grounds was 16.7% (10/60), 10% (6/60) at
landing sites and 13.3% (8/60) in market settings. Prevalence between sites were
comparable (P>0.05). Two of 12 ser. 42:r:- and six out of 30 S. waycross were resistant to
sulfamethoxazole, and one out of 12 ser. 42:r:- and six out of 30 S. waycross were resistant
to azithromycin. The results were not supported by the detection of gene aac(6')-laa
encoding for aminoglycoside resistance in the sequenced isolates. Resistance to
azithromycin could be associated with unknown chromosomal mutations in base pair in
regions of 16S RNA and 23S RNA detected in sequenced isolates. The WGS also revealed
ser. 42:r:- and ser. Fulica ST1208, ser. 42:r:- belonged to serogroup T. However, ser.
Fulica had an unidentified serogroup. In addition, four S. waycross ST2460, one S.
waycross ST3691, one S. wien ST2460 and S. wien ST3691 were reported. The S.
waycross serovars belonged to serogroup S, while S. wien had unidentified serogroup.
Moreover, S. waycross and S. wien had pathogenicity islands SPI-2 to SPI-5 with
associated virulent genes, but lacked in ser. 42:r:- and ser. Fulica. Plasmid replicon type
IncFII was only detected in S. waycross ST3691 and S. wien ST3691 and was not
associated with resistance or virulent genes. Furthermore, serovars of subsp. salamae had
unique esp and ompT genes for adaptation in aquatic environment, while S. waycross and
S. wien had a cluster of specific cit genes encoding citrate utilisation and oadAB for oxaloacetate decarboxylation in host cells. CSI phylogenetic analysis revealed a ser. 42:r:-
clonal relationship to ser. Fulica. Salmonella waycross were clonally related to each other,
while S. wien showed variations (Paper I and Manuscript II for SOBs I and II).
The overall prevalence of ESBL-producing E. coli in Nile perch from the Lake Victoria
was 4.4% (8/180). The isolates were resistant to sulphamethoxazole-trimethoprim (100%),
ampicillin/cloxacillin (100%), erythromycin 72.7% (8/11), tetracycline 90.9% (10/11) and
nalidixic acid 63.6% (7/11). The isolates carried resistance genes for sulphonamides (sul1
and sul2), trimethoprim (dfrA and dfrB), aminoglycosides (aac(3)-IId, strA and strB),
tetracycline (tet(B) and tet(D) and fluoroquinolones (qepA4). In addition, the isolates
harboured plasmid replicon types IncF, IncX, IncQ and Col and carried bla CTX-M-15 and
bla TEM-1B genes as well as resistance encoding genes. ESBL-producing E. coli isolates
formed three separate sequence type-phylogroup-serotype specific clades: C1, C2 and C3.
Clade C1 was composed of five isolates (maximum of 13 single nucleotide
polymorphisms [SNPs]) belonging to ST167, phylogroup A and serotype O9:H21. Two
C2 isolates (max. 11 SNPs) belonged to ST156, phylogroup B1 and serotype (O-
untypable) ONT:H28. Clade C3 was comprised of four isolates (max. 17 SNPs) of ST636,
phylogroup B2 and serotype O45:H7. The virulence gene gad encoding glutamate
decarboxylase was found in all isolates. In addition, C2 and C3 isolates harboured the
following virurence genes; iss encoding for increased serum survival, lpfA for long polar
fimbriae and nfaE encoding for diffuse adherence fimbrillar adhesin. The vat gene which
play role in vacuolating autotransporter toxin was found only in C3, and was responsible
for pathogenicity. A CSI phylogenetic analysis revealed ST167 were clonally related to
corresponding public genome of E. coli strains obtained from humans, animals and the
environment. The same scenario was reported for ST636 and ST156 of E. coli against their corresponding sequence types. In each sequence type, isolates showed the samev
genotypic resistance and virulence profiles. The present study highlights the occurrence of genotypic resistance and virulence profiles. The present study highlights the occurrence of
low virulence, multidrug resistant ESBL-producing E. coli in a highly commercialised fish
product obtained from Lake Victoria (Manuscript III for SOBs III and IV).
The study also revealed that salted sun-dried Nile perch products were not contaminated
with Salmonella spp. and ESBL-producing E. coli. However, total viable counts (TVCs)
of 4.5 log colony forming units (cfu)/g in fish heads with MCs of 38.0% and A w of 0.682
were reported in products sampled during the rainy season and were significantly higher
(P<0.05) than the corresponding samples collected during the dry season. The differences
attributed by high humidity and rainy condition which lowerd the drying temperature. Fish
chests collected during the rainy season had TVCs of 3.3 log cfu/g, MCs of 27.6% and A w
of 0.659, with no significant difference (P>0.05) to the values of samples dried during the
dry season. Fish belly flaps that were sampled during the rainy season had TVCs of 3.3
log cfu/g at 26.4% MCs and 0.669 A w , which were comparable (P>0.05) to those
collected during the dry season. Bacteria identified from TVCs included Staphylococcus
spp., Enterobacter spp., and Psychrobacter spp. (Paper IV SOB V).
The current study detected uncommon Salmonella serovars rarely associated with human
diseases. The serovars are apparently not of public health importance because they are less
virulent compared to other serovars frequently isolated from humans and animals. The
reported serovars could be normal flora in Nile perch fish. The study also revealed the low
prevalence of multidrug resistance ESBL-producing E. coli harbouring plasmids carrying
β-lactamases and antimicrobial resistance genes. The plasmids and resistance genes
reported in ESBL-producing E. coli showed no evidence of horizontal gene transfer
(HGT) between the isolates. The salted sun-dried Nile perch products were safe for human
consumption as the microbial parameters were within the acceptable limit set by the Tanzanian standards. However, high TVCs and identified contaminated bacteria in dried
products highlight the need to implement hygienic procedures during processing to ensure
the improved quality and safety of the products for consumers.