Antibiotic sensitivity and sodium chloride susceptibility patterns of Flavobacterium columnare isolated from clinical columnaris in cultured Clarias gariepinus

Antimicrobial resistance is a global concern, especially with the backdrop of the development of possible environmental and public health hazards. Flavobacterium columnare has been observed to be multi-drug resistant but highly susceptible to sodium chloride, hence the need for susceptibility profiling. Therapeutic efficacy of commonly used antibiotics and sodium chloride was tested in-vitro against four isolates of F. columnare using the disc diffusion and the pour plate methods, respectively. Comparative in-vivo testing was performed on experimentally infected Clarias gariepinus juveniles in 11 groups of 15 fish each. Nine groups were treated with 1.0, 2.0 and 3.0% NaCl for 5, 10 and 30 min each. A tenth group was exposed to 25 mg L -1 ciprofloxacin for 1 h by immersion, while the eleventh group was not treated (positive control). Growth inhibition was observed to be highest with ciprofloxacin followed by ofloxacin and tetracycline in that order, and also in all NaCl concentrations. There was no mortality in the infected fish groups treated with 25 mg L -1 ciprofloxacin, 1% NaCl, and 2 to 3% NaCl for 5 and 10 min only. However, 46.7 ± 9.4 % mortality observed in 3 % NaCl treatment for 30 min was significantly higher than the 23.3 ± 4.6% observed in 2 % NaCl for 30 min, but not significantly different from the positive control with 36.7±4.7 % mortality. Since short duration sodium chloride bath was found to be effective in the control of columnaris disease, 1% salt disinfection of fry and fingerling stock for 30 min could be incorporated into routine management in catfish hatchery, without the fear of environmental or public health hazards.


INTRODUCTION
Flavobacterium columnare has been observed by various researchers (Bernardet and Grimont 1989;Hawke and Thune, 1992) to have developed resistance to many antibiotics, the basis for which its isolation is done on medium made selective with the incorporation of antibiotics to which it is resistant such as neomycin, *Corresponding author.E-mail: gbolatomi@yahoo.com.Tel: 2348037978139.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License polymyxin B, and tobramycin (Hawke and Thune, 1992;Starliper, 2003).Therefore, where it is desirable to use antibiotics in the treatment of a disease, conducting antibiogram then becomes the rule rather than the exception.Blind use of antibiotics is not only wasteful but also constitutes potential environmental and public health hazard since 75% of the antibiotics still finds its way back to the environment and can further exacerbate the already worsening problems of antibiotic-resistant pathogens.
Common salt (NaCl) often referred to as the 'aspirin of aquaculture' is recommended in bath treatment of fish against some fish pathogens (Bernardet, 1989;Swann and Fitzgerald, 1991;Suomalainen et al., 2005;Noga, 2010).However, the use of NaCl in freshwater, stenohaline fish species will require standardization to ensure effective elimination of pathogens without serious adverse effects on the fish.This study evaluates antibiotic sensitivity profile of isolated F. columnare, and validates the efficacy of common salt as a topical disinfectant for the control of F. columnare infection.

Antibiotics sensitivity tests
Antibiotics sensitivity tests were carried out on four isolates of Flavobacterium columnare cultured from different clinical cases of columnaris disease.The isolates have been characterized in an earlier study, using species-specific primers (Oladosu et al., 2018).Colonies of 24 h pure culture of the isolates were inoculated into an aliquot of 5 mL cytophaga broth and incubated for 24 h at 25°C.Using a 10 mL automatic pipette and sterile pipette tips, 0.1 mL from each broth culture was spread onto different, newly prepared cytophaga agar plates, and stripped all over the plates to wet the entire surface.The plate for each isolate was properly identified using indelible marker.One antibiotic disc (for Gram-negative organisms) was then dispensed onto the surface of the plates, using sterile pointed forceps to ensure effective contact of the discs with the medium.The plates were then incubated at 25°C for 24 h.The diameter of the visible zone of growth inhibition around the various discs was then measured to determine whether the isolate was sensitive or resistant to which antibiotics.The antibiotics tested include gentamycin, ofloxacin, amoxicillin, ciprofloxacin, tetracycline, augmentin, chloramphenicol, nitrofurantoin, perfloxacin, and cotrimoxazole.

Salinity assay
Analytical grade of sodium chloride (Nacl; BDH -Anala R ) was weighed and added to cytophaga broth prior to sterilization by autoclave, to achieve different salinities of 0.5, 1, 2 and 3%.Ten milliliter aliquot of broth with the different salinities was dispensed into 4 bijou bottles for each salinity, and then autoclave at 121°C for 15 min, at 1 atm.Inoculum from each of the four F columnare isolates used for the antibiotic sensitivity test was introduced into one bijou bottle each for the different salt concentrations, and incubated with moderate agitation by shaking (150 rpm).
One hundred microliter (100 µL or 0.1 ml) of the inoculated broth was taken from each isolate in the different salinities and dispensed into a sterile petri-dish, to which sterilized cytophaga agar was poured and mixed in a plate-pour method.This was done at different duration of incubation; 5, 10, and 30 min for the different salinities.All the inoculated plates were then incubated at 25 o C, and growth examined in 24 and 48 h.

Antibiotic bath and salt disinfection trials
Apparently healthy 180 juveniles (9 weeks old) of Clarias gariepinus procured from a commercial fish farm in Ibadan were acclimatized at the rate of 1 fish per 3 L in 30 L water contained in twelve 50 L plastic tanks, for 2 weeks.They were earlier disinfected in 250 ppm formalin for 1 h, and then randomly examined and observed to be free from ectoparasitic infestation, lesions or clinical signs of any disease.Following acclimatization for 2 weeks, they were redistributed in a population of 30 fish in 30 L of water, at a density 3 times the normal for flow-through culture (300 m -3 ), and kept in static culture in 6 plastic tanks for 24 h to induce stress.
Thereafter, 81 ml of cytophaga broth culture estimated by McFarland 0.5 standard opacity tube to contain 3.3 x 10 7 CFU of F columnare was introduced into each of the 6 tanks with 30 juvenile C gariepinus, containing 30 L of water, to achieve 1.1 x 10 6 CFU/L.The fish in 5 tanks were exposed to the pathogen for one hour, while those in the 6 th tank were not infected (positive control).Fish in all tanks were later re-distributed at 15 fish per tank in 30 L of water in duplicate, and kept in static culture for 48 h before the daily water renewal was restored.Fish were thereafter subjected to treatments by topical bath with 25 mg L -1 of ciprofloxacin for 1 h, topical disinfection with sodium chloride at 1, 2 and 3% concentration for 5, 10, and 30 min duration each.The positive control group was not infected while the negative control group was infected but not treated.Two dead fish samples were collected where mortality was observed and inoculum from the skin, the gills and the dorsal kidney were cultured for possible re-isolation of F. columnare.

RESULTS
The antibiotic sensitivity pattern observed in this study revealed that all four isolates of F.columnare tested were resistant to gentamycin, cotrimoxazole, amoxycilin, augmentin, pefloxacin, ceftriazone, and nitrofurantoin as shown in Table 1.The zone of inhibition to the growth of F. columnare observed for ciprofloxacin ranged between 15 to 26 mm with an average of 21.25 mm.This was higher than what was observed for ofloxacin with a range of inhibition zone between 7 and 24 mm, and an average of 13.75 mm.The same trend was observed for tetracycline with a range of inhibition zone of zero to 18 mm and an average of 10.25 mm.
As shown in Table 2, the pattern of susceptibility of F. columnare to varied concentrations of sodium chloride indicated that all four isolates tested were resistant to salinity of 0 and 0.5% at all durations of exposure tested.Two isolates were resistant, and the other two susceptible to 1% salinity at 5 and 10 min exposure time, while all four isolates were susceptible to the same salinity at 30 min exposure period.Similarly, all four isolates were susceptible to 2 and 3% salinity at all duration of exposure tested.
As shown in Table 3, no mortality was recorded in infected fish group treated with ciprofloxacin, all duration of exposure (5,10, and 30 mins) for 1% salt bath, as well  as 5 and 10 min duration of exposure for 2 and 3% salt bath.However, mortality was observed even before the expiration of the 30 min exposure period for 3% salt concentration for 30 min exposure.The mortality rate of 46.6% recorded in infected group treated with 3% salt bath for 30 min was significantly higher (P < 0.05) than the 23.3% mortality rate recorded for the group treated with 2% salt bath for 30 min.No significant difference (P < 0.05) was however observed between these two groups and those that were infected but not treated, where 36.66 % mortality was observed.Moreover, all water quality parameters tested were observed to be within the acceptable limits, except ammonia levels which though higher than 0.05 mg L -1

Isolates
recommended, can only be toxic when present in the unionized form.

DISCUSSION
The in-vitro study on bacterial sensitivity to antibiotics and susceptibility to sodium chloride showed that F. columnare was susceptible to ciprofloxacin and sodium chloride concentration of 1% and above.The antibiotic bath and salt disinfection trials reflected the results obtained for the in-vitro susceptibility test also, as fish treated with 25 mg L -1 of ciprofloxacin and salt concentration of 1% and above were cured of columnaris disease.
The present study demonstrated the efficacy of ciprofloxacin against the four F. columnare isolates used, which were observed to be resistant to many frequently used antimicrobials.The bath treatment of infected fish used in this study was also observed to be very effective as no mortality was recorded in fish group treated with the antibiotic as against the 36.6%mortality recorded for the group that was infected but not treated (control).The efficacy of the ciprofloxacin bath treatment lends credence to the observed rapid uptake and tissue distribution of the drug in a pharmacokinetic study with C. gariepinus using 25 and 50 mg L -1 ciprofloxacin concentration (Oladele et al., 2011).
F. columnare had been reported to be susceptible to salt (NaCl) in-vitro, at various concentrations including concentrations above 0.1% (Suomalainen et al., 2005), 0.5% (Shamsudin and Plumb 1996) and 1% (Bernardet 1989).In-vitro studies conducted by Bernardet (1989) revealed that the growth of F columnare was inhibited at 10 ppt (1%) but not at 5 ppt (0.5%), while Suomalainen et al. (2005) observed that the use of high concentrations of salt and low pH can be a viable treatment option for columnaris disease.These observations formed the basis for the suggestion that salt bath may be an effective control measure against columnaris disease.However, the observations of Suomalainen et al. (2005) seem to prove otherwise.He reported that though 99% of F. columnare cells were eliminated in 1 h exposure to 4% salinity, bath treatment did not reflect such efficacy, as 100 % mortality was recorded in experimentally infected fish exposed for 15 min to 4% salinity.The reason adduced for the failure of the bath treatment was that the overlying mucus on the fish shielded the pathogen located on the skin from the disinfectant, based on the buffering capacity of the skin.
In the present study, the mortality recorded in the fish group that was infected but not treated was due to columnaris disease because of the clinical signs observed and the re-isolation of the pathogen from dead fish samples.Mortality rates of 23.3 and 46.6% observed in 30 min exposure to 2 and 3% salinity respectively could be ascribed to lethal NaCl concentrations at prolonged exposure.The absence of clinical signs, observation of signs of stress during bath treatment and commencement of mortality even before the end of the short bath treatment are all indicators of this assumption.Furthermore, F. columnare was not isolated from dead fish recovered from this treatments.It should be noted that C. gariepinus is a freshwater, stenohaline species, and do not survive or grow for a prolonged period in salinities much above 10 ppt (10 g L -1 or 1%).The freshwater stenohaline species are known to regulate their plasma ion such that the internal osmotic pressure of their fluids is equivalent to approximately 10 ppt salinity with a range of 2 ppt, depending on tolerance, regulating   capacity, and environmental salinity (Brett, 1979).Oladosu et al. (1999) observed the calculated median lethal salinity (MLS -96) of 7.8 ppt (0.78%) for fingerlings (wt = 1.33 g) of C. gariepinus, though ontogenetic variation in salinity tolerance, characteristic of freshwater stenohaline species, was observed as there was increase in MLS -96 from fertilized eggs to fingerlings.
From the aforesaid, it could be safely inferred that short duration salt bath not beyond 10 min can be used for the disinfection of C. gariepinus at salinities of 1 to 3%, to effectively control columnaris disease and concurrent parasitic infection.

Conclusion
The F. columnare isolates used in this study are multidrug resistant, indicating antibiotics mis-use in the aquaculture industry in Nigeria, and suggestive of the need for antibiogram prior to antibiotic treatment of fish.Furthermore, method of disposal of medicated water used in bath treatment of fish should take environmental protection into consideration.Also, since short duration sodium chloride bath was found to be effective in the control of columnaris disease, salt disinfection of fry and fingerling stock could be incorporated into routine management in catfish hatchery.

INTRODUCTION
According to World Health Organization (WHO, 2013), rabies is a fatal zoonotic disease that can affect all mammals.It is mainly transmitted through the bite and virus-containing saliva of an infected host.Other *Corresponding author.E-mail: piusynt@gmail.com.Tel: + 250788512536.. Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License transmission routes are contamination of mucous membranes, aerosol transmission, and corneal and organ transplantations (Leung et al., 2007).Human rabies is mainly transmitted by dog-bites (WHO, 2012).According to Rwanda Biomedical Centre, on average, 54 cases of human dog-bites are reported across Rwanda (Rwanda Focus, 2016).Rabies is caused by Rabies virus (RABV), which belongs to the Lyssavirus genus of the Rhabdoviridae family (World Organisation for Animal Health [OIE], 2011b).According to Malerczyk et al. (2014), 15 Lyssavirus species are known to exist.There is no recognised treatment for rabies, and once it manifests (Rupprecht et al., 2017), it leads to a deadly acute encephalitis or meningoencephalitis (Crowcroft and Thampi, 2015).Vaccination of dogs can help eliminate terrestrial rabies (Crowcroft and Thampi, 2015).To accomplish immunity in dogs, at least 70% of the population should receive rabies vaccination (WHO, 2013).The failure of rabies vaccination in animals might be higher, but it was rated at 0.025% (Oboegbulem et al., 1987;Tepsumethanon et al., 2016).A case-control study carried out in Finland on dogs received rabies vaccination over five years, revealed that 10.7% of the dogs had antibody levels lower than 0.5 IU/ml (Nokireki et al., 2017).According to OIE ( 2013), the cut-off point for antibody response to both canine and feline rabies vaccination is 0.5 IU/ml.
A study by Berndtsson et al. (2011) indicated that factors such as dog's breed and breed size, age at vaccination, brand of utilised vaccine and rabies vaccination number can influence outcome of canine rabies vaccination.Other factors such as nutrition status, sex and proper vaccine preservation (Kennedy et al., 2007;Jibat et al., 2015) as well as animal health status, amounts of antigens and way of application can also impact antibody response (Moore and Hanlon, 2010).Indirect enzyme-linked immunosorbent assay (ELISA) using rabies glycoprotein can be used to confirm whether dogs and cats which received rabies vaccination have seroconverted (Quinn et al., 2011).ELISA can replace neutralization tests for demonstrating antibodies (Moore and Hanlon, 2010).
The control of rabies disease in Rwanda involves vaccinating owned dogs annually and culling stray dogs (OIE, 2011a).In 2010, the number of both canine and feline population in Rwanda was approximated to 31,448 including 8,650 dogs vaccinated against rabies (Southern and Eastern African Rabies Group, 2011).The vaccination coverage for canine and feline rabies vaccination was at 27.5%.The report by Rwanda Agriculture Board revealed that dog population in Rwanda in 2016 was estimated to be 18, 117 and that only 11,375 were vaccinated against rabies and 2,870 culled.The rate of rabies vaccination coverage was at 62.7% (The New Times, 2017).There is no published data on dogs' response to rabies vaccines in field condition in Rwanda.Thus, this study assessed factors Ntampaka et al. 65 influencing rabies antibody titres in vaccinated and nonvaccinated pet dogs in Kigali City, Rwanda.The hypothesis assumed that dogs vaccinated against rabies in Kigali city, Rwanda seroconverted to protective antibody levels regardless of type of applied vaccine.

Ethics standards
The  2014).During data collection, the records of rabies vaccination of dogs at the district level were used.The map illustrating the study sectors is indicated in Figure 1.
Figure 1 illustrates 9 study sectors, namely Niboye, Gatenga, Kicukiro, Gisozi, Kacyiru, Kimironko, Kigali, Mageragere, and Nyamirambo.Blue points indicate households where dogs were identified.Three administrative sectors were chosen per district based on distribution of dogs and vaccination history.

Study design
This cross-sectional study involved collecting dogs' individual information as well as clotted blood from vaccinated and nonvaccinated pet dogs across Kigali city.Cluster sampling was used to select study sectors while snowballing sampling was used to identify households owning dogs.Computation of study dogs was done in accordance with a formula by Chomel et al. (1987).

Collection of individual dog data
Administrative local leaders helped the investigator in identifying and reaching households owning dogs and a total of 137 households owning at least a dog were identified, that is, a dog per household was recruited in the study.
A questionnaire was used to collect information on 137 dogs including each dog's rabies vaccination history (age, sex, date of vaccination, number of vaccinations and utilised vaccine type).Some dogs were aggressive while others died, thus, only 93 dogs including 80 received rabies vaccination and 13 non-vaccinated were sampled.

Collection of samples
Clotted blood samples were collected from restrained dogs.Figure 2 shows how dogs were restrained.
Clotted blood samples were collected in plain vacutainer collection tubes.The samples were preserved in a cooler box without ice packs.Laboratory analysis was held at Rwanda   National Veterinary Laboratory in Kigali.In the laboratory, the samples were centrifuged at 1,500 × g for 10 min at room temperature and the obtained sera kept frozen at -80°C until used.

Detection of rabies antibodies
Quantification of rabies antibody titres in dog sera involved a commercial indirect ELISA test.The assay was performed in accordance with the manufacturer's protocol (Platelia™ Rabies II; Bio-Rad, France).Optical densities in the microplate were read with a spectrophotometer at 450 nm, copied and transferred to the Bio-Rad conversion tool to obtain serum titres.A unit equivalent to the international units defined by seroneutralization (Equivalent Units per milliltre: EU/ml) was used to express the sera titres.The cut-off point for antibody was 0.5 EU/ml (OIE, 2013).The titres were interpreted based on four seroconversion levels which included undetectable (<0.125 EU/ml), insufficient (<0.5 EU/ml), sufficient (0.5-4 EU/ml) and high level of sero-conversion (˃4 EU/ml).

Data analysis
Statistical Package for Social Sciences (SPSS) Statistics version 20 was used to analyse computerised data.Both descriptive (frequency and geometric mean titre) and inferential (Chi-square tests and regression analysis) statistics were used to interpret impact of various factors (type of utilised rabies vaccine, number of rabies vaccinations, time elapsed between vaccination and sampling, dogs' age) on antibody titres in the studied dogs.The level of significance was set to 5%.

RESULTS
General information on study dogs is summarised in Table 1.
Table 1 summarises study dogs information regarding sex, rabies vaccination status, ages of the vaccinated dogs, number of vaccination and time elapsed between vaccination and sampling.The dogs were vaccinated by both private and public veterinarians using five types of rabies vaccines.Vaccines were coded as A: Rabies Veterinary Vaccine Inactivated B.P. (Vet.) by Indian immunological limited, India; B (Vaxipet R: inactivated vaccine, Laprovet, France); C: (Vaxipet DHPPi+LR:

Serological analysis findings
Table 2 shows that, of the vaccinated dogs, 65% had sufficient protective antibody levels (≥ 0.5 EU/ml), while 34 and 1% had inadequate and undetectable antibody levels, respectively.Ninety-nine percent of the vaccinated dogs seroconverted.Seven out 13 (53.8%)nonvaccinated dogs had detectable antibody titres varying between 0.133 and 0.238 EU/ml, while 46.2% did not have detectable antibodies (˂0.125 EU/ml).

Impact of various variables on production of antibodies
Vaccination status was found influential to rabies antibody response, that is, on average, the number of rabies antibodies per dog received vaccination (11.776059735) was eightfold than that of non-vaccinated dog (1.41579378).Considering other factors, the five vaccine types elicited diverse antibody levels and varyingly impacted the immunity in dogs received rabies vaccination.The highest mean titre (2.115) was attained by vaccine type E (Nobivac Rabies) while the mean titres for types B (Vaxipet R) and C (Vaxipet DHPPi+LR) were 1.850 and 1.261, respectively.The overall mean titre (1.071) was higher than 0.897 and 0.814 that were the lowest mean titres and produced by vaccine A (Rabies Veterinary Vaccine Inactivated) and D (Rabisin), respectively.Percentages of the dogs that were protected per vaccine are shown in Table 3. Table 3 indicates that, irrespective of sampling time, 41.7 and 36.3% of dogs which received vaccine types A and D were not protected while 33.4 and 22.2% of those vaccinated with vaccine types C and B were not protected.Approximately 11.1% of dogs vaccinated with vaccine type E were also not protected (Table 3).In terms of sampling time, the difference between protected and non-protected vaccinated dogs,  2 (9) =10, p= 0.350, was not statistically significant.This study found that time elapsed between vaccination and sampling influenced dogs' sera titres and the titres decreased with time, that is, 1.559, 0.949, and 0.934 in dogs sampled at 1-5, 6-9 and 10-12 months after vaccination, respectively.Approximately 42, 40 and 19% of the vaccinated dogs that were not protected (<0.5 EU/ml) were sampled at 6-9, 10-12 and 1-5 months following vaccination, respectively.The difference between sero-conversion levels,  2 (9) =12, p=0.213 was not statistically significant.Also, the difference between elapsed vaccinationsampling time and sero-conversion implication,  2 (9) =12, p=0.213 was not statistically significant.Antibody titres increased by number of vaccinations (Figure 3).
Figure 3 shows that the mean titres gradually augmented from the first to the fourth rabies vaccination while on at least fifth time, it slightly declined.Statistics of rabies protection status of the vaccinated dogs versus vaccination frequency are indicated (Table 4).

>2.5-4 years old
Table 4 shows that, the highest percentage (52%) of the vaccinated dogs that were not-protected was recorded in those vaccinated once.In terms of number of vaccinations, the difference between protected and nonprotected dogs,  2 (25) =30, p=0.224 was not statistically significant.The mean titres increased by dogs' age (Figure 4).
Figure 4 shows that, the lowest mean titre (0.638) was recorded in dogs that were under 1 year of age.The dogs aged at least five years had the mean titre of 1.227 and the mean was 1.515 in dogs aged >2.5 to 4 years old.Chi-square tests of association indicated that the relationship between seroconversion levels and ages of dogs,  2 (9)=11.509,p=0.242 was not statistically significant.Pearson's correlations coefficient showed that vaccination number at α=0.05, r=0.255, p=0.013, vaccination status at α=0.001, r=0.528, p<0.001, time elapsed between vaccination and sampling at α=0.001, r=0.391, p<0.001 positively correlated with antibody titres.Dogs' age negatively correlated with antibody titres at α=0.001, r=-0.281,p=0.006.At both α=0.05 and α=0.001, type of rabies vaccines did not correlate with antibody titres, r=-0.008,p=0.939.

DISCUSSION
The present findings show that 35% of vaccinated dogs had rabies antibodies below the protective levels (˂0.5 EU/ml) while 53.8% of non-vaccinated dogs had measurable rabies antibodies varying between 0.133 and 0.238 EU/ml.Such dogs whether vaccinated or not were at risk of contracting rabies infection.According to Oboegbulem et al. (1987) and Tepsumethanon et al. (2016) rabies vaccination failure in animals might be higher, but it was approximated to 0.025% in rabies vaccinated dogs.The 35% of vaccinated dogs that were not protected in this study was higher than 4.62, 12 and 30% which were reported by Ondrejková et al. (2015), Wallace et al. (2017) andFernandes et al. (2017).The high number of dogs that were not protected in this study could be related to time elapsed between vaccinationsampling.In the present study, dog blood samples were collected between 1 and 12 months after the dogs received vaccination while Ondrejková et al. (2015) sampled dogs on the 30th day following rabies vaccination.
In the study by Wallace et al. (2017), sampling interval varied from at least 3 days to over 270 days following rabies vaccination.The results are consistent with those of Cliquet et al. (2003), Kennedy et al. (2007), Minke et al. (2009), andBerndtsson et al. (2011) who reported that time elapsed between vaccination and sampling influenced rabies antibody titres.One of the other possible reasons behind the high number of vaccinated dogs that were not protected would be number of vaccination.The results were compatible with those of Cliquet et al. (2003), Berndtsson et al. (2011), and Watanabe et al. (2013) who reported that the number of antirabies vaccinations impacted rabies antibody titres.The present findings were also compatible with those of Kennedy et al. (2007) and Berndtsson et al. (2011) who found that age impacted on dogs' rabies antibody titres.However, the study disagrees with that of Salinas et al. (1992) who reported that number of vaccinations did not influence rabies antibody response.We think that the difference would be related to how the immune response was quantified.The study assessed the immune response of study dogs based on antibody titres, while Salina et al. (1992) evaluated the immune response of sample dogs based on optical densities.This could have influenced the cutoff point.
This study indicates that, of non-vaccinated pet dogs, 53.8% had detectable antibodies lower than the protective levels (<0.5 EU/ml).Probably, these dogs might have had an abortive infection.Ondrejková et al. (2015) reported 13.04% of non-vaccinated pet dogs that had measurable antibody titres while Cleaveland et al. (1999) found rabies antibody titres in 7.4% of nonvaccinated pet dogs.Detection of rabies virus antibodies in healthy animals was reported by Prager et al. (2012), though it was not known whether the animals had an abortive infection or recovered from rabies.A study by Watanabe et al. (2013) found maternal rabies antibodies in 15.3% of non-vaccinated study puppies younger than three month old; however, none of unvaccinated dogs involved in the present study was younger than three months old.A report by El-Sayed (2018) indicated that abortive animal and human rabies may be caused by non-encephalic rabies strains and such an infection does not result in health abnormalities.

Study limitations
The impact of dogs' breeds and breed size on rabies antibody titres could have been assessed, but the investigator had trouble recognising study dogs breeds, thus breed size data of the studied dogs was also not collected.

Conclusion
This study found that the vaccinated dogs produced diverse rabies antibody levels regardless of type of rabies vaccine applied.Some vaccinated dogs were poor responders to rabies vaccines and did not have protective rabies antibody levels.Some non-vaccinated studied dogs had detectable rabies antibodies below the cutoff point.Different factors including rabies vaccination status, number of rabies vaccination, age at vaccination, time elapsed between vaccination and sampling influenced rabies antibody titres of the studied dogs.Considering the high number of vaccinated dogs that were poor responders to rabies vaccination, further studies should be undertaken to investigate and understand the phenomenon.

INTRODUCTION
Brucellosis is one of the most widespread zoonotic diseases posing a serious obstacle to public health, food safety and security and, socio-economic development in most African countries (Mcdermott et al., 2013;Mcdermott and Arimi, 2002).Brucellosis affects many animal species, including such as cattle, sheep, and goats.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License Most human infections are acquired through direct contact with infected animals, placentas, foetal membranes, vaginal discharges or aborted foetuses, inhalation of infectious materials or via indirect transmission through untreated milk and milk products, undercooked meat or blood (Liu, 2015;Makita et al., 2008;Galińska and Zagórski, 2013).Brucellosis has wide-ranging impacts that include animal losses due to abortion, loss in milk production, death of infected animals and debilitating human illness causing reduced working capacity (Jergefa et al., 2009;Mcdermott et al., 2013).
Brucellosis in humans, also known as undulant fever, presents with intermittent fever, sweating, joint and low back pains, headache, fatigue, weight loss and general weakness persisting for a long time (Dean et al., 2012;Madut et al., 2018).These clinical signs resemble other febrile illnesses such as malaria, typhoid, rheumatic fever and viral diseases hence, the disease is often misdiagnosed and under-reported in most countries in sub-Saharan Africa (Bouley et al., 2012;Crump et al., 2013).Brucellosis is an occupational disease to farmers, herders, veterinarians, slaughterhouse workers, butchers and laboratory personnel who commonly get into contact with the animals or animal by-products in the course of their work (Gardner, 2005;Mirambo et al., 2018;Schneider et al., 2013).However, abattoir workers may be the more risky group compared to other occupations because they are overexposed to carcasses, viscera, body fluids and organs of infected animals (Mirambo et al., 2018;Mukhtar, 2010;Swai and Schoonman, 2009), and this is compounded by their proneness to knife cut wounds (Banjo et al., 2013).Consequently, abattoir workers are at greatest risk to get brucellosis through open wounds on bare hands, splashing of infected fluids in the conjunctiva and inhalation of aerosols in the slaughtering area (Aworh et al., 2013;Goncalves et al., 2006).The seroprevalence of human brucellosis in Tanzania has been reported to range from 14.1 to 28.2% (Chota et al., 2016;Mngumi et al., 2016) in pastoral and agropastoral settings.
In Tanzania, brucellosis in slaughterhouse personnel has been previously reported to be high, ranging from 19.5 to 42.1% in several slaughterhouses (Mirambo et al., 2018;Swai and Schoonman, 2009).The high prevalence may be attributed to lack of awareness on the disease among the workers and non-use of protective gears.It is well-known that a well-designed abattoir having separate sections from stunning to packaging with minimum exposure and coupled with strict adherence to biosafety measures and practices can reduce the chances of the occupational exposure to brucellosis (Zakaria et al., 2018).The objective of this study, therefore, was to determine the risk of occupational exposure to brucellosis among workers in Dodoma modern abattoir in Tanzania.The results of the study would be used to inform public health authorities on the magnitude of the problem so that they can design appropriate interventions for reducing the risk.

Study area
The study was conducted at Dodoma abattoir, one of among the most modern abattoirs in Dodoma Region, Tanzania.Dodoma Region is located between latitudes 6° 57 'S and 3° 82'S and between 36° 26' and 35° 26'E and, bordered by four regions, namely; Manyara, Iringa, Singida and Morogoro (Population and Housing Census of Tanzania [PHCT], 2013).Dodoma is the current capital city of the United Republic of Tanzania and is growing very fast in terms of both human population and economic activities resulting in increased demand for food supplies including meat.
Dodoma abattoir is the largest and most modern one in the central zone, currently the only in the country slaughtering animals for export.The animals slaughtered at the abattoir include cattle, goats and in a very few occasions sheep.Meat from goats is mainly exported to Comoro, Iraq, Vietnam, Oman, Qatar, Kuwait and United Arabs Emirates while beef is consumed locally.Dodoma abattoir has well established infrastructures from stunning area to packaging area which facilitates unidirectional flow of carcass.Animals slaughtered at Dodoma abattoir originate mainly from neighbouring regions of Iringa, Singida and Manyara as well as some from Lake and Western zone regions of Kigoma, Tabora, Mwanza, Shinyanga and Geita where brucellosis has been reported to be prevalent in livestock (Chitupila et al., 2015).

Study design
A cross-sectional study design was adopted.It was carried out from March to August 2018 to determine the occupational hazards and practices associated with risk of exposure of abattoir workers to brucellosis and assess their knowledge on the disease and other zoonotic diseases.
Where; N=Sample size, P=Expected prevalence, d=Precision level.By using the expected prevalence (P) of 12%, precision level (d) of 0.05 and confidence level of 95%.Thus, 324 samples were to be collected from all animal species.

Sampling procedure
A systematic random sampling method was used to select cattle and goats.By using a sampling fraction of 20% every fifth animal was sampled (Cadmus et al., 2006).Before sample collection, all the animals to be slaughtered in that same day were arranged in the crush and the first animal was randomly selected followed by every fifth animal counted from the first selected in a row (Cadmus et al., 2006).Abattoir workers above 18 years who had worked in the facility for at least one year were included in the study upon their consent to participate in the study.

Collection of blood samples
A total of 390 animals destined for slaughter comprising 200 goats and 190 cattle were bled for brucellosis screening.Furthermore, 62 human blood samples were collected from the abattoir workers who agreed and consented to participate in the study.The inclusion criteria of individuals to participate included age of at least 18 years and working in the abattoir for at least one year.However, prior to bleeding animals, verbal consent was sought from livestock traders.Prior to bleeding, individual animal information such as breed, age, sex and geographical origin were recorded.Sterile plain vacutainer tubes were used to collect approximately 5 ml of blood from the jugular vein in animals and brachial vein in humans.All blood samples were stored in a cool box with ice packs and transported to the Tanzania Veterinary Laboratory Agency station in Dodoma, where they were left at room temperature for six hours to allow separation of clear serum and blood clots.
A semi-structured questionnaire was used to assess the knowledge of abattoir workers about brucellosis as well as their attitude and practices with regard to the potential of occupational exposure to the disease.Social and demographic factors such as sex, age, marital status, education level, period of time worked at the abattoir, types of activities within and outside the abattoir were gathered.The questionnaire was administered in Swahili language because most of the workers were conversant with the same.

Laboratory analysis
Blood samples were centrifuged at 3000 g for 20 min for serum separation.Serum samples were collected in cryovials, well labelled and stored at -20°C until analysis.

Rose Bengal Plate Test (RBPT)
Four hundred and fifty-two serum samples were screened for antibrucella antibodies using Rose Bengal buffered antigen (ID Vet, France).The test detects both the IgM and IgG.The test procedure was performed as recommended by (OIE, 2016).Briefly, 25 µl of buffered antigen and 25 µl of test serum were pipetted on the white tile plate and then mixed thoroughly.The white tile plate was rocked for four minutes while observing the degree of agglutination.The sample was classified as positive if agglutination was observed and negative if there was no agglutination.Sample with some degree of agglutinations whether weak or strong was considered positive (Figure 1).
Indirect Enzyme-Linked Immunosorbent Assay (SVANOVIR® Brucella-Ab I-ELISA) was used to further test positive samples from the Rose Bengal plate test.Indirect ELISA measures the binding of secondary antibodies to a primary antibody isotype bound onto the Brucella lipopolysaccharides antigen.The test detects only the Ig G.The test was performed according to the manufacturer's instructions.Briefly, 90 µl of sample dilution buffer was added to each well to be used for serum samples and controls followed by adding 10 µl of positive control serum and 10 µl negative control serum, respectively to selected wells coated with Brucella abortus antigen.For confirmation purposes, the control sera were run in duplicates.
In the remaining wells coated with B. abortus antigen, 10 µl of serum samples was added, shaken thoroughly, sealed and incubated at 37°C for one hour.The plate was rinsed with PBStween buffer by filling up the wells at each rinse, emptied and taps hard to remove all remains of fluid.After rinsing, 100 µl of Horseradish peroxidase(HRP) conjugate was added to each well, sealed the plate and incubated at 37°C for one hour, followed by adding 100 µl of substrate solution to each well and incubation for 10 min at room temperature.The reaction was stopped by adding 50 µl of stop solution to each well mixed thoroughly and the optical density of the controls and samples were measured at 450 nm in a microplate photometer within 15 min.The optical density (OD) values were read in a microplate reader (ELISA reader, Multiskan Luwumba et al. 75 RC version 6.0,Thermo Labsystems,Helsinki). Strong positive (considered as 100% positivity) and negative standards were used.The results were expressed as per cent positivity (PP) of the sample tested in relation to the strong positive control.The criterion for considering a sample positive or negative was based on percent positivity calculated as follows: A sample was considered positive or negative using the cut-off values recommended by the manufacturer.Samples with equal or above 60% positivity were considered positive.

Data analysis
Data were stored and cleaned in Microsoft Excel spreadsheet and analyzed using R software version 3.4.3("Kite-Eating Tree" Copyright (C) 2017 The R Foundation for Statistical Computing platform).Descriptive statistics, particularly frequencies were computed for proportions of Brucella positive animals and human.Categorical dichotomous variables were computed and compared using Chi-square test at a critical probability of α ≤ 0.05.Odds ratio, 95% confidence interval, Chi-square and Fisher's exact tests were computed to determine the degree of association between Brucella seropositivity with knowledge, awareness and practices of abattoir workers.

Ethical consideration
The

The seroprevalence of brucellosis in abattoir workers and animals slaughtered at Dodoma abattoir
The seroprevalence of brucellosis in cattle, goats and abattoir workers was 7.3, 1.5 and 1.6% based on Rose Bengal plate test, respectively.The positive sera by Rose Bengal plate test were further tested by enzyme-linked immunosorbent assay and only nine (4.7%) cattle and one (1.6%)human samples were positive.All samples from goats were seronegative (Table 1).

Demographic characteristics of participants
A total of 62 (51.7%) out of the 120 abattoir workers volunteered for blood sample collection and of these, 49

Awareness of abattoir workers on brucellosis and other zoonotic diseases
In assessing the awareness level of abattoir workers on brucellosis and other zoonotic diseases, it was found that there is a significant difference in awareness on brucellosis among sex (p=0.001),occupation (p=0.0048),duration of work (p=0.05)groups while there was no significant difference in the education (p=0.519) and age categories (p=0.181).The participants were 14 times aware of brucellosis than other zoonotic diseases (OR 14, 95%CI 2.47,160.37,p=0.00046) (Table 3).However, there was no statistically significant difference in the knowledge of participants on other zoonotic diseases among gender, age, education, occupation and duration on the job categories (p > 0.05).Furthermore, 76% (n=37) of abattoir workers had knowledge of brucellosis and 61% (n=30) other zoonoses

DISCUSSION
Brucellosis is a zoonotic disease which affects both animals and humans who come in contact with infected animals and their products such as blood, milk, meat or foetus and placental fluids as well as other contaminated materials.Therefore, abattoir workers are at the greatest risk of acquiring brucellosis because they frequently come in contact with the animal and by-products during their routine activities.
The seroprevalence of 4.7% in slaughtered cattle in this study is an indication that brucellosis is prevalent in areas where the slaughtered animals originated.This poses a risk to abattoir workers who do not comply with basic biosafety principles in slaughter facilities.This prevalence recorded in this study is lower than those reported by other workers such as 12% in Tanga (Swai and Schoonman, 2012) and 21% in Karagwe (Kiputa et al., 2008).This finding may probably suggest that slaughter animals may have originated from areas with low infection rates.
Although the prevalence of brucellosis in humans was low (1.6%), it still indicates that workers at the abattoir were at risk of acquiring brucellosis in the course of their work, that is handling live animals and carcases.The observed prevalence was also low when compared to findings recorded in other studies in Tanzania (Mirambo et al., 2018;Swai and Schoonman, 2009) and elsewhere (Agada et al., 2018;Aworh et al., 2013;Cadmus et al., 2006;Mukhtar, 2010;Nabukenya et al., 2013;Osoro et , 2015).The difference may be attributed to wearing of protective gears, awareness through training and separation of various processes in the abattoir to enhance proper handling and cleanness.It was observed that all workers in the abattoir wear gloves, gumboots, overcoats and overhead covers, and a few also wore mouth covers and goggles.This was further supported by the fact that 59% of the workforce had post-primary education and 76% were aware of brucellosis.The interviewed participants were 14 times more aware of brucellosis than other zoonoses.The awareness on the occupational risk of brucellosis among the abattoir workers may have enhanced compliance with biosafety measures and practices leading to low prevalence found in this study.These findings concur with similar studies conducted elsewhere, which demonstrated that wearing personal protective gears and level of knowledge are protective factors for acquiring human brucellosis and other zoonoses (Ayoola et al., 2017;Islam et al., 2013;Madut et al., 2018).
In this study, it was observed that there was a statistical significant difference (p<0.05) in the knowledge of brucellosis among workers working in different sections.Sections having individuals with low awareness may be more prone to infection if the frequency of contact is high with susceptible infectious materials.The human seropositive case reported in this study was from the group of cleaners.Swai and Schoonman (2009) and Tsegay et al. ( 2017) argued that cleaners in the abattoir being the least educated group may lack knowledge on how brucellosis is transmitted.On the other hand, since they are most heavily involved in handling and disposing all condemned abattoir materials including aborted foetuses that usually have very high concentrations of Brucella pathogens; they are at a much higher risk of infection when compared to other groups.
Although the abattoir workers had some knowledge on brucellosis but it was observed that some of them had the tendency of opening the reproductive tract of slaughtered animals and aborted foetuses within the facility and sold them to vendors outside the abattoir premises.This practice may contribute to the contamination of the carcasses and environments resulting into increased risks of transmission of human brucellosis.Other studies have reported that consumption of gravid uteri increases the risks of humans to contract brucellosis (Adesokan et al., 2016).The reproductive tract is the predilection site for Brucella and it has to be removed from the animal without opening it in order to minimize the chances of dissemination of Brucella organisms thereby contaminating the meat and the surrounding environment.This increases the risk of transmission of the pathogen to the abattoir workers, meat vendors and the public in general.Therefore, training of all abattoir personnel on the proper practices, including proper disposal of reproductive organs should be emphasised.Despite the fact that the prevalence of human brucellosis was low in this study, the importance of human life underscores the need to undertake more systematic studies using the one-health approach in order to establish the magnitude of the problem country-wide as the basis for recommending practical control measures for the disease.
In conclusion, the present study has revealed that brucellosis is present both in animals slaughtered at the abattoir and humans working in the same facility.In view of this, there is a need for adopting biosafety measures such as wearing protective gear as well as educating the workers on the occupational risk of the disease in order to reduce its transmission.It is also recommended that integrated approaches be used in controlling the disease at the farm level in order to minimize transmission to abattoir workers.
numerous to count; min = minutes; % = percent or gram of salt per 100 mL of water.

Figure 1 .
Figure 1.Map of the study area.Source: Generated from GPS data by authors.

Figure 2 .
Figure 2. Restraining dogs during collection of blood samples.

Figure 3 .
Figure 3. Applied number of vaccinations and mean titres.

Figure 4 .
Figure 4. Influence of dogs' age on production of antibodies.

Table 1 .
Size of inhibition zone to Flavobacterium columnare growth observed with different antibiotics using disc diffussion method.

Table 2 .
Susceptibility pattern of Flavobacterium columnare isolates to varied concentration of sodium chloride (common salt) in-vitro, observed as bacterial colony count.

Table 3 .
Percent mortality observed during bath treatment and topical disinfection of fish experimentally infected with Flavobacterium columnare using ciprofloxacin and common salt.

Not infected Infected/ untreated 25 mg L -1 ciprofloxacin 1% Salt bath 2% salt bath 3% salt bath Duration of Exposure 1 h 5 min 10 min 30 min 5 min 10 min 30 min 5 min 10 min 30 min
University of Nairobi, Faculty of Veterinary Medicine's Biosafety, Animal Use, and Ethics Committee (FVM BAUEC/2017/126) and the Rwanda National Ethics Committee (Review Approval Notice: No. 115/RNEC/2017) approved this study.Before collecting samples from dogs, owners signed certificates of consent.
areaThe present study was conducted in the city of Kigali, in Rwanda between September 2016 and March 2017.Administratively, Kigali city is divided into 3 districts and 35 sectors which are also divided into cells and villages (National Institute of Statistics of Rwanda,

Table 1 .
Summary of general information on study dogs.

Table 2 .
Levels of sero-conversion in dogs that received rabies vaccination.

Table 3 .
Percentage of protection for vaccines' type and vaccination frequency.

Table 4 .
Statistics of non-protected dogs versus vaccination frequency.

Table 1 .
The seroprevalence of brucellosis in animals and abattoir workers at Dodoma abattoir.

Table 2 .
Demographic characteristics of workers at Dodoma abattoir.
were in the 20-30 years age group.Most of the participants were males 55.1% (n = 27) and 59.2% (n = 29) had post-primary education.The participants duration of work at the abattoir ranged from 1 to 14 years' duration of work at the abattoir ranged from 1 to 14 years with an average age of 5 ± 2.79) years (Table2).

Table 3 .
Awareness of abattoir workers on brucellosis and other zoonotic diseases.