Detection and characterisation of Leptospira spp. in dogs diagnosed with kidney and/or liver disease

Leptospirosis is a zoonotic bacterial disease that affects both humans and animals. In humans, a wide range of symptoms had been described but in dogs, it is commonly associated with kidney and/or liver disease. In Malaysia, information with regards to the common serovars causing leptospiral infection in dogs remains limited. Therefore, this study investigated the occurrences of leptospiral infection in 124 pet dogs diagnosed with kidney and/or liver disease in Malaysia. Based on microscopic agglutination test, 42.7% (53/124) of the dogs were seropositive for leptospiral infection. The predominant serovars detected were Bataviae (n = 12), Javanica (n = 10) and Icterohaemorrhagiae (n = 10). The direct detection using polymerase chain reaction showed that 33.9% (42/124) of the whole blood and 31.9% (36/113) of the urine samples were positive to pathogenic Leptospira spp. For tissue samples, 9.1% (2/23) of the kidney and 9.1% (2/23) of liver were positive for pathogenic Leptospira spp. Addition samples of abdominal effusion from four dogs were positive for pathogenic Leptospira spp. The species detected were L. interrogans, L. borgpetersenii, L. kirshneri and L. kmetyi by partial 16S rRNA sequencing. In this study, 11 Leptospira spp. isolated successfully from the eight dogs were further identied and characterised as Bataviae, Javanica and Australis. Unfortunately, the mortality rate of the infected dogs was high at 34.0% (18/53).

con rmatory tests for the direct or indirect identi cation of the pathogen, such as dark eld microscopy, polymerase chain reaction (PCR), bacterial culture and microscopic agglutination test (MAT) [1]. For serodiagnosis of acute leptospiral infection, MAT is still widely employed, despite its limitation and poor ability to predict the infecting serovar and may not distinguish between infection and vaccine-induced titres [15]. Conversely, PCR has been successfully used to con rm leptospiral infection at the early stages of infection [16,17], and sequencing PCR amplicon has enabled the identi cation of the different leptospiral species infecting dogs [18].
In Malaysia, the seroprevalence of canine leptospirosis among dogs ranged from 3.1-22.2% [19,20,21,22]. However, there is a limited report in dogs diagnosed with kidney and/or liver disease locally. Only one study reported in dogs with kidney disease by prevalence of 5.3% [23]. Therefore, this present study described the diagnosis of leptospirosis in dogs diagnosed with kidney and /or liver disease using multiple diagnostic strategies, such as serological, molecular, and bacteriological tests, along with the characterisation of the recovered leptospiral strains. The establishment of a panel of leptospiral strains circulating among dog populations remains the major strategy to support the development and commercialization of vaccines incorporating more speci c serovar compositions, which would hypothetically increase immunization effectiveness for local canine populations.

Results
A total of 124 dogs were recruited in this study. Out of 124 dogs, 68 dogs were diagnosed with both kidney and liver diseases, the remaining 34 dogs had kidney disease (BUN and/or creatinine elevated) and 22 dogs had liver disease (ALT and/or ALP elevated). The demographic data of the dogs were tabulated as shown in Table 1. The most common clinical signs observed in the dogs were inappetence followed by vomiting, lethargy, jaundice, and diarrhoea. All dogs presented were observed to have at least two clinical signs (refer Fig. 1).
Molecular detection and partial 16S rRNA sequencing Total molecular detection of leptospiral infection by PCR was 42.7% (53/124; 95% CI: 34.0% − 51.4%). The positive samples were obtained from 42 whole blood, 36 urine, tissue samples of two kidneys and two livers and four abdominal effusions. All of them were positive for pathogenic Leptospira sp. and four Leptospira species had been detected using partial 16S rRNA sequencing. L. interrogans (n = 62) was the most detected species followed by L. borgpetersenii (n = 17) and L. kirschneri (n = 6). Only one blood sample was detected as L. kmetyi as shown in Fig. 4. Out of 53 positive dogs, 18 dogs had died, therefore the mortality rate was 34.0% (95% CI: 21.2% − 46.7%).
Isolation and characterisation of Leptospira spp.
The Leptospira spp. were successfully isolated from eight out of 124 dogs (6.5%; 95%CI: 2.1% − 10.8%), where three dogs were vaccinated, and ve dogs were not vaccinated. From the eight dogs diagnosed with kidney and/or liver disease, 11 isolates were obtained from three blood, seven urine and one abdominal effusion samples. Serological characterisation using MAT showed eight of the isolates were Bataviae, two were Australis and one was Javanica. Molecular characterisation using PCR revealed all the isolates were pathogenic. Further analysis of the partial 16S rRNA sequencing using BLAST (www.ncbi.nlm.nih.gov/blast) con rmed 10 isolates had 95% identical towards L. interrogans and one isolate had 95% identical towards L. borgpetersenii. The sequences were submitted to the GenBank and accession number was obtained (refer Table 2). The phylogenetic tree showed three distinct clades related to pathogenic, intermediate, and saprophytic species. All the isolates obtained in this study were placed within the pathogenic group. Ten of the isolates were placed within species of L. interrogans while one isolate was placed within L. borgpetersenii as shown in Fig. 5. Three isolates namely L. interrogans strain UVH khor_sapy1, L. interrogans strain UVH khor_sapy3 and L. borgpetersenii strain khorsayo were found to diverge from the respective species.
Condensed tree was conducted with 50.0% cut-off point from original tree to further investigate the signi cance of the divergence and showed no signi cant divergence for all the isolates as shown in Fig. 6. From condensed tree, all 10 isolates were 61.0% closely related with all the representative species of L. interrogans. In addition, L. interrogans strain UVH khor_sapy1 and L. interrogans strain UVH khor_sapy3 were 99.0% related because those two isolates were isolated from one dog. The other isolate was 68.0% closely related to all the representative species of L. borgpetersenii.

Discussion
Leptospira spp. can be found worldwide regardless of climate and has been presumed to be the most widespread zoonoses [24]. Despite it being common, the diagnosis of canine leptospirosis is not often made unless the dog was presented with clinical manifestations such as fever, jaundice, renal and/or liver failure. Clinical diagnosis remains a challenge but with the aid of con rmatory laboratory tests, a diagnosis can be derived to allow immediate administration of therapeutic regime [25] especially for dogs diagnosed with leptospirosis. This study demonstrated the utility of direct detection using serological and molecular methods followed by bacterial isolation in dogs with kidney and/or liver disease and found that dogs can shed Leptospira spp., further contaminating the environment and poses a risk of infection to their owners.
Microscopic agglutination test is a sensitive assay, but because of the antigenic heterogeneity of Leptospira spp., the test requires many serovars as antigens [26]. The overall serological detection of leptospiral infection in dogs with kidney and/or liver disease was 42.7%, much higher compared to the previous studies locally. The reason could likely be due to the speci c selection of recruited dogs and serum tested against 20 leptospiral serovars selected. In comparison, previous studies investigated a larger population of apparent healthy shelter and working dogs [19], some investigated healthy dogs from a single location [21,22] and one study was carried out among the pet dogs [23]. All these studies only aimed to determine seropositive among apparently healthy dogs using a panel of 10 leptospiral serovars, unlike in this study.
The MAT titre frequently observed were at 1:100 (n = 21), and perhaps could be used as a potential cut-off titre for diagnosis of leptospirosis as these dogs were clinically sick with the supportive evidence of elevated kidney and/or liver pro les. Alarmingly, three dogs with a titre level of 1:100 did not survive despite being treated and serovar Bataviae was detected. These three dogs showed clinical signs such as inappetence, diarrhoea, vomiting and jaundice within three days with history of post-exposure to rats. Another four dogs died (titre of 1:200), in which serovars Australis (n = 2) and Icterohaemorrhagiae (n = 2) were detected. The highest MAT serological titre obtained were at 1:800 (n = 8) that could suggest a severe condition in these infected dogs. Only two dogs died with a titre of 1:800 (Javanica), whilst six other infected dogs survived post-treatment (2-Javanica; 1-Hardjobovis; 1-Pomona; 1-Copenhageni; 1-Malaysia, respectively). This shows that different level of titre detected does not correlate with the risk of mortality and prognosis. On the other hand, six dogs with high infection titres did survive and therefore perhaps the dogs' immunity level, infecting serovar and treatment initiated might all have played important roles in the survivability of infected dogs.
The panel of leptospiral serovars selected in this study was based on the important serovars circulating in Malaysia for human, rats and dogs as found in the previous studies [19,20,21,22,23,27]. It was observed that the three most frequent leptospiral serovars detected in diseased dogs were Bataviae (n = 12), Javanica (n = 10) and Icterohaemorrhagiae (n = 10), then followed by Ballum (n = 3), Australis (n = 3), Hardjobovis (n = 3), Malaysia (n = 3) and Pomona (n = 2). The least frequent leptospiral serovars observed in this study were Canicola, Lai, Pyrogenes, Copenhageni, Celledoni, Cynopteri and Autumnalis. Therefore, incorporating known local leptospiral serovars is highly recommended in the diagnostic workout to improve detection rate as these serovars had shown to cause disease among dogs. Locally, both serovars Bataviae and Javanica had been reported in dogs [19,21] and high serological detection of serovars Bataviae and Javanica (32 out of 53 seropositive dogs) was directly linked to direct contact with rats. This is not surprising as Benacer et al., states that Bataviae and Javanica are two leptospiral serovars circulating among the urban rats' population in Peninsular Malaysia [28]. Besides that, Icterohaemorrhagiae, Australis, Pomona and Canicola had been detected in rats within Kuala Lumpur [29] which further supports the ndings in the current study. In contrast, serovars Hardjobovis and Copenhageni were commonly reported in working dogs from livestock farms in New Zealand [30]. Thus, the dogs there were at greater risk of exposure to those serovars and detection was governed by local endemicity which was similarly observed in this study.
Direct PCR on specimens enables rapid and direct diagnosis, in both early and convalescent stages of infection. In this study, the overall molecular detection of leptospiral infection in dogs diagnosed with kidney and/or liver disease was 42.7%. Results were consistent with study by Miotto et [32.33]. Despite the similarity in large sample size with this study, the target population recruited were apparently healthy dogs which could explain the lower detection rate. The PCR can detect leptospiral DNA in the whole blood, down to extremely small amounts equivalent to the DNA content of about 10 leptospires or less [26]. In this study, 42 pathogenic Leptospira spp. (33.9%; 42/124; 95%CI: 25.5% − 42.2%) were detected in the whole blood and suggestive of leptospiraemia phase. Out of 42 dogs diagnosed with leptospirosis, 21 dogs were presented at the acute stage as Leptospira spp. with positive detection only from the blood samples. Sixteen dogs were positive for both PCR and MAT, which suggested that the dogs were in convalescent phase, where antibodies might have started to react with the antigens, and that could reduce the amount of leptospires circulating in the body, but still detectable using the molecular method.
The optimum time of Leptospira spp. evident in the urine of infected dogs were reported at seven or more days of clinical illness [26,34]. Positive detection of 36 pathogenic Leptospira sp. from urine samples suggested a leptospiruria phase in these diseased dogs. Twenty-three out of 36 dogs were presented at the convalescent phase. Twenty-ve dogs detected positive for pathogenic Leptospira sp. in both whole blood and urine samples and 14 of the diseased dogs were presented with chronic stage, either in the period of active infection and/or actively shedding. Using the conventional PCR method, blood samples allowed higher detection compared to urine samples. This could be related to samples obtained at different phases of infection or the biological feature of urine itself. There are limitations to PCR with regards to urine samples as urea (act as PCR inhibitor) may lead to polymerase degradation affecting the sensitivity of the assay or even leads to false-negative results [35], which is a challenging issue with direct detection from the clinical sample obtained.
Only two kidneys and two livers were positive for pathogenic Leptospira spp. Low detection from tissue samples could be due to the presence of PCR inhibitors such as haemoglobin and hormones [35]. However, to our knowledge, this study could be the rst to demonstrate all four abdominal effusion samples obtained from four different dogs were tested positive for pathogenic Leptospira spp. and perhaps abdominal effusion can be the preferred sample for molecular diagnostic investigation. Cerebrospinal uid (CSF) was not collected in this study, but leptospiral DNA has been expressed in CSF uids from both, human [36,37] and animal [38] studies. None of the dogs in this study showed neurological signs, The partial 16S rRNA sequencing performed after direct PCR revealed that the most common species detected were L. interrogans (n = 62) followed by L. borgpetersenii (n = 17), L. kirschneri (n = 6), and L. kmetyi (n-1). Leptospira interrogans were detected in all type of samples (blood, urine, abdominal effusion, kidney, and liver). Comparatively similar, L. kirshneri was detected in all sample types except in liver. The detection of L. interrogans and L. kirshneri were expected in this study and both species had been commonly associated with canine leptospirosis [1,31,39,40]. However, the detection of L. borgpetersenii was linked to contacts with rats as this bacterium is commonly shed by rats [28,41,42]. In this study, nine out of 17 samples from a total of six dogs had history of in contact with rats. Two out of the six dogs did not survive despite aggressive treatment therapy. Previous study in Germany reported three out of 200 healthy dogs shed leptospires of the species L. interrogans (n = 2) and L. borgpetersenii (n = 1) [8]. Even though L. borgpetersenii is not common in dogs but this species remains a contributing concern to canine leptospirosis with high mortality. Leptospira kmetyi was detected from a dog's blood sample and to our knowledge, could be the rst report of L. kmetyi identi ed in an animal. This could be associated with environmental exposure because L. kmetyi had been isolated from the environment in Malaysia [43,44].
Isolation and identi cation of leptospires are critical to con rm the speci c leptospiral serovar circulating in this group of dogs diagnosed with kidney and/or liver disease. However, culturing leptospires is challenging due to frequent contamination and the fastidious growth of the pathogen [1]. Leptospires are slow-growing bacteria in comparison to other bacteria. Even though the semisolid EMJH is the selective medium containing 5-FU to improve leptospiral isolation, however high number of contaminants greater leptospires in the samples may likely suppress the growth of the selected bacteria. Despite the 12 weeks challenges of continuous checking and sub-culturing, contaminated samples may alter the possibility of a result. Besides that, recovering leptospires from suspected dogs were limited due to early antibiotic therapy intervention, which is usually required after the disease is suspected [45]. Nonetheless, culturing leptospires still stands as the gold standard reference test for con rmation of leptospiral infection, and only serological characterisation of the isolated strains may provide reliable information regarding serovar identity [1].
Despite the challenges faced throughout the study, 11 Leptospira sp. isolates were recovered from whole blood (n = 3), urine (n = 7) and abdominal effusion (n = 1) samples. Comparing with previous studies, leptospires were successfully recovered only from urine sample of the diseased dogs [46], shelter and stray dogs [47] and farm dogs [20]. This supports that urine samples are superior samples for leptospiral isolation in dogs regardless of the target population. Leptospires can persist in the kidneys and colonize the renal proximal tubules, causing live bacteria excretion in the urine [48]. Leptospiruric dogs remain as potential shedders and therefore, may increase the risk of infection to their owners or from dog-to-dog within the same household. The previous study found that prolonged dog handler-dog contact time increases the risk of seropositivity, and the unknowing handling of infected dogs puts the dog handlers at risk from the leptospires shedding [19]. Sometimes, persistence leptospiruria due to inadequate antibiotic usage of failing to penetrate kidneys may ineffectively eradicate leptospires [49]. Therefore, the use of a sensitive molecular technique is recommended to investigate as to whether an infected treated dog is still leptospiruric and/or the post-antibiotics therapy is effective. Besides urine, leptospires were commonly isolated from whole blood in human [50,51,52]. According to Gompf, other body uids (besides blood and urine) might contain leptospires, but the opportunity to isolate them is slim [53]. Surprisingly, this study successfully cultured Leptospira sp. from abdominal effusion.
Further serological characterisation of the 11 isolates showed reaction towards three hyperimmune sera namely: Bataviae (n = 8), Javanica (n = 1) and Australis (n = 2). The molecular characterisation revealed that all isolates were pathogenic and further con rmed species of L. interrogans (n = 10) and L. borgpetersenii (n = 1) by partial 16S rRNA sequencing. These results strengthen the serological and molecular detection as discussed earlier. Phylogenetic tree analysis revealed that all isolates were within pathogenic clade as follows; eight isolates closely related to reference isolate of L. interrogans serovar Bataviae, two isolates closely related to reference isolate of L. interrogans serovar Australis, and one isolate closely related to reference isolate of L. borgpetersenii serovar Javanica. Phylogenetic analysis concurred with the ndings of previous reports that the pathogenic, intermediate, and saprophytic species each formed one clade [54,55].
The isolation of serovars Bataviae, Javanica and Australis could be alarming due to their absence in commercial vaccines. In Malaysia, bivalent (Canicola and Icterohaemorrhagiae) and tetravalent vaccines (Canicola, Icterohaemorrhagiae, Pomona and Grippotyphosa) were adopted based on the availability of imported vaccine and WSAVA guidelines [56]. Referring to the history obtained from pet owners, three diseased dogs were vaccinated with tetravalent vaccine and ve were not vaccinated. Dogs vaccinated annually remained at risk as vaccination does not provide cross-protection towards non-vaccinal serovars. In Australia, commercial vaccines for dogs containing serovars of Icterohaemorrhagiae and Australis have been marketed for use [56]. Meanwhile, trivalent (Canicola, Icterohaemorrhagiae, and Grippotyphosa) and tetravalent vaccines (Canicola, Icterohaemorrhagiae, Grippotyphosa, and Australis) have been licensed in European countries. The decision of in-cooperation incorporating serovars for optimal protection was made due to the observed shift in serovar prevalence in Europe and Australia with emerging serovar Australis [57].

Conclusions
Leptospirosis is not always diagnosed or being considered as a differential diagnosis in small animal practice, mostly due to the disease variable non-speci c clinical presentation. This study had shown that, one in three dogs presented with kidney and/or liver disease have the potential to be infected with leptospirosis and those patients with a history of in direct contact with rats have an increased risk. These results would help veterinarians to diagnose or even to consider leptospirosis as a differential diagnosis. The concurrent use of serological, molecular and isolation methods in this study allowed better disease understanding thus improving the diagnosis of leptospirosis especially in pet dogs diagnosed with kidney and/or liver disease. Proper clinical and laboratory diagnosis might increase the survivability of infected dogs as appropriate treatment can be administered immediately. The ndings in this study may indirectly increase the awareness of zoonotic risk for dog owners. The potential of zoonotic transmission of the leptospiral infection from dogs to their owners does exist due to the renal shedding of the bacteria and the close contact between humans and their pets, but the true extent or signi cance has still not been accurately investigated. To prevent the disease in pets and their owners, simple preventive measures can be applied focusing on reducing the chances of infection, including vaccination in dogs, good hygiene practices and avoid the exposure of their pet dogs to other infected animals and reservoir hosts. The detection of non-vaccinal serovars such as Bataviae and Javanica is a reminder of the limitation in vaccination. Proper characterisation of leptospiral isolates remain a crucial bottleneck to access the role of particular serovars or strains in the epidemiology of canine leptospirosis and may provide evidence-based knowledge to support the development and commercialisation of multivalent vaccines containing serovars that are circulating among local populations.

Sample collection and inclusion criteria
All the dogs diagnosed with kidney and/or liver disease either presented to the University Veterinary Hospital of Faculty of Veterinary Medicine, Universiti Putra Malaysia or from private veterinary clinics were recruited. Dog owner's permission was obtained prior to sample collection. Ethical approval was obtained from the Institutional Animal Care and Use Committee ( [58,59,60]. The recruited pet dogs were manually restraint for venipuncture (serum and whole blood) and urine samples were collected either via ultrasound guide cystocentesis or spontaneous micturition by experienced veterinarians. Blood samples collected using plain blood tube (BD Vacutainer®, USA) were allowed to clot at 4 °C then centrifuged for 10 minutes at 4000 rpm (80 − 2 Electronic Laboratory Medical Centrifuge, China). The serum was collected and stored at -20 °C for serological detection. Whole blood was collected in the blood tube containing anticoagulant ethylenediaminetetraacetic acid (EDTA) (BD Vacutainer®, USA) and urine samples were collected in sterile universal containers. Whole blood and urine samples were kept in the chiller at 4 °C until further analyses. After sample collection, the dog was monitored closely by the owner or veterinarian (if warded) and treated appropriately. If the dog was euthanised (procedure done by veterinarian using pentobarbitone intravenously) or found dead due to the sequelae of the disease, a necropsy was performed with permission. The kidney and liver were grossly examined and harvested. A section of the kidney and liver was cut into smaller pieces and homogenised with liquid Ellinghausen and McCullough modi ed by Johnson and Harris (EMJH) medium for bacterial isolation. If there was abdominal effusion, the uids were collected and stored appropriately.

Serological detection using microscopic agglutination test (MAT)
Using the serum that was obtained, MAT was performed to detect anti-leptospiral antibodies against a panel of 20 leptospiral serovars as shown in Table 3. The leptospiral serovars comprised of 19 pathogenic strains and one non-pathogenic strain (Patoc I). The endpoint titres were determined using Molecular detection using polymerase chain reaction (PCR) and partial 16S rRNA sequencing The whole blood, urine, abdominal effusion (if available) and tissue sample of kidney and liver were used to detect leptospiral DNA. DNA extraction from the samples and positive control of Leptospira interrogans serovar Canicola strain Hond Utrecht IV were performed using DNeasy® Blood & Tissue Kit (QIAGEN, Germany), as describe on the manufacturer's protocol. The end products (DNA template) were inspected using 1.5% agarose gel for purity.
With reference to previous studies [63,64], two sets of primers were selected and targeted the 16S rRNA and LipL32 genes (as shown in Table 4). Both genes were present in pathogenic Leptospira sp., but only 16S rRNA gene were present in the non-pathogenic Leptospira sp. [65]. The amplicons were analysed in tris-borate-EDTA (TBE) buffer at 80 volts for 1.5 hours by using 1.5% gel electrophoresis. The gel was pre-stained with SYBR® Safe DNA gel stain (Invitrogen™, North America) and examined using Gel Documentation (AlphaImager™, USA). The amplicons were identi ed by their band sizes. Figure 7 shows two bands at 541 bp and 756 bp were observed using the positive control of serovar Canicola.
After PCR analysis, the amplicons were subjected to a commercial facility for DNA sequencing together with a forward primer sequence. The sequencing data were compared to reference sequences deposited in GenBank using the BLAST tool (http://www.ncbi.nlm.nih.gov/BLAST/).
Isolation and characterisation of Leptospira spp.
Isolation of Leptospira spp.
The isolation of Leptospira spp. in this study was based on the protocol described by World Organization for Animal Health (OIE) [66]. Two drops of whole blood, urine, abdominal effusion and approximately 1 ml of homogenised tissue (kidney and liver) samples were inoculated into semisolid EMJH medium. The medium contained 200 µg/ml 5-uorouracil (5-FU). The inoculation of all the samples was performed within two hours after collection. Primary cultures were kept in the incubator at 30.0 °C for 12 weeks. The cultures were checked every fortnightly to observe for any presence of leptospires under the dark eld microscopy. The cultured medium was checked and sub-cultured into new semisolid EMJH medium for a few times to reduce contamination. If leptospires were observed within the 12 weeks, the positive cultures were transferred into liquid EMJH medium to enhance their growth and ltered by 0.45 µm (Millex®, Ireland) until pure isolates were obtained. The Leptospira spp. isolates were maintained in liquid EMJH medium and further characterised through serological and molecular characterisation. If negative for leptospires within the 12 weeks, the cultures were discarded after nal careful examination.
Serological characterisation of Leptospira spp. isolates The serovar of the isolates was determined by MAT using a panel of 18 hyperimmune sera as shown in Table 5. The isolates belonged to a particular leptospiral serovar when they reacted serologically to hyperimmune serum with speci c titre. The cut-off titre was based upon the reference value provided by Forensic and Scienti c Services, Department of Health, Leptospirosis Reference Laboratory, Queensland, Australia.
using Molecular Evolutionary Genetics Analysis Version 7.0 (MEGA7) and was inferred by using the Neighbour-Joining method with 1000 bootstrap based on General Time Reversible model [67].

Statistical analysis
Serological detection, molecular detection, and isolation of Leptospira spp. were represented using descriptive statistics with 95% con dence interval (CI) using IBM® SPSS® Statistics Version 23 (IBM®, USA). Mortality rate was calculated based on molecular detection and 95% CI was also applied.       Condensed tree generated from the original tree with 50.0% cut-off point