Infant rhesus macaques as a non-human primate model of Bordetella pertussis infection

Pertussis is an acute respiratory disease caused mostly by the gram-negative bacterium Bordetella pertussis (B.p). The basic illness is non-inflammatory in nature and occurs without significant fever. The disease is characterized by non-productive paroxysmal coughs followed by periods of total respiratory normalcy, which makes it different from all other infectious cough illnesses [1]. Severe respiratory failure is complicated by pulmonary hypertension, which may cause death, especially in infants [2]. Pertussis is transmitted directly from human to human, mostly via aerosolized respiratory droplets. Pertussis is a preventable disease, and its incidence decreased notably after vaccine immunization began. However, it has experienced a resurgence in several countries, even in countries with nearly universal vaccine coverage in the last 20 years [3‐5]. A deeper understanding of the mechanism of pertussis pathogenesis and the host response to both the pathogen and the vaccines is urgently needed to enable this important public health concern to be faced. Therefore, it’s urgent to develop suitable animal models for pertussis.

To establish animal models for pertussis, several studies have been carried out in mouse, rat, rabbit, and piglet models of pertussis [6, 7]. Unfortunately, these models have not been able to reproduce the full clinical spectrum observed in humans [8, 9]. In studies using non-human primate (NHP) models, a baboon model has been successfully established [10]. Low-grade fever, paroxysmal coughing, leukocytosis, a long-lived anti-pertussis toxin (PT) antibody response, protection against subsequent challenge, and transmission have been achieved in this baboon model, which makes the model crucial for studies on the pertussis pathogenic mechanism as well as for the development of new vaccines and therapeutics [11]. Another NHP, the rhesus macaque, has been evaluated for use as a pertussis model since 1929, but none of the studies have been able to completely replicate the human clinical disease [12‐14]. In contrast, 2 studies using Macaca (M.) cyclopis have investigated the similarity of the disease in these organisms to the human pertussis clinical syndrome [15, 16]. Baboons and macaques are Old World monkeys that were separated approximately 10 million years ago, and rhesus macaques are closely related to M. cyclopis. Several disease models have been established using macaques and/or baboons. While baboon models suffer from limited availability, high housing costs, and a lack of suitable reagents for use in these monkeys, rhesus macaques are more readily available and have low housing costs, and suitable reagents are available [17].

Pertussis is a vaccine-preventable childhood disease; however, there has been a resurgence in cases in recent years, including in countries with good vaccine immunization rates. A deep understanding of the immunology and epidemiology of this pathogen through studies on suitable experimental models, particularly NHP models, is important. Contrary to previous rhesus macaque challenge studies, in which most animals infected with pertussis have failed to develop obvious clinical manifestations of human pertussis [10, 14, 20], the present study achieved the pertussis clinical spectrum in a rhesus macaque model. First, the characteristic whooping cough of pertussis syndrome, which has thus far been achieved only in a baboon model, was confirmed to have developed. All 5 challenged macaques and 2 macaques cohoused with the challenged macaques developed severe coughs that persisted for over 4 weeks. The coughing appeared on day 2, peaked at day 15, and decreased gradually thereafter. At peak illness, the coughing became severe, lasting 10–20 s per episode. Second, the number of CFUs in the NPW peaked at day 10 after challenge, reaching 6.2 × 10⁶ CFU/mL, and returned to baseline levels after 35 days. Third, a higher serum antibody responses of PT-IgG than of FHA-IgG was observed in serum. This is the same dynamic antibody responses as that observed in human pertussis. In humans, symptomatic pertussis is characterized by an elevated serum PT antibody response, while asymptomatic infection is characterized by an elevated serum FHA antibody response [21]. Moreover, the antibody responses in NPW also indicated that both anti-PT IgA and anti-FHA IgA levels were significantly higher than the pre-challenge levels; these changes are also observed in humans, and measurement of these antibodies has been used as a sensitive method for the diagnosis of pertussis [22, 23].

Although rhesus macaque models have been successfully established for several infectious diseases, the models for B.p have been insufficient [12‐14]. Given the limited experimental details of the prior studies, which were performed in the 1920s–1950s, we speculated that the dose of infection, the method of infection, the age of the animal, and the infection strain may impact whether rhesus macaques can be successfully infected with B.p.

One of the most important factors for the pertussis model is the method of challenge. To date, nasal challenge, endotracheal intubation, in vivo injection, and aerosol challenge have been investigated in mouse models of pertussis [24, 25]. In 1929, Sauer, L.W. et al. [12] reported that typical pertussis manifestations were observed in 3 of 10 rhesus macaques after intranasal infection and in 5 of 18 Cebus monkeys after intralaryngeal infection. In 1935, Culotta, C.S. et al. [13] reported that 1 of 16 rhesus macaques were successfully infected with B.p via the intratracheal route, while none of five macaques were successfully infected via the intranasal route. However, the B.p strains and infection doses were not mentioned in these studies. We previously infected 3 rhesus macaques (4–5 years old) with strain 18,323 intranasally at a concentration of 5 × 10⁸ CFU/mL in a 2 mL volume, but none of these animals developed a cough or leukocytosis. Three animals were assigned randomly to euthanasia for scheduled necropsies at 1, 7, and 14 dpi. No pathological changes were found in the lungs, tracheas, or lung-draining lymph nodes. We observed only very low levels of bacteria in the trachea at 1 and 7 dpi, and no bacteria were observed at 14 dpi. However, no lung colonization was observed at the three scheduled time points. Recently, aerosol challenge has been used as a novel challenge method to accurately simulate natural infections and reduce animal stress [26]. Using an aerosol apparatus, we previously challenged mice with different concentrations for different periods of time and successfully established a B.p infection mouse model [27]. In the present study, using the aerosol apparatus, we achieved the whole spectrum of symptoms of pertussis. Compared to other methods of experimental infection, aerosol challenge has great value because it accurately simulates natural infections, exhibits superior reproducibility and results in predictable distribution of infection and pathology [28, 29].

The dose of infection should be considered during model establishment. In a recent B.p challenge experiment on wP-vaccinated adults, pertussis colonization exhibited dose dependence. The dose was gradually escalated from 10³ CFU (0% colonized) to 10⁵ CFU (80% colonized), while the minimum dose needed to induce colonization was 10⁵ CFU. In 1940, North, E.A. et al. [14] infected rhesus macaques with 2 mL of 5 × 10⁷ CFU/mL B.p via the intranasal route or with 1 mL of 1 mL 5 × 10⁷ CFU/mL B.p via the intratracheal route; however, the resultant infection was mild, and spasmodic cough and protective antibodies were not observed. Warfel, J.M. et al. [10] intranasally infected 4 rhesus macaques ranging in weight from 1.7 to 1.9 kg and ranging in age from 43 to 64 weeks with 0.5 mL of 10⁹ to 10¹⁰ CFU/mL B.p; 50% of the monkeys developed significant increases in WBC count, and one of the two monkeys with increased WBC counts developed a mild cough. However, since the infectious B.p strain and the animal age were not clear in North, E.A. et al.’s study, we cannot comprehensively compare our findings with those of North, E.A. et al. In the present study, infant monkeys were challenged with a recently clinically isolated strain via an aerosol apparatus in which the bacterial concentration was maintained at 10⁴–10⁵ CFU/mL for 60 min. The constant dose of B.p during the challenge may be one reason for the success of the present infectious pertussis model.

Another critical factor to consider when establishing an animal model is the age of the animals. A previous study using an enterovirus type 71 (EV71) rhesus macaque model showed that, of the challenged animals, a clinical spectrum similar to that of humans was observed only in young animals [30]. Moreover, studies have demonstrated that young baboons show severe disease signs, whereas adult baboons show mild signs [10]. In one baboon pertussis model, 5- to 6-week-old baboons all developed fatal pertussis; however, in juvenile baboons, the infection was not fatal [31]. Frequently, asymptomatic adults have been implicated in the spread of infection to susceptible children [32]. In our study, 5- to 6-month-old rhesus macaques were selected, and typical whooping cough, leukocytosis, bacteria-positive NPW, and transmission between animals were observed, similar to the results obtained in the baboon model. Therefore, we deduced that the challenge route as well as the age of the animals may influence B.p infection.

The strain of bacteria used for challenge is another factor that should be noted. Strain 18,323 was used successfully to establish pertussis in M. cyclopis models in the 1960s; however, it could not induce overt signs of disease in rhesus macaques [16]. A recent analysis of the global population structure of B.p indicated that strain 18,323 (genotype ptxP4/ptxA5/prn6/fim2–2/fim3–1) belongs to a branch containing a small number of strains that are evolutionarily far from the major prevalent branch and revealed that strain 18,323 diverged from the prevalent strain branch approximately 2000 years ago [33]. Thus, in contrast to previous macaque animal studies, our study used a strain recently isolated from a clinic in China, 2016-CY-41 (genotype ptxP1/ptxA1/prn1/fim2–1/fim3–1) (a common strain) and achieved the typical pertussis symptoms in infected macaques. Compared with that in the baboon model infected using strain D420 (genotype ptxP3/fim3–2), the peak symptom and disease progression in the macaque model infected using B.p strain 2016-CY-41 was delayed [10]. The prevalence of ptxP3, one of the major components of D420, has increased in many European counties, the US, and Australia in the past 25 to 30 years, rather than that of ptxp1; however, in China, ptxP1 has remained predominant [10, 34, 35]. A SNP in ptxP3 that lies in a binding site for the transcriptional regulator BvgA may result in a strong promoter and increase the level of transcription of the associated PT [36]. Strains harbouring the ptxP3 allele have been found to be more virulent than ptxP1 strains in a mouse infection model, and they may also be associated with severe disease in humans [36‐38]. Thus, we hypothesize that the genomic diversity of B.p may affect pertussis models. Further whole-genome sequencing experiments and analyses of virulence mechanisms and pertussis epidemiology should be performed in the future.

One of the possible reasons for the previous lack of success in establishment of rhesus macaque models of pertussis is high body temperature. The results of a temperature culture test in vitro showed that ACT protein levels are significantly lower in cells grown at 39 °C than in cells grown at 37 °C, supporting the hypothesis that high temperature (39 °C) may result in loss of ACT expression, in turn resulting in a lack of B.p infection [10]. However, pathogenic microorganisms cultured in vitro may also exhibit virulence factor loss due to the lack of host selective pressures. During non-random culturing, B.p can undergo spontaneous phase variation involving multistep disappearance of virulence factors in the following order: ACT, PT and FHA [39]. The in vitro results suggest that the reduced expression of ACT is caused by the elevated normal body temperatures of rhesus macaques [10]. In the present in vivo study, we observed a 20-fold increase in the anti-ACT antibody level, which was similar to the increase observed in the baboon model [40]. In addition, the rectal temperature was between 37.2 °C and 39.9 °C and did not exhibit significant changes. Thus, we deduce that body temperature may not be the only reason why previous rhesus macaque pertussis models have failed.

An infant rhesus macaque model of pertussis was established via aerosol challenge to provide a valuable alternative platform for research on pertussis pathogenesis and evaluation of vaccine candidates.