Antibody-based vaccine for tuberculosis: validation in horse foals challenged with the TB-related pathogen Rhodococcus equi

Immune correlates for protection against Mycobacterium tuberculosis (Mtb) infection and other intracellular pathogens are largely undetermined. Whether there is a role for antibody-mediated immunity is controversial. Rhodococcus equi is an intracellular pathogen causing severe pneumonia in young horse foals, eliciting a disease with many similarities to TB including intracellular residence, formation of granulomas and induction of severe respiratory distress. No purified vaccine antigens exist for R. equi or Mtb infections. Both express the microbial surface polysaccharide antigen poly-N-acetyl glucosamine (PNAG). In a randomized, controlled, blinded challenge trial, vaccination of pregnant mares with a synthetic PNAG oligosaccharide conjugated to tetanus toxoid elicited antibody that transferred to foals via colostrum and provided nearly complete protection against R. equi pneumonia. Infusion of PNAG-hyperimmune plasma protected 100% of foals against R. equi pneumonia. Vaccination induced opsonic antibodies that killed extracellular and intracellular R. equi and other intracellular pathogens. Killing of intracellular organisms was dependent on antibody recognition of surface expression of PNAG on infected macrophages, complement deposition and PMN-assisted lysis of infected macrophages. Protection also correlated with PBMC release of interferon-γ in response to PNAG. Antibody-mediated opsonic killing and interferon-γ release in response to PNAG may protect against disease caused by intracellular bacterial pathogens.


Abstract: 19
Immune correlates for protection against Mycobacterium tuberculosis (Mtb) infection 20 and other intracellular pathogens are largely undetermined. Whether there is a role for 21 antibody-mediated immunity is controversial. Rhodococcus equi is an intracellular 22 pathogen causing severe pneumonia in young horse foals, eliciting a disease with many 23 similarities to TB including intracellular residence, formation of granulomas and primates, have not led to an effective human vaccine (2,3) outside of the limited 46 efficacy of the Bacillus Calmette-Guerin whole-cell vaccine (3)(4)(5). Rhodococcus equi is a 47 Gram-positive, facultative intracellular pathogen that primarily infects alveolar 48 macrophages of foals following inhalation, resulting in a granulomatous pneumonia that 49 is pathologically similar to that caused by Mtb in humans (6). Like Mtb, R. equi also 50 causes extrapulmonary disorders including osseous and intra-abdominal lymphadenitis 51 (6)(7)(8). The disease is of considerable importance to the equine industry (6,8), and has 52 relevance to human health both as a cause of granulomatous pneumonia (9), and as a 53 platform for studying Mtb infections, due to the similar pathological and clinical findings 54 (6)(7)(8)10). Additionally, both Mtb and R. equi synthesize the conserved surface capsule-55 like polysaccharide, poly-N-acetyl glucosamine (PNAG), a target for the development of 56 a broadly-protective vaccine against many pathogens (11,12). These findings indicate 57 that a study evaluating vaccination against PNAG for induction of protective immunity to 58 R. equi challenge in foals could provide critical information and support for testing a 59 PNAG vaccine or PNAG-specific monoclonal antibody against human Mtb infections. 60 To justify the premises underlying a R. equi vaccine study, multiple considerations infections, and properties of natural and vaccine-induced immunity to PNAG had to be 63 taken into account. Many investigators consider effective immunologic control of these 64 intracellular pathogens to be primarily based on cell-mediated immune (CMI) responses 65 (13), a finding clearly in evidence for R. equi. Disease occurs almost exclusively in foals 66 less than 6 months of age, but by ~ 9 months of age most young horses become highly 67 resistant to this pathogen (6,8). This acquired natural resistance is obviously not 68 antibody-mediated inasmuch as the solid immunity to infection in healthy horses > 9 69 months of age, which obviously includes pregnant mares, is not transferred to 70 susceptible foals via antibody in the colostrum. Colostrum is the only source of 71 maternal antibody in foals and the offspring of other animals producing an 72 epitheliochorial placenta. Human development of TB often correlates with defective 73 CMI (1,14), particularly in those infected with the human immunodeficiency virus (14, 74 15). Nonetheless, considerable evidence has emerged to indicate that antibodies to Mtb 75 have the potential to be major mediators of protective immunity (16)(17)(18)(19)(20). Both Mtb and 76 R. equi survive within alveolar macrophages and induce granulomas, supporting the 77 use of R. equi as a relevant model for TB pathogenesis, immunity, and vaccine 78 development. 79 In regards to targeting the conserved PNAG surface polysaccharide, the production 80 of this antigen by many microbes induces natural IgG antibody in most humans and 81 animals (21,22), but natural antibody is generally ineffective at eliciting protection 82 against infection. These antibodies do not activate the complement pathway and cannot 83 mediate microbial killing (21)(22)(23). By removing most of the acetate substituents from the for the complement-fixing antibody to PNAG, which, along with added PF, representing the serum sample obtained after the booster immunization. Similarly, 132 total IgG and IgG subisotype titers were significantly higher in the colostrum obtained on 133 the day of foaling from vaccinated mares compared with controls (Fig. S2). Notably, 134 non-immunized mares had antibody titers to PNAG, representative of the natural 135 response to this antigen commonly seen in normal animal and human sera. 136 Successful oral delivery of antibody to the blood of foals born to vaccinated mares 137 (hereafter termed vaccinated foals) was shown by the significantly higher titers of serum 138 IgG to PNAG compared with foals from control mares at ages 2, 28, and 56 days, but 139 not 84 days (Fig. 1A). Foal serum concentrations of subisotypes IgG 1 , IgG 3/5 , and IgG 4/7 140 to PNAG were significantly higher at 2, 28, and 42 days of age (after colostral transfer) 141 in the vaccinated group compared with the control group, and subisotype IgG 1 (34)). 184

R. equi expression of PNAG in vitro and in vivo 185
Using immunofluorescence microscopy, we demonstrated that 100% of 14 virulent 186 strains of R. equi tested express PNAG (Fig. S4). Moreover, we found that PNAG was 187 expressed in the lungs of foals naturally infected with R. equi (Fig. S5a), similar to our 188 prior demonstration of PNAG expression in the lung of a human infected with Mtb (11).  (Fig. 3A). Notably, the natural 196 antibody to PNAG in sera of non-vaccinated, control mares and their foals did not 197 deposit C1q onto the PNAG antigen, consistent with prior findings that natural 198 antibodies are immunologically inert in these assays (21,22,35). Sera from vaccinated 199 foals on the day of R. equi infection mediated high levels of opsonic killing of extracellular R. equi whereas control foals with only natural maternal antibody to PNAG 201 had no killing activity (Fig. 3B), again demonstrating the lack of functional activity of 202 these natural antibodies to PNAG. 203 As some of the vaccinated foals developed small subclinical lung lesions that 204 resolved rapidly (Table S1, Fig. 2 infected cells (Fig. S6B). However, we found strong expression of PNAG both on the 220 infected MDM surface and within infected cells (Fig. S6C). Similarly, using a GFP- To show that antibody to PNAG, complement, and PMN represent a general 262 mechanism for killing of disparate intracellular pathogenic bacteria that express PNAG, 263 we used the above-described system of infected human macrophages to test killing of 264  Fig. 5B and fig. S9), although when analyzing data from all 5 of these 274 experiments combined there was a modest but significant release of LDH release with 275 antibody to PNAG and complement alone ( Fig. 5B and fig. S9). 276 To substantiate the specificity of this CMI reaction, we demonstrated that stimulation 290 of horse PBMC from vaccinated foals with an R. equi lysate treated with the enzyme 291 dispersin B diminished IFN-γ responses by ~90% (Fig. 6B). We also made a post hoc comparison of CMI responses between foals that remained healthy and foals that 293 developed pneumonia. In this analysis (Fig. 6C), foals that remained healthy (11 294 vaccinates and 1 control) had significantly (P < 0.05; linear mixed-effects modeling) 295

Maternal PNAG vaccination and antibody transfer to foals enhances in vitro
higher CMI responses at all ages, including age 2 days, than foals that became ill (1 296 vaccinate and 6 controls), suggesting that both innate and acquired cellular immunity 297 contribute to resistance to R. equi pneumonia. Overall, it appears the maternally 298 derived antibody to PNAG sensitizes foal PBMC to recognize the PNAG antigen and 299 release IFN-γ, which is a known effector of immunity to intracellular pathogens. 300 301

Discussion 302
In this study we showed maternal immunization against the deacetylated glycoform of 303 the conserved microbial surface polysaccharide PNAG induced antibodies that 304 protected 11 of 12 (91%) ~4-week-old foals from challenge with live, virulent R. equi. 305 The attack rate in the controls was 86% (6/7). A confirmatory passive transfer study 306 similarly showed antibody to PNAG infused into foal blood on the day of birth protected  (Table S1) due to R. equi whereas 10 of 11 non-immune 312 controls had a clinical diagnosis of pneumonia (P < 0.0001, Fisher's exact test; 313 protected fraction 90%, (95% C.I. 59%-98%, Koopman asymptotic score (34)). 314 Vaccine-induced antibody to PNAG deposited complement component C1q onto the purified PNAG antigen, mediated opsonic killing of both extracellular and intracellular R. 316 equi, and sensitized PBMC from vaccinated foals to release IFN-γ in response to PNAG. 317 The mechanism of killing of intracellular PNAG-expressing microbes was dependent 318 upon surface expression of this antigen, presumably intercalated into the plasma 319 membrane of the infected host cell following release and intracellular transport of 320 microbial surface vesicles (37). This mechanism of killing was shown to be applicable several weeks to months to develop (6,7). Pulmonary acquisition of Mtb followed by a 333 latent period prior to the emergence of disease occurs in humans exposed to this 334 pathogen (40). Thus, it cannot be predicted with certainty that the protective efficacy of 335 antibody to PNAG manifest in the setting of acute, bolus challenge will also be effective 336 when a lower infectious inoculum and more insidious course of disease develops. In this 337 setting, activation of immune effectors may not be of sufficient intensity to take advantage of the opsonic killing activity of antibody, thus allowing progression to 339 disease to occur. In the context of acute challenge, we noted that many of the 340 protected vaccinated foals developed small lung lesions after challenge that rapidly 341 resolved and no disease signs were seen. Finding such lesions by routine ultrasound 342 examination of foals that occurs on farms (41)  The use of animal models to predict vaccine efficacy in humans is fraught with 347 uncertainty, even when non-human primates are used (42)(43)(44). Thus, whether the 348 protective efficacy shown here for R. equi in foals is predictive of efficacy against human 349 infection with Mtb is unknown, even though these 2 organisms share significant 350 pathogenic properties. One difference between these microbes is their growth rates, 351 and this may impact the effectiveness of PNAG immunization of humans. R. equi grows 352 rapidly in laboratory culture but Mtb slowly. Also, while R. equi, Mtb, and other 353 pathogenic mycobacteria are reasonably closely related genetically, there are likely 354 some differences in surface composition and antigens that might impact the ability of 355 antibody to PNAG to kill mycobacteria, thus limiting the utility of this vaccine for human 356 Mtb. However, we did find that the antibody to PNAG could kill a variety of human 357 pathogens inside of infected macrophages, including a rapidly-growing mycobacterial 358 pathogen, dependent on antibody recognition of PNAG antigen intercalated into the 359 infected cell's plasma membrane. This finding not only provides an explanation for how 360 antibody to PNAG can mediate protective immunity to intracellular pathogens, but also supports the potential of antibody to PNAG for providing broad-based protective 362 immunity to many intracellular pathogens. 363 The protection studies described here for R. equi disease in foals has led to the 364 implementation of a human trial evaluating the impact of infusion of the fully human IgG 1 365 MAb to PNAG (21) on latent and new onset TB. The MAb has been successfully tested 366 for safety, pharmacokinetic, and pharmacodynamic properties in a human phase I test 367 Numerous investigators have studied how antibodies can mediate protection 382 against intracellular bacterial pathogens like TB (16,19,46), although specific 383 mechanisms of immunity are not well defined. The in vitro results we derived indicated that a cell infected with a PNAG-producing pathogen has prominent surface display of 385 this antigen that serves as a target for antibody, complement and PMN to lyse the 386 infected cell and release the intracellular organisms for subsequent opsonic killing. 387 Likely other bacterial antigens are displayed on the infected host cell as well, and thus 388 this system could be used to evaluate the protective efficacy and mechanism of killing of 389 antibodies to other antigens produced by intracellular organisms. Although we have not 390 investigated the basis for the appearance of PNAG in the plasma membrane of infected 391 host cells, we suspect that microbial extracellular vesicles, known to be released by 392 many microbes including Mtb (47), are a likely source of the plasma membrane antigen 393 due to trafficking from infected cellular compartments (37). indicating that an antibody-dependent cellular response to PNAG underlay the IFN-γ 399 response. As this cytokine is well known to be an important component of resistance to 400 human TB (13), it was notable that the maternal immunization strategy led to an IFN-γ 401 response. It also appears that the reliance on traditional T-cell effectors recognizing 402 MHC-restricted microbial antigens to provide components of cellular immunity can 403 potentially be bypassed by an antibody-dependent mechanism of cellular responses, 404 further emphasizing how antibody can provide immunity to intracellular pathogens. 405 intracellular pathogen via colostrum to immunologically immature offspring, the efficacy 408 and mechanism of action of antibody to PNAG in protective efficacy, and identification 409 of a role for antibody-dependent IFN-γ release in the response to immunization that 410 likely contributed to full immunity to challenge. The success of immunization in 411 protecting against R. equi challenge in foals targeting the broadly synthesized PNAG 412 antigen raises the possibility that this single vaccine could engender protection against 413 many microbial pathogens. While the potential to protect against multiple microbial 414 targets is encouraging, the findings do raise issues as to whether antibody to PNAG will 415 be protective against many microbes or potentially manifest some toxicities or 416 unanticipated enhancements of infection caused by some organisms. Thus, continued 417 monitoring and collection of safety data among animals and humans vaccinated against 418 PNAG is paramount until the safety profile of antibody to PNAG becomes firmly 419 established. Overall, the protective efficacy study in foals against R. equi has initiated 420 the pathway to development of PNAG as a vaccine for significant human and animal 421 pathogens, and barring unacceptable toxicity, the ability to raise protective antibodies to 422 PNAG with the 5GlcNH 2 -TT conjugate vaccine portends effective vaccination against a 423 very broad range of microbial pathogens.

Materials and Methods 425
Experimental Design 426 The objective of the research was to test the ability of maternal vaccination of horse 427 mares with a conjugate vaccine targeting the PNAG antigen to deliver, via colostral 428 transfer, antibody to their offspring that would prevent disease due to intrabronchial 429 R. equi challenge at ~4 weeks of life. A confirmatory study using passive infusion of 430 immune or control horse plasma to foals in the first 24 hours of life was also undertaken. 431 The main research subjects were the foals; the secondary subjects were the mares and 432 their immune responses. The experimental design was a randomized, controlled, 433 experimental immunization-challenge trial in horses, with pregnant mares and their foals 434 randomly assigned to the vaccine or control group. Group assignment was made using 435 a randomized, block design for each year. Data were obtained and processed 436 randomly then pooled after unblinding for analysis. Investigators with the responsibility 437 for clinical diagnosis were blinded to the immune status of the foals. An unblinded 438 investigator monitored the data collected to ascertain lack of efficacy and stopping of 439 the infections if 5 or more vaccinated foals developed pneumonia. A similar design was 440 used for the transfusion/passive infusion study, except for the stopping rule. 441

Samples size determination 442
The sample size for the foal protection study was based on prior experience with 443 this model 6,30,48 indicating a dose of 10 6 CFU of R. equi delivered in half-portions to the 444 left and right lungs via intrabronchial instillation would cause disease in ~85% of foals. 445 Thus, a control group of 7 foals, anticipating 6 illnesses, and a vaccinated group of 12 446 foals, would have the ability to detect a significant effect at a P value of <0.05 if 75% of vaccinated foals were disease-free using a 1-sided Fisher's exact test. A 2-sided test is 448 not feasible as one cannot realistically measure a disease rate in vaccinated foals 449 significantly greater than 85%, and any failure to reduce the disease rate would not be 450 indicative of vaccine efficacy. Thus, lack of reduction in disease would lead to rejection 451 of the hypothesis that vaccination against PNAG is effective in preventing R. equi 452 pneumonia. Similar criteria were applied to the passive infusion/protection study. All Samples were washed and mounted for immunofluorescent microscopic examination as 625 described 11 . 626 and blocked with skim milk as described above, dilutions of different horse sera added 655 in 50 µl-volumes after which 50 µl of 10% intact, normal horse serum was added. After 656 60 minutes incubation at 37 o C, plates were washed and 100 µl of goat anti-human C1q, 657 which also binds to equine C1q, diluted 1:1,000 in incubation buffer added and plates 658 incubated at room temperature for 60 minutes. After washing, 100 µl of rabbit anti-goat 659

Analysis of PNAG expression in infected horse tissues and human monocyte-
IgG whole molecule conjugated to alkaline phosphatase and diluted 1:1,000 in 660 incubation buffer was added and a 1-hour incubation at room temperature carried out. 661 Washing and developing of the color indicator was then carried out as described above, 662 and endpoint titers determined as described above for IgG titers by ELISA.
Immune (10) Immune-No C' were used as controls, as were tubes lacking PMN or complement (C') as indicated.
Bars represent means of technical replicates.