Attenuation of a Pathogenic Mycoplasma Strain by Modification of the obg Gene by Using Synthetic Biology Approaches

Animal diseases due to mycoplasmas are a major cause of morbidity and mortality associated with economic losses for farmers all over the world. Currently used mycoplasma vaccines exhibit several drawbacks, including low efficacy, short time of protection, adverse reactions, and difficulty in differentiating infected from vaccinated animals. Therefore, there is a need for improved vaccines to control animal mycoplasmoses. Here, we used genome engineering tools derived from synthetic biology approaches to produce targeted mutations in the essential GTPase-encoding obg gene of Mycoplasma mycoides subsp. capri. Some of the resulting mutants exhibited a marked temperature-sensitive phenotype. The virulence of one of the obg mutants was evaluated in a caprine septicemia model and found to be strongly reduced. Although the obg mutant reverted to a virulent phenotype in one infected animal, we believe that these results contribute to a strategy that should help in building new vaccines against animal mycoplasmoses.


FIG 1
Global strategy used to produce M. mycoides subsp. capri obg mutants. (A) Residues targeted in the M. mycoides subsp. capri Obg protein during this study. The amino acid changes targeted in this study (Gly80Glu, Asp85Asn, and Gly124Arg) were all located in the N-terminal end of the Obg protein. The Gly80Glu (mutant 7.1) and Asp85Asn (mutant 11.1) substitutions were produced using the nucleotide changes GGTϾGAG and GATϾAAC, respectively. The Gly124Arg substitution was produced using two different changes in the nucleotide sequence, either GGGϾAGA (mutant 15.2) or GGGϾCGT (mutant 69.3). Unexpectedly, one clone (43.7) from the plate for obtaining the Gly124Arg (GGGϾAGA) substitution showed three other mutations, two in the obg gene, Ile244Val (TATϾTGT) and Asp85Gly (GATϾGGT), and one in the neighboring nadE gene, Gln7Arg (CAAϾCGA). The latter mutation was repaired and resulted (Continued on next page) Lartigue et al. sequenced to confirm the expected substitutions on the obg gene. Altogether, six M. mycoides subsp. capri obg mutants were finally selected for phenotypic characterization (Fig. 1B, step 3), with four presenting a single-amino-acid substitution (mutants 7.1, 11.1, 15.2, and 69.3) and two presenting multiple substitutions (mutants 43.7 and 9.2) ( Fig. 1A and Table 1).
Phenotypic characterization of the M. mycoides subsp. capri obg mutants. The impact of the above-mentioned obg substitutions on M. mycoides subsp. capri growth was evaluated by combining culture assays and quantitative PCR (qPCR) analyses. All M. mycoides subsp. capri obg mutants were cultivated in standard SP5 mycoplasma medium at 32°C and 40°C together with the parental clone 30.1 (YCpMmyc1.1 with a wild-type [WT] obg region) as a control. The color change of the medium was first observed daily during 8 consecutive days (48-h and 144-h time points are shown in Fig. 2). Similarly to WT M. mycoides subsp. capri strain GM12 (data not shown), the control clone 30.1 grew slightly faster at 40°C than at 32°C and reached a maximum titer (Ͼ10 9 color-changing units [CCU] · ml Ϫ1 ) after 48 h of incubation at 40°C. Unexpectedly, the M. mycoides subsp. capri obg mutants 11.1, 15.2, and 69.3 with the FIG 1 Legend (Continued) in mutant 9.2. (B) Scheme of the general experimental strategy followed in this study. The YCpMmyc1.1 genome was previously cloned into the yeast strain VL6-48N (7). The replacement of the essential obg gene in the YCpMmyc1.1 genome was achieved using TREC-IN, a three-step method allowing the seamless knock-in of the target gene(s) on bacterial genomes cloned into yeast (10). After modification, M. mycoides subsp. capri (Mmc) obg-mutated genomes (YCpMmyc1.1-obgmod) were isolated from yeast and transplanted into restriction-free M. capricolum subsp. capricolum recipient cells (McapΔRE) to obtain M. mycoides subsp. capri obg mutants characterized by mutation at a precise position within the obg gene. Mutants and filter-cloned derivatives were characterized in vitro (genotypic and phenotypic analyses). The M. mycoides subsp. capri obg subclone that showed the most temperature-sensitive phenotype (t ϩ ) was selected for animal experiments using goats. single-amino-acid changes Asp85Asn (GATϾAAC) and Gly124Arg (GGGϾAGA or GGGϾCGT) grew as well as the control at both temperatures tested and therefore did not show a TS ϩ phenotype (Fig. 2). In contrast, the mutants (7.1, 43.7, and 9.2) with the single mutation Gly80Glu (GGTϾGAG) and with the multiple mutations Asp85Gly (GATϾGGT), Gly124Arg (GGGϾAGA), and Ile244Val (TATϾTGT), with or without a mutation in nadE (Gln7Arg, CAAϾCGA), showed a marked TS ϩ phenotype. Clone 43.7 was the most affected, with a 4-log lag behind the control clone 30.1 still after 144 h of culture at 40°C (Fig. 2). As the color change of the medium does not precisely reflect the cell concentration in a mycoplasma culture, we performed 16S rRNA gene qPCR analyses during the in vitro culture. While the qPCR curves confirmed the results obtained by the color change culture assay (Fig. 3), they provided a more precise measure of mycoplasma growth. At 32°C, we observed that the curves were very similar between the M. mycoides subsp. capri control clone 30.1 and the different mutants tested (Fig. 3A). All bacterial cultures reached their stationary phase after 72 h of incubation, except for mutant 43.7, which was slightly delayed and reached its stationary phase after 96 h (24 h later). At 40°C (Fig. 3B), no growth difference was observed between mutants 11.1, 15.2, and 69.3 and the control strain 30.1. They all reached the stationary phase at the 48-h time point (T 48 ), confirming that M. mycoides subsp. capri grows faster at 40°C than at 32°C, but an ϳ1-log decrease in the final bacterial concentration was consistently measured (ϳ10 8 cells · ml Ϫ1 at 40°C versus ϳ10 9 cells · ml Ϫ1 at 32°C) ( Fig. 3A and B). Interestingly, the growth curves of three other mutants did not follow the trend of the control strain. At 40°C, the doubling time of mutants 7.1 and 9.2 appeared to be greatly affected, since these cells reached the stationary phase at T 120 and T 144 , respectively. In addition, mutant 43.7 did not appear to grow at all, since the bacterial concentration remained consistently below 10 3 cells · ml Ϫ1 (Fig. 3B), even after 144 h of incubation. However, according to Fig. 2, a color change from red to yellow could be observed for this mutant up to a 10 Ϫ5 dilution, indicating a residual ability to grow. Consistent with this result, the qPCR measurements of the in vitro culture done with the 10 Ϫ8 dilution did not reveal any growth (Fig. 2, dotted rectangles). Based on these results, mutants 7.1, 9.2, and 43.7 were selected for further analysis.
Genotypic and phenotypic analyses of the selected M. mycoides subsp. capri obg subclones. In order to eliminate the possibility of having heterogeneous populations, the three mutants (7.1, 9.2, and 43.7) were submitted to three rounds of filter cloning (Fig. 1B). After this procedure, a total of six subclones were selected for each of the three selected mutants and cultured in SP5 medium supplemented with tetracy- cline prior to genotypic analysis. For mutants 7.1 and 43.7, all subclones showed an obg sequence identical to that of the parental clones. However, for mutant 9.2, the sequencing data of the 6 subclones revealed some sequence heterogeneity. While 4 subclones carried the expected the GATϾGGT (Asp85Gly), GGGϾAGA (Gly124Arg), and TATϾTGT (Ile244Val) substitutions, 2 others (9.2.2 and 9.2.6) presented a different mutation in one of the targeted codons, GATϾAGT (Asp85Ser). This suggested the coexistence of two populations in the initially selected 9.2 mutant. Consequently, subclones 9.2.2 and 9.2.6 with an unwanted additional mutation were not analyzed further.
Phenotypic analysis of the subclones was performed as described above, using two subclones from each parental mutant (Table 1). After evaluating the temperature sensitivity phenotype of the subclones by evaluating their growth at two temperatures ( Fig. S5), data from the phenotypic analysis of the subclones that showed the most pronounced phenotype (43.7.3E, 43.7.4B, 9.2.1A, and 9.2.3B) were confirmed using qPCR assays ( ). This discrepancy, most likely caused by the presence of a mixed population in mutant 9.2, suggested that the mutation GATϾAGT (Asp85Ser) detected in some 9.2 subclones probably does not confer a TS ϩ phenotype.
Whole-genome sequencing of subclones derived from the 30.1 control and from the 7.1, 9.2, and 43.7 mutants indicated a number of mutations other than those in the obg gene, but none of these correlated with the TS ϩ phenotype observed (Table S2).
Evaluation of the virulence of the 9.2.3B obg subclone in an experimental animal model. Among the four M. mycoides subsp. capri obg subclones (43.7.3E, 43.7.4B, 9.2.1A, and 9.2.3B) that were characterized as described above, subclone 9.2.3B, which carried 3 mutations in the obg gene and showed the strongest TS ϩ phenotype, was selected for subsequent in vivo testing of attenuation using a caprine infection model.
Two groups of eight male goats, housed in separated rooms, were infected transtracheally with either M. mycoides subsp. capri YCpMmyc1.1 (22, 23) (control group) or M. mycoides subsp. capri obg subclone 9.2.3B (obg group). After two injections, one at 0 days postinfection (dpi), with 10 8 CFU, and the other at 7 dpi, with 10 9 CFU, animal health status was monitored daily by measuring key parameters, such as body weight, heartbeat, breathing frequency, and body temperature (Fig. 4). The following endpoint criteria for animal euthanasia were applied: a body temperature higher than 40.5°C for more than 3 consecutive days, the appearance of clinical signs indicating moderate to severe pain or distress, weight loss of more than 10% within 7 days, and a breathing frequency higher than 50 breaths · min Ϫ1 for more than 3 days. In the control group, 5 goats out of 8 met the endpoint criteria before the envisaged end of the trial (i.e., 35 dpi) and were euthanized in the course of the experiment, whereas only 1 goat (animal CM135) out of 8 met the same criteria in the obg group ( Fig. 4A) (24). M. mycoides subsp. capri was cultured from 3 out of 8 animals that were infected with the control strain YCpMmyc1.1. The time points of M. mycoides subsp. capri isolation coincided with severe pyrexia. Bacterial titers ranged between 100 and 1,000 CFU · ml Ϫ1 (24). Animals that did not develop the disease had body temperatures ranging between 38°C and 39.5°C, which represents the physiological range for goats (Fig. 4B). Within the group infected with M. mycoides subsp. capri YCpMmyc1.1, those animals showing signs of disease developed fever (rectal temperature of Ͼ39.5°C) between days 3 and 10 after the first injection (Fig. 4B), and animal CM135 from the obg group The difference in survival rates (Kaplan-Meier curve) is significant according to both the Mantel-Cox test (P ϭ 0.0351) and the Gehan-Breslow-Wilcoxon test (P ϭ 0.0325). There were 8 goats per group. (B) Rectal temperature measured daily, shown as an average per group. The difference in average temperatures between the groups was significant on days 9 and 10 postinfection (day 9, P ϭ 0.0006; day 10, P ϭ 0.0053) using the Mann-Whitney test, coinciding with the peak of severity of disease in the group that received the M. mycoides subsp. capri YcpMmyc1.1 strain. (C) The white blood cell (WBC) count (measured twice weekly) was relatively stable throughout the trial, and no major difference between the groups was observed at any time point. (D) Both groups gained weight, measured three times weekly, continuously during the trial. (E) Respiratory rate, measured daily, remained within the normal range throughout the study. (F) Heart rate was measured daily and remained within the normal range for both groups. Arrows indicate days of infection (day 0, 10 8 CFU per animal; day 7, 10 9 CFU per animal). All graphs show the average measured parameters per group, with error bars showing SD.
showed fever starting at 13 dpi (data not shown). In most of the animals showing fever, body temperature reached 41°C in less than 3 days. In the obg group, the heart rate and breathing rate remained within the normal ranges for all animals, including animal CM135, throughout the study (Fig. 4E and F). We cultivated M. mycoides subsp. capri from the blood of the animals of the obg group, and only blood from the animal (CM135) that succumbed to disease gave a positive culture indicating bacteremia, with a titer of 10 3 CCU · ml Ϫ1 . Cultivation of M. mycoides subsp. capri from synovial fluid, urine, and pleural fluid of the same animal did not yield any positive cultures.
Necropsy of animals from the control group revealed severe and extensive necrosis, edema, and inflammation of the skin, subcutis, skeletal muscle of the neck, and the trachea around the site of inoculation in five animals. Histologically, there was extensive coagulation necrosis of these tissues surrounding the trachea in the vicinity of the inoculation site and a marked infiltration of neutrophilic granulocytes (Fig. S6a). These areas were surrounded by a rim of infiltration with mainly macrophages and lymphocytes. Regional lymph nodes showed purulent to necrotizing lymphadenitis, and there were multifocal neutrophilic infiltrates in the spleen. In one animal, there were multifocal, randomly distributed small areas of necrosis in neutrophilic infiltrates in the liver. These lesions were suggestive of septicemic spread of M. mycoides subsp. capri from the site of inoculation. Three animals of this group (animals CM045, CM047, and CM191) did not show macroscopic or histological lesions. Animal CM135 in the obg group had lesions similar to those in the control group (Fig. S6b). Overall, the data clearly demonstrated that even though one animal in the obg group developed disease and had to be euthanized, the virulence of M. mycoides subsp. capri obg subclone 9.2.3B was attenuated compared to its parental strain.
Genotypic analysis of the revertant iCM135. In order to find out the reason for the loss of one animal in the obg group (Fig. 4A), the mycoplasma strain isolated from blood samples of the euthanized animal (iCM135 strain) was subjected to further analyses. Whole-genome sequencing showed that iCM135 carried a single reversion in one of the target mutations within the obg gene. Indeed, compared to the original TS ϩ M. mycoides subsp. capri obg subclone 9.2.3B, the GATϾGGT (Asp85Gly) triplets reverted back to GAT (Gly85Asp) in the iCM135 strain. The temperature sensitivity of iCM135 was analyzed in vitro using culture assays at 32°C or 40°C along with the M. mycoides subsp. capri control subclone 30.1.1 (intact obg). The experiment clearly showed that the iCM135 strain lost its temperature sensitivity and recovered a nontemperature-sensitive (TS Ϫ ) phenotype that is similar to that of the parental strain (Fig. S7) and that the loss of the TS ϩ phenotype is probably due to the reversion observed at position 85 (Table S2). It should be stressed that a similar reversion of phenotype associated with reversion at position 85 was also observed in vitro during culture of subclone 9.2.1C at 40°C (Fig. S7).
Experimental challenge following infection with the 9.2.3B obg subclone. Since seven of eight goats infected with the 9.2.3B obg subclone survived after 35 dpi, we evaluated whether infection in these animals had triggered an immune response that would provide a certain degree of protection against an experimental challenge. The animals were challenged transtracheally with the M. mycoides subsp. capri GM12 strain (25). It should be noted that this strain is different from the M. mycoides subsp. capri YCpMmyc1.1 strain used as the control in the above-described experiment because it has not been submitted to a round of yeast cloning-genome transplantation followed by triple filter cloning. Therefore, the M. mycoides subsp. capri GM12 strain, at a low passage number, is known for its high pathogenicity, causing septicemia and death when administered at a high dose (27). At 0 days postchallenge (dpc), the seven animals were infected transtracheally with 10 9 CFU of the M. mycoides subsp. capri GM12 strain, and animal health status was monitored daily for 28 days by measuring parameters such as body weight, heartbeat, breathing frequency, and body temperature (Fig. 5). Between 4 dpc and 10 dpc, 6 out of the 7 goats met the endpoint criteria and were euthanized, with only 1 animal (CM118) surviving until the end of the experiment (Fig. 5A). As expected with this highly pathogenic and septicemic strain of M. mycoides subsp. capri, infection resulted in a peak body temperature reaching Ն41°C for 5 out of the 6 animals that were euthanized. The M. mycoides subsp. capri challenge strain GM12 was isolated from the blood of 6 of the 7 animals. The bacterial titers ranged from 10 4 to 10 10 CFU · ml Ϫ1 . The only animal for which the cultures remained negative was animal CM118, which survived until the end of the experiment, at 30 days postchallenge. Other clinical parameters were also affected although not uniformly among these animals (Fig. 5C to F).
Pathological lesions in 6 out of 7 of the M. mycoides subsp. capri GM12-inoculated animals were similar to those of the above-described control group, with massive necrosis and mainly neutrophilic inflammation in the soft tissues and the trachea of the neck extending from the inoculation site (Fig. S6c). All animals showed histopathological signs of septicemia, characterized as multifocal acute necrosis and neutrophilic infiltration in the spleen. In addition, three animals developed mild, purulent bronchopneumonia. Animal CM118 did not show any macroscopic or histological lesions.

DISCUSSION
Using synthetic biology approaches, the essential gene obg encoding a P-loop GTPase was mutated on the M. mycoides subsp. capri genome, cloned in yeast, and subsequently transplanted into Mycoplasma capricolum recipient cells to obtain a set of M. mycoides subsp. capri mutants carrying various mutations. The targeted modifications of obg reproduced mutations that are known to be associated with a temperature-sensitive phenotype in various bacteria, including in a commercial live vaccine of M. synoviae.
Among the single obg mutations that were previously associated with a TS ϩ phenotype in E. coli, B. subtilis, and M. synoviae, the only one that showed a significant TS ϩ phenotype in M. mycoides subsp. capri is Gly80Glu (mutant 7.1) (Fig. 2). This mutation has been described in detail in Caulobacter crescentus and was also characterized by a marked TS ϩ phenotype associated with alterations in the cell cycle (21). In contrast, the single mutation described in the TS ϩ vaccine strain of M. synoviae (Gly124Arg) did not, by itself, produce a TS ϩ phenotype in M. mycoides subsp. capri. Recently, complementation of the Gly124Arg substitution in M. synoviae with a wildtype copy of the obg gene did not result in a TS Ϫ phenotype (10), but some of the growth parameters, including growth rate and cell viability, were at least partially restored; it should be stressed that in this case, complemented clones expressed both wild-type Obg and the protein with the Gly124Arg substitution. In the obg mutants with more than one mutation, the complex interplay among them is probably due to structure constraints. For example, the reversion at position 85 (Gly85Asp) was associated with the loss of the TS ϩ phenotype in the 9.2.3B mutant, although the single mutation Asp85Asn (mutant 11.1) did not produce a significant TS ϩ phenotype in M. mycoides subsp. capri. As the replacement at this position has been done using two different amino acids (Gly and Asn), we cannot exclude the possibility that one change is more drastic than the other one regarding the growth of M. mycoides subsp. capri at a nonpermissive temperature.
The set of mutants that was obtained in this study finally resulted in a range of TS ϩ phenotypes. The mutant that was the most severely affected in its growth at 40°C was selected because the multiple mutations (at least the Asp85Gly and the Gly124Arg mutations) were considered to confer a higher level of safety against possible genetic reversion. These two mutations did not prove to be sufficient, since a single reversion at position 85 (Gly85Asp) was associated with the loss of the TS ϩ phenotype both in vivo (mutant iCM135) and in vitro (subclone A9.2.1C 40 ). This indicates that a structural change in the protein that leads to the TS ϩ phenotype occurs only when both of the two mutations (Gly85Asp and Gly124Arg) are present, possibly together with the third mutation (Ile224Val). Therefore, the reversion in one of them was enough to lose the TS ϩ phenotype in M. mycoides subsp. capri. Obviously, the reversion of these mutations is a problem if one envisages obtaining safe vaccine strains with this approach. In fact, this reversion to the TS Ϫ phenotype is already known for the vaccine strain of M. synoviae that was produced by chemical mutagenesis, but this attenuation is most likely due to other mutations in other genes, such as the ones encoding the oligopeptide permease (Opp) system (6,10,26).
A novel caprine challenge model has been established for M. mycoides subsp. capri GM12 (24,27), which we basically followed, with slight modifications, as reported recently (24). This challenge model proved to be robust and reproducible (24,27). Another caprine challenge model using an atypical M. mycoides subsp. capri strain was reported elsewhere (28), but as the number of infected animals was small and the outcome differed between groups markedly, we opted not to follow this challenge model. The model of infection performed here is a significant advancement, as there is a general lack of animal models for studying mycoplasma pathogenicity, which is mostly due to the high host specificity of mycoplasmas. However, this animal model has its own advantages as well as limits. Its main advantage is that it allows evaluation of the attenuation of mutants when derived from a strongly pathogenic strain such as M. mycoides subsp. capri GM12. Within a few days, the survival curves (Fig. 4) together with the data on bacteremia provide the main tools for evaluation of the outcome of infection. The weight of each clinical parameter, including body temperatures, is more difficult to evaluate because we used outbred goats that are genetically heterogeneous. In addition, as shown in this study, the high dose of infection with mutants provides the possibility of detecting pathogenic revertants. However, the limits of this animal model also need to be recognized. The transtracheal application of the 9.2.3B obg subclone was selected for comparison of virulence with the parental strain YCpM-myc1.1. This route was not meant to be used as prospective immunization route. However, we opted to challenge the 7 goats that survived the first phase of the experiment (35 dpi) because we expected the induction of an immune response, and it provides a comparison with similar results obtained with a GM12 glf mutant defective in capsule biosynthesis (24). Whereas 3 of the 8 animals infected with the glf mutant survived, only 1 of the 7 animals infected in this study with the obg mutant survived. This difference is not significant. It is difficult to discuss much more about this difference without a full analysis of the immune response in individual animals; our preliminary attempts using enzyme-linked immunosorbent assays (ELISAs) and immunoblotting did not provide meaningful results because of the high background of antibodies cross-reacting with M. mycoides subsp. capri antigens in the goats. This is likely due to the exposure of these animals to a number of parasites before being selected for this trial. Finally, it should also be stressed that very little is known about the immune determinants that would provide protection from challenge with highly virulent M. mycoides subsp. capri.
Until now, the tools from synthetic biology applied to Mycoplasma species have been mostly limited to building a minimal cell (20). However, we have shown that these tools are also useful for the elucidation of gene functions (29) and for the deciphering of host-pathogen interactions (30). The versatility of these approaches that combine in-yeast genome engineering and back-transplantation into a recipient cell is further demonstrated in the present study by showing our ability to generate targeted point mutations within the essential obg gene. Such an achievement was previously impossible using the genetic tools available for any Mycoplasma species. Since it is now possible to use synthetic biology tools for most of the ruminant-pathogenic Mycoplasma species that are closely related to M. mycoides subsp. capri (31), new strategies of vaccine design based on precise and multiple targets using genome engineering have become fully realistic. Within this global aim, the TS ϩ phenotype resulting from obg mutations described here is of interest if combined with (i) other mutations that hamper completely the possibility of reversion to a virulent phenotype and (ii) other deletion(s), such as deletion of the nonessential glpO gene known to be involved in the production of H 2 O 2 (32). Owing to this new animal model, we now have the possibility of evaluating the attenuation of virulence of different mutants resulting from genome engineering approaches and the immune response providing protection against an experimental challenge.
M. mycoides subsp. capri genome engineering in yeast. The genome of the modified M. mycoides subsp. capri clone YCpMmyc1.1 (strain GM12) was previously cloned into S. cerevisiae strain VL6-48N (22). Targeted nucleotides corresponding to the amino acids at position 80, 85, or 124 of the M. mycoides subsp. capri Obg protein ( Fig. 1A; see also Fig. S1 in the supplemental material) were mutated in the YCpMmyc1.1 genome cloned into yeast by using the TREC-IN (tandem repeat endonuclease cleavageinsertion) method as previously described (36). The experimental procedure used to obtain these engineered genomes is shown in Fig. 1B and detailed in the legend of Fig. S2 in the supplemental material (primers used during the procedure are listed in Table S1). At the end of the process, YCpMmyc1.1-obg genomes were obtained and contained mutations at either position 239/240, 253/255, or 369/371 of the obg gene, which corresponds to position 80, 85, or 124 of the Obg protein, respectively. To check this, total DNA was extracted from all selected yeast recombinants, and the DNA was analyzed by simplex PCR (Table S1), multiplex PCR, and pulsed-field gel electrophoresis (PFGE) as described previously (29) (Fig. S3). The presence of the mutations was confirmed by sequencing the 1,601-bp obg PCR fragment obtained using the primers obgF1 and obgR2 (Fig. S2). For multiplex PCR, a set of 11 different pairs of primers was used as previously described (22) (Table S1).
The TREC-IN method was also used to build a genome equivalent to the mutant 43.7 genome but with a repaired nadE sequence. Using the primer pair obgF1/obgR3 (Table S1) and mutant 43.7 genomic DNA, we amplified the obg gene with changes at positions 254, 369, 371, and 732 plus a 113-bp segment located upstream of the obg gene. This amplicon was then used as a template for PCR with primers cassette B-forward-F1 and cassette D-reverse-R1 to create a minicassette, named minicassette BC, with 60-bp overlaps with minicassette A at its 5= end and 60-bp overlaps with the sequence downstream of 5= Kan (positions 1 to 515) of the YCpMmyc1.1-⌬obg::URA3 genome, at its 3= end. Minicassettes BC and A were finally cotransformed into VL6-48N yeast clone 34 harboring the YCpMmyc1.1-⌬obg::URA3 genome (step 2 of the TREC-IN method). The next steps of the TREC-IN method as well as all genotypic tests were performed as described above.
Back-transplantation of the engineered YCPMmyc1.1-obgmod genomes into mycoplasma recipient cells to produce the M. mycoides subsp. capri obg mutants. In-yeast-engineered M. mycoides subsp. capri genomes were isolated in agarose plugs and transplanted back into restriction-free M. capricolum subsp. capricolum (McapΔRE) recipient cells as previously described (22,37). The resulting transplants were selected on SP5 medium plates supplemented with 5 g · ml Ϫ1 of tetracycline. After 3 successive passages in liquid medium, transplants were analyzed by PCR using M. mycoides subsp. capri-specific primers (Table S1), multiplex PCR (Table S1), and PFGE, as previously described (22). The obg gene was systematically amplified using the primer pair Obg-F1/ObgR1 (Fig. S2B) and sequenced in order to verify the presence of the desired mutations.
Mutants showing temperature-sensitive phenotypes were submitted to three successive rounds of filter cloning using 0.22-m sterilizing-grade filters (38). Subclones were once more assessed for their temperature-sensitive properties, and their whole genomes were sequenced using Illumina nextgeneration sequencers (see the supplemental material). Mutant and subclone names and main characteristics are summarized in Table 1.
Assessment of temperature sensitivity by serial dilution and qPCR. Mutants and subclones were grown at 32°C in SP5 medium until the pH reached 6.4. The temperature sensitivity of the samples was then analyzed by serial dilutions (10 Ϫ1 to 10 Ϫ9 ) of the cell cultures in SP5 medium, followed by an incubation period of 8 days at a permissive temperature (32°C) or a nonpermissive temperature (40°C). Culture color change was monitored daily and compared to that of a culture of WT M. mycoides subsp. capri strain GM12 grown under the same conditions. The cell concentration of each initial culture (pH 6.4) was determined by plating dilutions on SP5 plates. Just before putting the samples into the incubator (T 0 ) and then every 24 h during 6 days (T 24 to T 144 ), a 100-l aliquot was taken from the "10 Ϫ8 dilution tube" (for the mutants) and from the "10 Ϫ5 dilution tube" (for the subclones) for qPCR assays. These qPCR assays were performed using a protocol described previously (31). Briefly, qPCR measurements were conducted in 96-well plates using the SsoFast EvaGreen supermix (Bio-Rad) on the LightCycler 480 real-time PCR system (Roche) according to the manufacturers' instructions. Standard curves used to quantify the numbers of M. mycoides subsp. capri genomes were run independently for each assay by using purified genomic DNA (Wizard genomic DNA purification kit; Promega), with samples containing between 10 2 and 10 8 genomes. Measurements were done in triplicates for all samples. Data analyses were performed using GraphPad Prism statistical software version 7. All qPCR data were previously log transformed to meet model assumptions, and the Benjamini-Hochberg method (39) was used to adjust the false discovery rate (FDR) associated with the test. Analyses were done using an unpaired (independent) t test to compare the concentrations (means) of the different mutants with those of the M. mycoides subsp. capri controls (30.1 or 30.1.1) at each time point.
Animal studies. The animal experiments were performed under strict ethical rules (see "Ethics statement," below). Briefly, 16 male crossbred goats (Capra hircus), 1 to 2 years of age, were randomly selected from the International Livestock Research Institute (ILRI) ranch in Kapiti (see the supplemental material for details). Five days before experimental infection, all animals were transferred to the animal biosafety level 2 (ABSL2) unit and split randomly into two groups of eight animals (positive-control and obg groups), and each group was housed in a separate room. On day 0 postinfection (p.i.) and day 7 p.i., goats of the positive-control group and the obg group were infected transtracheally (the injection site was 5 to 10 cm distal from the larynx) with M. mycoides subsp. capri YCpMmyc1.1 and M. mycoides subsp. capri YCpMmyc1.1-obg, respectively. On day 0 p.i., each animal was infected with a 1-ml liquid culture containing 10 8 CFU, followed by a flush with 5 ml of phosphate-buffered saline (PBS); on day 7 p.i., this procedure was repeated, except that each animal was infected with a liquid culture containing 10 9 CFU. The animals were allowed to move freely within the ABSL2 unit and received hay and water ad libitum as well as pellets in the morning. Heart rate, breathing frequency, rectal temperature, and oxygen blood saturation were measured daily in the morning hours. Animals showing either a body temperature of Ͼ40.5°C for Ͼ3 consecutive days, moderate to severe pain or distress, weight loss of Ͼ10% within 7 days, or a breathing frequency of Ͼ50 breaths/min for Ͼ3 days were euthanized. More information regarding the group infected with YCpMmyc1.1 were described previously by Jores et al. (24).
Ethics statement. All protocols of this study were designed and performed in strict accordance with Kenyan legislation for animal experimentation and were approved by the institutional animal care and use committee (IACUC) (reference no. IACUC-RC2016-04, signed by Roger Pelle). Since 1993, the ILRI has complied voluntarily with the United Kingdom's Animals (Scientific Procedures) Act 1986 (http://www .homeoffice.gov.uk/science-research/animal-research/), which contains guidelines and codes of practice for the housing and care of animals used in scientific procedures. The study reported here was carried out in strict accordance with the recommendations in the standard operating procedures of the ILRI IACUC (ILRI IACUC ref no. 2016.06) and with adequate consideration of the 3 R's (replacement of animal with nonanimal techniques, reduction in the number of animals used, and refinement of techniques and procedures that reduce pain and distress). As this project also received financial support from the NSF (NSF proposal IOS-1110151 entitled BREAD: Toward Development of a Vaccine for Contagious Bovine Pleuropneumonia [CBPP]) awarded to the J. Craig Venter Institute (JCVI), the protocols of this study were also approved by the JCVI's own IACUC. The project was approved initially by the IACUC on 17 August 2011, reapproved on 6 October 2014 (no. 011-14), and signed by Mark Adams, JCVI Scientific Director.
Pathomorphological and histological analyses. Complete necropsy was performed on all animals. Tissue samples of the neck region around the inoculation site and all internal organs were fixed in 10% buffered formalin for 72 h and subsequently routinely processed for paraffin embedding. Tissue sections were cut at 3 m, stained routinely with hematoxylin and eosin (H&E), and evaluated by a board-certified pathologist.
Data availability. We declare that the data supporting the findings of this study are available within the paper and its supplemental material. The sequences from this study are available from the NCBI under SRA accession no. PRJNA510711.