Escherichia coli DH5α, M. smegmatis mc² 155, M. bovis BCG Danish (American Type Culture Collection, ATCC #35733) and their derivatives were used in the study. E. coli was used for the cloning steps and grown in Luria-Bertani (LB) (5 g/L yeast extract, 10 g/L tryptone and 10 g/L NaCl), (Sigma-Aldrich^®, Merck KGaA, St. Louis, MO, USA). M. smegmatis was grown in Middlebrook 7H9 (MB7H9) (Difco, Detroit, MI, USA), supplemented with 0.5% glycerol (Sigma-Aldrich^®) and 0.05% Tween 80 (Sigma-Aldrich^®) or plated on Middlebrook 7H10 agar (Difco) supplemented with 0.5% glycerol (MB7H10). BCG was grown in Middlebrook 7H9 supplemented with 10% of OADC (oleic acid-albumin-dextrose-catalase; BBL, Cockeysville, MD, USA), 0.5% glycerol and 0.05% Tween 80 (MB7H9-OADC) or plated on Middlebrook 7H10 agar supplemented with 0.5% glycerol and OADC (MB7H10-OADC). Mtb was grown in Middlebrook 7H11 (Difco) supplemented with OADC, glycerol and Tween 80 (MB7H11-OADC). E. coli and M. smegmatis were grown at 37°C. BCG and Mtb were grown at 37°C and 5% CO₂. When indicated, kanamycin sulphate (20 µg/mL) (Sigma-Aldrich^®), tetracycline hydrochloride (Tc) (200 ng/mL) (Sigma-Aldrich^®) and/or L-lysine (40 µg/mL) (Sigma-Aldrich^®) were added.

Chemically competent E. coli DH5α was prepared according to standard protocols (17). Electrocompetent mycobacteria were prepared as previously described (18). For transformation of mycobacteria, 300-500 ng of plasmid DNA were mixed with competent cells and electroporation performed using a Gene Pulser II device (BioRad, Hercules, CA, UK). Cells were recovered in MB7H9 and plated onto MB7H10 agar until the appearance of visible colonies. Kanamycin, tetracycline, or/and lysine were added to the media, as described above, when required.

Codon-optimized cas9 expression cassette, the Tet regulator (tetR) cassette and the tracrRNA sequences (Genscript, Piscataway, NJ, USA) were sequentially cloned into pJH152 (a kind gift from Dr. Stewart Cole, EPFL, Lausanne, France) using the restriction enzymes NotI/ClaI, HpaI and NheI (New England Biolabs, Ipswich, MA, USA), respectively, originating pKLM-CRISPR ( Figure 1 ). Selection of crRNA was made using the Cas-Designer tool at R-GENOME website (http://www.rgenome.net/cas-designer/). Each crRNAs sequence was inserted using BbsI (New England Biolabs) upstream of tracrRNA in pKLM-CRISPR, generating the sgRNA sequences. The expression of Cas9 and sgRNA are both controlled by tetracycline-inducible promoters (pUV15tetO). The T4g32 transcriptional terminator was added to each cassette end (T). Orientation of the cloned fragments according to Figure 1 were confirmed by Sanger sequencing. Vectors including crRNA aiming to knock-out lysA were named pKLM-CRISPR-lysA(x), where (x) represents different crRNA sequences. The list of plasmids and oligonucleotides used are shown in Supplementary Tables S1, S2 . Alignment of lysA from BCG and M. smegmatis as well as the relative position of sgRNA targets are represented in Supplementary Figure 1 .

The pAN71-ltak63 vector containing pAN promoter and driving low expression of the LTAK63 adjuvant was previously constructed by our group (5). A PCR-amplified lysA gene (containing a Shine-Dalgarno, SD), or a lysA expression cassette digested from pJH152 were used. The pAN71-ltak63 vector was digested with NotI/PvuII (New England Biolabs) and the lysA fragment inserted in tandem with ltak63 under the same pAN promoter, generating pAN71-ltak63-lysA(t) ( Figure 2A ). In this construct, the kanamycin resistance marker was removed by digestion with NsiI (New England Biolabs) and self-ligated. The pAN71-ltak63 vector was also digested with NotI/ClaI (New England Biolabs) to clone the lysA expression cassette, generating pAN71-ltak63-lysA(c) ( Figure 2B ). The kanamycin resistance marker was removed by digestion with ClaI (New England Biolabs) and self-ligated. Both vectors were electroporated into M. smegmatis mc² 1493 (lysine auxotroph) (19) and transformants plated onto MB7H10. Transformants were then grown in MB7H9 with and without kanamycin to confirm construction of the unmarked pAN71-ltak63-lysA vectors. Plasmid extraction from M. smegmatis was performed using Wizard Plus SV Minipreps DNA Purification kit (Promega, Madison, WI, USA) and the extracted plasmids were used to transform the lysine auxotrophic BCG (BCGΔlysA).

After transformation of M. smegmatis and BCG with pKLM-CRISPR-lysA(x), kanamycin-resistant colonies were recovered and cultured in 5 mL of MB7H9 and MB7H9-OADC, respectively. These cultures were used as pre-inoculum in a fresh culture starting at OD₆₀₀ 0.1. After 2 h (M. smegmatis) or 24 h (BCG) of incubation, tetracycline was added to induce expression of Cas9, and the culture was maintained at 37°C for 4-24 h (M. smegmatis) or 24-120 h (BCG). After the induction, bacteria were lysed by sonication using an ultrasonic processor GE100 (GE Healthcare, Chicago, IL, USA) and protein extracts separated by SDS-PAGE (BioRad, Hercules, CA, UK), transferred to PVDF membranes (GE Healthcare) and blocked with 5% non-fat dry milk at 4°C for 16 h. The membrane was probed using monoclonal anti-Cas9 antibodies (7A9-3A3, 1:1,000) (Santa Cruz Biotechnology, Dallas, TX, USA) incubated for 90 min and anti-mouse IgG conjugated with peroxidase (A6782, 1:1,000) (Sigma-Aldrich^®) for 60 min. E. coli DH5α transformed with pCas (20) was used as positive control for Cas9 expression. Chemiluminescent signal was developed using ECL Prime Western Blotting System (GE Healthcare) and images acquired with the LAS4000 digital imaging system (GE Healthcare).

After tetracycline induction the cultures were plated onto MB7H10 or MB7H10-OADC, both supplemented with lysine and kanamycin and incubated at 37°C until the appearance of visible colonies. Single colonies were then transferred to mirror plates either containing or not lysine, both in presence of kanamycin. Growth only in the lysine-containing plate revealed positive knock-out clones (Lysine-KO). The Lysine-KO clones were used as template for PCR amplification of the lysA region and sequencing. The lysA sequences from the Lysine-KO clones were compared with the reference sequences in UNIPROT database (M. smegmatis-Q9X5M1 and BCG-P9WIU7).

To assess reversion to the wild-type phenotype during in vitro culturing, rBCGΔlysA strains were serially passaged weekly for up to 8 passages when the cells were plated on MB7H10-OADC containing or not lysine. Additionally, the curing of pKLM-CRISPR-lysA(x) was assessed by plating rBCGΔlysA onto mirror plates containing or not kanamycin. Growth only in the plate lacking kanamycin indicated plasmid loss.

Competent BCGΔlysA were prepared as previously described and transformed with pAN71-ltak63-lysA(t) and pAN71-ltak63-lysA(c), generating rBCGΔ-LTAK63(t or c), respectively. Selected clones were grown in MB7H9-OADC until an OD₆₀₀ 1.0, when bacteria were recovered, and protein extracts used to detect the expression of LTAK63 by Western blot (20). Detection of LTAK63 was performed using anti-serum of mice previously immunized with rLTK63 (1:1,000) incubated for 60 min and an anti-mouse IgG antibody conjugated with peroxidase (A6782, 1:3,000 Sigma-Aldrich^®) incubated for 60 min. Additionally, growth curves of complemented auxotrophs were compared to wild-type BCG to determine whether rBCGΔ-LTAK63(t or c) would show altered in vitro growth.

All animal experiments were performed according to Brazilian and international guidelines on animal experimentation and approved by the Ethics Committee of Instituto Butantan, São Paulo-SP (CEUAIB), (Permit number 8591010817). Five to eight-week-old female BALB/c mice were obtained from the Central Animal Facility of the Instituto Butantan, SP, Brazil. BALB/c were immunized through the subcutaneous route (s.c.) with a single dose of BCG or rBCGΔ-LTAK63(t) (1x10⁶ CFU/100 µL), or saline and challenged after 90 days via the intranasal route (i.n.) with M. tuberculosis H37Rv (500 CFU/50 µL). Thirty days after challenge, the lungs were collected, homogenized, and 20 µL of serial dilutions (10^-1, 10^-2 and 10^-3) were plated on MB7H11-OADC at 37°C and 5% CO₂ for CFU counting.

Thirty days after Mtb challenge, lung tissue samples were collected, preserved in 10% neutral buffered formalin, embedded in paraffin, cut into 5-6 μm sections, and stained with hematoxylin and eosin (H&E). The severity of inflammation in the mouse lungs was assessed according to (21). In summary, H&E-stained lung sections were photographed at 40 x magnification using a microscope (Nikon Eclipse Ti-S) coupled to a digital camera (DS-Fi1c, Nikon). Image analysis software (ImageJ, National Institutes of Health, USA) was used to determine the pulmonary area affected. The inflammatory area was measured according to (21) and the functional lung area is represented by the intra-alveolar regions in the lung tissue determined using morphometric analysis according to (22). Briefly, five images at 40 x magnification per lung lobule, totaling 25 images per treatment, were randomly selected, and analyzed for the qualitative evaluation of the cell infiltrate and intra-alveolar regions. To measure the areas of interest the images were transformed into 8-bit and treated with threshold and percentage of the measured area. For leukocyte counting, the Color Deconvolution 2 plugin were used to visualize and separate nuclei from the cytoplasm. For cell counting, the Cell Counter plugin was used. This analysis is used to facilitate the differential counting of segmented and mononuclear nuclei.

Statistical analysis was performed using the GraphPad Prism (GraphPad Software, Inc.). The significance of differences among groups was calculated by unpaired parametric two-way Student’s t-test as described in the figure legends. Differences between mean values were considered significant when p < 0.05.

Transformation of M. smegmatis and BCG with pKLM-CRISPR showed that both were able to express Cas9, as determined by the Western blot (~160 kDa). In M. smegmatis, the highest expression of Cas9 was observed at 8 h after the induction in a concentration of 200 ng/mL of tetracycline ( Supplementary Figure 2 and Figure 3A ). In BCG, the highest expression of Cas9 was between 24-48 h after the induction, with decreasing levels being observed from that time until 120 h ( Figure 3B ). Attempts to express the wild-type S. pyogenes Cas9 in mycobacteria were unsuccessful (data not shown).

Mycobacteria were transformed with the gene editing vectors (pKLM-CRISPR containing the specific crRNAs to target lysA (n=3 for M. smegmatis and n=2 for BCG) ( Supplementary Table 2 ), and the expression of Cas9 was induced by tetracycline. After induction, the culture was plated on solid media containing kanamycin to obtain isolated colonies. These were transferred to mirror plates with and without lysine to detect the functional knock-out of LysA. The transformation of M. smegmatis with the vector containing crRNA-lysA820-Smeg, resulted in 2 out of 8 colonies (25%) that did not grow in the absence of lysine (SmegΔlysA) ( Figure 4A ). The induction with crRNA-lysA394-Smeg did not generate any functional knock-out (0/5), while crRNA-lysA123-Smeg did not produce any transformant. In BCG transformed with the vectors containing crRNA-lysA88-BCG, 2 out of 50 colonies (4%) did not grow without the supplementation of lysine (BCGΔlysA) ( Figure 4B ). No functional knock-outs were obtained using the vector containing crRNA-lysA20-BCG. In the genotypic analysis of M. smegmatis, we observed the addition of a single nucleotide (lysA_2b) and the deletion of two nucleotides (lysA_2c) next to the PAM site ( Figure 4C ). In BCG, we performed an additional round of induction (independent experiment), obtaining a total of five functional knock-outs. Sequencing analysis of BCGΔlysA demonstrated the deletion of two nucleotides (lysA_7 and lysA_9), deletion of larger fragments (lysA_10 and lysA_39), 83 and 107 bp ( Figure 4D and Supplementary Figure 3 ), and a single nucleotide deletion (lysA_43). The in-silico translation of these sequences revealed that the mutations results in the early interruption of translation or frameshifts impairing the correct translation of LysA ( Supplementary Figure 4 ). The possibility of reversion from the auxotrophic phenotype was evaluated by subculturing BCGΔlysA and observation of growth without supplementation of lysine. Even after 8 passages, without the supplementation of lysine, no prototrophic colony was observed ( Supplementary Figure 5 ).

Once reversion to the wild-type phenotype was excluded, the BCGΔlysA lysA_39 (mutant containing the deletion of 107 bp) was selected to be complemented. Before the complementation, the pKLM-CRISPR plasmid was cured. In a single incubation without kanamycin, 35% of the colonies showed plasmid loss ( Supplementary Figure 6 ). A single colony was selected and prepared as competent cells. The BCGΔlysA was then transformed with the complementation vectors, pAN71-ltak63-lysA(t) and pAN71-ltak63-lysA(c) and protein extracts used to detect the expression of LTAK63 ( Figure 5 ). More importantly, either when driven by the vectors with lysA in tandem or as an expression cassette, the level of LTAK63 expression was comparable to that observed in BCG transformed with the pAN71-ltak63 vector.

We investigated whether the complemented auxotrophic BCG would show distinct in vitro growth in comparison to wild-type mycobacteria. It can be observed that both recombinant BCG strains showed growth comparable to the wild-type ( Supplementary Figure 7 ). The rBCGΔ-LTAK63(t) was selected for the in vivo experiments, hereafter, named rBCGΔ-LTAK63.

BALB/c mice were immunized with BCG, rBCGΔ-LTAK63 or not immunized and were challenged with Mtb 90 days after. The group immunized with rBCGΔ-LTAK63 showed a 1.5 log CFU reduction in the lungs when compared to the control group and 0.5 log reduction when compared to the BCG group ( Figure 6A ). Histopathological findings typical in a Mtb infection in the mouse model showed a granulomatous inflammatory process spread across the infected lungs. Immunization with rBCGΔ-LTAK63 showed increased protection of the lung tissue compared to BCG as evaluated by the functional lung area ( Figure 6B ). The effect of rBCGΔ-LTAK63 on leukocyte migration differs from that of BCG, displaying a reduction in the inflammatory score ( Figure 6C ). The improvement of the histopathological lesions is shown in the representative lung sections. Mice immunized with rBCGΔ-LTAK63 showed fewer lesions consisting of perivascular and peribronchiolar inflammatory infiltrates when compared with BCG-immunized mice ( Figure 6D ). Characterization of the cellular infiltrate showed the presence of mononuclear cells, neutrophils, and alveolar macrophages in the perivascular and peribronchiolar infiltrates; a general decrease in the total number of these cell types in the lungs of the groups immunized with BCG or rBCGΔ-LTAK63 indicates improved protection against lung pathology upon vaccination ( Figure 7 ).