Bacteriophage Therapy for Tuberculosis Infections in Humanized Mice PDF

communications biology Article...

communications biology Article https://doi.org/10.1038/s42003-024-06006-x Bacteriophage therapy for the treatment of Mycobacterium tuberculosis infections in humanized mice Check for updates 1,2,3,8 1,2,3,7,8 1,2,3,8 Fan Yang , Alireza Labani-Motlagh , Jose Alejandro Bohorquez , Josimar Dornelas Moreira2,3, Danish Ansari1,2,3, Sahil Patel1,2,3, Fabrizio Spagnolo4, Jon Florence2,3, Abhinav Vankayalapati2,3, Tsuyoshi Sakai2,3, Osamu Sato2,3, Mitsuo Ikebe2,3, Ramakrishna Vankayalapati2,3, John J. Dennehy 5,6 , Buka Samten 2,3 & Guohua Yi 1,2,3 1234567890():,; 1234567890():,; The continuing emergence of new strains of antibiotic-resistant bacteria has renewed interest in phage therapy; however, there has been limited progress in applying phage therapy to multi-drug resistant Mycobacterium tuberculosis (Mtb) infections. In this study, we show that bacteriophage strains D29 and DS6A can efficiently lyse Mtb H37Rv in 7H10 agar plates. However, only phage DS6A efficiently kills H37Rv in liquid culture and in Mtb-infected human primary macrophages. We further show in subsequent experiments that, after the humanized mice were infected with aerosolized H37Rv, then treated with DS6A intravenously, the DS6A treated mice showed increased body weight and improved pulmonary function relative to control mice. Furthermore, DS6A reduces Mtb load in mouse organs with greater efficacy in the spleen. These results demonstrate the feasibility of developing phage therapy as an effective therapeutic against Mtb infection. Mycobacterium tuberculosis (Mtb), the causative pathogen of Tuberculosis people developed DR-TB, and 78% of these cases developed multidrug- (TB), infected 10.6 million people and caused 1.3 million deaths globally, resistant TB (MDR-TB; resistant to first-line drugs, rifampicin and iso- making it one of the leading causes of death by a single infectious agent. niazid) or extensively drug-resistant TB (XDR-TB; resistant to rifampicin, More importantly, one-fourth (~1.7 billion) of the world’s population isoniazid, and a second-line drug)6. MDR-TB and XDR-TB are more has been latently infected with Mtb (LTBI)1. When complicated with other challenging to treat and are associated with increased morbidity and co-morbid conditions, such as diabetes, HIV, and COVID-19, the morbidity mortality7. Therefore, the development of alternative therapies in addition and mortality of tuberculosis infection is further increased2. Although these to antibiotics is critical for the advancement of TB therapy in the era of Mtb devastating facts highlight the threat posed by this deadly bacterium, the drug resistance. emergence of drug resistant Mtb strains in recent years has worsened the Bacteriophage therapy has emerged as a renewed approach to elim- situation in terms of Mtb prevention and control. inate bacterial infections8,9. Bacteriophages (also known as phages), viruses Upon establishment of infection, Mtb exhibits remarkable abilities to that infect bacteria, are bacteria’s natural enemies and have been used to adapt to the local environment to evade the host’s immune responses due to control bacterial infections since even before the discovery of antibiotics10. In its unique and dynamic four-layer cell envelope structure, which can help it the age of antibiotic resistance, phage therapy has drawn tremendous adapt to hostile lung microenvironments and facilitate its entry into a non- attention. Recent clinical cases and trials demonstrated that phages can be replicating drug-tolerant persister state. In this state, the bacilli are well used to treat antibiotic-resistant bacterial infections with positive clinical protected against antibiotic therapy3,4, and are potentially a source of new outcomes11–16, and patients with drug-resistant M. abscessus and M. chelo- strains of drug-resistant Mtb (DR-TB)5. In 2019, around half a million nae have been successfully treated17–19. These prominent bodies of work on 1 Department of Medicine, The University of Texas at Tyler School of Medicine, Tyler, TX, USA. 2Center for Biomedical Research, The University of Texas Health Science Center at Tyler, Tyler, TX, USA. 3Department of Cellular and Molecular Biology, The University of Texas Health Science Center at Tyler, Tyler, TX, USA. 4Life Sciences Department, Long Island University Post, Brookville, NY, USA. 5Biology Department, Queens College of The City University of New York, Flushing, NY, USA. 6The Graduate Center of The City University of New York, New York, NY, USA. 7Present address: Center for Discovery and Innovation, Hackensack Meridian Health, Hackensack, NJ, USA. 8These authors contributed equally: Fan Yang, Alireza Labani-Motlagh, Jose Alejandro Bohorquez. e-mail: [email protected]; [email protected]; [email protected] Communications Biology | (2024)7:294 1 https://doi.org/10.1038/s42003-024-06006-x Article using phage therapy to treat other mycobacterial infections inspired us to Mtb OD600 on day 9 (Fig. 1C), while other phages did not affect Mtb OD600 pursue phage therapy as a potentially effective treatment for Mtb infection. on day 9, although D29 also showed the ability to kill Mtb in agar plates. We However, as to the phage therapy for Mtb treatment, the significant then plated the cultures from different time points to determine the CFU of advancements have focused on screening lytic phages for effective killing of the bacilli (Fig. 1D, E). The CFU results were consistent with the OD600 Mtb in bacterial culture plates, demonstrated by plaque formation20–22, but measurements. To exclude the possibility that the unsuccessful Mtb elim- no systematic study has characterized the effectiveness of phages in killing ination of D29 in liquid culture is due to the low infection titer, we used Mtb in primary human macrophages or in reliable animal models. There- MOIs of 1, 10, and 100 to infect the Mtb in 7H9 culture, and monitored the fore, extensive preclinical studies, especially using primary human macro- Mtb CFU over time. We found that even the highest MOI (MOI of 100) did phages or animal models that resemble human clinical settings, are of not kill the Mtb in broth culture (Supplementary Fig. 1). It may be that Mtb exceptional clinical importance in the development of phage therapy for rapidly acquires resistance to D29, leading to the replacement of the sus- Mtb treatment. ceptible Mtb in liquid cultures. Therefore, we conclude that the Mtb killing One of the major challenges to eliminating Mtb infection is Mtb’s ability of D29 in agar plates does not necessarily correlate with the Mtb induction of granuloma formation once host defenses fail to kill the bacteria. killing ability in liquid culture. While granuloma formation limits Mtb growth, it provides a survival niche We further confirmed this outcome using a GFP-expressing for Mtb replication when the immune system is weakened. Granulomas, the H37Rv strain, H37Rv-GFP, in a liquid culture condition as described hallmark of TB pathology, consist of macrophages, neutrophils, and lym- above. Due to the highly efficient expression of GFP gene, the live phoid cells, including T and B cells, and are formed upon Mtb infection in bacteria are visible under fluorescent microscopy. The microscopy patients23. Most commonly employed mouse models can be infected with images of the day 9 cultures clearly showed that there was no GFP Mtb, but Mtb can only form a granuloma-like structure in mice. Mouse Mtb expression in the wells with H37Rv-GFP co-cultured with DS6A, infections lack the caseous necrotic granulomas that are often observed in whereas H37Rv-GFP in the wells co-cultured with other phages remain TB patients24. The Kramnik’s mouse can form hypoxic, encapsulated fluorescent, further demonstrating the highly effective Mtb-killing granulomas with a caseous necrotic center following Mtb infection25,26, but ability of DS6A (Fig. 1F). generally these mouse models cannot fully recapitulate the human immune responses. Recently, humanized mice based on NOD-scid IL2RgammaKO Phage DS6A can efficiently eliminate Mtb H37Rv in primary (NSG) mice were developed. However, the innate immune systems in these human macrophages mice were deficient and transplanted human hematopoietic stem cells To test whether the phages can kill Mtb in infected primary human mac- (hHSCs) were generally not well-developed; thus, the human B, T, and rophages, which is critical for therapeutic effectiveness in clinical settings, we myeloid cells were immature, and the NK cells lost functions27,28. These isolated peripheral blood mononuclear cells (PBMCs) from the blood of a mouse models provide valuable tools to study Mtb infection24,29; however, healthy donor and isolated the hCD14+ monocytes using microbeads to the myeloid barrier in these NSG mice causes a relative lack of leukocyte above 95% purity as determined by flow cytometry analysis (Supplementary differentiation28. The lack of sufficient granuloma formations (which Fig. 2). The purified CD14+ cells were differentiated into macrophages by requires mature macrophages) makes these mice less able to recapitulate incubation of the cells with hM-CSF, hGM-CSF, and hIL-4 for 5 days. The Mtb infection in humans. macrophages were then infected with H37Rv at an MOI of 1, and treated These deficiencies are rectified in the newly developed NSG-SGM3 with different phages at an MOI of 10. On days 5 and 10, we plated the mice, which can transgenically express three human cytokine/chemokine macrophage lysates on 7H10 agar plates to determine the CFUs (Fig. 2A). genes IL-3, GM-CSF, and KITLG. The expression of these genes can The results showed no significant reduction in Mtb CFUs for all phage- enhance the differentiation and maturation of the myeloid-lineage treated groups on day 5 when compared to the control Mtb-infected mac- cells28,30–33. Moreover, these three transgenic genes can improve the home- rophages without phage treatment. However, on day 10, phage DS6A ostasis of human CD34+ HSCs, increase neutrophil and macrophage completely eradicated the bacilli from the infected macrophages, while other numbers and function, and stimulate hCD34+ cells to differentiate into phages still failed to show any reduction in Mtb growth in macrophages myeloid progenitor cells32,34. Thus, this humanized mouse model can gen- (Fig. 2B, C). erate sufficient and fully functional myeloid cells, including macrophages, to We further tested whether the phages can also kill Mtb in infected study Mtb infection in the context of an entire human immune system. macrophages from different donors. For this purpose, we isolated mono- In this study, we show that phage DS6A can efficiently kill wild-type cytes from four healthy donors, and repeated the above experiment in these Mtb H37Rv in vitro (in both 7H10 agar plates and 7H9 liquid culture), in four samples using phage DS6A, and phage D29 was used as a negative primary human macrophages, and in humanized NSG-SGM3 mice, control. We also wanted to see whether Mtb could be killed in a shorter time demonstrating the feasibility of developing phage DS6A as an effective rather than ten days, so we set the sampling time points to days 4 and 7. The therapeutic against Mtb infection. results showed that all four donors displayed the same pattern in which Mtb in the infected macrophages were killed within seven days by phage DS6A, Results but not in the control phage treatment (Fig. 2D, E). Interestingly, some of the Phage infectivity in M. smegmatis and in M. tuberculosis H37Rv phage-treated cultures had a significantly higher bacillary load than the We first tested three bacteriophages, D29, Chah and DS6A, for their lysis control bacteria cultures (Fig. 2B, D), which we address further in the capacities of M. smegmatis and Mtb H37Rv by plaque assay (Chah was used discussion section. as a negative control for Mtb killing35). Serially diluted phages (102–104 pfu) To ensure that the phage DS6A can kill the intracellular Mtb, we were incubated with M. smegmatis or H37Rv in 12-well plates with 7H10 infected the macrophages with H37R-GFP at an MOI of 1, and then treated agar. Infection was carried out at a multiplicity of infection (MOI, a ratio of them with phage DS6A. Infected macrophages without phage served as a phage to bacteria) between 0.001 and 0.1. The results showed that two phage positive control, and H37Rv-GFP uninfected macrophages were used as a strains, D29 and Chah lysed M. smegmatis (Fig. 1A) and formed different negative control. We confirmed that Mtb bacilli entered the macrophages at sized plaques, and two phage strains D29 and DS6A, lysed H37Rv and 4 h post-infection using fluorescence microscopy (Supplementary Fig. 3). formed plaques (Fig. 1B). We then observed the cells at 7 days post-infection (dpi) using confocal We then tested the Mtb killing ability of these three phage strains in miacroscopy. We found that at 7 dpi, Mtb-infected macrophages without liquid culture by measuring the optical density at 600 nm (OD600) of the phage treatment showed 20-30% GFP-positive signals (Fig. 2F, middle row), H37Rv suspensions at different time points after incubation with phages at while the Mtb-infected macrophages treated with phage were completely an MOI of one, by inoculation of 7H9 media with 5 × 104 CFUs of H37Rv void of GFP-positivity (Fig. 2F, bottom row). This result is consistant with and 5 × 104 pfu of phages. We found that only DS6A significantly reduced the CFU result shown in Fig. 2B–E, further lending support that phage Communications Biology | (2024)7:294 2 https://doi.org/10.1038/s42003-024-06006-x Article Fig. 1 | Phage infectivity of M. smegmatis and Mtb H37Rv in solid agar plates, and dilutions of on the 7H10 agar plates and cultured at 37 °C for CFU determination. C: in Mtb liquid culture. Serial dilutions (102–104 pfu) of different strains of bacter- OD600 of sampled liquid cultures was measured at different time points as indicated. iophages were mixed with 1 × 105 CFUs M. smegmatis (A) or H37Rv (B) in 7H9 D: Statistics of the CFUs of each sampled liquid culture sampled at various time medium and incubated at 37 °C with shaking for 1 h. The infection cultures were points. E: Titering assay of phage-Mtb liquid culture sampled at different time then mixed with 0.8% top agar and spread on 12-well 7H10 agar plates supple- points. The data are representative results of two independent experiments. F: Phage mented with 10% OADC enrichment. C–F: H37Rv (1 × 105 CFUs) were infected infection of H37Rv that expresses a GFP reporter gene. An MOI of 1 was used to with various bacteriophages at an MOI of 1 for 1 h, then inoculated into 20 mL 7H9 infect H37Rv-GFP. GFP expression was pictured under fluorescence microscopy media supplemented with 10% ADC and incubated at 37 °C for 9 days. The cultures after nine days of phage infection. were sampled at days 3, 6, and 9 post infection by plating 10 μL of 10-fold serial DS6A can enter the macrophages and eliminate the intracellular growth and maturation; thus the functional macrophages and dendritic cells of Mtb. are significantly elevated in the bone marrow compared with NSG recipients28,31; (3) There is a significant increase of regulatory T cells Mtb infection of humanized NSG-SGM3 mice (Treg)31, which are important in regulating Mtb pathogenesis36,37; (4) Compared to the commonly used Hu-PBL and BLT mouse models, B cells can be developed to mature phenotypes with the ability of the humanized NSG-SGM3 mouse model has several advantages: (1) class-switching27, which is important to evaluate the IgM and IgG Humanized NSG-SGM3 mice exhibit improved reconstitution with responses against phages; (5) They are relatively easy to generate by an increased general population of human immune cells (hCD45 + ); intravenous injection of human HSC into irradiated adult mice for (2) Cytokine expression supports human myeloid cell differentiation humanization. Communications Biology | (2024)7:294 3 https://doi.org/10.1038/s42003-024-06006-x Article Fig. 2 | Phage DS6A eliminates Mtb in infected primary human macrophages. A Experimental procedures of the assay. B, C: Half million macro- phages were infected with H37Rv at an MOI of 1. After 4 h of Mtb infection, three different phages (5 × 106, MOI of 10) were applied to the Mtb- infected primary human macrophages, and cultured at 37 °C. The cells were sampled at days 0, 5 and 10 to determine the bacillary load for the evaluation the Mtb-killing ability of the phages by plating the ultrasound-broken cells on 7H10 plates (supple- mented with 10% OADC) and the CFUs were counted on Day 8. B shows the statistics of the CFUs of all phages, and (C) shows the representative pic- tures (duplicated) of the 7H10 plates taken at day 5 and day 10. D, E: Testing the Mtb-killing ability of phage DS6A in Mtb-infected macrophages derived from four different healthy donors, and phage D29 was used as a negative control. The infected mac- rophages were sampled at different time points, sonicated, and plated on 7H10 agar plates using different dilutions to titer the bacillary load. D shows the statistics of five donors, and (E) shows the representative pictures taken from 10x diluted plates for days 4 and 7 cultures. F: Confocal macroscopy images show the Mtb-killing efficacy of phage DS6A in H37Rv-GFP infected macrophages. Cellmask plasma membrane stain (red) was used to localize the cell membrane, GFP (green) was used for tracking the Mtb bacilli, and DAPI (blue) was used to stain the nuclei of the macrophages. The red arrows show the Mtb in the macrophages. Unpaired student T-tests were used to analyze the differences between groups in Figure D. n = 5 independent donors. All statistical data are represented as mean ± SEM. Statistical significance was defined as *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001. We obtained NSG-SGM3 mice from the Jackson Laboratory and have were significantly expanded (Fig. 3B). Meanwhile, in both the lung and bred them in-house. For humanization, 4–6-week-old NSG-SGM3 mice spleen homogenates, we detected a high number of bacilli (Fig. 3C), indi- were irradiated at 150 cGy and then infused with human CD34 + HSC. Ten cating that the mice established Mtb reservoirs not only in the lungs but also to fifteen weeks after HSC injection, the animals developed a full comple- systemically. ment of human immune cell types, including CD4+ and CD8 + T cells, B cells, myeloid cells, and natural killer (NK) cells (Fig. 3A, Gating strategy is Phage DS6A improved pulmonary function and eradicated shown in Supplementary Fig. 4). splenic infection in Mtb-infected humanized mice To establish humanized NSG-SGM3 mouse model of Mtb infection, To investigate if phage DS6A can eradicate Mtb in vivo, we performed an we infected the humanized NSG-SGM3 mice with H37Rv via a Madison animal experiment in the humanized NSG-SGM3 mouse model of tuber- chamber to deposit about 50–100 CFUs per mouse lung. At 25 days post- culosis (Fig. 4A). We infected the humanized NSG-SGM3 mice with low infection, we sacrificed the mice and analyzed the immune cell profile of dose aerosolized H37Rv, as confirmed by an average ~100 bacilli per lung lung and spleen homogenates. Additionally, we determined Mtb growth in deposited at day one post-infection after sacrificing three animals to evaluate lungs and spleens. The results showed that, when compared with the the CFU in the lung homogenates (Fig. 4C). From day 3 of Mtb infection, we PBMCs analysis before infection, the spleen immune cells underwent treated the mice every other day with DS6A (1011 phage particles/dose) for a substantial changes after infection, and the CD8 + T cells and myeloid cells total of 10 doses via intravenous administration. Communications Biology | (2024)7:294 4 https://doi.org/10.1038/s42003-024-06006-x Article Fig. 3 | Mtb-infection of humanized NSG-SGM3 mice. A Reconstitution of human population was analyzed by flow cytometry. B Statistics of different immune cells immune cells in splenocytes (upper panels) and PBMC (lower panels) from before and after infection. C Five mice were sacrificed to check Mtb infection in lungs humanized NSG-SGM3 mice. NSG-SGM3 mice were irradiated and intravenously and spleens by determining the organ CFUs by plating serially diluted organ injected with 2 × 105 hCD34+ HSC. After 10–12 weeks, reconstitution of human homogenates on 7H10 agar plates as described above. T cells, B cells, myeloid cells, and natural killer (NK) cells within human CD45-gated Communications Biology | (2024)7:294 5 https://doi.org/10.1038/s42003-024-06006-x Article After the treatments, we determined the growth of Mtb in mouse lungs two of the four parameters (Ers and Rrs) tested in the phage-treated group and spleens. We also performed pulmonary function tests (PFTs) and had significantly improved, with less respiratory resistance and better evaluated overall mouse health by monitoring mouse body weight changes elasticity, demonstrating improved pulmonary function after treatments in association with changes in the bacillary burden of the mice. The results (Fig. 4D). The CT scan results showed that there were fewer high-density showed that DS6A-treated Mtb-infected mice gained significantly more areas with much cleaner lungs in the phage-treated group of mice, which weight than control Mtb-infected mice (Fig. 4B). The PFTs also showed that indicates less inflammation and other pathological changes (Fig. 4E). Communications Biology | (2024)7:294 6 https://doi.org/10.1038/s42003-024-06006-x Article Fig. 4 | Phage DS6A is effective at eradicating Mtb in humanized NSG- of each mouse CT scan figure shows the 3D image, the white areas represent the SGM3 mice. A Schematic presentation of the animal experiment procedures. H37Rv high-density scan (e.g., tissues), while the black areas represent low density scan (e.g., aerosol infected humanized mice were treated three days after infection with air). The three small figures in each mouse scan show different angles of scan results. 1 × 1011 pfu of phage DS6A per mouse or equal volume of sterile PBS every other day F The colonies from the plates that spread with lung or spleen homogenates of the for eight times. B The mouse body weights were monitored over time, and the body H37Rv infected control or phage-treated mice were picked and mixed with weight change over the initial body weight of the mice are shown as percentages 1 × 106 pfu of phage DS6A or a negative control phage (phage Chah) in 10 µl 7H9 (n = 9 for each group). C The growth of H37Rv in the mice lung and spleen media and incubated at 37 °C for 1 h. Then the infection cultures were pipetted onto homogenates together with initial infection dose (left panel in purple) are shown the 7H10 agar plates OADC and incubated at 37 °C for determination of CFUs. The (n = 3 for initial infection dose determination; n = 9 for Mtb control and phage- animal experiments were performed four times, and the 3rd and 4th experiments treated group). D Before termination of the experiment, the mice were subjected to were shown. Except the pulmonary functions were tested once (3rd animal pulmonary function testing. The measurements of elastance, compliance, total lung experiment, n = 3 for Mtb control group, and n = 4 for phage DS6A-treated group), resistance, and total lung volume were collected, and the statistics are shown (n = 3 the other data are shown as the combined results from the 3rd and 4th experiments for Mtb control group, n = 4 for phage-treated group). E CT scans were performed (n = 9 for each group). All the statistics are shown as mean ± SEM. Unpaired Student for each mouse, and two representative figures for each each group show the Mtb- T-test was used to analyze the differences between groups. Statistical significance was infected control mouse and phage-treated mouse lungs, respectively. The left panel defined as *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001. Furthermore, the lung volume of the phage-treated mice was larger than the cytokine levels in the spleen were higher than that of the lungs, and untreated Mtb control mice, even though the difference didn’t reach sig- that the phage treatment induced significantly fewer myeloid cell- nificance (p = 0.1782) (Fig. 4D). The CFU counts showed that the bacilli derived cytokines than observed in the control Mtb-infected mice. This were completely eradicated in the spleens in six of nine mice of the phage- observation is consistent with reduced bacilli burden in the mice treated treated group, and the average bacillary load in the spleen was significantly with phage. We did not see significantly elevated levels of T cell cytokines lower than in the Mtb-infected control group (Fig. 4C). Moreover, the lung in the mice after four weeks of Mtb infection, consistent with sig- bacillary load showed three-times higher in Mtb-infected control mice than nificantly increased CD14+ monocytes, but not lymphocytes, in the in phage-treated mice, while the difference didn’t reach significance humanized mice after Mtb infection. (Fig. 4C). These results suggest that, after intravenous administration, phage DS6A has the capacity to kill the replicating Mtb within/or outside the lung Discussion that are disseminated to the blood circulation. The development of alternative treatments to antibiotics for Mtb infection is To explain why the phage DS6A performed better in killing Mtb bacilli an urgent task in the era of antibiotic resistance. In this study, we have shown in spleens than in lungs, we determined the phage genome copies in the lung that bacteriophages D29 and DS6A can efficiently kill Mtb bacilli on 7H10 and spleen homogenates. The quantitative PCR result showed that the agar plates, while only phage DS6A significantly killed Mtb in liquid cul- spleen had three times more phage genome copies than the lungs (Sup- tures. However, only phage DS6A was able to kill Mtb bacilli in primary plementary Fig. 5), suggesting that the effectiveness in tissues may be human macrophages. In addition, we showed that phage DS6A could positively correlated to the phage copies, and that phages intravenously effectively eradicate Mtb bacilli in spleens and reduced the bacilli load in the administered was preferentially distributed to the filter organs such as spleen lungs in a humanized mouse model of Mtb infection. Thus, phage DS6A and liver more than to the lungs38, which may explain the organ-dependent appears to have the potential to be developed as an effective therapeutic for differences in Mtb eradication in humanized mice. Mtb treatment. Given that phage DS6A could not wholly eradicate the spleen bacteria Phage therapy has shown great potential in controlling some intran- in several mice, we wondered whether phage resistance developed in the sigent drug-resistant bacterial infections. Compared to typical antibiotic bacteria within these mice. For this purpose, we picked colonies from the treatment, phage therapy has advantages such as: host specificity, low plates that spread the lung and spleen homogenates of phage-treated mice, toxicity, the ability to lyse the bacteria, sustained therapeutic levels in situ, and then mixed them with phage DS6A to see if the phage could kill the due to phage replication, and a narrower potential for inducing resistance, bacteria in vitro. The phage Chah was used as a negative control. The results all of which render it an attractive treatment strategy in the era of antibiotic showed that the negative control phage could not kill bacteria colonies resistance. Several phage strains have shown potential in eliminating Mtb picked from DS6A-treated or untreated mice, while the DS6A could kill infection20. Unfortunately, there has been little progress in Mtb phage both (Fig. 4F), indicating that no phage resistance developed during the 20- therapy, with only two studies published to date that reported on the use of day phage treatment period. phage therapy for the treatment of Mtb in guinea pigs, with limited success39,40. Therefore, our study showing the effectiveness of phage therapy Phage therapy elicits antibody responses in humanized NSG- in treating Mtb infections of clinically relevant humanized mice represents a SGM3 mice without affecting the function of immune cells significant advancement in this field. There is always a concern that the immune responses, especially antibody Humanized mouse models provide versatile tools to study various responses, will stymie the effect of phage therapy. Therefore, we determined infectious diseases. For Mtb infection, the myeloid cell lineage is especially the phage DS6A-specific antibody responses in experimental mouse sera by important because the macrophages are the main targets of the bacilli. In ELISA. We detected a significantly higher level of human IgM in phage most NSG-based humanized mice, myeloid cell differentiation is not effi- DS6A-treated mouse sera (Fig. 5B). However, we could only detect weak IgA cient, and the macrophages are not mature, thus the physiological processes and IgG responses in these mice (Fig. 5A, C). We further tested if there are such as phagocytosis and endocytosis, which may be involved in Mtb any neutralizing antibody responses that may mitigate the phage treatment infection and bacteriophage uptake, may not be completed as properly as efficacy, but we could not detect any significant neutralizing antibody they are in humans. However, due to the transgenic expression of the three responses in any serum dilutions of all phage-treated humanized mice human cytokine genes, the myeloid cells were well-differentiated in NSG- (Supplementary Fig. 6). SGM3 humanized mice that we have used in this study. After Mtb infection, We further tested whether phage therapy could affect the cytokine the myeloid cells were successfully propagated (Fig. 3B). Moreover, after profiles associated with reduced Mtb growth and improved lung func- phage treatments, the phage was shown to be capable of killing Mtb in tion due to the phage treatment. We determined the levels of 27 cyto- human macrophages engrafted in the humanized mice, suggesting that kines including cytokines of human lymphocytes and myeloid cells in NSG-SGM3 mice can recapitulate human Mtb infections, and thus NSG- the lung and spleen homogenates of the mice by a multiplex cytokine SGM3 mice are an appropriate and clinically relevant animal model for Mtb assay (Supplementary Fig. 7). Overall, the cytokine results showed that phage therapy. Communications Biology | (2024)7:294 7 https://doi.org/10.1038/s42003-024-06006-x Article Fig. 5 | Humanized NSG-SGM3 mice developed antibody responses against repeated once, and the results from one experiment with five mice each group are phage DS6A. ELISA was performed to determine the phage DS6A-specific human shown. All the statistics are shown as mean ± SEM. Unpaired Student’s T-test was IgA (A), IgM (B), and IgG (C) titers in sera of the humanized mice infected with used to analyze the differences between groups. Statistical significance was defined as H37Rv and treated with phage DS6A or PBS buffer control. The experiment was *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001. Although D29 and DS6A showed the ability to kill Mtb on bacterial to significance. We reason that the less bacilli burden in phage-treated group lawns, only DS6A efficiently killed Mtb in liquid culture and in human (~8-fold less than in Mtb control group) contributed to the improved lung primary macrophages. Additionally, the proven capacity of DS6A to reduce function. However, we cannot exclude any other reason for the improved bacterial load in humanized mice suggests that DS6A is the most appro- lung function. priate phage therapeutic for Mtb infections. The reasons why D29 can kill To test the hypothesis that phage delivery to the lungs was restricted, we H37Rv efficiently on bacterial lawns, but not in liquid culture, are not fully compared phage copy number in spleens and in lungs, and the results clear (Fig. 1). The most likely explanation is that Mtb readily acquires indicated there are more phage copies in spleen than that in lungs. It is worth resistance to D29, but not to DS6A. When inoculated in liquid culture at an noting that our original dose contained 1011 phages per dose, while we only MOI = 1, it may be that at least some Mtb cells are genetically incapable of detected thousands of copies at the termination time point. This discrepancy being infected by D29, thus these resistant Mtb cells eventually dominate the may be due to the pharmacodynamical changes of the phages, as the phage’s culture. Additionally, it appears that D29 cannot easily break this resistance. half-life might be as low as 2.2 h and some studies have reported that phages Alternatively, it is possible that D29 is capable of lysogenizing Mtb, but the are undetectable after 36 h of injection45,46. Even though we need to further available evidence suggests that this is not the case. First, D29 does not form investigate the pharmacokinetics of the phage DS6A, our qPCR result turbid plaques on host lawns suggesting that, if it is capable of lysogeny, it offered a reasonable explanation of why the intravenous administration of does so at a very low rate41 (Fig. 1). Moreover, although D29 does have an phage is more efficient in spleens than in lungs. Nevertheless, our results intact attP-integration system, no repressor gene required for the main- suggest that optimization of administration will be helpful to improve lung tenance of lysogeny and superinfection has been identified41. While it is bacterial clearance (e.g., via intratracheal route or nebulizer delivery, etc.) as possible that D29 possesses a cryptic or unrecognizable repressor, a genome tuberculosis, in majority of cases, are infections of the lungs. analysis suggests that it does not. D29 is highly homologous to the myco- Phage resistance has always been a concern as it may dampen the bacteriophage L5, but has a 3.6 kb deletion relative to L5 that removes a part therapeutic efficacy of phage therapy. In our study, it seems not to be a case of its genome that corresponds to L5’s repressor gene41. Nevertheless, these for phage DS6A within the treatment period (Fig. 4f). Similarly, in a recent results indicate that liquid cultures, rather than bacteria lawns, appears to be compassionate clinical use of phage therapy for the treatment of 20 patients a better prediction model for the in vivo efficacy of Mtb phage therapy. with Nontuberculous Mycobacterium infection, no phage resistance was Of note, we also observed that once the phage cannot kill Mtb in liquid found in any of the patients47. However, in the clinical settings of Mtb phage culturing condition, it sometimes will stimulate Mtb growth, as shown in therapy, the treatment time may be as long as several months, and Mtb- Fig. 2. The observation that, in some cultures, bacterial load is higher in the resistant strains may still appear. Therefore, screening more effective bac- presence of phages D29 and Chah but not DS6A, than in phage-free controls teriophage strains to form a phage cocktail would be essential to overcome (Fig. 2B, D) is puzzling. We postulate that the reaction between the phages phage resistance by Mtb. (D29 and Chah) and macrophages may interfere with bacteria growth, or Moreover, our study showed that significant IgM but weaker IgA and this may be due to the selective pressure caused by phages that results in the IgG responses were induced after 20 days of phage treatment (Fig. 5). This fast propagation of phage-resistant cells42. However, the exact mechanism is may be because the short-term administration of phage may lead to IgM as worth investigating in the future study. dominant antibodies, and with the prolonged administration, the IgG Nevertheless, our experiments provide clear evidence that phage DS6A response will become dominant. Given that phage treatment still improved was effective in killing Mtb in the spleen via the intravenous route. However, disease progression in a human clinical case of M. chelonae despite a sig- the bacterial clearance in the lungs was not as effective as in spleens. This nificant anti-phage antibody response18, it may be that are not a great discrepancy is possibly due to the administration route, as this is one of the concern. Indeed, in our study, the antibody responses did not adversely most important factors that affects in vivo efficacy of phage therapy43. affect the efficacy of phage therapy to eradicate Mtb in the spleen. This may Through the intravenous route, phages can easily reach the spleen and liver be because our treatment regimen was not deployed long enough to enable through circulation as blood filtering organs, but the lungs are much more to the development of sufficient quantity and quality of antibody responses difficult to reach. A previous study showed that the intravenous delivery of a to counteract the high dose of phages we administered. However, in clinical pneumonia bacteria-specific phage led to a 2-log lower prevalence (