The contribution of microfluidics to the fight against tuberculosis

2.1 Culture of bacterial cells embedded in matrices

Different matrices have been used to trap the cells. Golchin et al. [21] used a thin layer of alginate hydrogel matrix to immobilize bacteria and promote planar growth, preventing the cells from going out of focus during long incubations. Dividing cells were recorded using time-lapse confocal imaging to observe phenotypic variations and antibiotic killing in individual mycobacteria cells. The nutrients were provided through media perfusion across a cellulose dialysis membrane from polydimethylsiloxane (PDMS) and a glass flow cell, sealed with a mechanical clamp (Figure 1 left). The group of Choi et al. utilized 3D agarose matrices to immobilize Mtb for rapid DST, developing the disc agarose channel (DAC) chip [22]. Mtb cultures are loaded in 24-well plexiglass chips, mixed with the liquid agarose, which is allowed to solidify (Figure 1 right). The antibiotics are flooded with the medium and allowed to diffuse in the agarose matrix; then, the chip is sealed and incubated to analyze the drug susceptibility. The 3D agarose structure allowed the immobilization and tracking of specific cells for 13 days with an automated imaging system. The susceptibilities of different strains (non-resistant/control, MDR, and XDR-TB strains) were tested with the DAC and compared with the results of liquid culture DST to validate the system. The agreement rates of specificity and sensitivity of different drugs obtained were around or above 90% when compared with liquid culture, thus proving its capacity for DST. The time required for the DST was 9 days using the DAC system. However, Mtb cells used to assess the validity were obtained from pre-cultured colonies on a liquid medium, which requires one or 2 weeks of culture to get a significant number of cells. Although the system reduces the turnover of conventional methods, which require 4–12 weeks (1–6 of sputum culture and 2–6 of DST), the sample culture still needs a long time.

Figure 1

Left: Schematic of the microfluidic device presented in ref. [21], based on the hydrogel matrix. On top of a coverslip, cells are embedded in the thin alginate hydrogel, covered by a cellulose dialysis membrane that allows media diffusion. Mechanical clamps and a glass lid seal the PDMS flow channel structure. Right: Sketch of the automatic disc agarose channel chip and trapping methodology in agarose matrix used in refs [18,22]. (leftmost) Top view of the chips within 24-well plate, with channels, inlet, and drug reservoirs depicted. (rightmost) The experimental process of loading Mtb cells within an agarose matrix, covering with media and diffusion of anti-TB drugs within the media can be observed. Images reproduced with permission from TB (left, [21]) and Applied Microbiology and Biotechnology (right, [22]).

Two years later, in 2018, the same DAC chip was used to overcome complications regarding the inoculum effect [18] (increment in the minimal inhibitory concentration (MIC) of antibiotic required when the size of the inoculum, i.e., the number of organisms inoculated, increases [23]). This effect constitutes one of the main problems of culture-based DST methods, as a standardized inoculum size is required to obtain reproducible DST results; additionally, it can cause a reduction in the drug effectiveness due to the increased number of bacteria [23,24]. Therefore, the DAC platform was used to evaluate the consistency of MIC values obtained from different inoculum sizes compared to the liquid culture reference method. Moreover, the critical concentration of 12 anti-TB drugs was determined from 254 clinical isolates (including pan-susceptible and resistant strains) without measuring the inoculum size. The tested anti-TB drugs gave consistent results with 96.3% overall agreement within 7 days. However, the isolates required a lengthy subculture on a liquid medium before the test, as in the previous study. In a recent 2021 study, the DAC system improved with a modified channel to increment the antibiotic diffusion into media to reduce turnaround times for DSTs. This fully automated DST system was named QuantaMatrix microfluidic agarose channel (QMAC) and integrated mycobacteria growth indicator tubes [25].

2.2 Monitoring in culture chambers

Culture chambers within microdevices have been widely used to analyze drug effects on Mtb. However, to achieve this, bacterial cells need to be encapsulated in the chambers. To this respect, Delincé et al. [26] developed InfectChip, consisting of a cellulose semipermeable membrane between a micropatterned coverslip and a PDMS chip. The coverslip is patterned with circular microchambers where the cells are trapped, grow, and divide, whereas the medium, is continuously flowing through a microchannel in the PDMS chip (Figure 2 left).

Figure 2

Microfluidic devices that use PDMS channels to trap cells. Left: 3D model of the microfluidic device from ref. [26], showing the semipermeable membrane that covers the patterned microchambers in the coverslip. The PDMS chip diffuses media through the membrane. Right: Representation of the microdevice used in ref. [27] to culture cells in PDMS grooves covered by a membrane, fed by diffusion through a flow channel encapsulated in a PDMS chip. Images reproduced with permission from Lab Chip (left, [26]) and Science (right, [28]).

This device was used to study the dynamics and interactions of single-cell behavior between Mtb and motile eukaryotic phagocytes in a host-cell model, providing a platform to visualize the different stages of infection. However, despite its advantages, it presents several limitations, including a limited lifespan of the macrophages, which constrains the visualization of the infection process.

The design shares similarities with the device developed by Wakamoto et al. [29], used for basic research (Figure 4). It can be used to test new antibiotics. For example, they tested faropenem [30], and it was also used to analyze the effects of inhibitors against drug resistance mutation for isoniazid [31].

Figure 3

Left: Picture of the high-throughput (48 × 48 array) microfluidic device used in ref. [32] (developed in ref. [33]) for the transcriptomic profiling for speeding drug development programs. The sample and reagents are introduced through the inlets, which are pressure-driven through valves to the integrated fluidic circuit, composed of fluid lines, valves, and microchambers in which the PCR takes place. Right: Cross-sectional view and picture of the chamber of the Raman platform developed in ref. [34]. Cells go through silicone tubes into a trapping chamber of quartz. Then, acoustic pressure is applied through a lithium transducer to make them levitate and aggregate to determine the Raman spectra. Images reproduced with permission from PLoS (left, [33]) and Communications Biology (right, [34]) journals.

Other systems have been developed to encapsulate bacterial cells in microfluidic devices. For example, Sala et al. [27] proposed a model to analyze the effect of different drugs on dormant cells using a previously developed microdevice [28]. The microdevice design includes a PDMS layer patterned with long grooves, covered by a transparent membrane, and sealed with a thick PDMS layer with flow channels. Cells get trapped in the grooves between the membrane and the PDMS layer (Figure 2, right). This system differs from the previous ones [26,29] as the cells are not inside culture chambers in the coverslip but directly growing on PDMS grooves.

2.3 Other methods for drug development

Time-lapse microscopy combined with microdevices for bacterial culture is the primary technique used in drug testing applications and basic research. Still, other methods are used in the field of drug development. Instead of focusing on the quantification of death rates and characterization of the persistent cells, this research analyzed the transcriptome and other markers to determine the response of the bacterial cells to the antibiotics applied.

Murima et al. [32] utilized a high-throughput microfluidic array developed in ref. [33] to obtain biomarker genes that could improve and shorten the traditional anti-TB drug-discovery schemes. This device contains an integrated fluidic circuit chip containing valves and 2304 chambers where the real-time polymerase chain reaction (PCR) occurs (Figure 3 left). After the fluids are pressure-driven in the corresponding chambers, the chip is placed inside the instrument, where the thermal cycling occurs. The high number of wells allows performing many simultaneous reactions, which were used to analyze transcriptomics of M. tuberculosis and M. bovis in the presence and absence of several growth inhibitors.

Figure 4

Schematic of the PDMS microfluidic chip mounted on a cellulose membrane and a glass coverslip for time-lapse imaging. (a) Top view depicting inlet, outlet, and parallel channels. (b) Cross-sectional view of the channels in the PDMS chip. (c) Closeup to the cells growing in the microchambers, covered by a membrane to allow media diffusion from the channels. Image reproduced with permission from Science [29].

Raman spectroscopy was used to analyze the effect of isoniazid in the lipidic expression of M. smegmatis [34]. The device consists of a flow cell with a trapping chamber, formed by laser-cutting the channel between quartz slides and sealed with transfer tape, as can be observed in Figure 3, right. The inlet and outlet are made with silicone tubes. The transparent lithium transducer allowed the acoustic pressure required for the cells to levitate and aggregate in the channel and to be trapped in the lateral direction to obtain the wavelength-modulated Raman spectra. The live-cell monitoring was carried out by acoustic trapping after 7 days of preculture.

4.1 Production of anti-TB vaccines

Vaccination programs can be essential for disease management by preventing active and latent TB. The first (and to the day, only) vaccine against TB is the Bacille Calmette Guérin (BCG), introduced in 1921 [52]. It prevents extrapulmonary TB in infants by newborn vaccination; however, it fails to stop the pulmonary form of the disease [53]. Due to these limitations, new vaccine candidates are continuously being proposed and reviewed. Some of the leading candidates and their clinical phases are evaluated in detail in refs [54,55].

Particularly, cationic liposomes were formulated as a gene delivery system that could be used in gene therapy and vaccine due to their ability to form complexes with DNAhsp65 [56–58]. The low yield production obtained with conventional methods and the escalation required led to the use of microfluidic devices to generate high lipid concentrations. Two PDMS devices were proposed based on single and double hydrodynamic focusing (SHF and DHF, respectively) to enhance vesicle formation and mixing [59], as depicted in Figure 7. These devices facilitate optimizing parameters affecting liposome production for the development and production of anti-TB vaccines. This process led to the high yield production of liposomes, suitable for gene delivery and vaccination due to their unilamellar nature and particle size.

Figure 7

Schematic and photography of the microfluidic devices developed in ref. [59] depicting the single (a) and double (b) hydrodynamic focusing. Lipids are diluted in ethanol and injected in the center channel with a flow velocity of 143 mm/s. Deionized water flows through the exterior channels to hydrodynamically compress the central flow. In the case of double focusing, another flow-focusing stream is added to each side. Image reproduced with permission from Chemical Engineering Journal [59].

This study demonstrates the possibility of taking advantage of microfluidic devices not only for low-scale analysis or diagnostics but also for scaled-up production of pharmaceutical compounds, especially in the cases of vaccines or antibiotics.

4.2 Diagnosis of coinfection of TB and HIV

TB is the leading cause of death in HIV-co-infected patients, with a proportion of approximately 8.6% of TB patients co-infected with HIV in 2019 [1]. Additionally, HIV infection in TB patients increments the probability of activating a latent TB state between 80- and 100-fold [60]. Nevertheless, its diagnosis remains challenging in HIV co-infected individuals due to a high frequency of smear-negative disease, high rates of extrapulmonary TB, and reduced pulmonary load [18]. Traditional techniques do not provide rapid and accurate results to manage HIV co-infected patients effectively [61]. Despite its importance, there are almost no microfluidic detection methods to detect both diseases simultaneously.

A multiplexed nucleic acid amplification test (NAAT) was developed by the Fudan University of Shangai, portraying a microdevice for simultaneous detection of HIV, TB, and Pneumocystis carinii pneumonia (PCP) using a loop-mediated isothermal amplification strategy [62]. The PDMS device consists of a multiple loop-mediated isothermal amplification (LAMP) with the shape of a five-pointed star (Figure 8). The sample is loaded in the middle of the chip and distributed to the pools located at the points of the star by capillary forces. Three of the amplification pools are coated with LAMP probes against HIV, Mtb, and PCP. The fourth one serves as a positive control (coated with β-actin LAMP probe) and the last one as negative control (uncoated). Once the amplification pools fill with the nucleic acid sample and LAMP buffer, the inlet is sealed using uncured PDMS that requires on-site incubation. Results can be checked within 90 min with the naked eye (supporting its utility in low resources areas) by the appearance of white residue generated during the LAMP reaction.

Figure 8

Schematic representation of the five-channel µLAMP-based microfluidic device. The amplification pools in the channels are filled with the sample and the LAMP reaction buffer, then sealed. Positive samples form a white magnesium pyrophosphate precipitate that can be assessed with the naked eye or using agarose gel electrophoresis. Image reproduced with permission from Talanta [62].

Additionally, green fluorescence is produced by SYBR green I and can be detected using agarose gel electrophoresis. The detection limit of the multiplexed chip is 50 copies/µL for HIV and 100/µL copies for Mtb and PCP, respectively. Moreover, no nonspecific amplification or cross-reactivity was detected when samples of the diseases were tested using nucleic acids of different viruses as templates.

This platform is presented as a proof-of-concept for resource-poor countries. However, the accuracy of the primers and the rates of false positives and negatives needs to be characterized.

5.1 Organ-on-a-chip and organoids

Advancing towards the elimination of TB requires improving the diagnosis and identification of drug resistance and a deeper understanding of the interactions between the pathogen and the host cells in the pulmonary tract during infection and drug treatment [63]. Mtb presents a sustained interaction over time with host cells [64], which displays a spatial organization of the host–pathogen interaction [65]. Moreover, the extracellular matrix influences the survivability of the infected host cells [66]. Additionally, interactions with the host cells and environment affect bacterial physiology, inducing stress-mediated variations in their genetic expression [67]. These factors disclose the importance of developing novel models representing the diversity of microenvironments and elements that mediate the interaction between pulmonary cells and Mtb [68]. Animal models or human 2D cellular models have been primarily used for drug development and testing, pharmacokinetics, and research in general. Basic 3D structures of lung tissues have also served to model the spatiotemporal features of the diseases [69,70].

Organoids and organ-on-a-chip are more complex models featuring 3D cellular organization that has emerged in the last decade to bridge this gap between 2D cell cultures, animal models, and the actual 3D microenvironments. These technologies and their potential applications in the pharmaceutical analysis have been extensively reviewed elsewhere [63,71–74]. However, despite their strong potential for TB research, barely any endeavors have been dedicated to this field yet. Most lung-on-chip models will require adaptation for infection and security requirements in general, particularly TB-specific conditions, while organoids have not been used to analyze Mtb infections to date [71]. On a related note, a microsphere system combined with a microfluidic plate has been developed to model the effect of dynamic antibiotic concentration on Mtb [75]. A suspended microfluidic platform has also been used as a granuloma infection model [76]. Another example is the use of a commercial lung-on-a-chip model to analyze the interactions between Mtb and lung at the early infection stage [77].

Regarding the former [75], 3D microspheres were prepared by bioelectrospray methodology [78] using primary human cells (monocytes and T cells), active Mtb, and type I collagen on the microfluidic device, leading to the formation of human granulomas after 14 days. The system modulated the drug concentrations over time to replicate the pharmacokinetics observed in vivo during patient treatment. The granuloma’s genetic expression and the pharmacokinetics of antibiotics were measured and compared with the liquid culture, demonstrating the differences in the treatment outcome. The microfluidic devices were manufactured from a plexiglass sheet by micromilling, with inlet and outlet channels, and were then introduced in the pits of a 24-well plate before microsphere seeding.

The granuloma environment was also mimicked in ref. [76] using an in vitro suspended microfluidic platform adapted from the previously described Stacks system [79]. This system is based on capillary pinning to enable the stacking of layers with different microenvironments and uses open culture wells, which also allow easy imaging. Furthermore, the system does not require pumps (as traditionally occurs in microfluidic devices) because it functions based on surface tension and capillary forces. In this study, the Stacks system was used to create an infection model to study signaling microenvironments, combining human monocyte-derived macrophages and M. bovis BCG. The stacks were suspended in a 3D collagen extracellular matrix (ECM) and cultured to initiate paracrine signaling (Figure 9 left).

Figure 9

3D organ-on-a-chip models for Mtb interaction with lung tissue. Left: Workflow showing the setup of a microfluidic device containing granuloma microspheres grown in a suspended microfluidics “Stack system” from ref. [76]. Immune and Mycobacteria cells arrange in collagen matrix into a 3D granuloma model. Right: Schematic representation of the lung-on-a-chip from ref. [77]. Alveolar epithelial cells and macrophages cover the upper face of the membrane in the air-filled alveolar channel, while endothelial cells reside in the lower, liquid-filled channel. The tissues able to maintain this interface are infected with Mtb and imaged with confocal microscopy. Images adapted with permission from Front Bioeng Biotechnol. (left, [76]) and eLife (right, [77]).

Another model using spheroids is presented in ref. [80], generated first by the combination of human monocyte and macrophage spheroids, subjected to different Mycobacteria strains. Then, the complexity of the model increased by incorporating peripheral blood mononuclear cells and lung fibroblasts. Finally, these 3D models were used to analyze the co-culture response to drugs such as Isoniazid and Rifampicin.

Concerning the lung-on-chip [77], commercial lung chips were manufactured in PDMS from Emulate (Boston, MA), then preactivated and coated with ECM solution according to the manufacturer’s instructions. Then, endothelial cells were seeded in the different channels on the bottom face of the porous membrane (the vascular channel) and alveolar epithelial cells (AECs) on the epithelial channel (Figure 9 right). The use of two different cell types allows recreating the air–liquid interface present at the alveoli. After organ maturation, macrophages are introduced to the epithelial channel and allowed to attach to the epithelium. The lung tissues seeded on the chip capable of maintaining the interface are then infected with Mtb, allowing the analysis of its development in macrophages and AECs. Time-lapse fluorescence microscopy is used to study the dynamics of early infection and quantify the intracellular progress of the bacteria. This lung chip was also used in ref. [81] with murine cells and time-lapse microscopy to elucidate the role of surfactants produced by the alveolar epithelial cells in reducing bacterial growth.

These platforms that are still being developed and optimized present a broad range of applications. In the field of TB, they offer the possibility to study pathogen behavior in a more realistic environment and strongly increase the capacities of DSTs. In this way, the effect of antibiotics on bacterial populations and pulmonary tissue grown in the chip or organoid can be quantified. Different research groups worldwide continue their research in these areas. Furthermore, these systems grant the opportunity to analyze and model the interactions between drugs, pathogens, and tissues. They could also be helpful for the study of coinfections, especially those of the respiratory airway.

Studying these coinfections has growing importance, as respiratory tract infections (i.e., avian influenza, coronavirus) have resulted in pandemic situations during the last decade. Although different meta-analyses have found connections between susceptibility to COVID-19 and the presence of TB [82,83], further studies will be required to investigate the correlations between the diseases to advance the fight against them. Thus, the development of these platforms (organ-on-a-chip and organoids) can significantly enhance the knowledge about the interactions between the diseases and speed up the research pipeline.

5.2 Conclusion

Microfluidic devices have been present in the biomedical field for some decades now, enabling the advance and development of multiple techniques and platforms due to their numerous benefits over traditional methods. They have roles in the basic research field and at the translational and clinical stage in diagnosis and drug testing and development.

In this manuscript, we have dissected the potential applications of microfluidics to improve treatment prescription or to increment our knowledge about TB disease. Microdevices can provide an advantage over traditional diagnosis methods due to their simplicity and ease of use without medical facilities. They can also be used to detect different resistant strains of the bacteria, allowing more personalized treatment for each patient.

Moreover, many of the reviewed platforms have POC testing potential, some of them optimized in terms of cost for their use in underdeveloped or resource-constrained countries. There is a growing tendency to reduce the prices, materials, and laboratory setting requirements, as can be observed in the development of lab-on-paper devices in combination with smartphones compared to more traditional PDMS chips. In this way, microdevices position themselves at the forefront of biomedical research, with promising potential for future developments and opportunities.

Several research groups over the globe have routinely incorporated microfluidic devices in their laboratories as a daily tool, constituting an important source of knowledge about the basic forms of the disease and their relationship with other illnesses that entail threats to humanity. Moreover, through the development of new organ-on-chip and other techniques, insights into the behavior of the bacteria within the human body and the host reactions can significantly improve our understanding of infection mechanisms. Microdevices also present new opportunities in therapeutic and drug development testing. They not only enhance the diagnosis and treatment guidance but also facilitate basic research and drug development (Table 1).