Figure. 3: Comparison of biofilm formed at ALI () and LLI () assessed by CFU/mL and the percentage of the attached to the total. (n=18, mean ± SEM) ** p<0.01 (using unpaired t-tests).
No significant differences were observed between the biofilms grown on ALI and LLI models when comparing the total CFU number. However, when comparing the proportion of attached bacteria relative to the total (attached + detached), a significantly higher proportion of biofilm formation was observed for biofilms growing on ALI than LLI. While the overall volume of biofilm in ALI and LLI were similar, biofilms grown on ALI are more resilient and more difficult to detach from the membrane. Conversely, biofilms grown on LLI are more vulnerable. Hence the percentage of detached bacteria is greater. Also, the variability for biofilms grown on LLI is wider, suggesting that these biofilms are more heterogeneous.
Biofilms can adapt their architecture to cope with different hydrodynamic conditions and nutrient availability [24]. The interface on which the biofilm grows specifically determines the availability and permeability of nutrients and thus will have an impact on the morphology and properties of the biofilm. Several studies have attempted to investigate the properties of biofilm formed on a particular interface. Wu et al. [25] conducted interfacial rheological measurements on ALI biofilms produced by Escherichia coli and found that hydrophobic curli fibres associated with the bacteria improve strength, viscoelasticity, and resistance to ALI biofilms. Another study by Rühsa et. at. [26] measured the transient elasticity and viscosity of the biofilm formed at the water-oil interface using interfacial rheology and pendant drop tensiometry and found that the ability to form biofilms against oil improved with increasing hydrophobicity. However, these studies are unable to provide comparative studies on the effect of the interface type using a single platform to control the variables, as most of the instruments are designed to investigate only one specific interface.
While microfluidic instruments have been developed to examine the effect of external forces such as the hydrodynamic shear force on the growth of biofilm [27], most of these devices are irreversibly bonded, which severely undermine the flexibility and potential of incorporating other end-point analysis approaches such as the approaches we adopted in our study. The dual-chamber microreactor presented in this study overcome these challenges by having a design that can be disassembled and with the capabilities to incorporate multiple interfaces with interchangeable membrane while keeping other variables (e.g., dynamic flow conditions) constant between experiments.
The morphology of 48h-old ALI and LLI PAO1 biofilms were also visualized using CLSM and SEM (Fig.4). The 3D reconstruction of z-stack CLSM images (Fig. 4 A and D) show that a layer of biofilm was formed over the membrane and the thickness of the biofilm was approximately 5μm (Fig. 4 B and E). Mushroom-shaped bacterial aggregates dispersed across the membrane were observed in SEM images in both ALI and LLI models for PAO1 biofilms (Fig. 4 C and F), and their shapes are more distinct in the ALI compared to the LLI. This is in good agreement with Moller, S., et al. [5] which demonstrated that biofilms grown under nutrient depletion are highly structured, and the ’mushroom-shaped’ microcolonies are interspersed between open water channels. In our study, biofilms have been grown in static conditions with a finite amount of culture media supplied, resulting in nutritional exhaustion conditions towards the later stages of bacteria growth. Moreover, in comparison to ALI culture, LLI culture has more nutrients coming from the top chamber. Thus, the mushroom-shaped microcolonies showed in our SEM images in both the LLI and ALI models, and the difference between them corroborated Moller ’s [5] discoveries.
Biofilms have also been known to have open water channels that facilitate the diffusion of nutrients and waste products between the external environment and biofilm and within the biofilm matrix [8]. These channels are also evident in our CLSM images (black space unstained with green and red) and SEM images (black holes in the biofilm). Comparing Fig. 4 B and E, the structure of the LLI model biofilm is more dispersed, while the ALI model biofilm is more aggregated. Also, more water channel holes can be seen in the LLI model biofilm SEM images. We hypothesize that the media transmission occurring in the LLI model is higher than the ALI model, enabling more water channels to be formed to facilitate the transport of nutrients and waste. Differences in the structure of the biofilm matrix could contribute to the differences in mechanical properties, which has been demonstrated through the scattered structures (LLI model biofilm), making it more vulnerable to disruption of physical forces. Consequently, the adhesion strength between the biofilm and the underlaid surface and the cohesive force within the biofilm is weaker for the LLI biofilm relative to the ALI biofilm. This further supports the findings in Fig. 3 that LLI biofilms are more susceptible to physical removal.