Rapid quality control for MAFs using FESEM images
To ensure successful, reproducible synthesis of MAFs, the main filter parameters should be checked for every synthesis batch. However, an easy and fast quality control for MAF has so far been missing. Mercury intrusion porosimetry was used successfully in the past, but no batch-to-batch control in a short timeframe is possible with this method. For this reason, the application of FESEM imaging analyzed with ImageJ to measure pore sizes and polymer globule diameter was tested. The following FESEM images for pore sizes (Fig. 5a) and polymer globule diameters (Fig. 5c) exemplarily show pristine MAFs from batches synthesized for this work. The observed mean pore size was 22.30 ± 6.30 μm corresponding well to the literature value of 22.5 ± 9.0 μm . Functionalized MAF-OH showed a mean pore size of 22.34 ± 5.58 μm corresponding well with the mean pore size of pristine MAFs. Thus, functionalization does not change the pore size. For the mean polymer globule diameter, a value of 4.61 ± 0.56 μm was found with a relatively homogeneous size distribution (Fig. 5d), indicating a continuous and homogeneous polymerization process. Thus, FESEM measurements allow for a rapid quality assessment for different MAF batches that is important to ensure a high reproducibility of the overall method.
Comparison of MAF functionalization types for P. aeruginosa filtration
Three different MAF functionalization types (MAF-OH, MAF-DEAE, and MAF-PmB) were tested under the respective standard conditions to identify the MAF functionalization with the most promising results for further experiments with P. aeruginosa. The standard conditions were MAF-OH: 10-L sample volume, pH 3, BEG elution buffer ; MAF-DEAE: 1-L sample volume, pH 7, BEG elution buffer ; MAF-PmB: 1-L sample volume, pH 4, carbonate elution buffer . An overview of recoveries is shown in Fig. 6. The number of P. aeruginosa cells on the MAF could not be calculated directly as no reliable concentrations of the filtrate could be determined by qPCR due to the low number of cells in the filtrate. Therefore, recovery was used to quantify the retention and elution process and, unless stated otherwise, calculated as the ratio of the total number of cells found in the eluate by qPCR after filtration and the total number of cells in the initial sample. Detailed information on how recoveries were obtained can be found in the ESM. Initial spiked concentration of P. aeruginosa was 108 CFU L−1. While MAF-DEAE and MAF-PmB only show a recovery of 0.04 ± 0.01% and 4.3 ± 0.3%, respectively, MAF-OH shows a remarkably higher recovery of 68.6 ± 7.4%. Unspecific retention of bacterial cells occurs only to a very minor extent as is evident by the recovery for MAF-DEAE (0.04 ± 0.01%) which is seen as a reference for unspecific binding in this work. Additional recovery experiments with MAF-DEAE and different elution buffers as well as different filtration techniques (one-time filtration, repeated filtration of the same sample, and circulating filtration) have been carried out (data in ESM) but did not yield significantly better results. MAF-PmB uses a surface-active antibiotic as the affinity ligand while MAF-DEAE and MAF-OH make use of electrostatic interactions between the MAF surface and the bacteria for retention of the analyte. The MAF-OH seems to offer the positively charged bacteria (due to acidification) high interaction possibilities for electrostatic interaction compared with the MAF-DEAE, where almost no interaction takes place between the MAF surface and the negatively charged bacteria surface (at neutral pH due to lipopolysaccharide (LPS) structures presented on the surface). The LPS structure in P. aeruginosa is typical for Gram-negative bacteria consisting of lipid A, inner and outer cores, and O-antigen with variations depending on the strain (detailed information in ) and thus is positively charged at pH 3 and negatively charged at neutral pH. The main interaction properties are stated to be hydrogen bonds. Based on these results, MAF-OH was chosen for further optimization of the filtration process.
Comparison of elution buffers
Five different elution buffers (BEG buffer, high-salt buffer, carbonate buffer, glycine, and Pluronic® F68 solution) were evaluated for filtration with MAF-OH. The applied MAF process was the same for all (10-L initial sample volume, sample pH 3, 20-mL elution volume). Spiked P. aeruginosa cell concentration was 108 CFU mL−1. Figure 7a shows an overview of the recoveries. While four buffers show relatively similar values with Pluronic® F68 solution (8.6 ± 0.3%), carbonate buffer (8.8 ± 0.7%), high-salt buffer (11.6 ± 0.6%), and glycine buffer (17.5 ± 1.5%), the combination of glycine with beef extract (BEG buffer) gives a significantly higher recovery of 57.0 ± 3.0%. This buffer combines the desorption properties of a change in pH (from pH 3 in the sample to pH 9.5 in the elution buffer) with proteins present therefore breaking the interactions between the cells and the monolith surface. While the change in pH changes the net charge of the bacteria, the protein uses van der Waals forces and hydrophobic interactions to remove the cells from the filter and therefore reduces the electrostatic interactions present in the adsorption process. Thus, BEG buffer was the elution buffer of choice for further experiments.
Optimization of initial sample volume
While the original MAF-OH filtration protocol for MAF with 3.86-cm diameter and 1-cm height states 10-L initial sample volume , it was aimed on finding a lower sample volume with reasonably high recovery to facilitate an easier sample collection. Here, three different sample volumes (1 L, 5 L, and 10 L) were tested under the same conditions (MAF-OH, pH 3 as sample pH, and BEG elution buffer, 108 CFU L−1 spiked P. aeruginosa). With increasing sample volume from 1 to 5 L, a notable increase in recovery from 23.0 ± 0.4% to 67.1 ± 1.2% was observable (Fig. 7b). One explanation for this is the distribution of the P. aeruginosa cells within the MAF: With a lower sample volume, the bacteria only attach to the first part of the cylindrical monolith structure while with higher sample volume, several absorption and desorption events occur before the cells finally stick to the MAF. This leads to an even distribution of the bacterial cells over the whole filter at higher sample volumes and thus higher recovery using the same elution protocol. However, a further doubling of the sample volume from 5 to 10 L only resulted in a minor increase in recovery up to 68.6 ± 7.4%. As is evident from the differing recoveries for the 10-L samples from both experiments, an interbatch variance can be seen, while the intracharge variance is relatively low. Therefore, it is important to perform quality control of each MAF charge to prevent variances in recovery values. Due to the lower filtration time, the easier handling, and the higher reproducibility (as indicated by the lower standard deviation), the optimal filtration volume for the assay was set at 5 L for P. aeruginosa filtration.
Adjustment of sample pH
To evaluate the efficiency of the filtration procedure at different pH values with MAF-OH, sample pH values between pH 3 and pH 4 (in 0.2 intervals) as well as pH 5, 6, and 7 were tested. Beforehand, culturability of P. aeruginosa under the tested conditions in tap water was investigated, as the goal was to find a filtration procedure that would allow subsequent detection via culture. This would then enable the use of cell culture as a confirmatory quantification method, as it is still the gold standard method despite the high time expenditure. For P. aeruginosa, growth on agar plates was observed for all pH values above pH 3.3 (overgrown after overnight incubation) and no growth was observed for pH values below pH 3.2 (no colonies visible), indicating a change in physiology at these pH values. This result is comparable with the result of another study investigating the effect of low pH values on the survival of P. aeruginosa  while a different study found good bacterial growth for pH higher than 3.8 . For all filtrations, the cultivated P. aeruginosa samples were acidified before filtration and the optimized protocol (MAF-OH, 5-L initial sample volume, BEG elution buffer) was used. Spiked P. aeruginosa cell concentration was 108 CFU L−1. As is visible in the relative recoveries with pH 3 set to 100% displayed in Fig. 8a, good retention and elution of the bacteria could be achieved for a sample pH of 3.0 only. Filtrations at pH 3.2 and higher showed significantly lower recoveries with the lowest recovery being at pH 4.0 (pH 3.2: 26.7% ± 5.3%; pH 3.4: 32.5% ± 12.72%; pH 3.6: 34.4% ± 15.45%; pH 3.8: 17.7% ± 8.4%; pH 4: 11.6% ± 9.3%). Relative recoveries for pH 5, pH 6, and pH 7 were 23.2%, 17.0%, and 10.6%, respectively. Thus, it was concluded that at filtration conditions where P. aeruginosa are culturable, no good retention and following elution is possible. The recoveries show higher values than for unspecific binding; hence, a specific interaction between the cells and the MAF surface takes place. Spiked cultivated P. aeruginosa which were non-culturable at pH 3 seem to be adsorbed on the surface of MAF-OH much better than culturable cells at higher pH. To evaluate the correlation between culturability and filtration efficiency, additional experiments using both heat-inactivated P. aeruginosa and viable cells for spiking were carried out (recoveries in Fig. 8b). For comparison, pH 3 with cells that were culturable prior to acidification was set at 100% recovery (standard deviation ± 27.2%). At a sample pH of 5, the relative recovery for heat-inactivated and culturable cells was 9.4% ± 4.1% and 8.5% ± 5.7%, respectively. This leads to the conclusion that the filtration efficiency correlates with the sample pH value but not the culturability status of the bacteria. A change in outer membrane chemistry and consequently a change in the interaction between the monolith’s surface and the outer membrane might induce the higher retainability at pH 3. As P. aeruginosa cells are not culturable after acidification to pH 3, a live-dead discrimination or detection via culture after filtration using the current setup is not possible, but culture-independent detection methods can be used for the successful quantification after filtration.
Calibration of the final setup
Based on the optimization process, a calibration using the best filtration conditions for highest recovery of P. aeruginosa was carried out. At a sample pH of 3, 5-L sample volume, filtration with MAF-OH, and elution with BEG buffer, five different concentrations of P. aeruginosa spiked in tap water (1·104–1·108 CFU/L) as well as a blank sample (tap water) were concentrated. The linear range of the calibration is displayed in Fig. 9. A concentration factor of 3·103 in under 1 h was achieved by reducing the initial volume of 5 L to a final volume of 1.5 mL after filtration and centrifugal ultrafiltration. The filtration efficiency (recovery rate of 67.1 ± 1.2%) is included in the calibration line so that it can be used to calculate the initial sample concentration from the total number of detected cells in the eluate with the quantification method of choice.