Basically, the simple idea underlying the experimental setup is to facilitate the contact between the bacteria in the superficial microlayer and a bioactive surface. This is accomplished by means of a step motor that provides a rotatory motion to the solid support bearing the bioactive surface, which is partially immersed into the bacterial contaminated urine samples. As a result, bacteria adhere to the bioactive surface onto a region precisely located at the interface between air, liquid, and solid substrate.

The device, outlined in Fig. 1, consists of different connected parts: a step motor that generates a rotatory motion applied to the capturing surface; a holder for the solid substrate with the capturing surface; a holder for the bacterial suspension; and a simple electronic controller to set the rotatory speed. The overall system is placed in a temperature-controlled environment.

Various factors influence the capturing method presented here. The first and most important factor is the bioactive surface employed in the process. Different types of bioactive surfaces exhibit different efficiency in promoting bacterial adhesion to them, so it is essential to use surfaces with high chemical affinity to the bacterial wall. However, even for low efficient bioactive surfaces, it is possible to optimize other factors to maximize their bacteria capturing ability. For any chosen bioactive surface, the number of bacteria captured is influenced by the rotation speed of the capturing surface, the capturing time, the temperature of the sample, and finally by the concentration of bacteria in the sample.

The experiments reported here were performed with two types of commercial microscope slides: Thermo Scientific${}^{TM}$ Polysine (Fisher Scientific - Milan - Italy) adhesion slides that are electrostatically and biochemically adhesive; and the positively charged Klinipath (Avantor delivered by vwr – Milan - Italy) slides. These slides are commercial products for histological applications and they have a certain grade of chemical affinity for cells. Both types of slides have shown similar capability to capture bacteria from their suspension with our system. Prior to their use, the slides were examined with an optical microscope to verify the homogeneity of their surface, i.e., an extremely smooth surface not exhibiting any type of structure visible under a 40$\times$ optical objective lens. The slides that did not pass this visual inspection were discarded. Subsequently, the slide was mounted onto a holder, which is connected to the step motor, and partially immersed in a sample of 70 mL of urine, paying attention to avoid bubble formation close to the slide surface. In the capturing phase, the following parameters were set: rotational speed of 60 rpm; capturing temperature 25 $\pm$ 1 ${}^{\circ }$C, and capturing time of 30 minutes.

The contaminated urine samples analyzed in this experiment were obtained from generally healthy individuals with urinary tract infections (UTI) caused mostly by Escherichia coli. The donors supplied the samples to a microbiology laboratory for assessment of UTI and were informed about the secondary use for research purposes of their urine samples and a consent was obtained. The samples’ bacteria concentrations were also determined with other two techniques: bacterial culture tests, and direct microscopic count. To perform the viable count to determine the number of bacteria in the culture, aliquots of urine dilutions were plated onto standard plates count agar (Avantor delivered by vwr – Milan - Italy), spread homogeneously over the plate, incubated at 37 ${}^{\circ }$C for 24 h, and the number of colonies counted manually, following the guideline of the European Association of Urology [23]. While, to perform the direct microscopic count were used a Bürker-Türk counting chamber, which has a volumetric grid divided into differently-sized cubes useful for accurately counting the number of bacteria in a cube by means of an optical microscope, and then calculating the concentration of the entire sample. Every bacterial culture tests were replicated three times at four different dilutions: 1/100, 1/1000, 1/10000, and 1/100000.

In Fig. 2 is displayed a capturing system working with three capturing surfaces simultaneously. After 30 minutes of rotation at 60 rpm, the slides were removed from the suspension and washed with 1 mL of MilliQ water obtained with a Millipore system, then gently dried with a flow of nitrogen, and lastly examined with an optical microscope using first a 5$\times$ optical objective lens to detect the whole bioactive surface. The bioactive surface, in correspondence with the air-liquid-solid interface, presents a clearly visible transverse band across the slide. Further investigations using a 40$\times$ optical objective lens and an atomic force microscope (AFM) have revealed that the band is made of bacteria. The washing step is essential to obtain a reliable bacterial count in the following phase of bacterial density determination. Since urine samples are rich in salts and environmental debris, washing the slides after the capture phase cleanses the bacteria band from most of the debris that may be confused with bacteria during the counting phase. The effectiveness of the washing step was validated by comparing 3D topographic images of the captured bacteria before and after the washing step. The 3D images were collected with an Atomic Force Microscope (AFM) Dimension 3100 Veeco, operating in Tapping Mode in air. The AFM images analysis confirmed both the effectiveness of the washing step and that the band identified with the optical microscope consists of bacteria, which shapes are easily recognizable in the AFM images. Subsequently, a set of at least 30 optical images along the bacteria band were collected using a 40$\times$ optical objective lens and a digital camera connected to the microscope. Moreover. all optical microscope images were collected using a green filter to improve their quality. These images were analyzed to evaluate bacterial density on the surface by counting the bacteria in the frame of the images by means of ImageJ, a freeware image processing software [24]. The surface density of bacteria obtained with our experimental setup was compared with the concentration of bacteria obtained using the bacterial culture method and the direct microscopic count method, to assess whether a relation between the amount of captured bacteria and their concentration in the sample occurs.

The study of direct effect of drying processes on microorganism has revealed high tolerance of bacteria to these processes [25]. To verify that the bacteria were alive after capture, a bacteria growing test was implemented. Two sets of slides were sterilized with bleach. The first series of slides were used to capture bacteria from contaminated urine samples, while the other series of slides were used in blank capturing experiments, i.e., using sterile samples. Those experiments were performed in parallel and handled in the same way and under highly controlled, sterilized conditions in order to avoid any environmental contamination. Sample sterilization was accomplished using a biological safety cabinet equipped with UV lamps. After the capturing step, the slides were placed in 50 mL sterile tubes filled with highly nutrient liquid medium, Tryptone Soya Broth (Avantor delivered by vwr – Milan - Italy), and incubated overnight at 37 ${}^{\circ }$C. The presence of viable bacteria was based on the growth of the captured bacteria and on the turbidity of the culture broth.

The most remarkable result emerging from the optical images analysis, collected under 5$\times$ magnification, is a clearly visible, bright band that runs across the slide and parallel to the free surface of the urine samples (Fig. 3).

The band is made of bacteria captured from the urine contaminated samples, easily identifiable in the zoomed round areas (see Fig. 4) obtained under a 40$\times$ optical objective lens. Moreover, with a fixed set of capturing parameters, the width of the bacteria band varies in relation to the bacterial concentrations in the sample. It can range from a few tens to several hundreds of micrometers. The images collected along the band with a 40$\times$ optical objective lens were used to measure the average bacteria surface density. All images have a frame size of 250 $\mu$m $\times$ 140 $\mu$m, thus it was possible to determine bacterial density in terms of number of bacteria per frame.

Atomic force microscopy (AFM) analysis was performed to verify that following the washing step the captured bacteria do not include environmental debris. This analysis compared AFM-images collected before and after the washing step. The AFM analysis confirmed that the particles captured into the band are bacteria, which are easily recognizable in the AFM images. The results of these measurements are shown in Fig. 5. The presence of environmental debris is evident in the AFM images collected before the washing step, highlighted by red arrows in Fig. 5A, while there is no evidence of significant presence of environmental debris in Fig. 5B collected after the washing step. In Fig. 5A and Fig. 5B the different scale range of the z-axes is due to the autosetting of the AFM acquisition software, which sets automatically the z-axis range in order to obtain the best final image. These results assure that, after the washing step, the majority of the particles forming the band are captured bacteria, i.e., it is clear that the debris greatly decrease with the washing process. The image processing software ImageJ (version 1.52r), which is a freeware image processing program developed at the National Institutes of Health and the Laboratory for Optical and Computational Instrumentation (LOCI, University of Wisconsin, USA), was used to count the number of bacteria per frame. Since bacteria captured onto the bioactive surface tend to form clusters, clusters comprising a maximum of six bacteria were also included in the counting procedure. ImageJ can count bacteria through the command “COUNT PARTICLES” on 32 bit black and white images and it allows setting the area range of the particles. This procedure is similar to the one reported in a previous work [26].

The average of the bacteria surface density (bacteria/frame), i.e., the average number of bacteria captured per image, is used to verify if a correlation between the bacteria concentration in the urine samples and the bacteria captured onto the slide surface exists. Therefore, to evaluate the existence of a possible correlation between the numbers of bacteria captured onto the slides and their concentration in urine contaminated samples, a set of 24 capturing experiments were performed. In 12 of the contaminated urine samples, bacterial concentration was determined by bacterial culture method, while for the other 12 samples it was determined by direct microscopic count method. For each of the 24 samples, the capturing experiment was replicated twice, once using the Thermo Scientific${}^{TM}$ Polysine adhesion slides, and once using the positively charged Klinipath slides, and the average value of the bacteria surface density was counted, since there were no significant differences in the number of bacteria captured with the two types of slides. Graphs in Fig. 6 show the dependence between the concentration of the bacteria in contaminated urine samples and the bacteria surface density determined with our method. Indeed, the implementation of a calibration curve based on bacteria concentration vs bacteria surface density, allows determining the unknown bacteria concentration of contaminated urine samples. In particular, Fig. 6A and Fig. 6B show the relations between the bacterial concentrations obtained with the culture bacteria method and with the direct microscopic count method, and the average bacteria surface density obtained with the capturing system/method here described. Ordinary least square regression was used to determine the best fit line for both sets of data, with R-squared of 0.99 and 0.90 respectively (dashed line).

Table 1 reports the dataset of measurements performed with contaminated urine samples used in the regression analysis. In contrast, the microscopic analysis of capturing slides from uncontaminated samples does not shown presence of bacteria. The detection limit of this technique depend on the capturing parameters and for the experiments here reported the detection limit was $\le$10${}^3$ CFU/mL. The uncertainties associated with these experiments are slightly different: in the case of the bacterial culture method the error is around the 20% [27], while the error associated with the direct microscope count method, using a Bürker-Türk counting chamber, is around 15% [28]. The error estimated for bacteria surface density measurements is 5%, calculated from the standard error of the mean of the bacteria count data.

To properly determine the most suitable values of the capturing parameters for our apparatus, i.e., speed, time and temperature, sets of experiments were performed to examine the effects of these parameters on the efficacy of the capturing phase.

The first parameter analyzed was rotation speed. From the capture experiments where only the capturing slide rotational speed was changed, it is evident that a directly proportional effect exists between speed and capture efficacy, below 150 rpm. Indeed, speeds higher than 150 rpm produce irregular bands of captured bacteria, as shown in the optical images of different bands of captured bacteria reported in Fig. 7A. Rotational speeds comprised between 30 rpm and 160 rpm lead to an evident increase of the superficial density of captured bacteria, as illustrated in Fig. 7D to Fig. 7B; while at higher rotational speed are evident irregularities of the bacterial surface density (see Fig. 7A). Therefore, a rotational speed between 50 rpm and 70 rpm seems to be the most suitable for our apparatus, whose target speed was set to 60 rpm. This rotational speed exhibited a positive correlation between the suspension bacterial concentration and the surface density of bacteria captured.

Additionally, it was observed that capturing time influences the number of bacteria captured, with greater numbers of bacteria being captured by the bioactive surface over longer capturing times. Nevertheless, a short capturing time is desirable to devise a quick, reliable method, thus making it necessary to find a balance point between the amount of bacteria captured and the time necessary to capture them. To find the most suitable capturing time, three bacterial suspensions with concentrations of 2.0$\cdot$10${}^5$ CFU/mL, 1.5$\cdot$10${}^6$ CFU/mL, and 8.2$\cdot$10${}^6$ CFU/mL, were used in a set of capturing experiments. Capturing temperature and rotational speed were held constant in all experiments, while capturing time was progressively reduced from 120 min., to 60 min., to 30 min., and to 15 min. The results of these tests are displayed in Fig. 8. In all cases where capturing time was greater than 15 min. it was possible to detect a clear band of captured bacteria, whereas the 15 min. capturing time did not allow for the formation of a captured bacteria band in the two less concentrated suspensions. Fig. 8 shows that the increase of capturing time produces a consequential increase in the amount of captured bacteria, and the time suitable for a rapid capturing phase can be chosen in the range of 30–60 minutes. Moreover, from tests performed at a longer capturing time, it clearly emerges that a sort of saturation point of the bioactive capturing surface is reached, that is, after 200 min. there are no further increases in bacteria surface density.

Finally, the capturing temperature influences positively the amount of bacteria captured [29, 30], i.e., increasing the temperature from 20 ${}^{\circ }$C to 40 ${}^{\circ }$C produces an increase on the amount of captured bacteria as it is possible to see in Fig. 9.

Likely this increase is due to two different effects: the first is that the increase in temperature produces an increase in convective thermal motion inside the bacterial suspension; and the second is associated with the bacterial growth which is higher at 40 ${}^{\circ }$C than at 20 ${}^{\circ }$C. Therefore, there is an increase of bacteria number in the suspension at 40 ${}^{\circ }$C during the capturing phase. The capturing temperature of 25 $\pm$ 1 ${}^{\circ }$C, close to the room temperature, has the double advantage of limiting the environmental thermal fluctuations and the bacterial growth.

Following the procedure previously described, the set of slides used to capture bacteria from contaminated samples showed bacteria growth, while the set used in the blank experiments did not show any trace of bacterial growth (Fig. 10). The sterilized environmental conditions in which all experiments were performed exclude that the cloudy appearance is due to external bacterial contamination. Indeed contamination would have led to bacterial growth also in the blank experiment samples, since the two sets of experiments were performed in parallel. Therefore, this result confirms bacterial viability after capture. Moreover, samples from both experiments were analyzed by means of optical microscope and the presence of bacterial growth was confirmed only for the slides used with the contaminated samples, while no presence of bacteria growth was detected in the blank experiments.