Immobilizing Siderophores on Solid Surfaces for Bacterial Detection

The article reports a proof of concept validation of bacterial detection using Siderophores. Desferrioxamine B (DesfB) was used as the Siderophore to capture Escherichia coli on gold coated microcantilever surface. Self-assembled monolayer based gold thiol chemistry was used for surface functionalization of the Siderophore on top surface of the microcantilever. The bacterial attachment to the siderophore was observed through Fourier Transform Infrared Spectroscopy (FTIR) and Fluorescence Microscopy. The DesfB coated surface allows only the live bacterial cells to be attached and it is evident through the FTIR band formation. In a mixture of live and dead E. coli cells, the Fluorescence Microscope image indicates green emission from live cells and a core-shell structure formation upon progress of time. For a sample dilution of 10 − 1 , the mass change of live E. coli bonded to Siderophores is four times higher than that of dead cells and 12 times higher to that of negative control on microcantilevers. Therefore this study should be considered as a foundation to build a miniaturized biosensing platform to distinguish between bacterial or viral infections in real time. The proposed platform could differentiate between bacterial and viral infections thus rendering it as Point of Care (PoC) diagnostic tool aiding Internet of Things (IoT) applications.

Current strategies, whether in the detection of bacterial agents in the environment (air, water, and biothreat/attack) or in clinical diagnosis, suffer from many limitations. 1,2 Many of the current methods rely on the detection of specific biomarkers rather than the intact pathogen itself. 3 In environmental detection, there are pathogenspecific biomarkers, whereas in clinical diagnosis both host and bacterial biomarkers are used. There is a need for pathogen-culture for confirmatory diagnosis for these methods. PCR and culture confirmation for viability are laboratory intensive procedures requiring at least 18 hours to complete. 4 Antibody based methods though preferred are associated with a high rate of false positives, possibly due to the rapid mutation rates of pathogenic strains. Further, antibodies cannot always distinguish between debris and viable pathogens. 5 In addition, antibodies suffer from instability upon prolonged storage. Fast and specific interrogation of bacterial viability is critical in many fields including food safety, rapid determination of species in infectious diseases, efficacy assessment after initial medical intervention to infection, and detection of exposure to a biological threat agent.
Siderophores are low molecular weight, iron-chelating ligands produced by microorganisms, such as bacteria and fungi but not viruses. 6 Siderophores (SDPs), molecules employed by the bacteria to sequester vital iron within a host, show a great promise to selectively interrogate viable bacteria. 7 Iron is a key nutrient for survival of the bacteria. Free iron is rare in nature to scavenge. Bacteria synthesize and release SDPs to bind iron with high affinity, up to 10 52 M -1 . 8 Iron-bound SDPs are recognized by specific cell surface receptors and internalized to the periplasm, where processing and degradation transfers the iron to the bacterial cell. This nutrient sequestration route is critical for bacterial survival. The hypothesis governing the SDP approach is that binding of iron-loaded SDPs to the bacterial cell is a high-affinity event that requires an intact cell membrane, aka viability, and can be extended to all pathogenic bacteria. Hence, SDPs can be used to 'fish' for viable bacteria. SDP based detection of bacterial viability has been demonstrated through fluorescence microscopy. 9,10 The limitations of this optical method include cost and the inability to detect multiple species concurrently.
The ability of microcantilever sensors to perform relatively fast, easy, small volume analyte requirement, and label-free analysis of biological species justifies the growing interest of the biosciences community in these devices. 11 The microcantilever sensors have been successfully applied in studies on single cell estimation, 12 modeling of drug action mechanisms, 13 cancer research, 14 and detection of pathogens and nerve agents. 15,16 In this study, the authors envision to demonstrate the sensing of bacterial viability through microcantilevers enabled by SDPs. Furthermore, the focus of this work to show that it is possible to implement SDP based differentiation of live bacteria from dead cell by resonant frequency change of the microcantilevers due to mass loading rather than being aimed at detecting very low concentrations (such as picomolar) of bacteria or to selectively differentiate certain bacteria from other microorganisms.
In this research, Desferrioxamine B mesylate salt (DesfB), a bacterial SDP characteristic of Streptomyces species, is used to detect live E. coli and differentiate it from dead bacterial cells. The binding of DesfB to the E. coli SDP receptor, FhuD, is well documented, including the reported crystal structure of DesfB bound to FhuD. 17 Mass loading on the microcantilever is taken as the sensor signal. The mass loading is recorded after each surface treatment of the microcantilevers. Human IgG (Immunoglobulin-γ) was used as the negative control. IgG, an antibody, is a protein and similar to the live bacterial cell surface receptor. To differentiate these two proteins, IgG was taken as negative control.

Materials and Methods
DesfB functionalization.-P-type silicon wafer (100) and microcantilever arrays (IBM, Zurich, Switzerland) consisting of eight identical cantilevers of dimension 750 × 100 × 1 μm 3 were used as samples. Due to the fact that all eight cantilever beams were attached in a single array, functionalization of all beams were done together and it assumed that all beams were functionalized uniformly. The first step of the functionalization process is to deposit metal on the cantilever. 5 nm chromium layer and 20 nm gold layer were deposited sequentially on both the wafer and microcantilever sample surface through E-beam evaporation system (model no: BC-300T). The silicon wafer was diced into 1 cm x 1 cm size samples by ADT dicing instrument. Then, individual Si wafer samples (1 cm 2 ) and the microcantilevers were cleaned with acetone, isopropyl alcohol and finally rinsed with ethanol and deionized water (DI water) separately. Next, the samples were submerged in ethanol solutions containing 1 mM of 11-mercapto undecanoic acid to form self-assembled monolayer (SAM) for 12hr. After the exposure period, the individual samples were washed in hot ethanol and DI water environment.
To covalently immobilize the DesfB to the SAM adsorbed on the gold surface, the SAM terminated surface was activated with a 1:1 volume mixture solution of 0.1 M N-hydroxysuccinimide (NHS) and HCl stabilized 0.4 M 1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide (EDC) for 1 hr at room temperature and finally washed with DI water. Each sample was immersed in a solution containing DesfB, sodium carbonate, and FeCl 3 (1 mM each) for 2 hr at room temperature and then subsequently rinsed with DI water. This iron chelated DesfB immobilization strategy utilizes the Siderophores' terminal amine, which is not critical to iron binding.

Preparation of live and dead bacterial suspension: attachment on
wafer.-E. coli was grown overnight in Luria Bertani media at 37 • C. For quantitative determination of bacterial population, a standard agar plate count method was performed with serially diluted bacterial stock suspended in sterile DI water environment. A wide series of dilutions (10 −1 -10 −10 ) were plated for the experiment and plates with bacterial colonies in the range 30-300 were accepted for further experimentation (considered to be in the range where one's data is statistically accurate). Of the accepted serial dilution, two sample dilutions (10 −4 and 10 −7 ) were chosen for the contrast in viable bacterial cells per ml. The number of bacteria per ml in 10 −4 and 10 −7 dilutions were 3000 and 310 respectively. The average number of bacterial cell per ml of stock solution is 157 × 10 7 CFU/ml, calculated from the measurement of distinct colony-forming units (CFUs) of viable bacteria of each agar plate. Additionally, 10 −1 dilution was also selected for its high bacterial load but due to its high bacterial concentration it was too numerous to count the CFUs. For dead bacterial dilution, the bacterial cells were harvested, re-suspended in phosphate buffered saline (PBS), and incubated with 70% isopropanol for 2 hrs after which the cell suspension was washed thoroughly with PBS. This method is a standard microbiology technique where bacterial cells are killed by way of internal protein coagulation. Cell viability study was confirmed by incubation (18 hr) of both live and dead bacterial suspensions in Luria Bertani agar media at 37 • C. Live and dead bacterial suspensions were labeled individually and in a mixture using a Live/Dead Bac-Light bacterial viability kit. The kit differentially stains living and dead bacterial cells.
It consists of two nucleic acid stains, SYTO 9 and Propidium Iodide. SYTO 9 (green) penetrates most of the membranes freely whereas Propidium Iodide (red) only penetrates the damaged membranes and competitively reduces SYTO 9 fluorescence when both the dyes are present. Thus, viable cells are dyed green and dead cells are dyed red when visualized under a fluorescence microscope.
To identify if the bacterial strain used in this experiment scavenges for SDPs, both live and dead bacterial suspensions were incubated on the DesfB functionalized wafer sample for 2 hrs. After immobilization the surface was thoroughly washed with PBS with 0.01% Tween-20 and PBS to remove the unbound bacteria as well as non-specifically associated components, while preserving the dye activity assay. It is known that DesfB is able to bind with live E. coli cell only. So in order to set controls, the live E. coli cell used in this experiment was taken as positive control and dead E. coli cell as well as human IgG (other than bacteria) was taken as negative control. Both human IgG and live E. coli cell surface receptor is a protein. To differentiate these two different proteins, human IgG was taken as another negative control.

Attenuated total reflectance fourier transform infrared spectroscopy (ATR-FTIR) study.-To
gain more detailed molecular bonding information regarding DesfB immobilization on gold coated samples, ATR-FTIR measurement was performed. The spectra were taken by averaging 20 scans at a resolution of 4 cm −1 . Reference spectrum was confined to the spectrum obtained with a bare silicon sample. All measurements were taken at room temperature.
DesfB bound E. coli fluorescence microscopy analysis.-The fluorescence images were collected from inverted Fluorescence Microscope (Olympus) using the respective bandpass filters for viewing both SYTO 9 and Propidium Iodide separately. Bacterial viability kit L7012, contains separately SYTO 9 dye (3.34 mM, Component A) and Propidium Iodide (20 mM, Component B) in 300 μL DMSO solution. Equal volumes (1:1) of component A and B were combined in a microfuge tube and mixed thoroughly before use. 3 μL of mixed dye mixture was added for every 1 mL of PBS buffer (pH 7.4).Exclusively, 20 μL of the dye mixture was coated on individual wafer sample which were already mounted by both positive and negative control for 15 minutes under dark ambience. Excess amount of unbound dyes were washed by PBS buffer twice.
Mass loading on microcantilever.-Experiments were performed in Cantisens CSR-801 system supplied by Concentris Inc. All measurements were done in DI water environment. The sample chamber has a volume of 5 μL and the temperature set in the chamber at 25 • C. DesfB functionalized microcantilever was inserted inside the chamber and the samples to be measured (here live E. coli, dead E. coli and Human IgG) were injected separately into the chamber. An automated pump was programmed to pick up the sample solution at a constant speed of 4 μl/s. DesfB functionalized microcantilevers interact with the sample solution and the resultant resonance frequency shift ( f) of the microcantilever recorded infers additional mass loaded on microcantilever due to DesfB and E. coli interaction. For the given dimension of microcantilever, the 19 th mode was selected as the experimental resonant mode for its high signal to noise ratio. Upon target binding, cantilever will deflect if operating in static mode or its resonant frequency will shift if operating in dynamic mode. In this present study continuous tracking of the resonant mode was performed with Phase Locked Loop (PLL) due to its higher sensitivity upon mass loading. Integral and Proportional gains of the PLL were adjusted to achieve real time accurate tracking of the resonant mode with minimum noise in signal.

ATR-FTIR spectra of attached DesfB on functionalized
surface.-FTIR analysis of 11-MUA SAM and surface functionalization with NHS/EDC chemistry in this study correlates well with the earlier published work. 18 In the context of surface functionalization, 11-MUA molecules bind to the Au surface through the S-Au bond at one end while the carboxyl group is free at the other end to react with the NHS/EDC. The intermediate NHS/EDC ester binds the iron chelated DesfB with the self assembled monolayer of 11-MUA covalently. The DesfB molecule consists of three bidentatehydroxamic groups to form a hexadentate ligand. The residual chain includes two secondary amide groups and an aliphatic chain. Upon iron chelation, the oxygen atoms of the hydroxamate groups bind to the iron molecule with deprotonation of three hydroxyls to form a three identical asymmetrical chelation rings, Ferrioxamine B molecule. 19 The open chain, ending with a saturated amine group on one edge of the linear molecule gives the molecule a positive charge. A comparison of the FTIR spectra of DesfB and its iron chelated complex Ferrioxamine B is presented in Fig. 1 and their consolidate wave numbers are tabulated in Table I. After SAM formation and prior to DesfB coating, the DesfB molecule was treated in a solution containing sodium carbonate and FeCl 3 (1 mM each) to form its iron complex, Ferrioxamine B. The broad band at 3317 cm −1 is assigned to the N-H of the secondary amide (Table I). Iron chelated DesfB has the secondary amide shifted from 3317 cm −1 to 3287 cm −1 corresponding to bond formation.
An intensive overlapping of vibrations was also exhibited due to the iron chelation seen in case of bonding with the -COOH group of 11-MUA. Furthermore, a single band at 3087 cm −1 assigned to C-N-H overtone appears in presence of two individual bands. During Ferrioxamine B formation, the band assigned to CH 3 at 2960 cm −1 shifts to lower wavenumber at 2950 cm −1 . The asymmetric CH 2 band is further shifted to the lower wavenumber at 2926 cm 1 due to the interaction of 11-MUA. It is clearly mentioned that pure DesfB exhibited very strong absorption of C=O at 1642 cm −1 which shifted upon iron chelation (treated in sodium carbonate and FeCl 3 mixed solution) to 1632 cm −1 in the Ferrioxamine B spectrum. The amide II band (C-NH), which appeared at 1563 cm −1 in Desf B was shifted to 1533 cm −1 upon iron chelation. This shifting of the FTIR bands to the lower wavenumber corresponds the bond formation. The band at 1167 cm −1 that was exhibited in the DesfB spectra, assigned to one type of CN vibration, disappeared in the spectra of Ferrioxamine B. Multiple small bands in the range of 859-968 cm −1 for DesfB were in Ferrioxamine B at 859-1068 cm −1 . Due to the transformation of the DesfB to its iron chelated part, there have the evidences of overlapping bands. 20,21 FTIR spectrum of DesfB bound to 11-MUA SAM is shown in Fig. 2. 11-MUA on Au has C=O band formation at 1650 cm −1 which further shifted to 1608 cm −1 and formed a strong band upon DesfB interaction. The appearance of a strong band at 1608 cm −1 instead of multiple bands demonstrates the bonding between NH 2 group of the ferric complex of DesfB and COOH group of the 11-MUA. The C-N stretch and N-H bend vibrationat 1276 cm −1 and C-H vibration  Fluorescence microscopy analysis.-The microbial cell is considered viable if the membrane is intact and it can be assumed to be dead if the membrane structure is ruptured. 22 In the experiment both the dye SYTO 9 and Propidium Iodide (PI) were equally mixed and treated with bacterial cells to check the membrane structural integrity. Both the dyes have their own staining properties to distinguish viable and non-viable bacterial cells. PI was added to the SYTO 9 mixture having different concentration of live/dead cells. Fig. 4a indicates the  viable cells showing green color due to the penetration of dye SYTO9. PI penetration induces red color emission of dead cells observed in Fig. 4b. Both green and red color emitting cells corresponds to mixture of live and dead bacteria as well (Fig. 4c).The plain DesfB coated (Fig. 4g) and Human IgG coated (Fig. 4h) glass slides were devoid of any emission, which indicates the sample does not have any background fluorescence from DesfB, Human IgG, and their interaction (as negative control). However, live bacteria are critically dependent upon iron sequestration. Iron is the key ingredient for survival and live bacteria synthesize and release siderophores in the culture media to bind iron with a remarkably high affinity. Iron-bound siderophores are recognized by specific cell surface receptors of bacteria. This is a highly conserved process that occurs only in intact bacteria.
The experiment was designed such a manner that DesfB coated glass slide is used to specifically capture viable bacteria from a mixture of viable and dead E. coli, as demonstrated by fluorescence microscopy. Fig. 4d depicts live E. coli bacterial cells decorated on DesfB coated glass slide. The image shows green cells uniformly distributed on the slide surface. In the case of dead E. coli cells coated on DesfB surface (Fig. 4e), the red coloration was missing indicating the absence of binding with DesfB. An interesting result was found in mixed equal population of live and dead E. coli cells treated on DesfB coated surface (Fig. 4f). From Fig. 4f, it shows a green emission zone which indicates that the DesfB coated surface can differentiate live cell from the dead one and further it concludes that DesfB coated surface can bind only to live cell.
Upon further time lapse, core shell like structure formed where the core and shell is denoted by green and red color respectively. Even in between this structure there is the existence of green emission zone which resembles the presence of a live bacterial cell. It indicates that live cells have the capability to attach with the DesfB coated surface. The red color emitting shell zone indicates the destruction of the outer membrane whereas the green color core demonstrates the intact viable cells.
Microcantilever study.-Frequency tracking was performed for the 19 th resonant mode of the iron chelated DesfB immobilized microcantilevers (Fig. 5). DesfB coated microcantilever was taken as background i.e. a stable baseline was allowed to be achieved for each DesfB coated microcantilever before the sample was injected. Since the liquid environment is DI water, there was no shift in frequency observed when DI water was injected. However, when E. coli (10 −1 dilution) dispersed in DI water was injected shift from the baseline frequency ( f) is clearly visible due to the E. coli attaching to DesfB coated microcantilever surface. Live and dead E. coli of different  dilutions was used as samples. It can be inferred from Fig. 6 that live E. coli produces a greater f shift compared to the other samples.
Since the E. coli is active, it readily scavenges the iron chelated DesfB and anchors itself on microcantilever surface. Dead E. coli does not have this scavenging characteristic and thus does not produce a large f. Adsorption of live and dead cells is possible since the microcantilevers' back side is not protected by an anti-fouling coating. Hence, a minute f is seen even in the case of dead cells. This correlates well with the negative control, as it too produces a small but negligible f. Higher concentration/dilution (10 −1 dilution) of bacterial cell was chosen for plotting because of higher number of bacterial cells corresponds to higher mass produced which shows better visible resonant frequency shift (leading to accuracy in mass calculation).
The loaded mass ( m) corresponding to f of different samples injected is tabulated in Table II. Clearly, iron chelated DesfB immobilized microcantilevers are able to distinguish between live and dead E. coli cells. There is significant decrease in live cell attached for increasing dilution, whereas no such significance in dead cell adsorption is seen for increasing dilution. This indicates that the live E. coli are actively scavenging for DesfB and as bacterial dilution increase number of scavengers in the dilution decrease. On the other side, dead bacteria have no scavenging activity as validated with fluorescence microscopy. Furthermore, dead bacteria adsorption is due to its sticky proteinaceous surfacetemporarily tethering to the microcantilever and thus can be treated as error due to dead cell adsorption.  original stock), the live cell m is 0.27 ± 0.05 ng (386 ± 71 bacterial cells), which is lower than that of dead cell m of the previous two dilutions and also, on level with the negative control (0.18 ± 0.15 ng) (257 ± 214 bacterial cells). In earlier section it is already mentioned that in 10 −7 dilution (3.1 × 10 9 CFU/ml bacteria of original stock) there is only 310 bacterial cell present whereas from cantilever study it shows 386 ± 71 bacterial cells. Thus, this system of differentiation between live cell and dead cell is valid upto 10 −4 dilution. In this current study we are less concerned about the number of bacterial cells as the objective of the work is to demonstrate the binding capability followed by sensing activity of Siderophore (DesfB) toward live E. coli cell compare to dead one. The authors did not investigate long-term stability of the platform. However, the platform appears to be stable for at least 3 months (during the experimental testing period) with reliable detection of live bacteria.

Conclusions
In summary, the recent work demonstrated immobilization of Siderophores (DesfB) on solid surfaces such as a cantilever surface to distinguish between live and dead E. coli. The use of self-assembled monolayer based gold thiol chemistry for surface functionalization helped attach DesfB covalently on top of the gold coated surface. FTIR study revealed the interaction between live bacterial cells to the DesfB coated surface. Simultaneously this attachment was further observed by Fluorescence microscopy image analysis. The live cell attachment to the DesfB coated surface has been reported and validated by bacterial cell mass differential. It is seen that live cells attach to the functionalized cantilever whereas dead cells do not have the tendency to do so.
This study can serve as a preliminary demonstration toward experimental data in constructing a biosensing platform to differentiate bacterial versus viral illness in real time for point-of-care diagnostics aiding Internet of Things (IoT) application.