3.1. Morphological Characterization
Cellulose acetate (CA) was electrospun into fibrous, nonwoven membranes then deacetylated to regenerated cellulose (RC), grafted with alkyne moiety (alkyne-cellulose), and finally clicked with azide-biotin conjugate (biotin-cellulose). SEM images of the nanofiber membranes at each reaction step are compared in
Figure 1. The rough surface of the cylindrical as-spun fibers (
Figure 1a) became smooth and round after deacetylation (
Figure 1b). Swelling of the cellulose fibers during the alkyne substitution and click reaction steps caused the irregular appearance observed in
Figure 1c,d, respectively.
Figure 1c depicts the two-step alkyne substitution sample but is representative of both the one- and two-step alkyne substitution processes; neither process negatively impacted the fiber morphology at the respective optimal reaction conditions.
Figure 1d illustrates a successful click reaction of the 10:1 AA:Cu ratio for 48 h and is representative for the click samples listed in
Table 1.
Temporary needle blockage caused side jets to form that reduced the diameter of the fibers and resulted in a skewed fiber diameter distribution (
Figure S1a) with a mean of 741 nm and median of 680 nm. The diameter of CA fibers from as-spun to deacetylated (
Figure S1b) was not significantly reduced, but the alkyne-cellulose fiber diameters (
Figure S1c) were significantly reduced relative to both the as-spun and deacetylated CA fibers. The change in hydrogen bonding, removal of the acetyl groups, and addition of alkyne groups contributed to the changes in fiber diameter. No trend was found between successful click reactions and significant change in fiber diameter (
Figure S2). Although no definite size was provided by the supplier, the azide-biotin conjugate had a molecular weight of 444.55 g/mol and was distinctly smaller than the diameter of the nanofibers. Therefore, the biotin addition was not the only factor in the change of fiber diameter.
3.2. Alkyne Substitution
CA nanofibers were modified to obtain alkyne-cellulose nanofibers, either in a one- or two-step process. In the one-step process, CA nanofibers were deacetylated and alkyne substituted in one pot. In the two-step process, CA nanofibers were first deacetylated to regenerated cellulose (RC) nanofibers, and the alkyne groups were attached to the RC nanofibers in a separate step. The nucleophilic substitution of the alkyne group onto the nanofiber surface was tested at three levels of temperature and six time intervals for both CA (one-step) and RC (two-step).
ATR-FTIR was used to monitor the changes in sample surface chemistry. Representative spectra for each step are shown in
Figure 2a. The characteristic absorption bands for CA appeared at 1740, 1226, and 1368 cm
−1, attributed to the C=0 and C-O stretching and C-H bending vibration of CH
3 of the acetyl group, respectively. The broad bands at 2850–2950 and 3400–3500 cm
−1 were attributed to the C-H stretching of CH
2 or CH
3 and -OH stretching of unacetylated cellulose, respectively. The RC spectrum showed changes in these characteristic peaks to resemble native cellulose. Specifically, the disappearance of the 1740, 1226, and 1368 cm
−1 peaks and increase in the 3100–3500 cm
−1 band confirmed the removal of the acetyl groups and exposure of the hydroxyl groups.
The alkyne peak at 2116 cm
−1 was small for FTIR, even at high concentrations of the alkyne group (see inset of
Figure 2a). So, for low concentrations, it was difficult to differentiate the alkyne substitution from the background noise. However, Raman spectroscopy showed an intense, sharp peak for the alkyne group at 2100 cm
−1 (
Figure S3). So, coupled with the low noise-to-signal ratio, Raman is a more reliable way to show that a sample contains an alkyne group. Raman is difficult to perform on nanofibers due to their curvature and, therefore, both FTIR and Raman spectroscopy were employed for the alkyne substitution. Care was taken to obtain reproducible spectra. Changes in the sharpness of the broad band between 3100 and 3600 cm
−1 confirmed the change in the bonding of the hydroxyl groups and disruption of the hydrogen bonding. The ratio of the peak height of the alkyne group at 2115 cm
−1 (FTIR) and 2100 cm
−1 (Raman) to the reference peak of the ether bridge at ~1155 cm
−1 was calculated to correlate the degree of alkyne substitution to reaction condition (
Figure 3).
For the one-step trials, the carbonyl peak of the acetyl group disappeared in the FTIR spectra within the first 3 h. The calculated peak height ratios were relative but showed trends in the alkyne substitution. The FTIR and Raman data followed the same trends for all trials except for the RC at 60°C. This trial exhibited a high peak height ratio and high standard deviation for the 3-h Raman, while all other trials started with a low substitution at 3 h and increased over time before reaching a maximum and subsequently decreasing the peak height ratio. The alkyne substitution reached a maximum peak height ratio for each temperature due to the competition between the nucleophilic substitution of the alkyne group, side reactions, and degradation of the nanofiber mat. The Raman showed that as the reaction progressed (3 to 18 h), the standard deviation decreased, and the fibers became more uniformly substituted. After 18 h, hydrolysis of the glucose units resulted in loss of the alkyne substitution and a decrease in the alkyne peak ratio.
FTIR and XRD were used to monitor the change in crystallinity after the deacetylation and alkyne substitution reactions. Amorphous as-spun CA fibers became more ordered upon removal of the acetyl groups and rearrangement of the inter- and intra-molecular hydrogen bonds to form the stable RC nanofibers. Increasing the deacetylation time led to increase in chain orientation. For the alkyne substitution, any order obtained during deacetylation was lost with increase in reaction time until an equilibrium was reached for the two-step RC (
Figure S4a). The amorphous as-spun CA in the one-step process transitioned into regenerated cellulose as the reaction proceeded and also reached an equilibrium (
Figure S4b). Therefore, the alkyne substitution reaction only took place at the surface, and the bulk of the fiber was unaffected.
RC membranes maintained sufficient strength and dimensional stability to process further in the click reaction step. The CA membranes, however, were weak and did not retain sufficient strength for further handling. Regenerating cellulose nanofibers and separately substituting the alkyne group was a milder process that reduced the overall reaction time from 48 h to 8 h. Optimal conditions chosen for alkyne substitution were RC in 20/80 (v/v) IPA/water at 50 °C for 6 h. Samples prepared under these conditions were used for the click reaction investigation of biotin immobilization.
3.3. Biotin Immobilization
Alkyne-substituted RC mats and azide-biotin conjugate participated in the copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC) click reaction. Copper (II) sulfate was used as the source of copper, and ascorbic acid acted as a reducing agent to reduce copper (II) to copper (I). The copper (I) subsequently catalyzed the click coupling of the alkyne attached to the nanofiber mat and the azide attached to the biotin conjugate to form the triazole ring that linked the biotin to the cellulose nanofibers. Then, 2, 5, and 10 equiv of ascorbic acid to copper sulfate were tested alongside six controls in the following permutations, listed in
Table 1, to elucidate the interactions of each component with the RC nanofiber membranes.
FTIR was used as first indication of successful biotin immobilization (
Figure 2b). The characteristic absorption bands for the biotin conjugate appeared around 1700, 1650, and 1550 cm
−1, due to amide I and II of the carbamide of biotin and amide of the peptide linkage of biotin to PEG section of the conjugate. Amide I band was due to C=O stretching vibrations of the peptide bonds whereas the amide II band was due to C-N stretching vibrations in combination with N-H bending. The azide group in the biotin conjugate at 2100 cm
−1 disappeared after the click reaction, but the 1700, 1650, and 1550 cm
−1 peaks remained in the successfully clicked samples. All samples with full reaction components and Control 5 (
Table 1) gained biotin peaks, whereas the remaining control samples did not show evidence of biotin (
Figure S5).
Elemental mapping was used to analyze the distribution of biotin (
Figure S6). Nitrogen, only present in the biotin conjugate, was seen along the fibers and not in the interstitial spaces. This confirmed that the biotin conjugate was not simply just entangled in the pores of the membrane but immobilized on the surface of the nanofibers. The nitrogen along the fibers was consistently seen in all the same trials as the ones with biotin FTIR peaks. No nitrogen was detected for the remaining samples.
For the biotin-cellulose nanofiber membranes to be used in a diagnostic device, the membranes must be able to selectively and rapidly bind streptavidin. Fluorescently tagged streptavidin (streptavidin-FITC) was used as the model binding molecule. The binding schematic is shown in
Figure 4.
Confocal microscopy was used to image the fluorescent emission of the bound streptavidin-FITC (
Table 1). Z stack images were taken to ensure the fluorescence or lack thereof was throughout the membrane and not solely on the surface of the membranes. The lack of fluorescence of Control 3 confirmed that no nonspecific binding of the streptavidin-FITC occurred when biotin was absent. Control 1 and 2 showed that the alkyne moiety was required for the biotin conjugate to be immobilized and bind with the streptavidin-FITC; the biotin conjugate did not physically absorb onto the fibers. Control 5 contained copper sulfate but not ascorbic acid to reduce the copper (II) to its catalyst copper (I) oxidation state, yet successfully bound some biotin conjugate. Control 6 contained no catalyst and could only react at uncatalyzed amounts. Therefore, both Control 5 and 6 had less intense fluorescence than the full reaction sample due to smaller initial yield.
The click reaction yield was quantified using HABA colorimetric assay and XPS.
Figure 5 describes the HABA assay process for the biotin-cellulose nanofiber membrane samples. Each sample was given sufficient time to allow for the streptavidin-HABA complex to diffuse through the porous membrane and reach all available biotin molecules bound to the surface of the nanofibers.
Surface-available biotin and degree of substitution were calculated using Equations (1) and (2), respectively. The HABA colorimetric assay was performed on the biotin-cellulose samples at 1.5, 3.5, and 5 months of aging in a desiccator, shown in
Figure 6. The 1.5-month samples were measured on four separate pieces of the original mat. The 3.5- and 5-month samples were only measured on one piece of original mat each. Controls 1–4 had no detected biotin and were omitted from
Figure 6. Control 6 had no detected biotin for the 3.5- and 5-month samples but is included in
Figure 6 to show the significance of the reaction solution’s pH.
The HABA assays were first conducted 1.5 months after the reactions were completed and showed that, within the 1.5-month sample set, the biotin yield increased with both increase in reaction time and ascorbic acid-to-copper sulfate ratio. The excess ascorbic acid ensured that enough Cu(II) was reduced to Cu(I) so as to not be the limiting step in the reaction. The ascorbic acid also lowered the pH of the solution, below the isoelectric point (pI) of the biotin (5.1), and can also denature proteins. At low pH, the protonated biotin had reduced its reactivity. Control 5 reaction did not contain ascorbic acid and ran at a pH above biotin’s pI. The measured biotin was higher than any of the trials with both components of the catalyst. The copper (I) was previously proven to be generated via alcohol oxidation and terminal alkyne homocoupling [
28]. Removing the need for a reducing agent will be important for bioconjugation applications, such as direct click of proteins or antibodies on the nanofibers, since ascorbic acid reduces the pH into the range that can disrupt the structure of biomolecules. The reduced form of ascorbic acid, dehydroascorbate, can react with certain amino acids that change the protein functionality and/or cause protein aggregation [
28]. Control 6 had no ascorbic acid or copper sulfate, so only a negligible amount of uncatalyzed click reaction was occurring. Control 6 further supported the assertion that Control 5 was being catalyzed via an alternative copper-reducing route.
All click reaction samples were stored in a desiccator for 5 months and were retested at the 3.5- and 5-month marks. Overall, surface-available biotin increased exponentially with storage time for all full-reaction samples. Control 5 maintained a more stable amount of surface-available biotin. Both alkyne and biotin have lower surface energy than cellulose, so they should stay on the surface and not migrate to the fiber interior. Since the click reaction was carried out in aqueous solution, the water molecules disrupted the intra- and inter-molecular hydrogen bonding of the cellulose chains, causing the cellulose fibers to swell in the aqueous solution. Once the cellulose membrane was removed from the aqueous solution, the hydroxyl groups remained in their conformation, which inhibited biotin availability to bind with the streptavidin in the HABA colorimetric assay. Over time in air, the cellulose chains can rotate the lower energy biotin to the surface to minimize the free surface energy. The reverse phenomenon was observed when samples after 5 months of storage were re-submerged in water for 24 h and 7 days. Both submersion times resulted in complete disappearance of the surface-available biotin in the HABA assays. Normalized FTIR spectra of the samples before and after water submersion showed that the characteristic biotin peaks had no change in absorbance (
Figure S7). This supported the hypothesis that the biotin did not leach away from the fibers when submerged in water, but instead rotated below the surface of the fibers due to the water molecules favorably hydrogen bonding with the cellulose hydroxyl groups. For the diagnostic device application, short contact times (up to one hour) did not affect the surface composition of the cellulose nanofibers.
The degree of substitution of the biotin after 5 months of storage was also calculated from the XPS results using Equation (3). The ratio of sulfur to carbon was calculated from the XPS survey scans.
Figure 7 shows the comparison of the degree of substitution of biotin calculated from HABA and XPS.
HABA quantified only the amount of biotin available to bind with the streptavidin in solution. The HABA assay did not account for any diffusion limitations, any biotin that was chemically modified, or any biotin that resided in a conformation that made it unavailable to bind with streptavidin. XPS detected all carbon and sulfur present on the sample near the surface. Therefore, it can detect total surface biotin regardless of configuration or chemical modification. Two main disadvantages to XPS for cellulose nanofibers are carbon contamination that can accumulate on the surface of the samples as well as the shadowing effect of the round nanofibers. Both sources of error can reduce the sulfur-to-carbon ratio and resulted in lower degree of substitution (DS) results as seen in the 2 ascorbic acid to copper sulfate (AA:Cu) ratio samples. Overall, the XPS and HABA DS calculations were in agreement.
The degree of substitution of biotin was calculated to be between 0.01 and 0.10 from both the HABA colorimetric assay and XPS. This magnitude of DS translates to one biotin molecule per every 10 to 100 AGU. The repeat unit of cellulose, cellobiose, contained two AGU and had a unit length of approximately 1.03 nm [
29]. Therefore, the biotin was, on average, spaced 5 to 50 nm apart. This biotin spacing is useful for filtration and diagnostic device applications as it allows for target and amplification molecules’ immobilization onto the pendant biotin. For example, streptavidin is approximately 5 nm and fits well within the biotin spacing [
30]. Amplification molecules rely on proper spacing of the first layer (i.e., biotin) so that subsequent molecules can bind without steric hindrance in sandwich assays.