Fast, Multiple-Use Optical Biosensor for Point-of-Care Glucose Detection with Mobile Devices Based on Bienzyme Cascade Supported on Polyamide 6 Microparticles
Round 1
Reviewer 1 Report
1. This type of biosensor show different results under different urine pH (maximum activity is almost 3 times higher than the minimum, figure 7), so it means that in biosensor construction we need provided the pH evaluation for each sample, otherwise, when analyzing real samples, significant errors are possible.
2. Can you explain in more detail how affects of GOx and HRP ratio on the kinetics and stability of the bienzyme cascade reaction was optimized in your study?
3. The article lacks the sensitivity of a biosensor with a confidence interval. The tables lack information on the number of experiments and the confidence level.
Minor editing of English language required
Author Response
Q1. This type of biosensor shows different results under different urine pH (maximum activity is almost 3 times higher than the minimum, figure 7), so it means that in biosensor construction we need provided the pH evaluation for each sample, otherwise, when analyzing real samples, significant errors are possible.
A1. Yes, Figure 7 (Fig. 6 in the revised manuscript) displays different activities of the GOx/HRP/TMB system at different pH, which is normal for enzymes. This, however, does not mean different read-outs of the sensor. Different activity means different time to reach the stable color of the sensor. According to the literature, the optimum pH for HRP is 6.0-6.5, while GOx shows maximum activity in the pH range of 6-8. Under our conditions (protein content, buffer strength, temperature) the GOx/HRP/TMB system showed maximum activity at pH 5-7. In general, the pH of urine of a healthy person varies exactly within these limits. All of the urine samples in the present study had a pH around 5.5 and the addition of glucose in the 0.01-3.0 M range (i.e., the working interval of the sensor) did not change that value at all. Therefore, we do not expect errors in the glucose concentration related to the pH value of the sample.
Moreover, we are convinced that the sensor can be used at pH values outside the above optimal limits (e.g., with more acidic urine samples of unhealthy people) without any negative effect on the glucose concentration read-out. In the present work (Figure 4), the detailed activity study of GH@PA-R and GH@PA-C in order to establish the time required to reach a stable color of the chromogen was intentionally done at pH = 4 and it was around 20s. In the more basic urine samples, the color stabilization is even faster.
In the caption of the revised Figure 4 we have indicated that the activity is made at pH 4.
Q2. Can you explain in more detail how affects of GOx and HRP ratio on the kinetics and stability of the bienzyme cascade reaction was optimized in your study?
A2. The enzyme ratio was explored in our previous article on the synthesis of polymer assisted organic/inorganic nanoflowers containing the same GOx/HRP pair of enzymes (please see https://doi.org/10.3390/molecules28020839). It was found in preliminary studies that the enzyme ratio for random co-immobilization should be 2:1 in favor of the GOx, so as to achieve maximum activity. In the present study, we adopted the same enzyme relation for immobilization. A line explaining this was introduced into the Experimental part, section 2.4, Immobilization of Gox and HRP on PA6 MP (lines 196-197 of the revised manuscript).
Q3. The article lacks the sensitivity of a biosensor with a confidence interval. The tables lack information on the number of experiments and the confidence level.
A3. We thank for drawing our attention to this important omission. We think that the sensors´ action in urine is the most important so we provided a new Table S1 in the Supplementary materials with all numerical and statistics data related to the standard calibration plot in Figure 7b. Moreover, to the existing Table 3 of the revised manuscript a new column we added with information about the 95% confidence interval of the sensor.
Reviewer 2 Report
This manuscript has provided lots of data for the development of mobile device based glucose biosensor, but the interpretations of the data are not clear, raised many questions. The authors should re-organize the writing and figures, carefully check the reaction mechanisms. It can be reconsidered after major revision.
1. The introduction section is too long, more than two pages. It should be shortened to around 1 page.
2. Page 4, line 192, Figure 1S should be Figure S1.
3. Page 5, line 235, “the specific activity is calculated normalizing by the total amount of protein used in the assay”, what does this mean? It may have grammar errors.
4. Page 5, line 245, Figure 2S should be Figure S2.
5. Page 7, figure 1, I couldn’t see difference among 1d, 1e, and 1f, how do the authors know the Gox and HRP were really immobilized by compartmentalization?
6. A figure of experimental scheme should be added in section 2, to show the entire process of PA6 synthesis, immobilization of GOx and HRP, biosensor construction, and glucose detection. Figure 3 can go to supporting materials.
7. In Figure 3, the HRP reaction is not shown correctly, the authors need to modify this part.
8. Figure 4 needs caption for 4a and 4b.
9. There are format problems in Table 1, “.” Should be “x”, line number 466 and 467 overlap with the table.
10. Figure 5 should be used as Figure 3. But the numbering method is not consistent here, “a”, “b”, “c”, and “d” should be used.
11. Section 3.4 Biosensor Preparation and Application should be moved to section 2 Materials and Methods.
11. Why did the curve in Figure 6b show linear fitting plot? What enzyme reaction mechanism did the authors use to fit a linear relationship here?
12. Figure 8 caption, “Relationship between color change and glucose concentration in human urine tested by GH@PA-C sensor by means of: (a) and (b) − processing of smartphone digital photographs”, here I don’t know what the authors want to say.
13. Page 15, line 650 to 651, “The fact that the color of the sensor disappears means that that the charge-transfer complex [TMB…DTMB] (see Figure 3) is destroyed producing free TMB and DTMB”, is there a grammar error here?
Several places have grammar errors or typos.
Author Response
General statement: This manuscript has provided lots of data for the development of mobile device based glucose biosensor, but the interpretations of the data are not clear, raised many questions. The authors should re-organize the writing and figures, carefully check the reaction mechanisms. It can be reconsidered after major revision.
General statement by the authors: We thank to the Reviewer 2 for the careful analysis of our manuscript and for the critical comments, notes and suggestions that helped improve our submission. We did most of the reorganizations suggested and provided comprehensive explanations for all queries. All the resulting corrections or additions are highlighted in red in the revised manuscript.
Q1. The introduction section is too long, more than two pages. It should be shortened to around 1 page.
A1. We respectfully disagree with this suggestion. If we shorten the introduction to around 1 page, it will be in detriment of the internal logic and coherence of the state-of-the-art description. As a result, the reader will lose the connection between what has been done so far and the contribution and novelty of the present manuscript to the area of optical glucose biosensors.
Q2. Page 4, line 192, Figure 1S should be Figure S1.
A2. Corrected as required.
Q3. Page 5, line 235, “the specific activity is calculated normalizing by the total amount of protein used in the assay”, what does this mean? It may have grammar errors.
A3. The sentence in question is: “To enable comparison, the specific activity is calculated normalizing by the total amount of protein used in the assay.” “Normalizing” here means “dividing”. Along with “Activity”, we have calculated also “Specific Activity” and “Relative activity”. These are different methods to express activity comparatively. Thus, “specific activity” is activity per 1 mg of protein, while “relative activity” is activity of the immobilized systems related to that of the free enzyme dyad, the latter being 100%.
These terms are well-known and accepted but to avoid any doubt we have included a line in the Experimental part with the above explanation (lines 236-239 of the revised manuscript).
Q4. Page 5, line 245, Figure 2S should be Figure S2.
A4. Corrected as required.
Q5. Page 7, figure 1, I couldn’t see difference among 1d, 1e, and 1f, how do the authors know the Gox and HRP were really immobilized by compartmentalization?
A5. This question has two aspects.
First, it is true that there is no difference between the 1d-1f micrographs (now 2d-2f of the revised manuscript), but we never said the opposite. We kept the original text to this figure saying: “The adsorption immobilization of GOx and HRP (simultaneously in GH@PA-R or first HRP then GOx in GH@PA-C complexes) did not produce any difference in the sample morphology that can be directly observed by SEM”. The additional synchrotron WAXS and SAXS measurements could not distinguish between the two complexes either, but clearly showed that the immobilized enzymes have much denser packing than the free ones and both are definitely within the pores of the support. It is the activity study in Figure 4ab that demonstrates that the GH@PA-C sample (a result of consecutive HRP and GOx immobilization) is twice as active as the GH@PA-R sample (random immobilization). This means that the two different immobilization strategies led to different organizations of the enzymes in the pores of the support. In the case of GH@PA-C this organization was much more efficient.
This brings us to the second aspect of the question: can we call the GH@PA-C sample compartmentalized? In the classical sense of this term, probably not, neither can we term it with “positional co-immobilization”. This is a particular case of co-immobilization with specific co-localization of the two enzymes. The term “compartmentalized sample” in this work is used conditionally as the closest accepted term reflecting relatively well the enzyme organization in sample GH@PA-C.
Q6. A figure of experimental scheme should be added in section 2, to show the entire process of PA6 synthesis, immobilization of GOx and HRP, biosensor construction, and glucose detection. Figure 3 can go to supporting materials.
A6. Corrected as required. The original Fig. 3 with the corrected schematic presentation of the GOx/HRP cascade is now Fig. S3 in the Supplementary data. The original Figure 5 is used now as Figure 3 (see also Q10).
Q7. In Figure 3, the HRP reaction is not shown correctly, the authors need to modify this part.
A7. The chemistry of the TMB transformations in the initial Figure 3 was modified and, as required, presented as Figure S3 in the Supplementary materials. The explanatory text to this scheme (lines 451-454 of the revised manuscript) were also changed accordingly. We thank to the Reviewer 2 for drawing our attention to the correction of this error.
Q8. Figure 4 needs caption for 4a and 4b.
A8. Corrected as required.
Q9. There are format problems in Table 1, “.” Should be “x”, line number 466 and 467 overlap with the table.
A9. The said line numbers are correctly formatted in the WORD version of the manuscript, this minor typo should have occurred during the conversion to pdf. We will see that Table 1 is correctly converted.
Q10. Figure 5 should be used as Figure 3. But the numbering method is not consistent here, “a”, “b”, “c”, and “d” should be used.
A10. Corrected as required.
Q11. Section 3.4 Biosensor Preparation and Application should be moved to section 2 Materials and Methods.
A11. Corrected as required.
Q12. Why did the curve in Figure 6b show linear fitting plot? What enzyme reaction mechanism did the authors use to fit a linear relationship here?
A12. The original Fig. 6ab (Fig. 5ab in the revised manuscript) show the relation between the color change ΔE (calculated according to Eq. 2 that was taken from Ref. 19) and the glucose concentration in water. The procedures described in section 2.8 are the following: 1. A digital photograph of the blank (sensor without glucose) is taken under controlled conditions; 2. Glucose solution is pipetted on the sensor´s interrogation area; 3. After 20 s the colored sensor is photographed again under the same conditions. 4. Then Adobe Photoshop is used to obtain the CIELab coordinates of the blank and of the colored sensor obtained exactly from the sane sensor area; 5. Then ΔE is calculated.
Fig. 5a presents the ΔE dependence in the whole 0.01-10 mM glucose interval. Fig. 5b displays only the interval of the sensor´s linear response. We can see here two intervals with different linear regressions. Fig. 5c presents the sensor´s color change as determined by UV/VIS that has a single liner interval within the whole concentration range studied. In both cases the enzyme reaction is the formation of the charge-transfer complex of TMB of the sensor as a function of glucose concentration. The new Fig. 7 (original Fig. 8) reflects the same relation in urine samples with different glucose concentration. In Figs. 7ab is constructed after color analysis in digital photographs, and Fig 7c – based on UV/VIS data. In fact, the latter spectral measurement was only made for control. In the real POCT equipment-less analysis with our sensor it is definitely not needed.
In this context, Figs 7 and 8 present the essence of this work.
Q13. Figure 8 caption, “Relationship between color change and glucose concentration in human urine tested by GH@PA-C sensor by means of: (a) and (b) − processing of smartphone digital photographs”, here I don’t know what the authors want to say.
A13. We slightly changed the caption of the new Fig. 7 to avoid any doubts. Now it says: “Relationship between the by GH@PA-C sensor´s color change and the glucose concentration in human urine determined by: (a) and (b) − processing of smartphone digital photographs; (c) − UV/VIS measurement. The colors of the sensor areas in Fig. 7b are after detection of the respective glucose concentrations”.
We hope that after the explanation to Q12 the issue of the present query is resolved.
Q14. Page 15, line 650 to 651, “The fact that the color of the sensor disappears means that that the charge-transfer complex [TMB…DTMB] (see Figure 3) is destroyed producing free TMB and DTMB”, is there a grammar error here?
A14. The second “that” is a typo and it was deleted.
Reviewer 3 Report
Review
In this work, the colorimetric biosensor detecting glucose in human urine was prepared. Highly porous polyamide microparticles were synthesized as support for the glucose oxidase (GOx) and horseradish peroxidase (HRP) dyad immobilized randomly or by compartmentalization. The results indicated that the new sensor had a large area of linearity between 0.01-3.0 mM of glucose (or 0.2-54.8 mg/dL) determined by CIELab treatment of digital photographs in human urine samples with a LOD value of 30.7 μM, which exhibited the excellent performance. This paper is thorough, well-structured and likely of interest to readers focused on non-invasive glucose determination. In my opinion, this article can be accepted after minor revision for publication in Polymers. Before publication, it is necessary to clarify following points.
Questions
1. Does it work for biosensor if the pH value ＜4.6 or pH ＞8?
2. The reaction principle should be further stated in ‘3.2. Description of the GOx-HRP Cascade Reaction’ part.
3. The reversibility capacity of samples were indicated in Table 3. The morphological and structural characterization after multiple use should be observed and shown in the manuscript.
4. There are ten paragraphs in the ‘Introduction’ part. The author should revise and reduce these.
Comments for author File: 
Comments.pdf
Minor editing of English language required.
Author Response
Q1. Does it work for biosensor if the pH value ＜4.6 or pH ＞8?
A1. Our intension was to construct a sensor for people predisposed to or already having diabetes, i.e., with increased glucose in urine. Increasing only the glucose concentration does not change the pH value of urine, which is normally in the 5-7 pH interval. That is why we studied a little broader interval 4-8.
From our previous studies https://doi.org/10.3390/molecules28020839 we know that the immobilized GOx/HRP/TMB system could work at pH = 3 at a significantly lower activity. As regards pH > 8, we do not have such studies. It can be hypothesized that the sensor will work but at lower activity, i.e., more time will be necessary to reach the final color of the sensor.
Q2. The reaction principle should be further stated in ‘3.2. Description of the GOx-HRP Cascade Reaction’ part.
A2. One of the other reviewers suggested to move the reaction scheme of the GOx-HRP Cascade Reaction to the Supplementary materials and we did so. In the revised manuscript, all comments related to this mechanism were kept at the end of Section 3.2., lines 448-458. The most important step there is the formation of the charge-transfer complex. For this we adopted the already published reaction scheme discussed in detail in previous spectral studies – refs. 28 and 29 of the manuscript.
Q3. The reversibility capacity of samples were indicated in Table 3. The morphological and structural characterization after multiple use should be observed and shown in the manuscript.
A3. In the present submission we only disclose the possibility for multiple use of one and the same sensor. Apparently, this finding is related to the reversibility of the [TMB...DTMB] charge-transfer complex formation, when in close contact with GOx/HRP. A lot of additional structural, morphological and spectral studies will be necessary, which completely justifies a separate paper. We hope the reviewer will accept these arguments.
Q4. There are ten paragraphs in the ‘Introduction’ part. The author should revise and reduce these.
A4. We agree that the Introduction part became long. However, there exist a number of different studies on biosensors based on the GOx/HRP/TMB system. Shortening the Introduction part will be in detriment of the internal logic and coherence of the state-of-the-art description. As a result, the reader will lose the connection between what has been done so far and the contribution and novelty of the present manuscript to the area of optical glucose biosensors. We hope the reviewer will accept these arguments.
Round 2
Reviewer 2 Report
The revision has made some progress, but still need additional work before it reach the level of publication. The authors should re-organize the introduction section, reduce any contents that are not directly related to this research work. Other issues are listed below.
1. Page 5, line 235, “the specific activity is calculated normalizing by the total amount of protein used in the assay”, The correct grammar should be “…is calculated by normalizing …”.
2. Page 7, figure 1, I couldn’t see difference among 1d, 1e, and 1f, how do the authors know the Gox and HRP were really immobilized by compartmentalization?
Please replace the words “compartmentalized” and “compartmentalization” with other words. They are too misleading.
3. Why did the curve in new Figure 5 and Figure 7 show linear fitting plot? What enzyme reaction mechanism did the authors use to fit a linear relationship here?
By asking this question, I mean what type of reaction is the enzymatic reaction belonging to? For example, is it a surface binding/unbinding reaction? Is it first-order or second-order reaction?
Author Response
General comment of Reviewer 2: The revision has made some progress, but still need additional work before it reach the level of publication. The authors should re-organize the introduction section, reduce any contents that are not directly related to this research work. Other issues are listed below.
General answer: Having in mind the suggestion of Reviewer 2, the Introduction section was shortened from 1392 words to 1035 words, i.e., with >25%. In the revised version the corrected paragraphs are highlighted in blue.
Q1. Page 5, line 235, “the specific activity is calculated normalizing by the total amount of protein used in the assay”, The correct grammar should be “…is calculated by normalizing …”.
A1. Line 235 of the original manuscript was eliminated at the previous stage of revision. Already in the first revised revision (lines 236-239) we had placed a new text, which does not contain the above phrase. This same text is in lines 208-211 of the second revised version submitted now.
Q2. Page 7, figure 1, I couldn’t see difference among 1d, 1e, and 1f, how do the authors know the Gox and HRP were really immobilized by compartmentalization? Please replace the words “compartmentalized” and “compartmentalization” with other words. They are too misleading.
A2. We accepted this suggestion. The terms “compartmentalized” and “compartmentalization” had 6 appearances in the original manuscript – 2 in the abstract and 4 in the main text. They were either substituted by “consecutive immobilization” or eliminated. The changes in the text are highlighted in blue. (See the following lines in the revised manuscript: 16, 171, 175, 344-345, and 444).
Q3. Why did the curve in new Figure 5 and Figure 7 show linear fitting plot? What enzyme reaction mechanism did the authors use to fit a linear relationship here? By asking this question, I mean what type of reaction is the enzymatic reaction belonging to? For example, is it a surface binding/unbinding reaction? Is it first-order or second-order reaction?
A3. At this point of time, it is difficult to give a concise scientific answer to the theoretical questions asked above. The reason is that the detection of glucose by the GOx/HRP enzymatic pair in the presence of TMB is not a simple enzymatic reaction. It represents a classic example of a sequential enzyme cascade. First, GOx as a bi-substrate enzyme, oxidizes glucose with oxygen and produces gluconolactone and H2O2. The latter, along with TMB, becomes one of the substrates in the next reaction step catalyzed by HRP. The overall kinetics of this cascade reaction in the presence of both free and immobilized enzymes has been well studied by other researchers, including assessment of the induction period, the influence of the distance between enzymes and the product´s diffusion, the presence/absence of substrate channeling etc. Examples:
https://pubs.acs.org/doi/10.1021/nn402823k and https://www.nature.com/articles/ncomms13982
In our own recent paper (https://doi.org/10.3390/molecules28020839) we performed kinetic studies on the GOx/HRP/TMB enzymatic system organized in the form of organic-inorganic nanoflowers. It was found that the cascade reaction follows the Michaelis-Menten kinetics.
In the present manuscript, kinetic studies were beyond the scope. The aim here was to present a practical application of the GOx/HRP enzyme cascade, developing a biosensor for glucose detection. It is of paramount importance for a biosensor that the transducer produces a reliable signal that is proportional to the concentration of the analyte. This is exactly the purpose of the studies presented in Fig. 5 and 7, namely: to establish the interval of concentrations, in which there is a linear dependence between the color change of the TMB complex ΔE (i.e., the transducer´s signal) and the concentration of the analyte (glucose). The presence of linearity in the ΔE = f [Glucose] curves) can be used to construct glucose biosensors. Many other authors have used such an approach – see for instance ref. 17,19,20 of the manuscript.
Round 3
Reviewer 2 Report
This version has improved a lot. It can be published after careful proofreading.
