Iteration 1

Design

The designing for plasmids started with in silico docking and modeling of human serum albumin (HSA) -nanobody complexes with the goal of identifying nanobodies with oxidation stable and unstable epitopes. Using in silico models, we aimed to find suitable nanobodypairs to use in our enzyme fragment complementation assay (EFCA).

Since most of our nanobodies did not have experimentally solved structures or exact binding sites, we used homology modeling (SWISS-MODEL) to predict their 3D structure based on their amino acid sequence by Shen et al. [1]. We then used crosslink mass spectrometry data of the nanobody-HSA complexes, also from Shen et al. [1] as constraints to run protein-protein docking (HADDOCK).

Next, we consulted Professor Olli Pentikäinen and Doctoral Researcher Paola Moyano-Gómez from the University of Turku Institute of Biomedicine, experts in molecular modeling and protein-protein interactions. Based on the modeling and discussions with experts, we identified nanobodies with oxidation prone epitopes and ones with epitopes less likely to undergo structural changes due to oxidation to form our EFCA pairs (Table 1). We decided to use periplasmic expression to promote proper folding of the nanobodies.

Table 1. Potential nanobody pairs for the EFCA. Blue means one of the fusion proteins in the pair binds to an unstable epitope while the other fusion protein binds to a stable epitope. These pairs can be used to measure the concentration of reduced HSA. Green means both of the fusion proteins in the pair bind to stable epitopes of HSA. These pairs can be used to measure the concentration of total HSA. Black means the pair is impossible because the fusion proteins compete for the same epitope.

Build

We designed 10 DNA constructs for expressing 10 novel recombinant fusion proteins. The proteins contain an anti-HSA-nanobody that is connected to a NanoLuc fragment with a flexible GGGGSx4 linker. The expression is directed to periplasm with the PelB signal peptide sequence. pET-IDT C His was chosen as a backbone. 2xHis were added to the insert so the proteins would in the end have 8xHis. A TEV site was added before the His-tag for easy His-tag removal in case it would have hindered NanoLuc reconstitution. Sequences for nanobodies, NanoLuc fragments, linker, and TEV site can be found on the Design page.

The plasmid designs were reviewed by Professor Urpo Lamminmäki and our PI Pauli Kallio.

Test

We placed an order for the constructs on IDT. 8 of the constructs failed the NGS (next-generation sequencing) test. We received the pET-IDT-C-His-NanoChuck and pET-IDT-C-His-Nb126-SmBit plasmids. The plasmid maps for these plasmids can be found on the Design page.

Learn

We realized that our constructs had multiple faults. The linker sequence is GC% rich, which is a potential reason for the synthetization failure. Some nanobody sequences also have a GC% rich area right before the linker sequence which further amplified the problem. The insert sequences also included areas where codons repeated ≥3 times, which increases genetic instability and the probability of the polymerase sliding off.

Iteration 2

Design

We seeked assistance from Tommi Riihinen, the ABOA 2024 co-leader. A new linker nucleotide sequence was designed manually. The revised linker sequence does not have codons repeating over 2 times.

Build

We decided to order the 8 plasmids that got previously cancelled on both, IDT and Twist Bioscience. pET-IDT C His was again chosen as a backbone plasmid for the constructs ordered from IDT. For constructs ordered from Twist Bioscience, we chose pET-21(+) as a backbone. We dropped the added 2xHis from the inserts for the pET-21(+) plasmids because there would have been other amino acids between the 2xHis in the insert and the 6xHis of the backbone because of the structure of the backbone. The constructs were all optimized with codon optimization tools provided by IDT, and the constructs ordered from Twist Bioscience were additionally codon optimized by the tools provided by Twist Bioscience.Therefore, the nucleotide sequences for the same fusion proteins differ between the IDT and Twist Bioscience constructs. The sequences and plasmid maps can be found on the Design page.

Test

From IDT, we received the pET-IDT-C-His-Nb29-NanoLuc plasmid while the rest did not pass the NGS test. From Twist Bioscience, we received all 8 plasmids we had ordered. We transformed the plasmids in E. coli BL21(DE3)pLysS (Figure 1). pET-IDT-C-His-Nb126-SmBit plasmid failed at the first experiment, but was successful in the second. pET-IDT-C-His-Nb29-NanoLuc we did not try to transform because we had already had a successful transformation with pET-21(+)-Nb29-NanoLuc.

Figure 1. Successful transformation plates for pET-IDT-C-His-NanoChuck (A), pET-21(+)-Nb118-SmBit (B), pET-21(+)-Nb80-LgBit (C), pET-21(+)-Nb29-SmBit (D), pET-21(+)-ALB8-LgBit (E), pET-21(+)-Nb13-SmBit (F), pET-IDT-C-His-Nb126-SmBit (G), pET-IDT-C-His-Nb29-NanoLuc (H), pET-21(+)-Nb80-NanoLuc (I), and pET-21(+)-Nb77-SmBit (J). The plasmids with a pET-IDT C His backbone are plated on LB-agar plate with 50 µg/mL kanamycin and the plasmids with pET-21(+) are plated on LB-agar plate with 100 µg/mL ampicillin.

Learn

The competent E. coli BL21(DE3)pLysS cells we used for transformation were made by ourselves. The protocols used for making the competent cells and doing transformations with them can be found here. We had done a test transformation prior to our actual transformations and learned that the transformation efficiency for our cells was about 0.7 transformants / ng of DNA. Thus, it is no wonder our first transformation experiment with pET-IDT-C-His-Nb126-SmBit failed the first experiment. Because we received less DNA from Twist Bioscience than IDT, we decided to do the transformation with more DNA to increase the probability of a successful transformation. The second experiment with pET-IDT-C-His-Nb126-SmBit transformation was also performed with double the amount of DNA to increase the odds.

Iteration 1

Design

For our first iteration at expression, extracting, and purifying our 10 proteins we decided to use a protocol we found on protocols.io (https://www.protocols.io/view/protein-expression-and-extraction-of-hard-to-produ-261geoxjwl47/v1) for expression and periplasmic extraction and a protocol provided by our PI Pauli Kallio for batch spin IMAC purification as starting points.

Build

We combined these two protocols to make the first version of the protocol containing the protocols for expression, periplasmic extraction, and IMAC purification.

Test

We decided to do Bradford assay on microtiter plates for periplasmic extracts, flow-throughs, and all 5 purification fractions for each protein for a quick check to see if the step was successful. Visual assessment of Bradford assay gave indication that only 1 one of all the purification fractions contained any protein, that being the first fraction of Nb77-SmBit (Figure 2). Second, third, fourth and fifth fractions for all proteins contained no protein. For flow-throughs, only Nb29-NanoLuc seemed not to contain any protein, with Nb80-LgBit and NanoChuck being unsure.

Figure 2. Bradford assay for first, second, and third fractions and flow-throughs performed on 11.7.25. A row contains the first fractions, B row the second fractions, C row the third fractions, and D row the flow-throughs. Column 1 contained Nb13-SmBit samples, column 2 and 12 MQ water, column 3 Nb80-LgBit, column 4 Nb77-SmBit, column 5 Nb80-NanoLuc, column 6 Nb29-NanoLuc, column 7 Nb29-SmBit, column 8 NanoChuck, column 9 Nb118-SmBit, column 10 ALB8-LgBit, and column 11 Nb126-SmBit. Nb118-SmBit flow-through was pipetted to E9 well instead of D9 well. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Periplasmic extracts contained protein for all proteins (Figure 3).

Figure 3. Bradford assay for periplasmic extractions performed on 10.7.25. A1: Nb13-SmBit, A2: MQ water, A3: Nb80-LgBit (pipetting mistake), A4: Nb77-SmBit, A5: Nb80-NanoLuc, A6: Nb29-NanoLuc, A7: Nb29-SmBit, A8: NanoChuck, A9: Nb118-SmBit, A10: ALB8-LgBit, A11: Nb126-SmBit, A12: MQ water, B1: Nb80-LgBit. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 125 µL.

We performed SDS-PAGE for some of the most promising samples based on Bradford assay. We also included a few samples that seemed to not contain any or only a little protein based on Bradford to be sure. We decided to load samples of Nb126-SmBit flow-through, NanoChuck flow-through, ALB8-LgBit flow-through, ALB8-LgBit first fraction, Nb77-SmBit flow-through and first fraction, Nb29-SmBit flow-through, and Nb118-SmBit flow-through (Figure 4).

Figure 4. SDS-PAGE performed on 11.7. for first fraction and flow-through samples. The gel used was Mini-Protean TGX and it was run at 200 V for about 1 h. The tape at the bottom of the gel was forgotten to be removed. Gel was stained with PageBlue Protein Staining Solution (Thermo Fisher) and imagined with Gel Doc EZ Imager (Bio-Rad). The ladder was Precision Plus Protein Dual Color (Bio-Rad). The colored boxes indicate the area where the band for the protein should be. The calculated protein size is in brackets after the sample name. FT: flow-through, F1: first fraction.

Unfortunately, we had forgotten to remove the tape from the gel before starting the run. This resulted in a hard-to-read gel. However, we were able to verify the purifying success of Nb77-SmBit. The faint, slightly larger band in the lane could be Nb77-SmBit with the signal peptide still intact. A quite large portion of Nb77-SmBit seemed also to have not bound to Co-NTA resin as indicated by the band in the Nb77-SmBit flow-through lane. NanoChuck flow-through also seemed to contain NanoChuck protein that had not bound to Co-NTA resin. A similar issue might have occurred with Nb126-SmBit, Nb29-SmBit, Nb118-SmBit and ALB8-LgBit as well though it can not be conclusively said based solely on the gel. The first fraction of ALB8-LgBit seems to not contain any protein.

Learn

Some of the flow-throughs loaded to the gel seemed to include our proteins, which is why we decided to recharge the Co-NTA resin prior to IMAC purification for the next iteration. As the gel is a bit wonky because we forgot to remove the tape, it is however hard to get any definitive proof whether the flow-throughs contain our proteins.

Based on the visual assessment of Bradford assay of the periplasmic extracts, 3 contained only a little protein. To our annoyment, we realized too late that we should have saved some of the periplasmic extracts to load to the gel. If we had done that, we could have verified whether the protein was simply lost during purification or if the extraction and/or expression were unsuccessful.

This iteration also left us wondering whether we should:

• modify the culture growth time and temperature, because the shakers could not achieve the low temperature of 18 °C proposed by the protocol
• use a different IPTG concentration, because the 1 mM IPTG concentration proposed by the protocol seemed a bit high compared to many other protocols for protein expression
• use another method for periplasmic extraction, as the TSE buffer used for the osmotic shock method contained 1 mM EDTA and 200 mM Tris-HCl, both which could affect IMAC purification results
• use Ni-NTA resin instead of Co-NTA resin to increase the yield though risk losing some purity
• do the periplasmic extraction and IMAC purification in a sterile manner to decrease the risk of losing our protein during storage
• change imidazole concentrations, because the 250 mM for elution buffer seemed a bit high compared to many other protocols for IMAC purification

With these ideas of possible modifications in mind, we decided to proceed to the next iteration to address the issues of this iteration.

Iteration 2

Design

We decided it would be best to consult experts regarding what parts of the protocol we should modify. We had conversations with doctoral researcher Sami Oksanen, PhD Eeva-Christine Brockmann, and docent Tuomas Huovinen about our previous iteration and the results it yielded. We decided to modify our previous protocol by doing lysis alongside periplasmic extraction, modifying the culture conditions of the expression cultures, modifying buffer pHs, and decreasing IPTG concentration.

We decided to work with Nb77-SmBit, Nb80-LgBit, Nb29-SmBit, Nb29-NanoLuc, and NanoChuck for this iteration. This decision was made based on the previous iteration’s success while keeping the uniqueness of the protein and thus its importance to our project in mind. With these proteins, we have two proteins that contain the functional enzyme and 3 proteins that contain only a fragment of the enzyme. These 3 proteins amount to 2 different pairs that can be used for our test VeriFied.

Build

Based on our own conclusions and the conclusions made during our discussions with experts, we made a new revised protocol. This protocol included lysis, modified culture conditions, modified buffer pHs, and a lower IPTG concentration.

Test

Bradford assay was performed on microtiter plates for periplasmic extracts (Figure 5), lysates (Figure 6), flow-throughs (Figure 7), and all 3 fractions (Figure 8) for each protein sample to see if the step was successful.

Figure 5. Bradford assay for periplasmic extractions performed on 31.7. C1: Nb29-NanoLuc, C2: Nb77-SmBit, C3: Nb80-LgBit, C4: NanoChuck, C5: Nb29-SmBit. The A row contains a bovine serum albumin standard that was too concentrated. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Figure 6. Bradford assay for lysates performed on 31.7. A1: Nb29-SmBit, A2: Nb77-SmBit, A3: NanoChuck, A4: Nb29-NanoLuc, A5: Nb80-LgBit. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Figure 7. Bradford assay for flow-throughs performed on 31.7. A1: MQ water, A2: Nb29-SmBit, A3: Nb29-SmBit, A4: Nb77-SmBit, A5: Nb77-SmBit, A6: Nb80-LgBit, A7: Nb80-LgBit, A8: Nb29-NanoLuc, A9: Nb29-NanoLuc, A10: NanoChuck, A11: NanoChuck, A12: MQ water. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Figure 8. Bradford assay for flow-throughs and fractions performed on 31.7. Flow-throughs on A row are no longer readable. The readable version is depicted in Figure 7. B row contains the first fractions, C row the second fractions and D row the third fractions. All column 1 and 12 contain MQ water, column 2 and 3 contain Nb29-SmBit samples, columns 4 and 5 contain Nb77-SmBit samples, columns 6 and 7 contain Nb80-LgBit samples, columns 8 and 9 contain Nb29-Nanoluc samples and columns 10 and 11 contain NanoChuck. Protein Assay Dye Reagent Concentrate (Bio-Rad) was not pipetted to wells D1, D3, D5, D7, D9, D11, and D12. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Based on Bradford assay, all lysates and periplasmic extracts contained protein, lysates significantly more as is to be expected. All flow-throughs contained protein, with Nb80-LgBit and NanoChuck containing only a little. None of the fractions contained any protein.

We used SDS-PAGE to analyze the periplasmic extractions of every protein, flow-throughs of Nb77-SmBit and Nb80-LgBit, lysates of Nb77-SmBit and Nb80-LgBit, and first fractions of Nb80-LgBit, Nb29-NanoLuc and Nb77-SmBit from this batch and also from the previous batch for reference (Figure 9).

Figure 9. SDS-PAGE performed on 1.8. for first fraction, periplasmic extract, and flow-through samples. The gels used were Mini-Protean TGX and they were run at 200 V for about 20 minutes before the voltage was changed to 150 V for about 5 minutes. Gels were stained with PageBlueProtein Staining Solution (Thermo Fisher) and imagined with Gel Doc EZ Imager (Bio-Rad). The ladder was Precision Plus Protein Dual Color (Bio-Rad). The colored boxes indicate the area where the band for the protein should be. FT: flow-through, PE: periplasmic extract, F1: first fraction.

Despite Bradford assay implying no protein content for Nb77-SmBit first fraction, SDS-PAGE showed a clear band of the calculated size in the lane. The lane was more intense than the first fraction of Nb77-SmBit from 11.7., though it remains unsure if this is because of a more successful purification or because of the older fraction having degraded during storage. The band for Nb77-SmBit was also visible in the periplasmic extract. The flow-through for Nb77-SmBit did not seem to contain any protein of the correct size. These results seem to indicate that the resin recharging enhanced the purification results.

Periplasmic extracts for NanoChuck and Nb29-SmBit seemed to contain protein of the correct size. Nb29-NanoLuc and Nb80-LgBit extracts also seemed to contain protein of the correct size, but this seems to be of the same size of a native E. coli BL21(DE3)pLysS protein found in every lane. Thus, it is hard to say whether we managed to produce Nb29-NanoLuc or Nb80-LgBit. First fractions for Nb29-NanoLuc and Nb80-LgBit did not contain any protein. Lysate samples analyzed for Nb77-SmBit and Nb80-LgBit contain many proteins of various sizes, which makes it hard to conclusively say if they contain the target protein. Nb80-LgBit flow-through has a faint band of correct size, so some of it could have been lost during purification.

To further verify whether we had produced our proteins, and to see whether our proteins could produce a luminescence signal, we planned a quick experiment on 5.8. to measure the luminescence signal produced by our periplasmic extracts when Nano-Glo luciferase assay reagent is added. As all of our tested samples produced significant luminescence compared to noise, we designed a more thorough luminescence measurement on 8.8. to verify these promising results.

All the samples were added into a flat bottom white microtiter plate for luminescence measurements using three technical replicates per sample. Unfortunately, since the tests had to be done with unpurified periplasmic extracts, protein concentrations are not known, making comparisons and thorough data analysis difficult or impossible.

We wanted to measure the background signal of all the fusion proteins on their own in addition to some basic controls such as MQ water. To test the function of the EFCA, we tested both Nb80-LgBit+Nb29-SmBit and Nb80-LgBit+Nb77-SmBit with and without pure HSA and in various volumes. Additionally, we had a fusion protein consisting of the NanoLuc LgBit and SmBit fragments connected with a linker that we called NanoChuck. To measure the signal produced by the whole NanoLuc-enzyme, we used Nb29-NanoLuc which has the full NanoLuc enzyme connected to Nb29 with a GGGGSx4 linker.

The background signal produced by individual fusion proteins tested reaches a maximum of approximately 16 000 RLUs as seen with Nb29-SmBit (Figure 10). We also measured the background signal caused by TSE (200 mM Tris-HCl pH 7.5; 500 mM sucrose, 1 mM EDTA), the buffer used to extract the proteins from the periplasm, HSA and MQ water (Table 3). Due to uncertainty with protein concentrations and the amount of variables in all tests, we chose to not reduce any amount of background signal from measurements.

Figure 10. Luminescence signal produced by different volumes of periplasmic extractions of Nb80-LgBit (a), Nb29-SmBit (b), and Nb77-SmBit (c). Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained either 70, 50, 30, 20, 10 or 5 µL of periplasmic extract, 0.02 M phosphate buffer pH 8 and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 101 µL. The signal for t=0 min was measured when the substrate had not been added to the wells. The fusion proteins used were extracted on 31.7. Variance reported in ± SD of three technical replicates. Measurements were performed on 8.8. The Y axis has been scaled to highlight the difference between the luminescence signal produced by proteins with only a fragment of the enzyme and the pair that can form a functional enzyme.

Table 3. Background signal of MQ water, HSA and TSE (200 mM Tris-HCl pH 7.5; 500 mM sucrose, 1 mM EDTA) buffer controls. Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained either 40 µL of TSE, 1 µg human serum albumin (HSA), or 80 µL of MQ water, 0.02 M phosphate buffer pH 8 and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 101 (HSA) or 100.5 (TSE, MQ) µL. Variance reported in ± SD of three technical replicates. Measurements were performed on 8.8.

Sample	Highest luminescence reading recorded (RLU) ± SD
TSE	209 ± 29
MQ water	238 ± 68
1 µg HSA	986 ± 246

Based on literature we expected the LgBit NanoLuc fragment to have significantly higher activity when compared to SmBit but based on our measurements Nb29-SmBit produces the largest signal compared to Nb80-LgBit and Nb77-SmBit. Our hypothesis is that this signal is due to large differences in protein concentrations.

The peak signal for Nb80-LgBit and Nb29-SmBit pair without HSA was detected between 45 and 60 minutes after addition of the Nano-Glo luciferase assay reagent (Figure 11). This 45-60 time frame for maximum luminescence was later used in measurements where the reaction was not tracked as a function of time. When compared to the background signal of individual NanoLuc fragments with a maximum luminescence of 16 000 RLUs, the signal from both NanoLuc fragments together is orders of magnitude higher even without the presence of HSA. This serves as preliminary proof that the EFCA is functioning. The fact that the proteins produce significant luminescence even without the presence of HSA implies that the protein concentrations were high enough for the fragments to combine even despite their low affinity.

Figure 11. Luminescence signal produced by Nb80-LgBit and Nb29-SmBit pair (A) and Nb80-LgBit and Nb77-SmBit pair (B) without HSA. Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained 10 µL of periplasmic extracts for both proteins of the pair, 0.02 M phosphate buffer pH 8 and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 101 µl. The fusion proteins used were extracted on 31.7. Variance reported in ± SD of three technical replicates. Measurements were performed on 8.8. HSA: human serum albumin.

With HSA present, the signal is lower for both of the pairs (Figure 12). This could be because of HSA binding the complementary proteins too far from each other to form a functional enzyme thus decreasing the luminescence signal.

Figure 12. Luminescence signal produced by Nb80-LgBit and Nb29-SmBit pair (a) and Nb80-LgBit and Nb77-SmBit pair (b) with HSA. Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained 20, 10 or 5 µL of periplasmic extracts for both proteins of the pair, 0.02 M phosphate buffer pH 8, 1 µg HSA, and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 101 µl. For the Nb80-LgBit and Nb29-SmBit pair, the Nano-Glo luciferase assay reagent used contained 2 % of Nano-Glo® Luciferase Assay Substrate in the Nano-Glo® Luciferase Assay Buffer while the reagent used with Nb80-LgBit and Nb77-SmBit pair contained 1 % of substrate in the buffer. The signal for t=0 min was measured when the substrate had not been added to the wells. The fusion proteins used were extracted on 31.7. Variance reported in ± SD of three technical replicates. Measurements were performed on 8.8. HSA: human serum albumin.

NanoChuck has both of the NanoLuc enzyme fragments separated by a linker, whereas Nb29-NanoLuc has the functional NanoLuc enzyme connected to Nb29 with a linker. Thus, it was hypothesized that the non-split enzyme would produce more luminescence. While Nb29-NanoLuc does produce more luminescence than NanoChuck, the signal increases when volume decreases (Figure 13). It seems that perhaps the wells that should contain 5 µL of Nb29-NanoLuc contain 20 µL of Nb29-NanoLuc because of a pipetting mistake. This mistake went unnoticed while pipetting but it seems like a logical explanation.

Figure 13. Luminescence signal produced by Nb29-NanoLuc (a) and NanoChuck (b). Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained 20, 10 or 5 µL of periplasmic extracts, 0.02 M phosphate buffer pH 8 and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 100.5 µL. The Nano-Glo luciferase assay reagent used contained 1 % of Nano-Glo® Luciferase Assay Substrate in the Nano-Glo® Luciferase Assay Buffer. The signal for t=0 min was measured when the substrate had not been added to the wells. The fusion proteins used were extracted on 31.7. Variance reported in ± SD of three technical replicates. Measurements were performed on 8.8.

Learn

During this iteration, we managed to purify only Nb77-SmBit again. As we also loaded the previously purified Nb77-SmBit to the gel, we were able to compare the band intensities. The band for Nb77-SmBit purified on 1.8. was of higher intensity than the one purified on 11.7. Though it remains unclear if this is due to the resin recharging, Nb77-SmBit degrading in the storage, or something else. Other lanes for 1st fractions were fully empty of any kind of protein as was hypothesized based on the Bradford results.

The lanes for unpurified lysates were really full of all kinds of proteins so it was hard to make sense of the bands. As the periplasmic extracts were much purer compared to the lysates and we still had not optimized the purification by batch spin IMAC, we decided to drop the lysis for future iterations.

From SDS-PAGE, we were able to see bands in the periplasmic extracts for NanoChuck, Nb29-SmBit and Nb77-SmBit. Whether we had successfully expressed Nb29-NanoLuc and Nb80-LgBit remained unclear as their bands were on the same level with some native protein found in all periplasmic extracts.

With the luminescence measurements, we were able to confirm that there were luminescent proteins in the periplasmic extracts for Nb80-LgBit, Nb77-SmBit, Nb29-SmBit, NanoChuck, and Nb29-NanoLuc. With SDS-PAGE verifying the existence of Nb77-SmBit, Nb29-SmBit, and NanoChuck, the luminescence signal seemed to indicate strongly that the NanoLuc fragments had folded functionally.The luminescence signal also gave some indication of Nb29-NanoLuc and Nb80-LgBit having expressed.

The proteins containing only a fragment of NanoLuc produced only background signal when compared to the signal produced by the pairs. The luminescence signal produced by proteins containing the full NanoLuc enzyme produced more signal than the pairs as is to be expected. However, comparing the signals produced by the proteins can not conclusively be done as they were periplasmic extracts and thus the precise concentrations of our proteins remains unknown.

With this cycle, we got a clear reminder about the need of optimizing HSA to protein ratio. It seemed like we had used a HSA concentration that was actually hindering the functionality of the proteins.

Iteration 3

Design

With the previous iteration, we had gotten clear success with 3 of our proteins with a NanoLuc fragment. We decided that it would be best to focus on them as the results indicated that the expression and folding of the enzyme fragments was successful. Previously, we could have only tested 2 different pairs as we only had one protein with the LgBit. Thus, we decided to include ALB8-LgBit for this iteration. We decided to not express the proteins with full NanoLuc enzyme anymore as they are not used for the actual VeriFied test but rather as controls or supporting measurements.

Build

We designed the third and final version of the protocol for conducting the expression and purification on a larger scale. The larger scale could result in bigger yields and protein concentrations for IMAC purification which could in turn increase the purification success. This protocol also included a control strain BL21(DE3)pLysS which did not have any of our plasmids in it. Control strain was included so we could verify which bands seen in SDS-PAGE are from native proteins, and also to verify that the luminescence signal produced by the periplasmic extracts was because of our fusion proteins.

Test

We performed Bradford assay on microtiter plates for periplasmic extracts (Figure 14), flow-throughs (Figure 15), and fractions (Figure 16). All periplasmic extracts contained protein, though Bradford assay indicated that the protein content was lower for the control strain BL21(DE3)pLysS that did not have any of our plasmids transformed in it. All flow-throughs contained proteins as well. None of the fractions had protein in them.

Figure 14. Bradford assay for periplasmic extracts performed on 14.8. Column 1 contains MQ, column 2 ALB8-LgBit, column 3 Nb77-SmBit, column 4 Nb29-SmBit, column 5 Nb80-LgBit and column 6 control strain BL21(DE3)pLysS. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Figure 15. Bradford assay for flow-through performed on 14.8. Column 7 contains MQ, column 8 ALB8-LgBit, column 9 Nb77-SmBit, column 10 Nb29-SmBit, and column 11 Nb80-LgBit. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

Figure 16. Bradford assay for fractions performed on 14.8. A and B rows and C row wells from 1 to 3 contain a too concentrated BSA standard. MQ water is in wells C4-6. D row contains fractions for ALB8-LgBit, E row for Nb77-SmBit, F row for Nb29-SmBit, and G for Nb80-LgBit. Columns 1 to 3 contain the first fractions, columns 4 to 6 second fractions, and columns 7 to 0 third fractions. The wells contain 25 µL Protein Assay Dye Reagent Concentrate (Bio-Rad) and 10 µL protein sample in a final volume of 135 µL.

For every protein, we decided to analyze the periplasmic extract and the first fraction with SDS-PAGE (Figure 17). Additionally, we analyzed the flow-throughs, second or third fractions for some of the proteins so that all proteins had at least 3 samples loaded to the gel. Periplasmic extract of the control strain BL21(DE3)pLysS was also analyzed.

Figure 17. SDS-PAGE performed on 15.8. for samples of Nb29-SmBit, Nb77-SmBit, ALB8-LgBit, control strain BL21(DE3)pLysS, and Nb80-LgBit. The gels used were Mini-Protean TGX and they were run at 150 V for about 10 minutes before the voltage was changed to 125 V for about 50 minutes. Gels were stained with PageBlue Protein Staining Solution (Thermo Fisher) and imagined with Gel Doc EZ Imager (Bio-Rad). The ladder was Color Prestained Protein Standard (New England Biolabs). The colored boxes indicate the area where the band for the protein should be. The calculated protein size is in brackets after the sample name. PE: periplasmic extract, FT: flow-through, F1: first fraction, F2: second fraction, F3: third fraction.

The periplasmic extracts for Nb29-SmBit and Nb80-LgBit had bands of correct sizes, indicating a successful expression and extraction of these proteins. While the periplasmic extracts for Nb77-SmBit and ALB8-LgBit contain protein in the area where the proteins should be, there are no distinct bands indicating a presence of our fusion proteins. SDS-PAGE verified no protein in the fractions as had already been indicated by Bradford assay. The flow-through of Nb29-SmBit seems to contain Nb29-SmBit which means it did not bind to the Co-NTA resin. The control strain periplasmic extract has a band of about 40 kDa that also appears in all of the periplasmic extracts. Thus, it has now been verified as a native BL21(DE3)pLysS protein.

We again performed bioluminescence measurements. This time we also tested the background signal of the control strain. The highest reading was 513 RLU with a 172 standard deviation when measured with three technical replicates. The periplasmic extract of the control strain was measured with 1 µg HSA as well. The highest reading was 332 RLU with a 79 standard deviation when measured with three technical replicates. It is now safe to say that the previous luminescence signals we had measured with the periplasmic extracts were caused by our fusion proteins.

This time, we used our four different pairs with extracted bloodstains of different ages to see if the signal would decrease as the bloodstain got older as we had hypothesized when designing our test VeriFied (Figure 18). We used the periplasmic extracts extracted on 14.8. Based on SDS-PAGE, these periplasmic extracts might not contain Nb77-SmBit or ALB8-LgBit.

Figure 18. Bloodstain age measurements. Luminescence signal was measured 50-60 minutes after adding Nano-Glo luciferase assay reagent. Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained 10 µL of periplasmic extracts for both proteins, 10 µL of extracted bloodstain (ages 1, 2, 3, 4, 5, 8, 11, 15, 22 or 29 d), 0.02 M phosphate buffer pH 8 and 0.5 µl Nano-Glo luciferase assay reagent in a final volume of 100.5 µl. The fusion proteins used were extracted on 14.8. Variance reported in ± SD of three technical replicates except Nb77-SmBit and Nb80-LgBit pair had only 2 replicates for 1 d. Measurements were performed on 22.8.

The Nb77-SmBit and Nb80-LgBit produces a high luminescence signal indicating a successful enzyme fragment complementation. The luminescence signal seems to decrease from day 1 to 5 before plateauing. The signal begins to decrease after day 15 until day 22. After day 22, the signal seems to begin to increase, though it can not be said conclusively as there are no older bloodstains tested as of 29-days-old. There is a sudden drop in signal on day 3 in all three replicates. The cause of this drop would need to be verified in future studies.

The Nb29-SmBit and ALB8-LgBit pair only produced background signal, with the highest measured signal being 364 RLU with the standard deviation of 21. This signal is comparable to the signal produced by MQ water or the control strain for example, and it is much lower than the signals we got previously from single proteins with only an enzyme fragment. Thus, it seems likely that the periplasmic extracts for ALB8-LgBit and Nb29-SmBit do not contain any of our fusion proteins or only contain a little of them.

Nb29-SmBit and Nb80-LgBit and Nb77-SmBit and ALB8-LgBit pairs also produce low luminescence signal when compared to the Nb77-SmBit and Nb80-LgBit pair. As they still produce a little signal, the signal could be caused by a single fusion protein of either Nb77-SmBit or Nb80-LgBit. The concentrations could also just be too low. With these pairs, we see a similar trend between the luminescence signal and the bloodstain age as with the higher luminescence signal producing pair Nb77-SmBit and Nb80-LgBit. From day one to day five, the luminescence signal seems to steadily decrease before plateauing before day 15 when the signal will begin to drop again. After day 22, the luminescence signal seems to increase.

Besides testing if our fusion proteins could be used to measure bloodstain age, we also wanted to start optimizing the HSA concentrations which we had noticed previously affecting the luminescence signal. We measured all pairs with two different amounts of HSA (Figure 19).

Figure 19. Human serum albumin (HSA) concentration’s effect on the luminescence signal produced by four different pairs. Luminescence signal was measured 50-60 minutes after adding the Nano-Glo luciferase assay reagent. Measurements were performed with Hidex Sense Microplate reader on a white Nunc Maxisorp 96-well plate. The wells contained 10 µL of periplasmic extracts for both proteins, 0.25, 0.5, 1 or 2 µg HSA, 0.02 M phosphate buffer pH 8, and 1:200 Nano-Glo luciferase assay reagent in a final volume of 100.5 µL. The final volume was 103.5 µL for Nb80-LgBit and Nb29-SmBit with 2 µg HSA. Periplasmic extracts were extracted on 14.8. Measurements were performed on 22.8.

With all the tested pairs and HSA concentrations, it is clear that the HSA concentration affects the luminescence signal. However, whether the signal decreases or increases as HSA concentrations increase and how big of a change it is, change with the different pairs. Most likely all pairs have an optimal HSA concentration, which can not be conclusively stated based on these findings as we only tested two HSA concentrations per pair.

Learn

We had hoped that producing our fusion proteins at a larger scale and decreasing elution buffer volume could enhance the IMAC purification results as the protein concentrations would be higher. However, this meant that the amount of periplasmic extract added on top of the pellet was higher, which meant using bigger tubes as well. This made the Co-NTA resin pellet even more susceptible to resuspending further hindering the purification process.

However, SDS-PAGE indicates that the problem might not simply be because of purification, because some of the periplasmic extracts did not seem to contain our fusion protein. This could be because of issues during the cultures. The shakers used with precultures could not reach 37 ℃ so the precultures had been growing at 32 ℃ during the night before they were put to a shaker that could reach 37 ℃ but not 250 rpm. Approximately half of the expression cultures were also grown in the shaker that could only reach about 210 rpm.

While this iteration did have worse success for protein expression and purification than previously, we were able to conduct preliminary measurements on bloodstain age measurements and HSA concentration optimization. Though these results are excitingly indicating that there is a trend between luminescence signal produced by at least one of our pairs and bloodstain age, these results still need further verification before they can be deemed conclusive. There were doubts about some periplasmic extracts not containing any fusion proteins. Older bloodstains should also be measured as should there be more bloodstains of different ages in general for verifying the decrease between days 15 and 22 for example. Biological replicates of bloodstains should be measured as well.

HSA concentration optimization still needs further measurements. It seems that the optimal HSA concentration might also be different for each pair.

More in-depth considerations about what we would optimize in the future can be found here.

Iteration 1

Design

Our team conducted preliminary research on the possible methods for purifying HSA from whole blood. Established protocols could not be found, because hemoglobin-related issues are normally circumvented by using plasma or serum, which do not contain hemoglobin. However, to preserve experimental authenticity, we felt that use of whole blood was necessary in our case. From literature it was determined that ammonium sulphate precipitation guided by protein isoelectric points (pI) could potentially be used to remove hemoglobin and purify HSA.

Build

A protocol, purifying HSA by ammonium sulphate precipitation v.1 for precipitating hemoglobin and isolating HSA was prepared. The protocol combined ammonium sulphate precipitation with pH adjustments, first to the pI value of hemoglobin and then to the pI of HSA. First, hemoglobin would be precipitated by adjusting the pH to ∼7.1 and increasing ammonium sulphate concentration to 50 %, followed by centrifugation. After this, HSA would be precipitated by lowering the pH to ∼4.7 and increasing the ammonium sulphate concentration to 65 %. After centrifuging, HSA was expected to be concentrated in the pellet.

Test

The dried blood samples were extracted following bloodstain extraction for downstream purification v. 1 protocol and purified according to the designed protocol. The pH at different steps was monitored with pH paper. The success of the purification was measured with Bradford assay for total protein tracking and spectrophotometric hemoglobin assay at 420 nm to quantify hemoglobin levels based on Beer-Lambert Law. These measurements also allowed us to observe changes in the protein concentrations between fresh and frozen blood samples, and so assess the effect of our blood handling protocol on the samples. The bovine serum albumin (BSA) standard curve for quantification of Bradford assay results is shown in Figure 20. The results of these measurements are shown in Figure 21.

Figure 20. Standard curve made from bovine serum albumin (BSA) for Bradford assay. The measurements were done on 10.6.2025 with Lambda Bio 40 spectrophotometer (PerkinElmer).

Figure 21. Total protein concentrations and hemoglobin concentrations of blood samples after different steps of ammonium sulphate precipitation. Total protein concentrations were measured with Bradford assay. Hemoglobin concentrations were measured with a spectrophotometer and quantified using Beer and Lambert Law. All values are means of replicates. FE: Fresh extracted blood, FA-P1: Fresh Extracted - Pellet from first fractionation step, FA-P2: Fresh Extracted - Pellet from second fractionation step, 1dFE: 1d frozen extracted blood, 1dF-P1: 1d frozen extracted blood - pellet from first fractionation step, 1dF-P2: 1d frozen extracted blood - pellet from second fractionation step and P-P: plasma pellet. Results for total protein are presented as means and variance reported in ± SD (n=2-5). Hemoglobin measurements are represented as means of two technical replicates. Measurements were done on 9.6. and 13.6.2025.

Additionally, SDS-PAGE was used to analyse purity at different purification steps. The gel picture is shown in Figure 22. Commercial HSA was analyzed alongside the blood samples to identify its molecular mass.

Figure 22. SDS-PAGE performed on 17.6. for samples from the first iteration of HSA purification by ammonium sulphate precipitation. The gel used was Mini-Protean TGX and it was run at 200 V for 30 minutes. Gel was stained with PageBlue Protein Staining Solution (Thermo Fisher) and imagined with Gel Doc EZ Imager (Bio-Rad). The ladder was Color Prestained Protein Standard (New England Biolabs). The upper yellow and the lower red arrow indicate the size of Human serum albumin (66 kDa) and hemoglobin monomer (16 kDa) respectively on the gel. HSA: Human serum albumin (Sigma-Aldrich, ≥99 %), 1dFE: 1d frozen extracted blood, FA-S: Fresh extracted, purification supernatant, 1dF-S: 1d frozen extracted blood, purification supernatant, 1dF-P2: 1d frozen extracted blood - pellet from second fractionation step, FA-P2: Fresh Extracted - Pellet from second fractionation step, P-P: plasma pellet. P-P sample ended up in two wells due to a broken well.

Learn

The designed purification protocol managed to decrease the hemoglobin levels of the samples to some extent, however, not sufficiently for further measurements. This was further supported by the SDS-PAGE, which showed intense bands corresponding to hemoglobin monomers for each sample. With pH measurements it was noted that the pH values at different steps did not correspond to the target pH, which affects the precipitation of proteins.

The measurements of the fresh extracted sample (FE) provided the possibility to validate the performed measurements, as the concentrations could be compared to values found in literature for total protein concentration and hemoglobin concentration. The total protein concentration of 152.4 mg/mL was lower than the estimated concentration of about 220 mg/mL in whole blood. The total protein value was estimated by adding together the plasma serum concentration of 73.2 mg/mL [2] and hemoglobin concentration of 150 mg/mL in hemolyzed blood, as we held the hypothesis that blood would hemolyze. The hemoglobin concentration of 12.4 mg/mL was tenfold lower than the literary values of 150 ± 20 mg/mL for men and 135 ± 15 mg/mL for women [3]. The source of this discrepancy remains unclear to us. However, our goal was the removal of hemoglobin and not its quantification, so for our purposes, a proportional change is just as useful as absolute change.

Iteration 2

Design

As Iteration 1 did not sufficiently remove hemoglobin, we conducted further research on possible purification methods for whole blood. Most published ammonium sulphate purification methods rely solely on HSA precipitation by altering ammonium sulphate concentrations, without pH adjustments. We therefore designed a modified ammonium sulphate precipitation protocol.

Additionally, we discovered an alternative method for purification: heat precipitation. This utilises the higher thermal stability of HSA compared to other blood proteins. By heating the blood samples in the presence of ammonium sulphate and pH optimisation to the pI of hemoglobin, other blood proteins should precipitate while HSA stays soluble.

We also consulted Professor Saara Wittfooth about interpreting the results from the first iteration. She pointed out that the used 1x Tris 10 mM buffer was too weak to keep the pH stable during the purification and this was the reason for the inaccurate pH values.

It was decided to perform the 1dFE hemoglobin measurement again, as the last iteration provided lower than expected concentrations. Additionally, the team adopted the bromocresol green (BCG) assay, a classical colorimetric assay widely used for quantification of serum albumin, to assess purification success. As the available BCG reagent was aged and there was a potential for chemical instability, the protocol had to be optimized for our measurements.

Build

Two new protocols for purification were designed: ammonium sulphate precipitation v.2 and precipitation by heating. The revised ammonium sulphate protocol used alternating ammonium sulphate concentrations to precipitate hemoglobin and HSA separately from the extracted blood sample, without pH adjustments. After reconstitution, the HSA should have been free of most other proteins present in the blood. As there was no available data on how this works with whole blood instead of plasma, multiple ammonium sulphate concentration levels were tested. Based on literature, a method of first increasing ammonium sulphate concentration to 54 % to precipitate most proteins and then increasing it to 70 % to precipitate HSA was found [4]. In addition to this, four other concentration changes were tested. The method exploits the high solubility of HSA in relation to other plasma proteins [5].

The heat purification precipitation protocol uses ammonium sulphate and the pH is controlled to be at the pI of hemoglobin. Both the pH manipulation and the addition of ammonium sulphate were done to promote protein aggregation, and the heating was applied to accelerate the process. The protocol was tested with five different heating times in 60 °C: 60 minutes, 90 minutes, 105 minutes, 120 minutes and 150 minutes.

From the suggestion of Professor Saara Wittfooth, a stronger Tris-HCl buffer of 50 mM was utilized instead of the 10 mM buffer used previously.

The protocol for BCG assay (Measuring HSA concentration with bromocresol green v. 2) was adapted from three commercial sources with bovine serum albumin (BSA) used as the reference standard.

Test

Dried blood samples were extracted following bloodstain extraction for downstream purification v. 2 protocol. The extracted blood samples were then purified according to both designed protocols. The success of the heat purification was measured with Bradford assay for total protein tracking, spectrophotometric hemoglobin assay and bromocresol green assay for measuring HSA. The results are shown in Figure 23. The quantification of total protein was done using the BSA standard curve on Figure 20.

Figure 23. Human serum albumin (HSA), total protein and hemoglobin concentrations for heat purified blood samples heated for different times (60, 90, 105, 120 or 150 minutes). HSA concentrations were measured using Bromocresol green measurement and total protein concentrations with Bradford assay. Hemoglobin concentrations were measured with a spectrophotometer and quantified using Beer and Lambert Law. The measurements were done 14.7., 8.7. and 22.7.2025.

Initially during the bromocresol green measurements abnormally low transmittance values for blank were observed, which hinted to possible turbidity or optical interference. To create the standard and the method for BCG measurements the following changes were made:

• Increased sample volume from 20 µL to 40 µL, which resulted in stronger and more consistent absorbance reading.
• Added a 37 °C incubation step, which was used in many older BCG protocols. This appeared to improve signal stability.

Despite these improvements we observed that the relationship between absorbance and albumin concentration was closer to logarithmic rather than expected linear response. This was likely caused by the instability of the reagent or the use of BSA instead of HSA in the creation of the standard. After consultation with our PI Pauli Kallio, we decided to proceed with linear standard curve, which achieved an R2 value of 0.9679 despite some deviation at lower concentrations. This was deemed sufficient for generating semi-quantitative data for our purification attempt.The BSA standard curve for quantification of bromocresol green assay results is shown in Figure 24.

Figure 24. Standard curve for bromocresol green assay quantification made with bovine serum albumin (BSA). Variance is reported in ± SD (n=3). Measurements were done on 15.7.2025 with Lambda Bio 40 spectrophotometer (PerkinElmer).

For the ammonium sulphate purification samples only hemoglobin measurements were performed, as it was visually noticeable that the hemoglobin had not precipitated from the final supernatant due to its red color. The hemoglobin measurements were performed for ammonium sulphate precipitation series 1 and 3. The results are shown in Figure 25.

Figure 25. Hemoglobin concentrations of samples from different steps of ammonium sulphate precipitation. Hemoglobin concentrations were measured with a spectrophotometer and quantified using Beer and Lambert Law. AS1: ammonium sulphate purification series 1, AS3: ammonium sulphate purification series 3, the number indicates the concentration of ammonium sulphate used in a given step and S: final supernatant. The measurements were done 14.7.2025.

Hemoglobin measurement from extracted and untreated 1dFE sample. The measurement was done with two technical replicates, which when quantified with Beer-Lambert Law gave the hemoglobin concentrations of 116.4 mg/mL and 115.7 mg/mL.

Learn

Heat precipitation gave the most promising results from the two methods tested. The best total protein to hemoglobin ratio was achieved with 120 minutes heating time in 60 °C. However, from this experiment it is unclear whether HSA retains its native form after the heat treatment. Bromocresol green (BCG) assay did not provide reliable quantitative results for HSA concentration as they are higher than expected. Due to the lack of reliability, the total protein concentrations were used rather than HSA concentrations for estimating the success of purification. If the purification was successful, most of the protein measured in Bradford measurement should be HSA, however, further confirmation is required to confirm this assumption. BCG data can however be seen to confirm the presence of HSA in the samples after heat treatment, despite its limitations in accuracy. The method for BCG assay was decided to be used in the BCG measurements in the future iterations to give indicative data and compare HSA levels in different samples.

The ammonium sulphate precipitation was unsuccessful, as unexpectedly the supernatant contained the highest hemoglobin concentration, even though it should have precipitated. It is possible that the 100 % ammonium sulphate stock used to increase the ammonium sulphate concentration during precipitation was not made properly and this was the reason for the method not providing expected results. Based on these findings, we chose to abandon the ammonium sulphate method and focus on optimizing the heat purification.

The hemoglobin concentrations of 116.4 mg/mL and 115.7 mg/mL are closer to the expected 150 mg/mL concentration compared to the results from last iteration. From this it was determined that the measurement protocol is capable of giving concentrations in the correct range. It is, however, unclear what mistake led to the abnormal results in the first iteration.

Iteration 3

Design

Due to the promising results of heat precipitation in Iteration 2, we wanted to optimize the method further. Since 120 minutes gave the most optimal ratio between total protein and hemoglobin levels, more heating times at around 120 minutes were tested.

Build

For this iteration of the protocol the heating times were decided in the range of 110 minutes to 150 minutes, with 5 minutes increment between times.

Test

The dried blood samples were extracted following bloodstain extraction for downstream purification v. 2 protocol. The extracted blood samples were then purified according to the designed protocol. To compare the success of purification at different heating times the following measurements were performed: Bradford assay for total protein tracking, spectrophotometric hemoglobin assay (Figure 26) and bromocresol green assay for measuring HSA (Figure 27).

Figure 26. Human serum albumin (HSA), total protein and hemoglobin concentrations for heat purified blood samples heated for different times. Total protein concentrations were measured with Bradford assay and quantified with the BSA standard curve on Figure 20. Hemoglobin concentrations were measured with a spectrophotometer and quantified using Beer and Lambert Law. Two technical replicates are shown for each data point. The measurements were done on 25.7. and 28.7.2025.

Figure 27. Human serum albumin (HSA) concentrations for heat purified blood samples heated for different times. HSA concentrations were measured using Bromocresol green measurement. Quantification was done with BSA standard on Figure 23. The measurements were done 24.7.2025.

Learn

The heating time of 115 minutes gave the best balance between total protein yield and hemoglobin removal. Although 110 minutes gave a higher yield of HSA, we opted on choosing 115 minutes as BCG measurements provided inconsistent results between different samples. In this iteration, the error was likely caused by too small pipetting volumes. Longer heating of 150 increased the removal of hemoglobin, while giving a decent protein yield, but there was worry about the long heating time causing conformational changes in HSA. Overall, the data suggest that 115 minutes heating at 60 °C provided the most promising results for purification, although the possibility of structural changes taking place in HSA remains possible.

Iteration 4

Design

After the last iteration, the refined purification protocol was decided to be used on aged blood samples that had been stored frozen. This was to assess robustness and see if any trend in protein concentrations is visible in different age samples.

Build

Frozen blood samples of varying ages were extracted and purified according to the protocol in Iteration 3.

Test

The dried and frozen blood samples of different ages were extracted following bloodstain extraction for downstream purification v. 2 protocol. The extracted blood samples were then purified according to the protocol from Iteration 3, with the heating time of 115 minutes at 60 °C. To compare the success of purification at different heating times the following measurements were performed: Bradford assay for total protein tracking, spectrophotometric hemoglobin assay and bromocresol green assay for measuring HSA concentration. The results are shown in Figure 29. The BSA standard curve for quantification of Bradford assay with plate reader is shown in Figure 28.

Figure 28. Bovine serum albumin (BSA) standard for Bradford assay quantification. Variance is reported in ± SD (n=3). Measurements were done on 22.8.2025 with Infinite plate reader (Tecan).

Figure 29. Human serum albumin (HSA), total protein and hemoglobin concentrations for heat purified blood samples of different ages. The heating time used was 115 minutes. HSA concentrations were measured using Bromocresol green measurement and total protein concentrations with Bradford assay, and quantified using BSA standard curves on Figures z and Figure q. Hemoglobin concentrations were measured with a spectrophotometer and quantified using Beer and Lambert Law. Variance reported in ± SD (n=4). The measurements were performed 20.8., 22.8. and 25.8.2025.

Additionally, SDS-PAGE was performed to assess the success of the purification by comparing the relative band intensities of HSA and hemoglobin in extracted and purified blood samples of varying ages. The gel picture is shown on Figure 30.

Figure 30. SDS-PAGE performed on 26.8. for samples from the final HSA purification by heating for blood samples of different ages. The gel used was Mini-Protean TGX and it was run at 150 V for 50 minutes. Gel was stained with PageBlue Protein Staining Solution (Thermo Fisher) and imagined with Gel Doc EZ Imager (Bio-Rad). The ladder was Color Prestained Protein Standard (New England Biolabs). The upper yellow and the lower red arrow indicate the size of Human serum albumin (66 kDa) and hemoglobin monomer (16 kDa) respectively on the gel. Number before “d” tells the number of days the blood sample had been aged, E: extracted, S: final supernatant from purification, letter A or B tells the replicate used.

Learn

The combination of spectrophotometric measurements and SDS-PAGE confirmed for us that the purification protocol could reduce the hemoglobin level in the aged frozen blood samples across all aging times. The hemoglobin levels after purification are consistently lower than the levels in extracted blood. This is also identifiable in SDS-PAGE in the relative intensities of the HSA and hemoglobin monomer bands between purified and extracted blood samples. Even though the purified samples are overloaded on the gel, the intensity differences between HSA and hemoglobin bands can be compared. The purified samples (FAS and FBS samples) show a strong HSA band and fainter hemoglobin monomer band. In comparison, the extracted blood samples (FEA and FEB) feature a more intensive hemoglobin monomer band and a fainter HSA band. Both types of samples show indistinct bands at around 30 kDa. These could be the result of other blood proteins or partial degradation of HSA.

The overall decrease in protein concentrations according to all three measurements suggests protein degradation. The data from the spectrophotometric measurements can be viewed as giving preliminary results, however, there are uncertainties with assay linearity as not all of the samples fit into the calibration line of measurements, and therefore the concentrations cannot be viewed as strictly quantitative. Even though the BCG assay overestimated the HSA concentrations, a trend in lowering concentrations in older stains is visible. A further iteration would be needed to quantify the results. It was realised that one reason for bromocresol green measurement overestimating the HSA concentration could be due to hemoglobin and HSA-bound heme groups absorbing light at used 630 nm wavelength in addition to the issues identified previously. Despite the need for further optimization Ellman’s assay was tested with the purified blood samples. Read more about the Ellman’s assay in Results.