Introduction

VeriFied is a homogeneous enzyme fragment complementation assay (EFCA) based on the interaction of nanobodies with human serum albumin (HSA), which is found in bloodstains. The assay has two bifunctional fusion proteins, each with a different nanobody and a complementary half of a luciferase enzyme, NanoLuc, attached by a peptide linker. Upon interaction of both nanobodies with HSA, the enzyme fragments are brought to a close proximity of each other. This results in complementation of the enzyme and therefore, the restoration of enzyme activity and a bioluminescent signal. The signal is dependent on the binding ability of the nanobodies. As bloodstains age, oxidation-mediated structural changes in HSA disrupt epitope integrity and therefore also the binding of the nanobodies, leading to a decrease in signal intensity.

Enzyme fragment complementation assay

EFCAs are commonly used to detect protein-protein interactions both in vivo and in vitro. The technology is based on split-enzyme engineering where the reporter enzyme is split into two or more complementary fragments that are inactive on their own but able to reconstitute into an active whole enzyme [1]. In homogeneous in vitro applications, low intrinsic affinity of the enzyme fragments is essential to avoid analyte-independent complementation and to establish a high signal-to-noise ratio [2]. The intrinsic affinity should be substantially lower than that of the interacting proteins of interest (K_D in μM) [3]. Furthermore, the accuracy and analytical sensitivity of the assay are dependent on the relative concentrations of assay components, which need to be optimized experimentally [4].

The homogeneous system offers a platform for a rapid mix-and-read application, suitable for an on-site solution. Additionally, the bioluminescent signal provides fast quantitative data that could be either used to compare bloodstains and rule out the irrelevant ones or possibly connect the signal intensity to the bloodstain age.

Biomarker for estimating the age of bloodstain

Our aim was to design our test while taking into consideration the real-world application. We began the design by going through several possible biomarkers and assessing their suitability for an on-site test that would be easy-to-use, noncontaminating, nondestructive and nonprofiling. We evaluated the biomarkers based on their abundance, degradation time and method, possible ways of detection and interindividual variability.

In crime scene investigation, the available sample volume is often small and the samples should be kept intact for further downstream forensic analysis. For an on-site and easy-to-use test, the required processing of the sample and the complexity of the test equipment should be minimal. Therefore, we sought to find an abundant biomarker that could be tested from a whole blood sample and would be detectable even in small sample sizes. In addition, for further forensic analysis integrity of the genetic information is crucial. The test should not contaminate the crime scene with foreign genetic material.

Several biomarkers have been studied for bloodstain age determination. The most researched ones are hemoglobin and genetic markers such as DNA [5], different RNAs [6]. Other studies include hormones related to the circadian rhythm [7], metabolomics [8], plasma proteins [9] and enzyme activity. Although the existing studies show promise, none of the developed methods we found fully met the objectives we had identified. Therefore, we decided to make an effort towards biomarker discovery and developed a method based on human serum albumin oxidation.

Due to the uncontrolled nature of the environment and the lack of circumstantial information, crime scenes present a challenge for diagnostics. Without contextual data, it is difficult to predict how the biomolecular composition of bloodstains change over time. Factors such as bloodstain size, surface properties like porosity, light exposure, temperature, humidity, microbial activity, and the presence of cleaning agents are expected to affect the natural aging of bloodstains.

Human serum albumin

In comparison to other proposed biomarkers, we chose HSA due to its high abundance in blood, the possibility of an immunological approach, and moderate inter-individual variability, more notably the variability related to sex [10]. Ruling out the identification of sex with the assay is an ethical consideration but also helpful in standardising the results, and in cases where the parties are unidentified.

Our hypothesis is that in dried bloodstains the degradation of HSA happens due to oxidation. Most notable structural changes happen in domain I due to the oxidation of the free thiol in cysteine 34. Other significant impairments happen in surface exposed methionines. After identifying the specific areas susceptible for oxidative damage, we faced a challenge: how to determine a standard when the original volume of the sample is not known? Rather than us using a separate biomarker for the standard, Professor Urpo Lamminmäki suggested that we take advantage of the divergent oxidative modifications and measure two qualities from the same biomarker. This way we can use the other result as a point of comparison, both as the standard, and the control.

VeriFied includes two separate assays, meaning two fusion protein pairs (Figure 1.). One where both of the nanobodies target stable areas on the HSA structure and another, where one nanobody targets a stable area and the other targets an unstable area. The stable-stable pair measures the total amount of HSA and the signal from the stable-unstable pair correlates with the amount of reduced HSA. Comparing the two, we can determine the amount of oxidized HSA in the sample and thus evaluate the age of the bloodstain.

Figure 1. The mechanism behind VeriFied assay. VeriFied consists of a pair of bifunctional proteins. The nanobody part of the proteins bind to human serum albumin, while the NanoLuc fragment part produces a bioluminescent signal when it meets its complementary fragment. One of the nanobodies target a stable epitope of human serum albumin, while the other targets an oxidation prone epitope (1). NanoLuc fragments bind to each other and produce a bioluminescent signal when both of the nanobodies bind to human serum albumin (2). When HSA is oxidized, the oxidation prone epitopes have degraded, which in turn inhibits nanobody binding and reduces produced signal (3). VeriFied test includes a control, in which both nanobodies of the pair bind to stable epitopes. Thus, the control produces a luminescent signal even with oxidized human serum albumin (4).

The structure of nanobodies

Nanobodies, discovered in 1993 by Hamers-Casterman et al., are small heavy-chain variable domains (VHHs), originally sourced from camelid antibodies [11]. Uniquely among mammals, camelids possess homodimeric heavy-chain antibodies (HCAbs), meaning the antibodies consist of only the heavy-chain [12]. Compared to conventional antibodies that are formed from heavy- and light-chains, the homodimeric nature of HCAbs allows the isolation of only the VHHs (Figure 2.), which work as functional binding units capable of binding various antigens with high affinity [12].

Figure 2. Differences between nanobodies and antibody fragments. Yellow indicates variable domains, blue indicates constant domains. Fab: fragment antigen-binding, scFv: single-chain variable fragment.

Benefits of using nanobodies

The single domain nature of nanobodies in addition to an extra disulfide bond, accounts for a more rigid structure with a higher thermal stability when compared to multidomain antibody fragments [13]. Additionally, the surfaces of nanobodies are more rich in hydrophobic amino acid residues compared to conventional antibody fragments, which increase their solubility and reduce risk of aggregation [13]. This risk of aggregation and the relatively more complex structure of multidomain antibody fragments leads to difficulties in using bacterial expression systems in their production despite continued attempts and improvements [14]. Functional nanobodies with their simpler and more stable structure can instead usually be expressed in high yields in bacterial systems such as E. coli [15]. Additionally, the small size of nanobodies (12-15 kDa) makes them ideal for homogeneous immunometric tests using enzyme fragment complementation, since they are less likely to cause issues related to steric hindrance than other, larger antibody fragments.

Origin of our nanobodies

When looking for anti-HSA nanobodies, we found a nanobody library by Shen et al. 2021 that contains ~70 high affinity anti-HSA nanobodies with varying epitopes [16]. The exact epitopes of the nanobodies were not detailed in the study, so we used protein-protein docking software to analyze the epitopes based on cross-link mass spectrometry data provided by the article. In addition to the nanobodies from the article by Shen et al. 2021, we used the nanobody ALB8 (PDB:8Z8V), which was sourced from PDB and had a structure solved by X-ray crystallography. We chose suitable nanobodies to use in our project based on predicted epitope stability and steric hindrance.

To emit a signal using enzyme fragment complementation, two nanobodies are required, one for each enzyme fragment. These two nanobodies must not compete for the same epitope and steric hindrance should not stop them from binding or the enzyme complementation can not work. To gather accurate information about the age of the bloodstain, we need to measure the amount of oxidized HSA, but we also need to measure the total HSA in the sample to use as a reference to calculate the ratio of oxidized to reduced HSA. A simple illustration of nanobody mediated EFCA can be seen in Figure 3.

Figure 3. Cartoon of enzyme fragment complementation assay. The heart indicates human serum albumin with domains I, II, and III shown in different colors, the blue shoulders indicate two nanobodies with different epitopes, the yellow hands indicate the NanoLuc enzyme fragments. Therefore, one arm from shoulder to hand represents a nanobody-NanoLuc fragment fusion protein.

For the test measuring oxidized HSA, the two nanobodies consist of a nanobody with an oxidization stable epitope, and another with an epitope unstable to oxidation based structural changes. The other test, measuring total HSA in the sample, consists simply of two nanobodies with oxidation stable epitopes. To improve the odds of finding a working pair of nanobodies for each pair, we chose at least two nanobodies to represent each role. Since both tests need at least one nanobody with an oxidation stable epitope, this nanobody can be the same for both tests.

Nanobodies with epitopes that are more stable to oxidation

Based on protein-protein docking and modeling to assess steric hindrance, two nanobodies were chosen as options to represent the stable epitope that is shared between the oxidized HSA and the total HSA tests. These were ALB8 and Nb80, two nanobodies that have epitopes on domain II (Figure 4. a, b) with Nb80 having a slightly concave binding site rich in stable alpha-helixes.

Figure 4. Human serum albumin (PDB:1AO6) with nanobody epitopes. Blue: domain I, yellow: domain II, red: domain III, cyan: Cys34, green: nanobody epitope. (a) ALB8, (b) Nb80, (c) Nb126, (d) Nb77, (e) Nb29, (f) Nb13, (g) Nb118. Figure made in pyMOL.

Additionally, another nanobody with an oxidation stable epitope to pair with either ALB8 or Nb80 is needed for the test measuring total HSA. We chose Nb126 with an epitope in an alpha-helix rich area of HSAs domain III (Figure 4. c) and Nb77 with an epitope between domains II and III (Figure 4. d).

Nanobodies with oxidation prone epitopes

For the nanobodies with oxidation unstable epitopes, we chose three nanobody options that all bind to different sides of domain I of HSA (Figure 4. e, f, g). Nb29 has an epitope directly on top of the oxidation prone amino acids Cys34, Met87, and Met123 with potential direct interactions with the residues. Nb13 binds to a side of domain I close to domain II and far from Cys34.The epitope of Nb118 is located mostly on domain I with interactions with all three domains and an alpha-helix connected directly to Cys34.

Choosing split enzyme

NanoLuc was chosen as the split reporter enzyme due to its small size and brighter signal compared to other reporter enzymes [4]. Its compact size helps to minimize steric hindrance compared to other larger signaling enzymes. There is also an established protocol for using NanoLuc in split form (NanoLuc Binary Technology, NanoBiT, by Promega) which made it suitable for our use. The NanoLuc enzyme is split into two fragments called LgBit (18 kDa) and a complementary peptide SmBit (1.3 kDa) [4]. These have been engineered to have low affinity to each other (K_D 190 = μM) and minimal enzyme activity when unbound [4]. Enzyme activity is restored only when the two fragments reassemble. In our design, the nanobodies have much higher affinity to HSA compared to the enzyme fragments’ affinity. When the nanobodies bind to their respective epitopes on HSA, the enzyme fragments are brought to close proximity to each other and can reassemble. Split NanoLuc has also previously been used with nanobody fusions, further supporting its applicability [4].

NanoLuc works by producing a bright luminescent signal in the presence of its substrate, furimazine. We opted for bioluminescence as a signal type due to its high sensitivity which makes it better suited for use with blood samples where background interference can be an issue with colorimetric detection [1]. Furthermore, the signal could potentially be measured on-site with a portable luminometer, making it a suitable signal type for the end product.

Linker

The nanobodies and enzyme fragments are connected using a flexible GGGGS linker repeated four times. Our structural modelling indicated that a length of three repeats was needed between the enzyme fragment and the nanobody to ensure that the linker would not hinder the enzyme fragment binding. After consulting Professor Urpo Lamminmäki, it was decided that longer linker length 4xGGGGS would be used as the additional length would likely not cause any issues and would ensure the fragments’ ability to bind. To maintain consistency across all of our constructs and focus on testing other aspects of our design, the same linker length was used for all of the fusion proteins.

Proteins

Both VeriFied assays, the test and the control, require two bifunctional proteins (Figure 5). The nanobody part of the protein binds to HSA, while the enzyme fragment part produces bioluminescence when brought into a close enough proximity of its complementary fragment.

Figure 5. Human serum albumin (HSA) with bound fusion proteins. Blue: HSA (PDB:1AO6), orange: Nb29, green: ALB8 (PDB:8Z8V), pink: NanoLuc LgBit, yellow: NanoLuc SmBit, white: linkers. Figure made in pyMOL.

Ten different proteins were designed in total. Seven different proteins were designed as possible proteins for the VeriFied test. Four of these bind to a predicted stable epitopes on HSA and contain either the LgBit or SmBit fragment of the NanoLuc enzyme (Table 1). Two proteins with LgBits were designed in total.

Table 1. Proteins that can be used in both the test and the control and should bind HSA no matter the oxidation state. pIs and molecular weights (Da) acquired from Benchling “Analyze as translation” tool.

Name	Structure	Predicted epitope oxidation stability	Expression in	pI	Da
ALB8-LgBit	(PelB) - ALB8 - linker - LgBit - TEV-site - 6xHis	A stable epitope	periplasm	4.93	34,279.53
Nb77-SmBit	(PelB) - Nb77 - linker - SmBit - TEV-site - 6xHis	A stable epitope	periplasm	8.96	20,246.62
Nb80-LgBit	(PelB) - Nb80 - linker - LgBit - TEV-site - 6xHis	A stable epitope	periplasm	5.10	36,678.17
Nb126-SmBit	(PelB) - Nb126 - linker - SmBit - TEV-site - 8xHis	A stable epitope	periplasm	8.43	20,351.72

Three of the seven proteins designed for the VeriFied bind to predicted unstable epitopes on HSA (Table 2). All of these three proteins contain the SmBit fragment of the NanoLuc enzyme.

Table 2. Proteins that can be used in the VeriFied test and should bind only to reduced HSA. pIs and molecular weights (Da) acquired from Benchling “Analyze as translation” tool.

Name	Structure	Predicted epitope oxidation stability	Expression in	pI	Da
Nb13-SmBit	(PelB) - Nb13 - linker - SmBit - TEV-site - 6xHis	An unstable epitope	periplasm	6.58	20,974.27
Nb29-SmBit	(PelB) - Nb29 - linker - SmBit - TEV-site - 6xHis	An unstable epitope	periplasm	8.77	19,640.06
Nb118-SmBit	(PelB) - Nb118 - linker - SmBit - TEV-site - 6xHis	An unstable epitope	periplasm	8.40	20,508.65

Additionally, we designed two proteins that could be used in heterogeneous assays (Tables 3 and 4). These proteins contain a nanobody, binding either to a predicted stable or unstable epitope on HSA, and full functional NanoLuc enzyme. There are two slightly different Nb29-NanoLuc proteins as we received a plasmid for expressing it from two different manufacturers. In the envisioned heterogenous assay, HSA would be bound to the bottom of the wells of a microtiter plate, these proteins would be added and the solution would be incubated for a while before rinsing. Thus, we would be able to get signal only from proteins bound to HSA. The experiment would be done with reduced and oxidized HSA, as our hypothesis is that the nanobodies that bind to unstable epitopes could only bind to reduced HSA.

Table 3. Proteins that can be used alone in heterogenous assay and should bind HSA no matter the oxidation state. pIs and molecular weights (Da) acquired from Benchling “Analyze as translation” tool.

Name	Structure	Predicted epitope oxidation stability	Expression in	pI	Da
Nb80-NanoLuc	(PelB) - Nb80 - linker - NanoLuc - TEV-site - 6xHis	A stable epitope	periplasm	5.65	38,043.83

Table 4. Proteins that can be used alone in heterogenous assay and should bind HSA no matter the oxidation state. pIs and molecular weights (Da) acquired from Benchling “Analyze as translation” tool.

Name	Structure	Predicted epitope oxidation stability	Expression in	pI	Da
Nb29-NanoLuc (in pET-21(+))	(PelB) - Nb29 - linker - NanoLuc - TEV-site - 6xHis	An unstable epitope	periplasm	6.12	37,395.39
Nb29-NanoLuc (in pET-IDT C His)	(PelB) - Nb29 - linker - NanoLuc - TEV-site - 8xHis	An unstable epitope	periplasm	6.35	37,538.49

One protein was designed to ensure that the NanoLuc fragments can form a functional enzyme (Table 5). This protein, called NanoChuck, contains both of the NanoLuc fragments separated by a linker.

Table 5. Proteins designed to ensure that the fragments can form a functional enzyme. pIs and molecular weights (Da) acquired from Benchling “Analyze as translation” tool.

Name	Structure	Predicted epitope oxidation stability	Expression in	pI	Da
NanoChuck	LgBit - linker - SmBit - TEV-site - 8xHis	Does not bind to HSA	cytoplasm	4.71	21,336.94

Potential pairs

With seven different proteins designed for the VeriFied test, 8 different pairs for EFCA exist (Table 6). Four of these pairs were envisioned to be used to measure the concentration of reduced HSA so the bloodstain’s age. Three of these pairs have nanobodies that both bind to stable epitopes and thus the pair could be used as a standard to measure the concentration of all albumin. One pair is hypothesized to be impossible as previous data by Shen et al. suggest that the nanobodies compete for the same epitope [16].

Table 6. Potential nanobody pairs for the VeriFied test. 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.

The hypotheses on which pairs would work together was done based on in silico modeling of the nanobody epitopes on the surface of HSA. Using modeling, we chose nanobody pairs with epitopes far enough apart to ensure steric hindrance would not reduce affinity or make the simultaneous binding of both nanobodies impossible.

Periplasmic expression in E. coli BL21(DE3)pLysS

Escherichia coli is one of the most common model organisms used in scientific research. Thus, it is no wonder it has also been used to produce a myriad of proteins. As E. coli is such a common host, finding multiple articles about nanobody production in E. coli is no hard task. Producing antibody fragments that require no glycosylation, such as nanobodies, in E. coli offers many advantages over other organisms, such as easy production with low cost [14].

The E. coli BL21(DE3) strain is derived from the E. coli B strain which lacks the ion protease and the outer membrane protease OmpT. The BL21(DE3) strain carries the λDE3 lysogen containing T7 RNA polymerase gene under lacUV5 promoter. To transcribe the T7 promoter regulated target gene, IPTG is used to induce the T7 RNA polymerase cascade system [17].

The E. coli BL21(DE3)pLysS strain is used with T7 promoter-based expression systems like pET [17]. All of the nanobodies used in our project except ALB8 have been successfully produced in the BL21(DE3) strain with a pET-21 expression system [14]. BL21(DE3)pLysS strain carries a pLysS plasmid unlike the BL21(DE3) strain. The pLysS plasmid codes for T7 lysozyme which inhibits T7 RNA polymerase. This inhibition reduces the basal expression of the gene of interest. Additionally, pLysS confers resistance to chloramphenicol and has the p15A origin. As the BL21(DE3)pLysS has a tighter expression control than just the BL21(DE3), it is recommended to reduce leakage [17].

One of the biggest issues with nanobody production in E. coli is the formation of inclusion bodies. Therefore, conventionally nanobodies are produced in the periplasm where the oxidizing conditions aid disulfide bond formation. Disulfide bonds help stabilize the nanobody structure [15]. By attaching a signal peptide sequence to the protein sequence, the expression can be directed to periplasm.

Sequences

We got the amino acid sequences for all our nanobodies except ALB8 from the study by Shen et al [16]. ALB8 nanobody amino acid sequence was acquired from PDB file 8Z8V [18].

We got our NanoLuc fragment sequences from two different places. We used the SmBit sequence found in the Registry of Standard Biological Parts (Part:BBa_K1761006) [19]. We got the LgBit amino acid sequence from the study by Dixon et al. [3].

The TEV sequence is from the New England Biolabs website [20]. We used the PelB sequence found in the Registry of Standard Biological Parts (Part:BBa_J32015) [21]. For linker we decided to use the flexible linker sequence GGGGS x n. We used four repeats of the sequence because according to our modeling work that would be sufficient length-wise.

Plasmids

We have used two backbone plasmids. Sequences for proteins NanoChuck, Nb29-NanoLuc and Nb126-SmBit are in a pET-IDT C His backbone [22]. Sequences for proteins ALB8-LgBit, Nb13-SmBit, Nb29-NanoLuc, Nb29-SmBit, Nb77-SmBit, Nb80-LgBit, Nb80-NanoLuc and Nb118-SmBit are in a pET-21(+) backbone [23]. You can find more information regarding this choice here.

pET-IDT C His backbone is made for bacterial expression. It includes a gene for kanamycin resistance (KanR) for antibiotic selection. pET-21(+) backbone is also designed for bacterial expression purposes. It includes a gene for ampicillin resistance (AmpR) for antibiotic selection. Both plasmids have a lac operator, T7 promoter and 6xHis. We added 2xHis to C-terminus for the inserts in pET-IDT C His backbones. All inserts have a C-terminal TEV-site sequence for His-tag cleavage. As pET-21(+) does not include a RBS, we added the RBS sequence from pET-IDT C His plasmid to our constructs ordered in a pET-21(+) backbone.

Our plasmids code for three types of fusion proteins: ones with SmBit, ones with LgBit and ones with whole NanoLuc. All of them have a N-terminal PelB sequence for periplasmic expression. There is also the exception NanoChuck, which consists of NanoLuc fragments connected via GGGGS x 4 linker. Nanochuck does not contain a PelB sequence and is thus expressed in the cytoplasm.

As many of our sequences were provided as amino acid sequences, we used codon optimization tools provided by IDT and Twist Bioscience to turn them to nucleotide sequences. You can find more information regarding the codon optimization here.