Analytical Studies of Antimicrobial Peptides as Diagnostic Biomarkers for the Detection of Bacterial and Viral Pneumonia

Pneumonia remains one of the leading causes of infectious mortality and significant economic losses among our growing population. The lack of specific biomarkers for correct and timely diagnosis to detect patients’ status is a bane towards initiating a proper treatment plan for the disease; thus, current biomarkers cannot distinguish between pneumonia and other associated conditions such as atherosclerotic plaques and human immunodeficiency virus (HIV). Antimicrobial peptides (AMPs) are potential candidates for detecting numerous illnesses due to their compensatory roles as theranostic molecules. This research sought to generate specific data for parental AMPs to identify viral and bacterial pneumonia pathogens using in silico technology. The parental antimicrobial peptides (AMPs) used in this work were AMPs discovered in our previous in silico analyses using the HMMER algorithm, which were used to generate derivative (mutated) AMPs that would bind with greater affinity, in order to detect the bacterial and viral receptors using an in silico site-directed mutagenesis approach. These AMPs’ 3D structures were subsequently predicted and docked against receptor proteins. The result shows putative AMPs with the potential capacity to detect pneumonia caused by these pathogens through their binding precision with high sensitivity, accuracy, and specificity for possible use in point-of-care diagnosis. These peptides’ tendency to detect receptor proteins of viral and bacterial pneumonia with precision justifies their use for differential diagnostics, in an attempt to reduce the problems of indiscriminate overuse, toxicity due to the wrong prescription, bacterial resistance, and the scarcity and high cost of existing pneumonia antibiotics.


Introduction
Pneumonia is challenging because of its significant impact on global health [1]. It is characterized by specific infections with a distinct clinical source, presentation, epidemiology, and pathogenesis [2]. The disease arises from an infection in the lower respiratory tract, starting with inflammation and impaired tissue functioning [3]. Diagnosis of the disease may be difficult for an established respiratory condition, cardiac failure, and non-pneumonic respiratory infection because there is no evidence-based consensus on its clinical signs and predictive symptoms [4]. The diagnostic biomarkers used at present are produced in response to other conditions such as HIV and atherosclerosis [5]. The use of antibodies has remained questionable despite its high sensitivity due to challenges such as cross-reactivity [6]. The discovery of more specific and sensitive biomarkers for detecting pneumonia in patients' samples is an ultimate goal of researchers to reduce the high mortality arising from the disease. The present research aimed to use the parental anti-pneumonia AMPs generated using the HMMER machine learning tool from our previous studies demonstrated using in silico binding studies as templates to create derivative AMPs with potential greater affinity for the receptors. The site-directed mutagenesis technique was used to increase the binding efficacy of these potential anti-pneumonia AMPs. Increasing the binding affinity between the ligand and the receptor could improve the possible test specificity during molecular validation. A test kit's ability to detect a pathogen at very low concentrations with high specificity and sensitivity is paramount in creating a perfect test kit to differentiate one pathogen from another and then bind to it with high affinity.

Identification of Hotspot Residues
Before the commencement of in silico site-directed mutagenesis, essential amino acids have to be identified. The Knowledge-based FADE and Contacts (KFC) Server ( (accessed on 2 February 2019) was employed for the precise identification of hotspot residues or mutation-sensitive residues to perform in silico site-directed mutagenesis effectively Representative codes for the generation of the parental putative antimicrobial peptides from the HMMER algorithm, a machine learning tool, where (i) indicates the generation of Clustal for the training datasets which yields the msf file; (ii) generates the hmm file from the msf file using hmmbuild; (iii) calibrates the hmm profile; (iv) searches for matching motifs in the test or negative datasets; and (v) searches for motifs in proteomes of other organisms for the discovery of new antimicrobial peptides.

Identification of Hotspot Residues
Before the commencement of in silico site-directed mutagenesis, essential amino acids have to be identified. The Knowledge-based FADE and Contacts (KFC) Server ((accessed on 2 February 2019) was employed for the precise identification of hotspot residues or mutation-sensitive residues to perform in silico site-directed mutagenesis effectively ( Figure 2). The KFC Server uses two predictive models, shape specificity features and biochemical contacts, to display the hotspots' predictive accuracy in identifying mutationsensitive amino acid residues [18].

In Silico Site-Directed Mutagenesis (SDM)
An in silico SDM approach was used to generate derivative AMPs that displayed increased predicted binding affinity to the receptor proteins of the bacterial and viral pneumonia by substituting amino acid residues at regions distinct from the peptide hotspots using the parental AMPs against viral and bacterial pneumonia as templates. All amino acids were substituted to generate derivate AMPs and to maintain the predicted activity and functioning of the AMPs.

Knowledge-Based FADE and Contacts (KFC) Analysis
The Knowledge-based FADE and Contacts (KFC) Server was used to identify hotspots or the subset of residues that determined the receptor proteins' interface binding free energy. The results are tabulated in Table 1, showing the hotspot residues for each interaction between a specific ligand and its receptor. Any amino acid hotspot residues identified within the interaction site of the parental AMPs cannot be substituted during SDM analysis to maintain their functionality.

In Silico Protein-Protein Interaction Study
The docking interaction analysis of the antibacterial and antiviral pneumonia AMPs was carried out against their respective protein receptors, as identified in our previous study [15]. The online protein-protein interaction server PATCHDOCK was employed for this analysis [23]. Docking of each derivative AMP 3D structure was carried out against its specific protein receptor's 3D structure, setting the RMSD as 4.0. Visualization of the 3D structure was carried out using PyMol version 1.3 ( Figure 2) [22].

Knowledge-Based FADE and Contacts (KFC) Analysis
The Knowledge-based FADE and Contacts (KFC) Server was used to identify hotspots or the subset of residues that determined the receptor proteins' interface binding free energy. The results are tabulated in Table 1, showing the hotspot residues for each interaction between a specific ligand and its receptor. Any amino acid hotspot residues identified within the interaction site of the parental AMPs cannot be substituted during SDM analysis to maintain their functionality.  Table 1 above represents the amino acid residues of the bacterial and viral receptors used against the respective putative AMPs (parental AMPs) at the hotspot positions which play critical roles in the binding process and must not be substituted as they may change the functionality of the AMPs. BOPAM is the given name of the AMPs, with the following two letters being derivatives from the pathogens, for example, Klebsiella pneumoniae = KP.

In Silico Site-Directed Mutagenesis of the AMPs
After identifying the hotspot residues at the interaction sites of the anti-pneumonia AMPs and their receptors, site-directed mutagenesis was performed with these parental anti-pneumonia AMPs as input ( Table 2). All amino acids substituted to generate derivate AMPs had similar characteristics to the amino acids present in the parental AMPs to maintain the predicted activity and functioning of the AMPs. BOPAM-AB1, 2, and 3 were mutated at position 10, where lysine was changed for arginine, while BOPAM-AB4 and 5 had leucine and isoleucine substituted for methionine at position 16. Moreover, BOPAM-SP1, 2, 3, 4, and 5 had serine or asparagine substituted for aspartate at position 25, while lysine was substituted for arginine in BOPAM-SP6 and 7 at position 36. Cysteine and tryptophan of BOPAM-KP1, 2, 3, and 7 were mutated at position 2, while glycine and serine of BOPAM-KP4, 5, and 6 were mutated at position 9, all with asparagine.

Physicochemical Properties of the Derivative AMPs
Following the substitution of amino acids distinct from the hotspot residues, the mutated AMPs were again subjected to physicochemical analysis using APD3 to predict molecular masses, hydrophobicity, isoelectric points, and the Boman index, and BACTIBASE to predict the common amino acids, net charges, and half-lives. This analysis was carried out to ensure that the mutated AMPs retained the same characteristics as the parental ones. From the results observed in Table 3, all the derivative antibacterial pneumonia AMPs had improved performance in terms of the Boman index and hydrophobicity compared to the parental AMPs. However, BOPAM-KP1-3 and BOPAM-SP7 had slightly reduced hydrophobicity, with a higher Boman index. A similar result was generated for the site-directed mutagenesis of anti-HIV AMP8 to yield AMP8.1 used for HIV diagnosis with reduced hydrophobicity, but an improved binding potential to the HIV p24 protein [17]. An amino acid was substituted with one carrying a higher positive charge, and the overall charge of the AMP and the Boman index were increased. Therefore, it is expected that the derivative AMP would have a greater binding affinity to its receptor than its parental counterpart due to the specific amino acid substitution. However, the results indicate that BOPAM-SP3.1, 4.1, 5.1, and 6.1 had negative charges following SDM, while BOPAM-KP1.1 was neutral. With the discovery of negatively charged AMPs such as maximin H5, which contradicts the belief that AMPs with high antimicrobial activities are cationic, there is an indication that the charge might not affect an AMP's activity [8].    Table 3 below.  Table 4 below shows the quality output from the I-TASSER server after predicting the derivative anti-pneumonia AMPs using the C-score, TM-score, and RMSD as measures. The C-scores of all derivative AMPs were within the range of −5 to 2, indicating accurate predictions from the templates used with the correct topology. The C-scores of the derivative AMPs did not deviate from those of the parental AMPs from which they were derived. The TM-scores were greater than 0.17, showing that the 3D structures of the derivative AMPs were not random predictions. The RMSD values were above 1 Å, further proving that the AMPs were correctly predicted.   Table 4 below shows the quality output from the I-TASSER server after predicting the derivative anti-pneumonia AMPs using the C-score, TM-score, and RMSD as measures. The C-scores of all derivative AMPs were within the range of −5 to 2, indicating accurate predictions from the templates used with the correct topology. The C-scores of the derivative AMPs did not deviate from those of the parental AMPs from which they were derived. The TM-scores were greater than 0.17, showing that the 3D structures of the derivative AMPs were not random predictions. The RMSD values were above 1Å, further proving that the AMPs were correctly predicted.

Docking Interaction Analysis of the Derivative AMPs with Bacterial and Viral Receptors
Using PATCHDOCK Table 5 below shows the binding scores of the antibacterial and antiviral pneumonia AMPs from the PATCHDOCK server with their respective receptors identified in the previous study. All binding scores of the derivative AMPs were compared with the parental AMPs for selection as candidate anti-pneumonia AMPs to be used in a lateral flow device (LFD). BOPAM-AB1.1 and BOPAM-AB5.1 returned higher binding scores than the parental AMPs against the outer membrane receptor protein of Acinetobacter baumannii. Moreover, BOPAM-SP1.1, 3.1, 4.1, and 7.1 yielded improved binding scores against the pneumolysin of Streptococcus pneumoniae, while a similar result was observed for BOPAM-KP1.1, 3.1, 5.1, and 6.1 against the iron-regulated outer membrane protein of Klebsiella pneumoniae.    All the derivative AMPs above (Table 4) had increased binding affinity to their receptors compared to their respective parental AMPs (Figure 4). In other words, they bound more tightly when compared to the separate parental AMPs at the same positions. The mutation due to amino acid substitution in these derivative AMPs did not change their 3D structures compared to those of the parental AMPs and displayed a better-predicted fit at the binding pockets of bacterial and viral pathogens' receptors compared to the parental AMPs.
There was no change in the binding region or orientation of the derivative AMPs with their receptors after substituting the amino acid residues of the parental AMPs, as indicated in the previous studies. These peptides' diagnostic tendency could be attributed to their small size, allowing them access to membrane receptors' biological surfaces. The receptors used in this research were carefully identified with residues seated at the membrane's extracellular surface for accessibility during binding. All the derivative AMPs above (Table 4) had increased binding affinity to their receptors compared to their respective parental AMPs (Figure 4). In other words, they bound more tightly when compared to the separate parental AMPs at the same positions. The mutation due to amino acid substitution in these derivative AMPs did not change their 3D structures compared to those of the parental AMPs and displayed a better-predicted fit at the binding pockets of bacterial and viral pathogens' receptors compared to the parental AMPs.   There was no change in the binding region or orientation of the derivative AMPs with their receptors after substituting the amino acid residues of the parental AMPs, as indicated in the previous studies. These peptides' diagnostic tendency could be attributed to their small size, allowing them access to membrane receptors' biological surfaces. The receptors used in this research were carefully identified with residues seated at the membrane's extracellular surface for accessibility during binding. The

Discussion
Various shortcomings have been suffered from using different diagnostic systems of bacteria and viruses. Antibodies, which are currently the gold standard, also suffer from limitations such as cross-reactivity and sequence coverage. Other shortcomings include

Discussion
Various shortcomings have been suffered from using different diagnostic systems of bacteria and viruses. Antibodies, which are currently the gold standard, also suffer from limitations such as cross-reactivity and sequence coverage. Other shortcomings include overlapping epitopes between the detected and natural host antibodies, resulting in false negative results, complex sample preparation, time consumption, and non-specific binding to non-target molecules [24]. Antimicrobial peptides have shown tremendous potential in circumventing the drawbacks of these systems and appear to be the favorite choice in diagnostic development due to their numerous properties that overcome these shortcomings, such as their small size, ease of modification, high stability, and minor/nontoxicity. The HMMER machine learning tool was employed to construct predictive models to identify sensitive and specific AMPs against Streptococcus pneumonia, Klebsiella pneumonia, Acinetobacter baumannii, respiratory syncytial virus, and the influenza A and B viruses in our previous studies [15,16]. This research employed the predicted AMPs as parental peptides to develop derivative AMPs using site-directed mutagenesis.
The Knowledge-based FADE and Contacts (KFC) Server was used to identify the amino acid residue "hotspots" that contributed to the receptor proteins' interface binding free energy to differentiate the mutation-sensitive residues. This means that these amino acid residues contribute the most to the free energy involved in the binding to their respective receptors. The use of the KFC Server offers two predictive models for accuracy dependent on the shape specificity features or the biochemical contacts [18]. After that, SDM analysis was performed to generate derivative AMPs to identify the ones that would lead to a higher predicted binding energy to the respective receptors of the pathogens [25].
The amino acid residues were substituted using SDM to maintain the predicted activity and functionality of the AMPs. This ensures that the enormous potential of the AMPs is realized to achieve the desired goal of sensitive identification of the pneumonia pathogens.
Physicochemical analysis of the derivative AMPs was conducted to ascertain their conformity to the known AMPs using criteria such as the charge, polarity, hydrophobicity, and Boman index. The reduced hydrophobic values of BOPAM-KP4.1, BOPAM-SP1.1, 2.1, and 7.1, BOPAM-INFA1.1, 3.1, 4.1, 5.1, 6.1, and 8.1, and BOPAM-RSV2.1, 7.1, 9.1, 10.1, and 13.1 are related to a reduced peptide helicity, poor self-associating ability in an aqueous phase, and reduced antimicrobial activity. Peptides with increased hydrophobicity can penetrate deeper into organisms, causing severe hemolysis by forming channels [26]. Thus, mutated antibacterial and antiviral AMPs with increased hydrophobicity could potentially penetrate the membrane core. More recently, AMPs from sugar-functionalized phosphonium polymers were reported to require the hydrophilic domains for their molecular structure to exert antibacterial activities against Gram-negative Escherichia coli and Grampositive Staphylococcus aureus [27]. There is a positive correlation between cationic AMPs and antimicrobial properties (Table 3), which shows conformity with ideal AMPs [28]. Nonetheless, the net charge of mutated AMPs with negative charges does not interpret an absence of antimicrobial activities since some negatively charged AMPs have recently been discovered: for example, surfactant-associated anionic peptides in the APD3 database (AP00528) with a net charge of −5 which have antibacterial activity, and maximin H5 with a charge ranging between −1 and −7 which displays bacterial growth inhibition against Listeria monocytogenes [29]. The isoelectric point is a function of individual amino acid residues in the peptides, which relates to their solubility properties in acid or alkaline solutions regardless of their charges. A negative Boman index directly correlates with hydrophobic peptides with a high protein binding potential. Notwithstanding, the positive Boman index in some peptides has been accounted for with the capacity to distinguish HIV in a lateral flow device [30]. All the mutated peptides exhibited a short half-life because they are not stable, and half-life values as low as one have been reported for AMP molecules used for HIV diagnosis [31].
The structure prediction results utilizing the C-scores from I-TASSER relate to the modeling of a template's certainty for predicting the structure quality, that is, the anticipated model's distance and its local structures [32]. TM and RMSD scores estimate structural model closeness and accuracy in two structures when the local structure is known. The outcomes of the peptides' structures demonstrated that they conformed to known AMPs. This outcome relates to the work of [15,16], where binding geometry scoring was utilized as the criteria for determining applicant AMPs for pneumonia diagnostics. These perceptions were additionally affirmed utilizing an in-house lateral flow device in which the putative AMPs were utilized to recognize HIV in patient samples [30].
The results from this research show that all the putative peptides displayed in Table 5 could be pursued for molecular validation as applicant specialists for the detection of bacterial and viral pneumonia pathogens. Furthermore, the docking interaction results display only the mutated peptides with increased binding potential and atomic contact energy to the respective receptors. As shown in this analysis, these parameters are significant for the determination of novel peptides for detecting diseases through the development of an LFD device [30]. This research is in line with Williams et al. [30], where parental anti-HIV antimicrobial peptides were mutated using SDM analysis to increase their binding potential against the HIV nucleoprotein, and subsequent molecular validation was carried out for the construction of an LFD device. Thus, the putative anti-pneumonia AMPs discovered in this research have better binding energy than the parental AMPs. This offers promising perspectives for sensitive diagnosis and differentiation of bacterial from viral pneumonia in patients' blood samples. It also allows health practitioners to forecast timely and correct treatment regimens for patients with this condition.

Conclusions
This research work used parental anti-pneumonia AMPs as templates to identify putative derivative AMPs that can function in the potential differential diagnosis of pneumonia. The derivative AMPs have an improved binding potential to the pneumonia receptors, with greater affinity, demonstrated through in silico binding studies. Thus far, 10 antibacterial and 13 antiviral pneumonia AMPs with improved binding specificity and accuracy against pneumonia pathogens have been identified. These AMPs offer promising perspectives to compete with current commercially available pneumonia diagnostic biomarkers to mitigate their shortcomings.

Future Work
Future work will include the in vitro study of the anti-pneumonia activity of the mutated peptides. Moreover, the EC 50 of all the AMPs and their diagnostic or selective index will be assessed for optimization. The anti-pneumonia activity of these derivative AMPs will be assessed on different pseudotypes of the pneumonia pathogens to determine their broad-spectrum activity. Finally, the binding complex formed between the pathogen receptors and derivative AMPs will be solved using structural biology to validate the in silico binding study's observations. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/bioengineering9070305/s1, the parental antimicrobial peptides along with the databases' accession numbers.