Shotgun metagenomic sequencing in culture negative microbial keratitis

European Journal of Ophthalmology Page 4 of 17 3 https://mc.manuscriptcentral.com/ej-ophthalmology 1Title Shotgun metagenomic sequencing in culture negative microbial keratitis Authors Borroni, Davidea; Bonzano, Chiarab; Sánchez-González, José-Maríac; Rachwani-Anil, Rahuld; Zamorano-Martín, Franciscoe; Pereza-Nieves, Jorgef; Traverso, Carlo Enricob; García Lorente, Maríaf; Rodríguez-Calvo-de-Mora, Marinaf; Esposito, Alfonsog; Godin, Fernandoh; Rocha-de-Lossada, Carlosi, j, k, l a Department of Ophthalmology, Riga Stradins University, Riga, Latvia b DiNOGMI, University of Genoa and IRCCS San Martino Polyclinic Hospital, Genoa, Italy c Department of Physics of Condensed Matter, University of Seville, Seville, Spain d Hospital of Antequera, Malaga, Spain e Hospital Clinic of Barcelona, Barcelona, Spain f Hospital Regional Universitario Málaga, Málaga, Spain g International Centre for Genetic Engineering and Biotechnology, Trieste, Italy h Department of Ophthalmology, Universidad El Bosque, Bogotá, Colombia i Opththalmology Department, VITHAS Almería Hospital, Almería, Qvision, Spain j Ophthalmology Department, VITHAS Málaga, Málaga, Spain k Hospital Regional Universitario de Málaga, Plaza del Hospital Civil, Málaga, Spain l Departamento de Cirugía, Universidad de Sevilla, Área de Oftalmología, Doctor Fedriani, Seville, Spain Page 5 of 17 European Journal of Ophthalmology 4 https://mc.manuscriptcentral.com/ej-ophthalmology Abstract Purpose: To evaluate the microbiota of culture negative Corneal Impression Membrane (CIM) microbial keratitis samples with the use of shotgun metagenomics analysis. Methods: DNA of microbial keratitis samples were collected with CIM and extracted using the MasterPure™ Complete DNA and RNA Purification Kit (Epicentre). DNA was fragmented by sonication into fragments of 300 to 400 base pairs (bp) using Bioruptor® (Diagenode, Belgium) and then used as a template for library preparation. DNA libraries were sequenced on Illumina® HiSeq2500. The resulting reads were quality controlled, trimmed and mapped against the human reference genome. The unmapped reads were taxonomically classified using the Kraken software. Results: 18 microbial keratitis samples were included in the study. Brevundimonas diminuta was found in 5 samples while 6 samples showed the presence of viral infections. Cutibacterium acnes, Staphylococcus aureus, Moraxella lacunata and Pseudomonas alcaligenes were also identified as the presumed putative cause of the infection in 7 samples. Conclusions: Shotgun sequencing can be used as a diagnostic tool in microbial keratitis samples. This diagnostic method expands the available tests to diagnose eye infections and could be clinically significant in culture negative samples. Author keywords brevundimonas diminuta; culture negative; keratitis; Metagenomics European Journal of Ophthalmology Page 6 of 17 3 https://mc.manuscriptcentral.com/ej-ophthalmology 2 3 53 Introduction: 4 5 54 Microbial keratitis may cause severe visual impairment and vision loss (1). It remains a major 6 7 55 cause of blindness in undeveloped countries and, due to the high prevalence of contact lens 8 9 56 wearers, there is an increasing incidence in developed countries as well (2; 3). Moreover, new 10 11 57 emerging pathogens are difficult to diagnose and may cause serious eye diseases with an 12 13 58 unusual course and poor response to standard treatment (4). However, the diagnostic and 14 15 59 therapeutic approaches of microbial keratitis management have remained unchanged for 16 17 60 over half a century (5). Despite the magnitude of this disease, the traditional methods based 18 19 61 on microbiological cultures fail in detecting a pathogen in over the 40% of the samples (6). 20 21 62 This limitation is probably due to the fact that less than 1% of all microorganisms of the 22 23 63 available environmental microbiota may be cultured with the conventional diagnostic 24 25 64 methods (CDM) available (7). Metagenomics is a new high-yield DNA sequencing method 26 27 65 that is able to provide taxonomic and functional profiles of microbial communities without 28 29 66 the need to culture microbes in the laboratory. Its main aim is to resolve the current 30 31 67 limitations of the CDM (8; 9). Next Generation Sequencing (NGS) offers multiple advantages 32 33 68 over CDM in sample analysis (10). The advantages provided by this approach are threefold: 34 35 69 i) the sampling, as only<2 μL volume of the sample may be enough to obtain a full 36 37 70 sequencing result (11); ii) the speed, as Wilson et al., 2015 reported that NGS protocols could 38 39 71 be completed in less than 48 hours, being more time-efficient compared to many of the 40 41 72 culture-dependent assays (12); and iii) the hypothesis-free approach, as it it overrules a long 42 43 73 empiric treatment which is common in CDM (13). Recent advances in metagenomics have 44 45 74 proved the practical applications and the potential of NGS(14; 15). Moreover, while marker- 46 47 75 based metagenomics is employed to determine the taxonomic profiles of bacteria or fungi, 48 49 76 shotgun metagenomics also allows the simultaneous study of viruses, bacteriophages, 50 51 77 archaea and eukaryotes (16). In addition, shotgun metagenome increases the diagnostic 52 53 78 yield as it amplifies the targeted sample from very low volumes of samples (17). Hence, 54 55 79 shotgun metagenomics analysis could be useful in samples with negative CDM where only 56 57 80 an empiric approach is usually carried out in clinical practice (18). 58 59 81 The aim of this study is to identify the most abundant microbial species in CDM negative 60 Page 7 of 17 European Journal of Ophthalmology 4 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 82 samples using NGS shotgun metagenomic analyses, aiming to detect the probable cause of 4 5 83 the disease. 6 7 84 8 9 85 Material and Methods: 10 11 86 Participants over 18 years of age were recruited for the study between January-July 2019 at 12 13 87 Clinica Oculistica, University of Genoa, Italy. All patients were informed about the details and 14 15 88 risks of the procedure and signed the consent form for data treatment. All patients received 16 17 89 a patient information sheet. This study was performed in agreement with the Helsinki 18 19 90 Declaration tenets. A 9 mm. Corneal Impression Membrane (CIM) (Millipore cell culture 20 21 91 inserts, pore size 0.04 um nr. HAWP04700 ) was used to collect the sample. The CIM was 22 23 92 immediately placed in a DNA/RNA-Shield (Zymo Re-search, Irvine, CA) to preserve the 24 25 93 integrity of the nucleic acids in the sample and was later transferred to a -80° C freezer for 26 27 94 storage. 28 29 95 30 31 96 DNA extraction and library preparation 32 33 97 A pre-study was carried out to determine which DNA extraction kit had the best 34 35 98 performance. CIM were inoculated with 105, 103 and 101 stock concentrations of P. 36 37 99 aeruginosa PAO1 and S. aureus Newmans by pipetting 55 µL of the stocks directly onto the 38 39 100 CIM disks. The CIMs were placed into 7ml bijou containers and air dried. To assess the effect 40 41 101 of the CIM on DNA extraction, 5 µL of each stock concentration was also pipetted directly 42 43 102 into empty 7ml bijou containers. All experiments were performed three times. DNA 44 45 103 extraction was performed using two extraction kits, a Masterpure™ Complete DNA and RNA 46 47 104 purification kit and a Qiagen UCP Pathogen Mini Kit. For the Masterpure™ extraction, 300 48 49 105 µL of tissue and cell lysis solution containing 1 µL of proteinase K was added to each 7 ml 50 51 106 bijou container and the DNA extraction was carried out according to the manufacturer’s 52 53 107 instructions in the bijou container until the precipitation of the total nucleic acid stage, when 54 55 108 the supernatant was placed into a 1.5 mL Eppendorf tube. For the Qiagen UCP Pathogen 56 57 109 mini kit extraction, the 500 µL of Buffer ATL was added directly into the 7 ml bijou container. 58 59 110 Subsequently, the DNA extraction was carried out according to the manufacturer’s 60 111 instructions. The DNA yield was very low for the Qiagen UCP Pathogen kit hence an exact European Journal of Ophthalmology Page 6 of 17 5 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 112 quantification of DNA concentration could not be done. Therefore, the use of this kit was 4 5 113 dismissed and the Masterpure™ Complete DNA and RNA purification kit was selected for 6 7 114 the study. 8 9 115 DNA was then extracted using the MasterPure™ Complete DNA and RNA Purification Kit 10 11 116 (Epicentre) according to the manufacturer’s instructions. DNA was fragmented by sonication 12 13 117 into 300 to 400 base pairs (bp) using Bioruptor (Diagenode, Belgium) and then subjected to 14 15 118 sequencing library preparation using the KAPA LTP Library Preparation Kit (Kapa Biosystems, 16 17 119 Wilmington, MA, USA), following the manufacture’s protocol. 18 19 120 20 21 121 Sequencing and Data analysis 22 23 122 DNA libraries were sequenced on the Illumina® HiSeq2500 sequencer using HiSeq PE 24 25 123 Cluster Kit v4 and HiSeq SBS V4 250 cycle kit (Illumina, San Diego, CA, USA) and the initial 26 27 124 processing was performed by CASAVA software (v1.8.2, Illumina). Conversion to FastQ files 28 29 125 and demultiplexing of the reads, according to their respective index, was performed using 30 31 126 the bcl2fastq software tool. The resulting reads were quality controlled, trimmed and 32 33 127 mapped against the human reference genome. The unmapped reads were taxonomically 34 35 128 classified using the Kraken2 software (19). Metagenomes were assembled from unmapped 36 37 129 reads using CLC genomics workbench and its microbial genomics module. Functional 38 39 130 profiles were generated for these metagenomes separately for best Blast hit annotation, 40 41 131 protein families (Pfam) domain annotation and GO term annotation, respectively. 42 43 132 44 45 133 Statistical analysis 46 47 134 Statistical analysis was performed with SPSS statistical software (version 26.0, IBM Corp, 48 49 135 Armonk, New York, USA). Descriptive analysis was carried out with the DNA extracted from 50 51 136 the samples. Number and percentage of pathogen reads were also calculated. For all tests, 52 53 137 level of significance was established at 95% (p-value < 0.05). 54 55 138 56 57 139 Results: 58 59 140 Eighteen microbial keratitis samples were collected for the study. Table 1 resumes the DNA 60 141 sample yield, measured using the QBIT fluorometric assay, and DNA purity, calculated as the Page 7 of 17 European Journal of Ophthalmology 6 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 142 260/280 ratio by nanodrop. CDM were all negative. No DNA was quantified from the 4 5 143 extracted negative controls which included 3 empty bijou containers and 3 CIMs, each 6 7 144 placed into an empty bijou in a sterile manner to mimic the sampling process in clinical 8 9 145 practice. The most abundant pathogens and number of reads for each sample are reported 10 11 146 in Table 2. 12 13 147 Six out of 18 samples showed the presence of a viral DNA. The patients were therefore 14 15 148 treated with standard 400 mg Acyclovir, 5 times a day, and the infections resolved within 3 16 17 149 to 5 weeks. 18 19 150 20 21 151 Discussion: 22 23 152 Diagnosis of microbial keratitis has always been a challenging task (20; 21) as it traditionally 24 25 153 relies on CDM for pathogen identification with the well-known limitations reviewed by Ung 26 27 154 et al., 2019 (6). In a 12-year retrospective study, Ting et al. reported that only 37.7% of corneal 28 29 155 scrapes samples were culture-positive in infectious keratitis (22). Similar outcomes have been 30 31 156 reported regarding staining procedures. Consequently, empiric treatment has been the 32 33 157 standard procedure with debatable results (23; 24). 34 35 158 Polymerase Chain Reaction (PCR) is a newer, faster, and more sensitive approach than 36 37 159 traditional stains and cultures (25). However, PCR demonstrates lower specificity rates 38 39 160 compared to the later. A combination of PCR and CDM has shown to increase the overall 40 41 161 correct diagnostic rate (26). However, the lack of valuable results given by CDM or staining 42 43 162 and the current limitations of PCR, such as relying on the knowledge of specific primer pairs 44 45 163 and targeting single microorganisms that could be only a partial answer to a complex, 46 47 164 polymicrobial disease, have guided the ophthalmology community to find novel and 48 49 165 innovative approaches (6). 50 51 166 Metagenomics, a high-yield DNA sequencing methodology, may provide taxonomic and 52 53 167 functional profiles of microbial communities (27). For instance, this type of analysis reports 54 55 168 species living in any given environment at high resolution (i.e. at a subspecies, strain or even 56 57 169 clone level), and its potential physiological features, such as its ability to metabolize 58 59 170 compounds, to express virulence factors or to establish symbiotic relationships (28). 60 171 Metagenomics has originally been associated to environmental studies. However, together European Journal of Ophthalmology Page 8 of 17 7 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 172 with the Human Microbiome Project, it is now been applied in clinical settings (29-32) and 4 5 173 many researchers are applying NGS technologies to obtain valid results in microbial keratitis 6 7 174 (33-35). In eye infections, shotgun metagenomic analysis could be highly advantageous as 8 9 175 it targets the whole microbiome while requiring minimal sample amount for successful 10 11 176 sequencing (36). 12 13 177 The innovative shotgun approach reported the potential to take over as a valid alternative 14 15 178 to CDM(32). For instance, a detailed taxonomy and the genomic determinants of antibiotic 16 17 179 resistance in bacterial infections stand as a requirement for targeted treatments (33). In the 18 19 180 current study, the application of shotgun sequencing showed the ability to give a clinically 20 21 181 useful result after CDM testing failed. 22 23 182 After DNA extraction (Table 1), it is possible to evaluate the results. Due to the difficulties in 24 25 183 data interpretation, we have simplified the results reporting the most abundant pathogen 26 27 184 (Table 2). Six out of 18 samples showed the presence of a viral infections, and the treatment 28 29 185 with acyclovir showed to be effective hence proving that the diagnosis was correct. As it has 30 31 186 previously been reported, coinfection is among the major misleading causes of medical 32 33 187 treatments. NGS metagenomics showed its ability to overcome this current limitation (37; 34 35 188 38). 36 37 189 Interestingly, Brevundimonas diminuta was found in 5 samples (Example: Sample nr.2 – 38 39 190 Figure 1). This result is unusually high considering the current literature (39; 40). We 40 41 191 hypothesise that Brevundimonas diminuta is probably one of the most difficult pathogens 42 43 192 to grow in the eye and it is suggested that it could be an emerging global opportunistic 44 45 193 pathogen (41). Brevundimonas diminuta has reported its ability to surpass sterilizing filters, 46 47 194 which can allow them to cause potentially harmful infections or even death. It is reported 48 49 195 that it should not be overlooked as a possible cause of nosocomial infections. Therefore, 50 51 196 NGS may provide clinically significant data in infections caused by unculturable organisms 52 53 197 before they become devastating (40). 54 55 198 Considering the possibility of coinfections, the need of simplifying the results making it more 56 57 199 understandable for clinicians is an important task that researchers need to address. 58 59 200 Therefore, we have considered the most abundant pathogen to be the putative cause of the 60 201 infection (Example: sample nr. 6 - HSV Simplex virus 2 - Figure 2). This is debatable and a Page 9 of 17 European Journal of Ophthalmology 8 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 202 possible limitation of the study, as less represented pathogens could play an active role of a 4 5 203 specific infection. Thus, the development of bioinformatic software, tailored to the 6 7 204 ophthalmologic field, is highly needed. The eye microbiome is among the least studied. The 8 9 205 knowledge on this peculiar niche still lags compared to the better studied gut microbiome 10 11 206 or other human compartments, such as the skin or the urogenital tract. One of the challenges 12 13 207 is that the eye microbiome presents a low microbial biomass, and therefore it is hard to 14 15 208 determine what a healthy microbiome is. On the other hand, a software that may study the 16 17 209 eye microbiome could surely improve data interpretation, especially the relationship 18 19 210 between the number or reads and the key-players of infection increasing diagnostic 20 21 211 accuracy. Other than the low number of samples and the presence of a coinfection, another 22 23 212 limitation of the study was the absence of a negative control. This is one of the major issues 24 25 213 nowadays in metagenomic applications. We are trying to solve the current issue creating a 26 27 214 normative database with healthy patients (42). 28 29 215 30 31 216 5. Conclusions 32 33 217 Shotgun sequencing is able to interrogate all microbial genomes in the eye microbiome in 34 35 218 keratitis samples. The possibility of a clinically useful result in culture negative samples is 36 37 219 expanding the available tests to diagnose eye infections. Similarly, the simplification of data 38 39 220 interpretation for clinical purpose will expand the use of metagenomics in ophthalmology. 40 41 221 Further investigations with NGS technologies for diagnostic purposes are required to 42 43 222 consolidate the methods and implement its use, preferably according to validated and 44 45 223 standardized protocols. 46 47 224 48 49 225 50 51 226 52 53 227 54 55 228 56 57 229 58 59 230 60 231 European Journal of Ophthalmology Page 10 of 17 9 https://mc.manuscriptcentral.com/ej-ophthalmology 1 2 3 232 References: 4 5 233 6 7 234 1. Urwin L, Okurowska K, Crowther G, Roy S, Garg P, et al. 2020. Corneal Infection 8 9 235 Models: Tools to Investigate the Role of Biofilms in Bacterial Keratitis. Cells 9 10 11 236 2. Ghosh A, Kaur H, Gupta A, Singh S, Rudramurthy SM, et al. 2020. Emerging 12 13 237 Dematiaceous and Hyaline Fungi Causing Keratitis in a Tertiary Care Centre From 14 15 238 North India. Cornea 39:868-76 16 17 239 3. Liesegang TJ. 1997. Contact lens-related microbial keratitis: Part I: Epidemiology. 18 19 240 Cornea 16:125-31 20 21 241 4. Sahay P, Goel S, Nagpal R, Maharana PK, Sinha R, et al. 2020. Infectious Keratitis 22 23 242 Caused by Rare and Emerging Micro-Organisms. Curr Eye Res 45:761-73 24 25 243 5. Ung L, Chodosh J. 2021. Urgent unmet needs in the care of bacterial keratitis: An 26 27 244 evidence-based synthesis. Ocul Surf 28 29 245 6. Ung L, Bispo PJM, Shanbhag SS, Gilmore MS, Chodosh J. 2019. The persistent dilemma 30 31 246 of microbial keratitis: Global burden, diagnosis, and antimicrobial resistance. Surv 32 33 247 Ophthalmol 64:255-71 34 35 248 7. Low L, Fuentes-Utrilla P, Hodson J, O'Neil JD, Rossiter AE, et al. 2021. Evaluation of 36 37 249 full-length nanopore 16S sequencing for detection of pathogens in microbial 38 39 250 keratitis. PeerJ 9:e10778 40 41 251 8. Borroni D, Rocha de Lossada C. 2020. Microbial keratitis: the clinical impact of 42 43 252 metagenomic next-generation sequencing (mNGS). Arch Soc Esp Oftalmol (Engl Ed) 44 45 253 95:621-3 46 47 254 9. Borroni D, Romano V, Kaye SB, Somerville T, Napoli L, et al. 2019. Metagenomics in 48 49 255 ophthalmology: current findings and future prospectives. BMJ Open Ophthalmol 50 51 256 4:e000248 52 53 257 10. Su Mun L, Wye Lum S, Kong Yuiin Sze G, Hock Yoong C, Ching Yung K, et al. 2021. 54 55 258 Association of Microbiome with Oral Squamous Cell Carcinoma: A Systematic Review 56 57 259 of the Metagenomic Studies. Int J Environ Res Public Health 18 58 59 260 11. Gallon P, Parekh M, Ferrari S, Fasolo A, Ponzin D, Borroni D. 2019. Metagenomics in 60 261 ophthalmology: Hypothesis or real prospective? Biotechnol Rep (Amst) 23:e00355 Page 17 of 17 European Journal of Ophthalmology 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/ejophthalmology Figure 1: Brevundimonas diminuita keratitis 70x57mm (300 x 300 DPI) European Journal of Ophthalmology Page 18 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 https://mc.manuscriptcentral.com/ejophthalmology Figure 2: HSV Simplex virus 2 keratitis 69x58mm (300 x 300 DPI)