Preliminary exploration on the serum biomarkers of bloodstream infection with carbapenem‐resistant Klebsiella pneumoniae based on mass spectrometry

Abstract Background Carbapenem‐resistant K. pneumoniae (CRKP) bloodstream infections (BSI) must be rapidly identified to improve patient survival rates. This study investigated a new mass spectrometry‐based method for improving the identification of CRKP BSI and explored potential biomarkers that could differentiate CRKP BSI from sensitive. Methods Mouse models of BSI were first established. MALDI‐TOF MS was then used to profile serum peptides in CRKP BSI versus normal samples before applying BioExplorer software to establish a diagnostic model to distinguish CRKP from normal. The diagnostic value of the model was then tested against 32 clinical CRKP BSI and 27 healthy serum samples. Finally, the identities of the polypeptides used to establish the diagnostic model were determined by secondary mass spectrometry. Results 107 peptide peaks were shared between the CRKP and normal groups, with 18 peaks found to be differentially expressed. Five highly expressed peptides in the CRKP group (m/z 1349.8, 2091.3, 2908.2, 4102.1, and 8129.5) were chosen to establish a diagnostic model. The accuracy, specificity and sensitivity of the model were determined as 79.66%, 81.48%, and 78.12%, respectively. Secondary mass spectrometry identified the Fibrinogen alpha chain (FGA), Inter‐alpha‐trypsin inhibitor heavy chain H4 (ITIH4) and Serum amyloid A‐2 protein (SAA2) as the source of the 5 serum peptides. Conclusions We successfully established a serum peptide‐based diagnostic model that distinguished clinical CRKP BSI samples from normal healthy controls. The application of MALDI‐TOF MS to measure serum peptides, therefore, represents a promising approach for early BSI diagnosis of BSI, especially for multidrug‐resistant bacteria where identification is urgent.


| INTRODUC TI ON
Among Gram-negative bacteria, K. pneumoniae represents one of the most prevalent micro-organisms found in health care and ICU acquired infections, 1 as well as one of the most frequent pathogens in clinical bloodstream infections (BSIs). 2 Furthermore, with the widespread application of antibiotics over the last 20 years, carbapenem-resistant K. pneumoniae (CRKP) strains have sharply increased along with the continuous emergence of new drug resistance mechanisms. [3][4][5][6][7] Carbapenem is a generic β-lactamase characterized by resistance to almost all β-lactam antibiotics, including cephalosporins and carbapenems. The most common carbapenems of Enterobacteriaceae are class A carbapenases (KPC), class B metalβ-lactamases (IMP, VIM, NDM), and class D oxacillinases (OXA-48). [7][8][9][10][11][12] Notably, outbreaks of CRKP have been reported around the world, including Southern Europe, China, South America, some parts of North America and many other countries and regions. 13 The increased incidence of BSIs is associated with inappropriate empirical therapy and the increasing use of carbapenems. Indeed, CRKP has become an independent risk factor for nosocomial death due to difficulty in infection control and high mortality. 13,14 Thus, CRKP provides serious problems for the diagnosis and treatment of BSIs, and constitutes a major threat to public health. [15][16][17][18] Rapid and cost-effective microbiological diagnostic methods are badly needed to reduce treatment failure and optimize treatments for BSI patients with CRKP. 18 Consequently, research efforts in this area have intensified to improve the early diagnosis of CRKP BSI and also to provide effective treatments. It is well known that lipopolysaccharide (LPS) from gram-negative bacteria acts to stimulate host cells to produce a number of pro-inflammatory mediators, including tumor necrosis factor (TNF), interleukin-1 alpha (IL-1α), and interleukin-1(IL-6). [19][20][21][22] However, whether CRKP BSI induce the production of specific proteins or invoke specific signaling pathways is currently unclear. It is reasonable to speculate that CRKP BSI produce peptide fragments or proteins related to CRKP virulence and drug resistance genes. Even though such peptides belong to some common and non-specific proteins, we hypothesized that the combination of these peptides could be used for the early diagnosis of CRKP BSIs.
Matrix assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) is often applied to identify new disease biomarkers. This approach has proved successful for identifying many new diagnostic markers, for example, in cancers of the breast, lung, and liver along with autoimmune diseases and other diseases. [23][24][25] However, to our best knowledge, it has not been used previously to identify specific biomarkers associated with drug-resistant bacterial infections. Considering the complex backgrounds of clinical BSI patients, we first employed a mouse BSI model to avoid confounding factors that could affect the biomarker discovery process.
We then used MALDI-TOF MS analysis to identify serum peptides identified from the BSI mouse model combined with analysis of CRKP BSI patients, with results providing a new direction for the rapid diagnosis and biomarker discovery of drug-resistant bacteria.

Determination of infection concentration and time of infection
Clinically isolated K. pneumoniae strains were purified by plate separation and cultured in Luria-Bertani (LB) liquid medium for 12 h.
Thereafter, 100 μl culture was diluted in 10 ml LB liquid medium for a further 4-6 h. The LD 50 concentration of K. pneumoniae was obtained according to Ma's study, 26 and mice were infected with 1/2 LD 50 concentration of bacteria to establish the BSI model. Serum samples were collected 12 h after infection when the symptoms of infection were most evident.

Group infection and serum/tissue collection
A total of 70 mice were randomly divided into treatment groups designated CRKP BSI (30 mice), carbapenem-sensitive K. pneumoniae (CSKP) BSI (30 mice) and a normal control group (10 mice).
Infection groups were injected with 0.1 ml/10 g body weight of the corresponding bacterial culture (CRKP or CSKP), while the normal group were injected with sterile phosphate-buffered saline (PBS).
Mice were intraperitoneally injected with chloral hydrate before drawing blood samples from the retro-orbital socket 12 h after infection. Blood samples were divided for routine determination (20 μl), pathogen identification (100 μl was absorbed by capillary and coated on Chinese blue medium), while the remaining sample was centrifuged at 5000 rpm for 20 min, and the serum stored at −80℃. Finally, the mice were sacrificed by spinal dislocation, and the visceral tissues, such as liver, lung, and kidney were dissected under aseptic conditions in order to observe the size, color, and presence of visible lesions. Some visceral tissues were fixed in 10% formaldehyde, and some were fresh frozen at −80℃ refrigerator for later analysis.

| Clinical serum samples
A total of 32 serum samples were collected from CRKP BSI patients and 27 serum samples from healthy people who underwent physical examination in the PLA Hospital between August 1, 2020 and February 28, 2021 (Table 1). Blood samples were collected at the same of blood cultures were submitted for examination, and CRKP BSI patients were screened after drug sensitivity results were obtained. Serum samples were collected by centrifugation at 3000 g for 15 min and stored at −80℃ for analysis. The study was approved by the Ethical Committees of PLA Hospital (no. S2018-207002). The study was conducted in accordance with the Declaration of Helsinki, which all participants provided informed consent to participate in.

| Extraction of serum polypeptides
Serum samples were thawed on ice for prior to extraction peptide extraction based on a previously described non-proteolytic procedure. 27 Briefly, different amounts of serum were combined with weak cation exchange (WCX) magnetic beads and WCX magnetic bead binding buffer into 0.2 ml eppendorf (EP) tubes before mixing and incubation for 0.5 h at RT. Afterward, a magnetic bead separator was used to collect the WCX magnetic beads before removing the supernatant and washing the beads a further 3 times with cleaning buffer, thereafter, eluting the peptides and transferring the samples to new EP tubes. The samples were immediately analyzed by MS or frozen at −20℃ for later mass spectrometry.

| MALDI-TOF MS analysis
One μL of each sample was transferred onto a Clin-TOF II target plate (Bioyong Tech) in three times and dried at room temperature before coating each spot with 1 μl of HCCA matrix solution (8 mg/ml in 0.1% trifluoroacetic acid/50% acetonitrile). The profile spectra were obtained from approximately 100 laser shots (m/z 1000 −10,000 Da) via MALDI-TOF-MS on a Clin-TOF II instrument (Bioyong Tech).
Five external standard peptides with molecular weights of 1533.8, 2465.7, 5733.5, 6181.48, and 8476.64 Da were involved in quality control measures (The average molecular weight deviation was controlled within 100 ppm). The mass spectrum was recalibrated after every 24 samples were tested. The obtained spectrums were analyzed by BioExplorer™3.0 (BioYong Tech) and the default parameters are used for normalization, baseline correction, and smoothing of the spectrum. In addition, the mass shift should not exceed 0.1% to ensure the align of the spectra.

| Peptide identification by LC-MS
The nano-liquid chromatography electrospray ionization-tandem mass spectrometry (nano-LC/ESI-MS/MS) was used to identify amino acid sequences of the candidate peptides. After serum peptides were extracted with MB-WCX beads, each peptide sample was desalted on a C18 column, dried in vacuum, and resuspended in 10

| Statistical analysis
Origin 2018 was used for figure processing and partial data analysis, with p < 0.05 as the difference is statistically significant.

| Manifestations of infected mice
All infected mice showed disheveled hair, listlessness, reduced activity, and weight loss 6-12h after inoculation. Blood culture analysis along with antibiotic sensitivity testing confirmed the successful introduction of CRKP or CSKP strains to provide a model of BSI.
Furthermore, evidence of widespread inflammatory cells infiltrating major organs (liver, kidney, and lung) was found using H&E staining ( Figure 1). Here, comparisons with the control (PBS) group showed BSI animals exhibited obvious alveolar ectasia, and no obvious inflammation was observed in the liver and kidney.

| Serum peptide profiling in the BSI mouse model
After denoising, baselining and smoothing of the original spectra by BioExplorer mass spectrometry software, 107 peptide peaks were found in the mass charge ratio range of 1000 to 10,000 Da in the normal and CRKP groups (Figure 2), of which 17 peptide peaks were found to be upregulated and 1 peptide peak (m/z 6630.8) was found to be downregulated in the CRKP group (Log2 fold change >1.5 and p < 0.05) ( Table 2). Comparisons between the CRKP and CSKP groups showed 152 peptide peaks in common, but no differences were detected (Supplementary   Table).

| Establishment and validation of bacterial bloodstream infection model
Using

| Peak identification
We obtained the identities of the five peptides in the diagnostic model using secondary mass spectrometry (  In addition to m/z 6630.8, other differential peptides were upregulated compared with the control group.

TA B L E 2
Upregulated and downregulated polypeptide peaks in CRKP BSI group compared with normal control group (x ± s) procalcitonin, IL-6 and C-reactive protein are not specific for BSI, and neither can these distinguish between drug-sensitive and drugresistant cases. Therefore, a rapid diagnostic method for BSIs is urgently needed to solve the clinical challenge provided by CRKP BSIs.
Differential peptides produced with the occurrence of immune response and bacterial resistance are expected to be used as indicators for follow-up monitoring and auxiliary diagnosis of BSI. 26 We speculated that CRKP BSI produces peptide fragments or proteins related to CRKP virulence and drug resistance genes, which could be potentially used for the early diagnosis of CRKP BSIs. Murine models of bacterial BSI were used in our study considering the complex background of clinical BSIs. Based on previous methods, 26,30 we directly purified serum peptides on WCX magnetic bead and analyzed the differences between CRKP BSI and normal groups using MALDI-TOF MS.
The experimental results show that differential serum peptide fingerprints were determined between the CRKP BSI and normal groups, of which 17 peptide peaks were significantly different. A machine learning algorithm approach was used to define a diag- Therefore, we speculated that these three proteins might serve as the potential molecular markers of CRKP BSIs due to their upregulation during CRKP bloodstream infection.
Peptide peak m/z 1349.8 was identified as a fragment of FGA.
FGA is a major component of mammalian extracellular matrix, which is mainly involved in the coagulation cascade and stabilizes the lesion in the early stage of wound repair. 31,32 Additionally, FGA deposition is also associated with infection. 33 Herwald et al. 34  Peptide peak m/z 2091.3 was identified as a fragment of ITIH4 which belongs to the inter-alpha-trypsin (ITI) family. ITIH4 consists of an acute phase response plasma glycoprotein that is primarily expressed by the liver. 36,37 ITIH4 is highly expressed in early liver development and plays an important role in liver formation. 38 Recently, serum ITIH4 fragments were reported to in connection with several malignancies. 39 Muk et al. 40 found that the clinical signs of sepsis and neuroinflammation in bacterial-infected piglets were associated with changes of ITIH4. In our previous study, we found ITIH4 to be significantly increased in BSI, and its sensitivity was higher than determined for other prospective BSI markers including complement C3, SAA2, and Kininogen1. 41 Its elevation in bacteremia is mainly due to regulation by IL-6 or lipopolysaccharide through the JAK/ STAT pathway. 42 Therefore, ITIH4 represents a strong candidate biomarker for the rapid diagnosis of BSIs.
Peptide peak m/z 2091.3 and 4102.3 were both identified as the fragments of SAA2, which encodes a member of the serum amyloid A family of apolipoproteins. SAA2 is a major acute phase protein and highly expressed in inflammation and tissue damage responses. 43 Derebe et al. 44  Above all, our study focused on the analysis of serum peptide fingerprints found in CRKP BSIs. Through MALDI-TOF MS analysis, common and characteristic peptide peaks were identified and used to construct a model to diagnose CRKP BSIs with high accuracy. Among the diagnostic peptides, FGA, ITIH4, and SAA2 were identified by LC-MS. Although the relationships between these biomarkers and drug-resistant bacterial infections are not clear, the current study provides a strong foundation for future mass spectrographic analyses to continue the search to identify specific biomarkers of CRKP BSIs. We will continue to explore the underlying scientific mechanisms associated between changes in these proteins and CRKP BSIs in future research. Moreover, the promising results obtained here also suggest the serum polypeptide mass spectrometry approach will be useful in diagnosis of bloodstream infections by other drug-resistant bacteria.

ACK N OWLED G M ENTS
I would like to express my sincere gratitude to the teachers and students of the PLA General Hospital and Xuzhou Medical University.
This study did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this article, further inquiries can be directed to the corresponding author. F I G U R E 5 STRING protein-protein interaction network for the three proteins identified from serum polypeptide fingerprinting