Multimodal fusion network for ICU patient outcome prediction.

Over the past decades, massive Electronic Health Records (EHRs) have been accumulated in Intensive Care Unit (ICU) and many other healthcare scenarios. The rich and comprehensive information recorded presents an exceptional opportunity for patient outcome predictions. Nevertheless, due to the diversity of data modalities, EHRs exhibit a heterogeneous characteristic, raising a difficulty to organically leverage information from various modalities. It is an urgent need to capture the underlying correlations among different modalities. In this paper, we propose a novel framework named Multimodal Fusion Network (MFNet) for ICU patient outcome prediction. First, we incorporate multiple modality-specific encoders to learn different modality representations. Notably, a graph guided encoder is designed to capture underlying global relationships among medical codes, and a text encoder with pre-fine-tuning strategy is adopted to extract appropriate text representations. Second, we propose to pairwise merge multimodal representations with a tailored hierarchical fusion mechanism. The experiments conducted on the eICU-CRD dataset validate that MFNet achieves superior performance on mortality prediction and Length of Stay (LoS) prediction compared with various representative and state-of-the-art baselines. Moreover, comprehensive ablation study demonstrates the effectiveness of each component of MFNet.

Mycobacterium tuberculosis PE_PGRS45 (Rv2615c) Promotes Recombinant Mycobacteria Intracellular Survival via Regulation of Innate Immunity, and Inhibition of Cell Apoptosis.

Tuberculosis (TB), a bacterial infectious disease caused by Mycobacterium tuberculosis (M. tuberculosis), is a significant global public health problem. Mycobacterium tuberculosis expresses a unique family of PE_PGRS proteins that have been implicated in pathogenesis. Despite numerous studies, the functions of most PE_PGRS proteins in the pathogenesis of mycobacterium infections remain unclear. PE_PGRS45 (Rv2615c) is only found in pathogenic mycobacteria. In this study, we successfully constructed a recombinant Mycobacterium smegmatis (M. smegmatis) strain which heterologously expresses the PE_PGRS45 protein. We found that overexpression of this cell wall-associated protein enhanced bacterial viability under stress in vitro and cell survival in macrophages. MS_PE_PGRS45 decreased the secretion of pro-inflammatory cytokines such as IL-1ß, IL-6, IL-12p40, and TNF-α. We also found that MS_PE_PGRS45 increased the expression of the anti-inflammatory cytokine IL-10 and altered macrophage-mediated immune responses. Furthermore, PE_PGRS45 enhanced the survival rate of M. smegmatis in macrophages by inhibiting cell apoptosis. Collectively, our findings show that PE_PGRS45 is a virulent factor actively involved in the interaction with the host macrophage.

[Research progress on the effect of Mycobacterium tuberculosis heat-resistant antigen (MTB-HAg) on γδ T cells].

Mycobacterium tuberculosis heat resistant antigen (MTB-HAg) is a polypeptide antigen released from Mycobacterium tuberculosis (MTB) H37Ra to the supernatant after being autoclaved at 121 DegreesCelsius for 20 minutes. γδ T cells are unconventional T cells widely distributed in non-lymphoid tissues. They secrete cytokines such as TNF-α and IFN-γ, and play an important role in the immune response against MTB infection. MTB-HAg can effectively stimulate the proliferation and activation of γδ T cells in vitro, enabling them to participate more efficiently in the anti-TB infection process. Therefore, MTB-HAg is a potential target for the design of novel TB vaccines or drugs. However, due to its complex composition, it is still unknown which component of MTB-HAg stimulates the anti-MTB infection function of γδ T cells.

Graph-guided deep hashing networks for similar patient retrieval.

With the rapid growth and widespread application of electronic health records (EHRs), similar patient retrieval has become an important task for downstream clinical decision support such as diagnostic reference, treatment planning, etc. However, the high dimensionality, large volume, and heterogeneity of EHRs pose challenges to the efficient and accurate retrieval of patients with similar medical conditions to the current case. Several previous studies have attempted to alleviate these issues by using hash coding techniques, improving retrieval efficiency but merely exploring underlying characteristics among instances to preserve retrieval accuracy. In this paper, drug categories of instances recorded in EHRs are regarded as the ground truth to determine the pairwise similarity, and we consider the abundant semantic information within such multi-labels and propose a novel framework named Graph-guided Deep Hashing Networks (GDHN). To capture correlation dependencies among the multi-labels, we first construct a label graph where each node represents a drug category, then a graph convolution network (GCN) is employed to derive the multi-label embedding of each instance. Thus, we can utilize the learned multi-label embeddings to guide the patient hashing process to obtain more informative and discriminative hash codes. Extensive experiments have been conducted on two datasets, including a real-world dataset concerning IgA nephropathy from Peking University First Hospital, and a publicly available dataset from MIMIC-III, compared with traditional hashing methods and state-of-the-art deep hashing methods using three evaluation metrics. The results demonstrate that GDHN outperforms the competitors at different hash code lengths, validating the superiority of our proposal.

Mycobacterium tuberculosis PE8 (Rv1040c) Promotes the Intracellular Survival of Recombinant Mycobacterium by Regulating Host Inflammatory Cytokines and Inhibiting Cell Late Apoptosis.

Tuberculosis is an important chronic and often fatal infectious disease mainly caused by the bacterium Mycobacterium tuberculosis (Mtb). Mtb is one of the most successful pathogens that harbors several potential virulence factors not found in nonpathogenic mycobacteria. As the Mtb cell envelope is closely associated with its virulence and resistance, it is very important to understand the cell envelope for better treatment of causative pathogen. There is increasing evidence that Pro-Glu (PE) and Pro-Pro-Glu (PPE) proteins are the major effectors of virulence and persistence encoded in the Mtb H37Rv genome. However, the function of PE8 has not been explored to date. In this study, we heterologously expressed PE8 in nonpathogenic, fast-growing M. smegmatis to investigate the interaction between PE8 and the host to determine its possible biological functions. We found that recombinant M. smegmatis cells expressing PE8 were less susceptible to sodium dodecyl sulfate-induced surface stress compared with those expressing the empty vector, suggesting that PE8 may be involved in stress responses. In addition, macrophages infected with PE8-expressing M. smegmatis produced obviously lower levels of the proinflammatory factor IL-1ß, IL-6, and TNF-α and higher levels of the inhibitory factor IL-10. We further found that PE8 promoted M. smegmatis survival within macrophages by inhibiting late apoptosis of macrophages. Collectively, selective targeting of the PE/PPE protein family offers an untapped opportunity to the development of more effective and safer drugs against Mtb infection.

[Establishment of a THP-1 macrophage model infected by recombinant Mycobacterium smegmatis expressing green fluorescent protein].

Objective To establish a THP-1 macrophage model infected by Mycobacterium smegmatis expressing green fluorescent protein (GFP), to quickly locate and visually detect Mycobacterium smegmatis, and to provide a tracer tool to identify the pathogenesis of tuberculosis and develop new tuberculosis vaccines. Methods The enhanced green fluorescent protein (EGFP) gene sequence was amplified by PCR using pEGFP-N1 plasmid as a template to obtain the coding gene of EGFP, and the amplified product was cloned into the vector pALACE to establish the recombinant plasmid pALACE-EGFP. Electroporation transformed the pALACE-EGFP into Mycobacterium smegmatis, and recombinant Mycobacterium smegmatis clones were screened by hygromycin resistance. After expanded culture, the smears were observed by fluorescence microscopy. The THP-1 macrophages were infected with recombinant Mycobacterium smegmatis, and the expression of EGFP was observed. Results The recombinant plasmid pALACE-EGFP was constructed appropriately. The recombinant Mycobacterium smegmatis was observed under fluorescence microscope. And it was confirmed that EGFP was expressed in recombinant Mycobacterium smegmatis, and THP-1 macrophages emitted green fluorescence after infection. Conclusion The successful establishment of recombinant Mycobacterium smegmatis expressing EGFP protein provides insights for investigating infection and pathogenesis of Mycobacterium tuberculosis.

[Prokaryotic expression and identification of PPE15 protein from Mycobacterium tuberculosis and preparation of rabbit polyclonal antibodies].

Objective To clone, express and purify PPE15 recombinant protein from Mycobacterium tuberculosis (H37Rv), as well as prepare and characterize its rabbit polyclonal antibody. Methods By The PPE15 gene was amplified from the genome of Mycobacterium tuberculosis H37Rv by PCR, and the His-tagged prokaryotic PPE15 prokaryotic expression plasmid pET28a-PPE15 was constructed by homologous recombination cloning technique, and transformed into E. coli BL21 (DE3). PPE15 expression was induced by isopropyl-ß-D-thiogalactopyranoside (IPTG). Recombinant PPE15 was identified by SDS-PAGE, and further purified by affinity chromatography with a Ni-NTA column. The renaturation purified PPE15 protein was used to immunize New-Zealand rabbit to prepare polyclonal antibodies. The antibody specificity was analyzed by Western blot analysis, and antibody titer was determined by indirect ELISA. Results Recombinant prokaryotic PPE15 protein was successfully expressed and purified with a molecular weight of 38 kDa. The purified PPE15 protein exhibited positive reaction with the serum of TB patients and the PPE15 protein, the titer of the polyclonal antibodies reaches more than 1:1 300 480. Conclusion The recombinant protein PPE15 was successfully expressed and purified, and high titer rabbit-derived polyclonal antibody was prepared which provided an experimental basis for further functional studies of PPE15 protein.

Codon Optimization, Soluble Expression and Purification of PE_PGRS45 Gene from Mycobacterium tuberculosis and Preparation of Its Polyclonal Antibody Protein.

Studies have demonstrated that PE_PGRS45 is constitutively expressed under various environmental conditions (such as nutrient depletion, hypoxia, and low pH) of the in vitro growth conditions examined, indicating that PE_PGRS45 protein is critical to the basic functions of Mycobacterium tuberculosis. However, there are few reports about the biochemical function and pathogenic mechanism of PE_PGRS45 protein. The fact that this M. tuberculosis gene is not easily expressed in E. coli may be mainly due to the high content of G+C and the use of unique codons. Fusion tags are indispensable tools used to improve the soluble expression of recombinant proteins and accelerate the characterization of protein structure and function. In the present study, His6, Trx, and His6-MBP were used as fusion tags, but only MBP-PE_PGRS45 was expressed solubly. The purification using His6-MBP tag-specific binding to the Ni column was easy to separate after the tag cleavage. We used the purified PE_PGRS45 to immunize New Zealand rabbits and obtained anti- PE_PGRS45 serum. We found that the titer of polyclonal antibodies against PE_PGR45 was higher than 1:256000. The result shows that purified PE_PGRS45 can induce New Zealand rabbits to produce high-titer antibodies. In conclusion, the recombinant protein PE_PGRS45 was successfully expressed in E. coli and specific antiserum was prepared, which will be followed by further evaluation of these specific antigens to develop highly sensitive and specific diagnostic tests for tuberculosis.

Effects of excessive nitrogen on nitrogen uptake and transformation in the wetland soils of Liaohe estuary, northeast China.

It remains unclear whether excessive nitrogen additions lead to the degradation of Suaeda salsa (S. salsa) by affecting the nitrogen pool, enzyme activities, and bacterial community structure of wetland soils. This study investigated the effect of five added nitrogen concentrations (0, 1, 2, 4, and 6 mmol L-1 N with NH4NO3 = group C, group L, group M, group H, and group G, respectively) on nitrogen uptake by S. salsa and nitrogen transformation in the wetland soils of the Liaohe estuary. The height, weight, and total nitrogen (TN) of S. salsa in group G was significantly lower than in the other groups (p <0.05). The NH4+-N concentration in the soil tended to increase with increasing nitrogen addition, but the TN concentration in the soil tended to decrease. The nitrogenase, protease, urease, ammonia monooxygenase (AMO), nitrous oxide reductase (NOR), and dehydrogenase (DHA) activities increased with increasing nitrogen addition within the range of 0 to 4 mmol L-1. We identified 30 phyla and 48 known genera across all five groups. The predominant phyla were Proteobacteria (52.68%), Bacteroidetes (22.58%), and Planctomycetes (3.94%). The most abundant genus was Acinetobacter (13.38%), followed by Proteiniphilum (11.88%) and Brevundimonas (6.03%). The total number of soil bacterial species increased with increasing nitrogen addition. Group G had lower soil bacterial activity and diversity than the other groups. It was concluded that appropriate levels of nitrogen addition could promote nitrogen uptake by S. salsa and nitrogen transformation in the wetland soils of the Liaohe estuary by affecting soil enzyme activities and soil bacterial activity, diversity, abundance, and composition, while excessive nitrogen additions may be one of the reasons for the degradation of S. salsa.