A Comparison Between GPT-3.5, GPT-4, and GPT-4V: Can the Large Language Model (ChatGPT) Pass the Japanese Board of Orthopaedic Surgery Examination?

Introduction Recently, large-scale language models, such as ChatGPT (OpenAI, San Francisco, CA), have evolved. These models are designed to think and act like humans and possess a broad range of specialized knowledge. GPT-3.5 was reported to be at a level of passing the United States Medical Licensing Examination. Its capabilities continue to evolve, and in October 2023, GPT-4V became available as a model capable of image recognition. Therefore, it is important to know the current performance of these models because they will be soon incorporated into medical practice. We aimed to evaluate the performance of ChatGPT in the field of orthopedic surgery. Methods We used three years' worth of Japanese Board of Orthopaedic Surgery Examinations (JBOSE) conducted in 2021, 2022, and 2023. Questions and their multiple-choice answers were used in their original Japanese form, as was the official examination rubric. We inputted these questions into three versions of ChatGPT: GPT-3.5, GPT-4, and GPT-4V. For image-based questions, we inputted only textual statements for GPT-3.5 and GPT-4, and both image and textual statements for GPT-4V. As the minimum scoring rate acquired to pass is not officially disclosed, it was calculated using publicly available data. Results The estimated minimum scoring rate acquired to pass was calculated as 50.1% (43.7-53.8%). For GPT-4, even when answering all questions, including the image-based ones, the percentage of correct answers was 59% (55-61%) and GPT-4 was able to achieve the passing line. When excluding image-based questions, the score reached 67% (63-73%). For GPT-3.5, the percentage was limited to 30% (28-32%), and this version could not pass the examination. There was a significant difference in the performance between GPT-4 and GPT-3.5 (p < 0.001). For image-based questions, the percentage of correct answers was 25% in GPT-3.5, 38% in GPT-4, and 38% in GPT-4V. There was no significant difference in the performance for image-based questions between GPT-4 and GPT-4V. Conclusions ChatGPT had enough performance to pass the orthopedic specialist examination. After adding further training data such as images, ChatGPT is expected to be applied to the orthopedics field.

Simultaneous measurement of nascent transcriptome and translatome using 4-thiouridine metabolic RNA labeling and translating ribosome affinity purification.

Regulation of gene expression in response to various biological processes, including extracellular stimulation and environmental adaptation requires nascent RNA synthesis and translation. Analysis of the coordinated regulation of dynamic RNA synthesis and translation is required to determine functional protein production. However, reliable methods for the simultaneous measurement of nascent RNA synthesis and translation at the gene level are limited. Here, we developed a novel method for the simultaneous assessment of nascent RNA synthesis and translation by combining 4-thiouridine (4sU) metabolic RNA labeling and translating ribosome affinity purification (TRAP) using a monoclonal antibody against evolutionarily conserved ribosomal P-stalk proteins. The P-stalk-mediated TRAP (P-TRAP) technique recovered endogenous translating ribosomes, allowing easy translatome analysis of various eukaryotes. We validated this method in mammalian cells by demonstrating that acute unfolded protein response (UPR) in the endoplasmic reticulum (ER) induces dynamic reprogramming of nascent RNA synthesis and translation. Our nascent P-TRAP (nP-TRAP) method may serve as a simple and powerful tool for analyzing the coordinated regulation of transcription and translation of individual genes in various eukaryotes.

The flexible N-terminal motif of uL11 unique to eukaryotic ribosomes interacts with P-complex and facilitates protein translation.

Eukaryotic uL11 contains a conserved MPPKFDP motif at the N-terminus that is not found in archaeal and bacterial homologs. Here, we determined the solution structure of human uL11 by NMR spectroscopy and characterized its backbone dynamics by 15N-1H relaxation experiments. We showed that these N-terminal residues are unstructured and flexible. Structural comparison with ribosome-bound uL11 suggests that the linker region between the N-terminal domain and C-terminal domain of human uL11 is intrinsically disordered and only becomes structured when bound to the ribosomes. Mutagenesis studies show that the N-terminal conserved MPPKFDP motif is involved in interacting with the P-complex and its extended protuberant domain of uL10 in vitro. Truncation of the MPPKFDP motif also reduced the poly-phenylalanine synthesis in both hybrid ribosome and yeast mutagenesis studies. In addition, G→A/P substitutions to the conserved GPLG motif of helix-1 reduced poly-phenylalanine synthesis to 9-32% in yeast ribosomes. We propose that the flexible N-terminal residues of uL11, which could extend up to ∼25 Šfrom the N-terminal domain of uL11, can form transient interactions with the uL10 that help to fetch and fix it into a position ready for recruiting the incoming translation factors and facilitate protein synthesis.

Direct visualization of translational GTPase factor pool formed around the archaeal ribosomal P-stalk by high-speed AFM.

In translation elongation, two translational guanosine triphosphatase (trGTPase) factors EF1A and EF2 alternately bind to the ribosome and promote polypeptide elongation. The ribosomal stalk is a multimeric ribosomal protein complex which plays an essential role in the recruitment of EF1A and EF2 to the ribosome and their GTP hydrolysis for efficient and accurate translation elongation. However, due to the flexible nature of the ribosomal stalk, its structural dynamics and mechanism of action remain unclear. Here, we applied high-speed atomic force microscopy (HS-AFM) to directly visualize the action of the archaeal ribosomal heptameric stalk complex, aP0•(aP1•aP1)3 (P-stalk). HS-AFM movies clearly demonstrated the wobbling motion of the P-stalk on the large ribosomal subunit where the stalk base adopted two conformational states, a predicted canonical state, and a newly identified flipped state. Moreover, we showed that up to seven molecules of archaeal EF1A (aEF1A) and archaeal EF2 (aEF2) assembled around the ribosomal P-stalk, corresponding to the copy number of the common C-terminal factor-binding site of the P-stalk. These results provide visual evidence for the factor-pooling mechanism by the P-stalk within the ribosome and reveal that the ribosomal P-stalk promotes translation elongation by increasing the local concentration of translational GTPase factors.

Arabidopsis ROOT PHOTOTROPISM2 Is a Light-Dependent Dynamic Modulator of Phototropin1.

The Arabidopsis (Arabidopsis thaliana) blue light photoreceptor phototropin1 (phot1) is a blue light-activated Ser/Thr protein kinase that mediates various light responses, including phototropism. The function of phot1 in hypocotyl phototropism is dependent on the light induction of ROOT PHOTOTROPISM2 (RPT2) proteins within a broad range of blue light intensities. It is not yet known however how RPT2 contributes to the photosensory adaptation of phot1 to high intensity blue light and the phototropic responses under bright light conditions. We show that RPT2 suppresses the activity of phot1 and demonstrate that RPT2 binds to the PHOT1 light, oxygen or voltage sensing1 (LOV1) domain that is required for its high photosensitivity. Our biochemical analyses revealed that RPT2 inhibits autophosphorylation of phot1, suggesting that it suppresses the photosensitivity and/or kinase activity of phot1 through the inhibition of LOV1 function. We found that RPT2 proteins are degraded via a ubiquitin-proteasome pathway when phot1 is inactive and are stabilized under blue light in a phot1-dependent manner. We propose that RPT2 is a molecular rheostat that maintains a moderate activation level of phot1 under any light intensity conditions.

Switch of the interactions between the ribosomal stalk and EF1A in the GTP- and GDP-bound conformations.

Translation elongation factor EF1A delivers aminoacyl-tRNA to the ribosome in a GTP-bound form, and is released from the ribosome in a GDP-bound form. This association/dissociation cycle proceeds efficiently via a marked conformational change in EF1A. EF1A function is dependent on the ribosomal "stalk" protein of the ribosomal large subunit, although the precise mechanism of action of the stalk on EF1A remains unclear. Here, we clarify the binding mode of archaeal stalk aP1 to GTP-bound aEF1A associated with aPelota. Intriguingly, the C-terminal domain (CTD) of aP1 binds to aEF1A•GTP with a similar affinity to aEF1A•GDP. We have also determined the crystal structure of the aP1-CTD•aEF1A•GTP•aPelota complex at 3.0 Šresolution. The structure shows that aP1-CTD binds to a space between domains 1 and 3 of aEF1A. Biochemical analyses show that this binding is crucial for protein synthesis. Comparison of the structures of aP1-CTD•aEF1A•GTP and aP1-CTD•aEF1A•GDP demonstrates that the binding mode of aP1 changes markedly upon a conformational switch between the GTP- and GDP-bound forms of aEF1A. Taking into account biochemical data, we infer that aP1 employs its structural flexibility to bind to aEF1A before and after GTP hydrolysis for efficient protein synthesis.

The ribosomal stalk protein is crucial for the action of the conserved ATPase ABCE1.

The ATP-binding cassette (ABC) protein ABCE1 is an essential factor in ribosome recycling during translation. However, the detailed mechanochemistry of its recruitment to the ribosome, ATPase activation and subunit dissociation remain to be elucidated. Here, we show that the ribosomal stalk protein, which is known to participate in the actions of translational GTPase factors, plays an important role in these events. Biochemical and crystal structural data indicate that the conserved hydrophobic amino acid residues at the C-terminus of the archaeal stalk protein aP1 binds to the nucleotide-binding domain 1 (NBD1) of aABCE1, and that this binding is crucial for ATPase activation of aABCE1 on the ribosome. The functional role of the stalk•ABCE1 interaction in ATPase activation and the subunit dissociation is also investigated using mutagenesis in a yeast system. The data demonstrate that the ribosomal stalk protein likely participates in efficient actions of both archaeal and eukaryotic ABCE1 in ribosome recycling. The results also show that the stalk protein has a role in the function of ATPase as well as GTPase factors in translation.

Binding of translation elongation factors to individual copies of the archaeal ribosomal stalk protein aP1 assembled onto aP0.

Ribosomes in all organisms contain oligomeric and flexible proteins called stalks, which are responsible for the recruitment of translational GTPase factors to the ribosome. Archaeal ribosomes have three stalk homodimers (aP1)2 that constitute a heptameric complex with the anchor protein aP0. We investigated the factor binding ability of aP1 proteins assembled onto aP0, by gel-retardation assays. The isolated aP0(aP1)2(aP1)2(aP1)2 complex, as well as the form bound to the Escherichia coli 50S core, as a hybrid 50S particle, interacted strongly with elongation factor aEF2, but weakly with aEF1A. These interactions were disrupted by a point mutation, F107S, at the C-terminus of aP1. To examine the ability of each copy of aP0-associated aP1 to bind to elongation factors, we constructed aP0·aP1 variant trimers, composed of an aP0 mutant and a single (aP1)2 dimer. Biochemical and quantitative analyses revealed that the resultant three trimers, aP0(aP1)2I, aP0(aP1)2II, and aP0(aP1)2III, individually bound two molecules of aEF2, suggesting that each copy of the aP1 C-terminal region in the aP0·aP1 trimers can bind tightly to aEF2. Interestingly, the unstable binding of aEF1A to each of the three aP0·aP1 trimers was remarkably stabilized in the presence of aEF2. The stability of the aEF1A binding to the stalk complex may be affected by the presence of aEF2 bound to the complex, by an unknown mechanism.

Functional role of the C-terminal tail of the archaeal ribosomal stalk in recruitment of two elongation factors to the sarcin/ricin loop of 23S rRNA.

Two types of elongation factors alternate in their binding to the factor-binding center of the ribosome. Both binding events are accompanied by GTP hydrolysis and drive the translation elongation cycle. The multicopy ribosomal protein family, termed the stalk, contributes actively to the elongation process. Recent evidence indicates that the mobile C-terminal tail of archaeal stalk aP1 directly interacts with both the elongation factors aEF1A and aEF2. To investigate the functional significance of these interactions in recruitment of elongation factors to the factor-binding center of the ribosome, we substituted the archaeal stalk complex aL10•aP1 for the bL10•bL12 stalk complex in the Escherichia coli 50S subunit. The resultant hybrid ribosome accessed archaeal aEF1A and aEF2 in a manner dependent on the C-terminal tail containing the hydrophobic residues Leu103, Leu106 and Phe107. Bases G2659 and A2660 in the sarcin/ricin loop (SRL) of 23S rRNA were protected against DMS modification by both factors as was A1067 by aEF2. Mutagenesis indicated that this protection was dependent on the intact C-terminal tail of aP1. The results suggest a crucial role for the interactions between the stalk C-terminal tail and elongation factors in their recruitment to the SRL of 23S rRNA within the ribosome.