Metabolic dysfunction-associated steatohepatitis (MASH) is a chronic liver disease characterized by inflammation and fibrosis, with enhanced interleukin-18 (IL-18) signaling. IL-18-binding protein (IL-18BP) neutralizes IL-18, but its therapeutic potential in MASH is unclear. We find elevated IL-18BP and IL-18 levels in patients with MASH and mice, with free IL-18 correlating with disease severity. IL-18 stimulates interferon-gamma (IFNγ) production in CD4 T cells, increasing hepatic IL-18BP. IL-18BP-deficient mice show worsened liver inflammation and fibrosis. We develop a human IL-18BP biologics (APB-R3) and inject it to mice to evaluate its pharmacologic efficacy. APB-R3 significantly improves MASH in reducing fibrosis and inflammation and inhibits hepatic stellate cell activation via the cGMP pathway. This study proposes that abrogation of IL-18 signaling by boosting IL-18BP can strongly inhibit the development of MASH-induced fibrosis, and our engineered IL-18BP biologics can become a promising therapeutic candidate for curing MASH.

Thyrogen, a recombinant human thyroid-stimulating hormone (rhTSH), has a short half-life in the bloodstream, which necessitates multiple doses during treatment. Therefore, we developed a new long-acting rhTSH using anti-serum albumin Fab-associated (SAFA) technology to validate its biological activity through in vitro assays, pharmacokinetic studies in healthy mice and pharmacodynamics studies in a thyroid-stimulating hormone (TSH)-suppressed mouse model. SAFA-TSH was produced using a Chinese hamster ovary expression system. To verify its biological activity, we generated Nthy-ori 3-1 cells stably overexpressing TSHR and measured the production of cyclic adenosine monophosphate (cAMP). In a rat study, slow-release triiodothyronine (T3) pellets were implanted 3 days before administering Thyrogen or SAFA-TSH to measure the amount of thyroxine (T4) release alone resulting from exogenous administration. SAFA-TSH increased cAMP production dose-dependently, but less effectively than Thyrogen at similar concentrations. SAFA-TSH required six times the dose of Thyrogen to achieve similar cAMP levels, likely due to differences in molecular weight and relative bioactivity. In a rat study, SAFA-TSH produced elevated thyroid hormone levels well after the decline in the response to Thyrogen. SAFA-TSH had significantly higher cumulative effects on T4 and free T4 levels compared with Thyrogen, as observed by a more than two-fold higher average area under the effect curve of 262.56 vs 118.89 μg × h/dL and 127.47 vs 60.75 μg × h/dL, respectively. SAFA technology created successful long-acting TSH that demonstrated bioactivity. These findings endorse the continued development of SAFA-TSH for clinical use, highlighting its potential as a significant advancement treating thyroid cancer patients.

Background: Recombinant human follicle-stimulating hormone (rhFSH) is commonly used to treat female infertility, but its short half-life necessitates multiple doses. Even corifollitropin alfa, with an extended half-life, requires supplementary injections of rhFSH after 7 days. This study aimed to develop and evaluate a long-acting follicle-stimulating hormone (FSH) formulation using anti-serum albumin Fab-associated (SAFA) technology to avoid additional injections and enhance ovarian function.Methods: SAFA-FSH was synthesized using a Chinese hamster ovary expression system. Its biological efficacy was confirmed through assays measuring its ability to stimulate cyclic adenosine monophosphate (cAMP) production, estradiol synthesis, and the expression of human cytochrome P450 family 19 subfamily A member 1 (hCYP19α1) and human steroidogenic acute regulatory protein (hSTAR) in human ovarian granulosa (KGN) cells. To evaluate the effects of SAFA-FSH, we compared its impact on serum estradiol levels and ovarian weight increase with that of rhFSH in Sprague-Dawley (SD) rats using the modified Steelman-Pohley test.Results: The results indicated that SAFA-FSH induces cAMP synthesis in KGN cells and upregulates the expression of hCYP19α1 and hSTAR in a dose-dependent manner. Female SD rats, aged 21 days, receiving daily subcutaneous human chorionic gonadotropin injections for 5 days exhibited a significant increase in serum estradiol levels and ovarian weight when administered SAFA-FSH on the first day or when given nine injections of rhFSH over 5 days. Notably, the group receiving SAFA-FSH on the first and third days demonstrated an even greater rise in serum estradiol levels and ovarian weight.Conclusion: These findings suggest that SAFA-FSH presents a promising alternative to current rhFSH treatments for female infertility. However, further research is essential to thoroughly assess its safety and efficacy in clinical contexts.

Large Language Model (LLM) changes our lives, while it requires unprecedented computing resources, especially it requires large memory capacity and high bandwidth to process weights. However, while the logic process was developing, the speed of development of the memory process could not keep up, causing problems that resulted in the performance of LLM being hindered by memory. Samsung have introduced breakthrough Processing-in-Memory/Processing-near-Memory (PIM/PNM) solutions that enhance the main memory bandwidth. With the HBM-PIM-based GPU-cluster system and LPDDR5-PIM-based system, the performance of transformer-based LLMs improved by up to 1.9× and 2.7×, respectively. The CXL-based PNM solution serves memory-centric computing systems by implementing logic inside the CXL memory controller. This results in a performance gain of over 4.4× with an energy reduction of about 53% with PNM. Furthermore, we provide PIM/PNM software stacks, including an AI compiler targeting the acceleration of AI models.

Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease characterized by progressive inflammation and fibrosis around intrahepatic and extrahepatic bile ducts leading to severe hepatic cirrhosis and high mortality. Although there is an urgent clinical unmet need for PSC, no effective medical therapy has been developed to delay the disease progression until today. IL-18 binding protein (IL-18BP) is well-known to be a natural negative feedback regulator for IL-18, and we have developed a recombinant long-acting IL-18BP referred to as APB-R3 as a therapeutic agent to treat IL-18-related inflammatory diseases. Here, we aimed to study whether disrupted IL-18 signaling by APB-R3 treatment can inhibit PSC injuries in the experimental DDC diet-induced PSC rodent model. First, we found that the amounts of free IL-18 are augmented under PSC condition with increased expression of biliary IL-18 receptors. Administration of APB-R3 effectively attenuated key diagnostic parameters of PSC such as plasma ALP and GGT levels as well as bile acids levels. We also observed that blockade of IL-18 suppressed ductular reactive and proliferative phenotypes of cholangiocytes. Additionally, APB-R3 significantly ameliorated DDC diet-induced periductal fibrosis and transcriptional expressions of pro-fibrotic marker genes. Enhanced senescence associated secretory phenotype (SASP) markers in cholestatic liver disease were diminished by APB-R3 treatment. Our findings clearly demonstrate that the administration of IL-18BP biologics, APB-R3, effectively alleviates DDC diet-induced biliary injuries in rodent PSC model, implying APB-R3 can be a promising therapeutic reagent which warrants clinical human trials as new therapeutic options.

Given the clinical success of cytokine blockade in managing diverse inflammatory human conditions, this approach could be exploited for numerous refractory or uncontrolled inflammatory conditions by identifying novel targets for functional blockade. Interleukin (IL)-18, a pro-inflammatory cytokine, is relatively underestimated as a therapeutic target, despite accumulated evidence indicating the unique roles of IL-18 in acute and chronic inflammatory conditions, such as macrophage activation syndrome. Herein, we designed a new form of IL-18 blockade, i.e., APB-R3, a long-acting recombinant human IL-18BP linked to human albumin-binding Fab fragment, SL335, for extending half-life. We then explored the pharmacokinetics and pharmacodynamics of APB-R3. In addition to an extended serum half-life, APB-R3 alleviates liver inflammation and splenomegaly in a model of the macrophage activation syndrome induced in IL-18BP knockout mice. Moreover, APB-R3 substantially controlled skin inflammation in a model of atopic dermatitis. Thus, we report APB-R3 as a new potent IL-18 blocking agent that could be applied to treat IL-18-mediated inflammatory diseases.

IntroductionAdministration of follicle-stimulating hormone (FSH) has been recommended to stimulate spermatogenesis in infertile men with hypogonadotropic hypogonadism, whose sperm counts do not respond to human chorionic gonadotropin alone. However, FSH has a short serum half-life requiring frequent administration to maintain its therapeutic efficacy. To improve its pharmacokinetic properties, we developed a unique albumin-binder technology, termed "anti-serum albumin Fab-associated" (SAFA) technology. We tested the feasibility of applying SAFA technology to create long-acting FSH as a therapeutic candidate for patients with hypogonadotropic hypogonadism. MethodsSAFA-FSH was produced using a Chinese hamster ovary expression system. To confirm the biological function, the production of cyclic AMP and phosphorylation of ERK and CREB were measured in TM4-FSHR cells. The effect of gonadotropin-releasing hormone agonists on spermatogenesis in a hypogonadal rat model was investigated. ResultsIn in vitro experiments, SAFA-FSH treatment increased the production of cyclic AMP and increased the phosphorylation of ERK and CREB in a dose-dependent manner. In animal experiments, sperm production was not restored by human chorionic gonadotropin treatment alone, but was restored after additional recombinant FSH treatment thrice per week or once every 5 days. Sperm production was restored even after additional SAFA-FSH treatment at intervals of once every 5 or 10 days. DiscussionLong-acting FSH with bioactivity was successfully created using SAFA technology. These data support further development of SAFA-FSH in a clinical setting, potentially representing an important advancement in the treatment of patients with hypogonadotropic hypogonadism.

Hybrid memory has become a promising new solution for meeting ever growing memory capacity demands in a cost-effective way. In hybrid memory systems, the fast and high bandwidth memory is used to store performance-critical data, while the slow and low bandwidth memory provides capacity backup. In supporting such hybridization, virtual memory is the key mechanism, which can combine different memory components to a single memory view. For efficient translation for virtual memory, page size has been growing. However, the hybrid memory support requires fine-grained migration to quickly move only necessary memory portions to the precious fast memory. To address the challenges posed by the conflicting goals in the hybrid memory support based on virtual memory, this paper investigates decoupling of address translation into a two-step process. With the two-level translation, the critical core-side TLBs perform the translation to an intermediate address space, and the memory-side translation provides the actual physical location in memory devices. As the second-level translation handling page migration across different memory types, is decoupled from the first-level translation, it allows dynamic adjustment of its mapping granularity to improve the efficiency of translation and data reuse in the fast memory. This paper proposes a hardware architecture which identifies the memory access behavior of an application online and selects the best mapping granularity for the second-level translation.

Off-label use of a human granulocyte colony stimulating factor (hG-CSF) has been allowed to treat dogs and cats with neutropenia. However, repeated administration of hG-CSF induces undesirable anti-drug antibody (ADA) responses, implying the necessity of animal-derived G-CSF as a therapeutic reagent, preferably with a long-acting capability. Herein, we generated a recombinant fusion protein by genetically combining FL335, a chimeric Fab specific for feline serum albumin (FSA), and feline G-CSF (fG-CSF), with the ultimate goal of developing a long-acting therapeutic fG-CSF for cats. The resulting FL335-fG-CSF fusion protein, referred to as APB-F1, was produced well as a functional form in a Chinese hamster ovary (CHO) expression system. In in vitro analyses, APB-F1 bound to FSA at high affinity (KD = 400 pM) and possessed 0.78 × 107 U/mg G-CSF biological activity, clearly proving its biological functionality. Pharmacokinetic (PK) and pharmacodynamic (PD) studies using healthy cats revealed that the serum half-life (t1/2) of APB-F1 was increased five times compared with that of fG-CSF (t1/2 = 13.3 h vs. 2.7 h) in subcutaneous (SC) injections. Additionally, APB-F1 induced a profound and sustained increase in white blood cell (WBC) and actual neutrophil count (ANC) up to 10 days, which was far superior to other G-CSF preparations, including filgrastim (Neupogen™) and even pegfilgrastim (Neulasta™). Conclusively, a long-acting fG-CSF with potent in vivo bioactivity was successfully created by using FL335; thus, we provided evidence that our "anti-serum albumin Fab-associated" (SAFA) technology can be applied reliably in developing valuable long-acting biologics in veterinary medicine.

With the proliferation of applications with machine learning (ML), the importance of edge platforms has been growing to process streaming sensor, data locally without resorting to remote servers. Such edge platforms are commonly equipped with heterogeneous computing processors such as GPU, DSP, and other accelerators, but their computational and energy budget are severely constrained compared to the data center servers. However, as an edge platform must perform the processing of multiple machine learning models concurrently for multimodal sensor data, its scheduling problem poses a new challenge to map heterogeneous machine learning computation to heterogeneous computing processors. Furthermore, processing of each input must provide a certain level of bounded response latency, making the scheduling decision critical for the edge platform. This article proposes a set of new heterogeneity-aware ML inference scheduling policies for edge platforms. Based on the regularity of computation in common ML tasks, the scheduler uses the pre-profiled behavior of each ML model and routes requests to the most appropriate processors. It also aims to satisfy the service-level objective (SLO) requirement while reducing the energy consumption for each request. For such SLO supports, the challenge of ML computation on GPUs and DSP is its inflexible preemption capability. To avoid the delay caused by a long task, the proposed scheduler decomposes a large ML task to sub-tasks by its layer in the DNN model.

Human serum albumin (HSA) has been used to extend the serum half-lives of various protein therapeutics through genetic fusion because HSA exhibits an exceptionally long circulation time as a result of neonatal Fc receptor (FcRn)-mediated recycling. As another serum half-life extender, the human antibody Fab SL335 that strongly binds HSA was developed. When SL335 was fused to a protein therapeutic, SL335 was shown to prolong the half-life of the drug. Despite the significance of SL335-HSA binding in the extension of drug circulation time, it remains unclear how SL335 interacts with HSA at a molecular structural level. To reveal the structural basis of HSA recognition by SL335, we determined the crystal structure of the SL335-HSA complex at a resolution of 2.95 Å. SL335 binds HSA at a 1:1 stoichiometry. SL335 uses the exposed loops of its heavy and light chains to specifically recognize the IIa and IIb subdomains of HSA. The SL335 epitope is located on the opposite side of the FcRn-binding site and does not overlap with it, suggesting that SL335 extends the serum half-lives of itself and its fusion partner through an FcRn-dependent recycling mechanism.

The availability of large pages has dramatically improved the efficiency of address translation for applications that use large contiguous regions of memory. However, large pages can be difficult to allocate due to fragmented memory, non-movable pages, or the need to split a large page into regular pages when part of the large page is forced to have a different permission status from the rest of the page. Furthermore, they can also be expensive due to memory bloating caused by sparse accesses to application data. In this work, we enable the allocation of large 2MB pages even in the presence of fragmented physical memory via perforated pages. Perforated pages permit the OS to punch 4KB page-sized holes in the physical address range allocated to a large page and re-map them to other addresses as needed. This not only enables the system to benefit from large pages in the presence of fragmentation, but also allows for different permissions to exist within a large page, enhancing sharing flexibility. In addition, it allows unused parts of a large page to be used elsewhere, mitigating memory bloating. To minimize changes to the system, perforated pages reuse the 4KBlevel page table entries to store the hole locations and translates holes into regular 4KB pages. For performance, the proposed technique caches the translations for hole pages in the TLBs and track holes via cached bitmaps in the L2 TLB. By enabling large pages in the presence of physical memory fragmentation, perforated pages increase the applicability and resulting benefits of large pages with only minor changes to the hardware and OS. In this work, we evaluate the effectiveness of perforated pages with timing simulations under diverse and realistic fragmentation scenarios. Our results show that even with fragmented memory, perforated pages accomplish 93.2% to 99.9% of the performance achievable by ideal memory allocation, and 2.0% to 11.5% better performance over the conventional system running with fragmented memory.

The support for the heterogeneous memory in the conventional virtual memory has an inherent problem. For the efficient translation in the critical translation lookaside buffers (TLBs), the page size has been growing. However, the heterogeneous memory management requires a nimble fine-grained migration mechanism to quickly move necessary memory portions to the precious fast memory.To address the challenges posed by the conflicting goals in the heterogeneous memory support, this paper proposes to decouple the address translation into a two-step process. The decoupling resolves the conflict as the critical core-side TLBs perform the translation to an intermediate address space, and the memory-side translation provides the actual physical location of the memory devices.