ZYP1 is required for obligate cross-over formation and cross-over interference in Arabidopsis

The synaptonemal complex is a tripartite proteinaceous ultrastructure that forms between homologous chromosomes during prophase I of meiosis in the majority of eukaryotes. It is characterized by the coordinated installation of transverse filament proteins between two lateral elements and is required for wild-type levels of crossing over and meiotic progression. We have generated null mutants of the duplicated Arabidopsis transverse filament genes zyp1a and zyp1b using a combination of T-DNA insertional mutants and targeted CRISPR/Cas mutagenesis. Cytological and genetic analysis of the zyp1 null mutants reveals loss of the obligate chiasma, an increase in recombination map length by 1.3- to 1.7-fold and a virtual absence of cross-over (CO) interference, determined by a significant increase in the number of double COs. At diplotene, the numbers of HEI10 foci, a marker for Class I interference-sensitive COs, are twofold greater in the zyp1 mutant compared to wild type. The increase in recombination in zyp1 does not appear to be due to the Class II interference-insensitive COs as chiasmata were reduced by ∼52% in msh5/zyp1 compared to msh5. These data suggest that ZYP1 limits the formation of closely spaced Class I COs in Arabidopsis. Our data indicate that installation of ZYP1 occurs at ASY1-labeled axial bridges and that loss of the protein disrupts progressive coalignment of the chromosome axes.

The synaptonemal complex (SC) is a proteinaceous ultrastructure that forms between homologous chromosomes (homologs) during midprophase I of meiosis and plays a critical role in coordinating the repair of programmed DNA double-strand breaks (DSBs) to form cross-over (CO) products (1, 2). At the onset of leptotene, the sister chromatids are organized into linear looped chromatin arrays conjoined at the loop bases by a protein axis that runs along the chromosomes (3, 4). Early steps in the recombination pathway enable the loose alignment of homolog axes at a distance of ∼400 nm (5). Formation of the SC then initiates and continues throughout zygotene via progressive installation of transverse filaments (TFs) that run perpendicular to the aligned homolog axes (referred to as lateral elements in the context of the SC), ultimately bringing them into close apposition along their entire length at a distance of ∼100 nm (2, 5). Installation of the TFs starts at multiple synapsis initiation sites that correspond to future Class I COs in Saccharomyces cerevisiae (6). In species with larger chromosomes such as Sordaria macrospora, synapsis initiates from CO-designated sites as well as additional sites whose distribution also appears sensitive to interference (1, 5). In Arabidopsis thaliana, 20 to 25 synapsis initiation sites per cell indicate a ∼2- to 2.5-fold excess over COs and in barley 76 synapsis initiation sites, versus 17 chiasmata reveal a ∼4.5-fold excess (7, 8). Full synapsis denotes the onset of pachytene and is maintained throughout this stage during which time CO formation is completed. As prophase I progresses to diplotene/diakinesis, the SC is disassembled.TFs have been described in a variety of organisms, and in most cases, they are composed of a single protein. These include Zip1 in budding yeast, C(3)G in Drosophila melanogaster, SYCP1 in mouse, ZYP1 in A. thaliana (encoded by duplicated genes, ZYP1a and ZYP1b), ZEP1 in rice (Oryza sativa), and ZYP1 in barley (Hordeum vulgare) (9–16). Caenorhabditis elegans is an exception that possesses six TF proteins (SYP1-6) required for normal synapsis (17–22). Despite a striking lack of homology between the TFs at the primary amino acid sequence level, they share very similar structures, comprising a globular N-terminal domain linked to another globular domain at the C terminus via a long alpha helical central region that is able to form large stretches of parallel, in-register, homodimeric coiled coils (23). Studies have shown that the TFs are oriented such that the C termini are associated with lateral elements potentially interacting with DNA, while the N-terminal domains localize to the central region (2, 24). Evidence suggests that the overall three-dimensional macromolecular organization of the SC is also somewhat conserved. Analyses in mouse, Drosophila, and H. vulgare (barley) strongly suggest that these organisms form SCs with a bilayer of TFs (25–28). A multilayered structure is also supported by studies in Blabs cribrosa (beetle) (29, 30). However, key aspects of the organization of the TFs within the SC remain a matter of debate. Initially, analysis of zip1 mutants in S. cerevisiae suggested that the TFs comprise a tetramer of two opposing Zip1 dimers with their N termini forming overlapping interactions in the central region of the SC (31). X-ray crystallographic studies of the human TF, SYCP1, report that the protein forms a tetrameric building block that self-assembles into a zipper-like lattice through “head-to-head” N-terminal interactions in the SC central region and “back-to-back” interactions between adjacent C-terminal dimers at the lateral elements (24). In contrast, analysis of the mouse SC using electron tomography has led to the proposal that the SC has a more dynamic structure with TF dimers forming a variety of less regimented interactions as part of an irregular single plane. However, this model appears inconsistent with other studies in mouse which support a more ordered structure (25, 26).Mutant analysis has demonstrated that TF proteins are essential for assembly of the SC central region and thus homolog synapsis. These also confirm an important role in the control of CO formation but with some variation between organisms. Studies of zip1 mutants in S. cerevisiae have shown that the Zip1 protein is a member of the ZMM group of proteins comprising Zip1, Zip2, Zip3, Zip4, Msh4, Msh5, and Mer3 that are required for the formation of Class I interfering COs (32). CO interference is a patterning mechanism that ensures even spacing of COs along the chromosomes (33–35). In S. cerevisiae and Arabidopsis, Class I COs account for ∼85% of total COs and the remaining Class II COs (∼15%) are randomly distributed (36–38). However, in plants, Zip1 orthologs appear to be functionally independent of the other ZMM proteins for CO formation (14–16). Genetic analysis of S. cerevisiae zip1 deletion mutants revealed a modest reduction in CO formation ∼30 to 40% with residual COs no longer exhibiting CO interference leading to the suggestion that the SC may mediate this process (10). Subsequent studies based on a molecular analysis of recombination intermediates in zip1 and other zmm mutants argue against a role for the SC in mediating interference as they indicate that the fate of DSBs is designated at an early stage in the recombination pathway prior to installation of the SC (32, 39). In female Drosophila lacking the TF protein C(3)G, DSB formation is thought to be reduced and they fail to form COs, although SC formation is independent of recombination (12). These authors also report that analysis of flies expressing a mutant version of the protein reveals that a complete SC is not required for CO interference (12). A major reduction in COs of ∼90% is also observed in mouse sycp1 mutants although DSB formation appears normal (13). Similarly in C. elegans (in which SC installation occurs at pairing centers), syp-1 and syp-2 null mutants recombination is initiated but COs do not form (17, 18). A further study in which the SC central region was partially depleted by RNA interference (RNAi)–induced SYP-1 knockdown found that CO interference was reduced leading to an increase in COs, suggesting a role for the SC in limiting COs (40).TFs have been studied in several plant species including Arabidopsis, barley, and rice (14–16). Analysis of Tos17 insertion mutants of the rice TF gene ZEP1 demonstrated that in common with other organisms, it is essential for SC formation and affects CO formation (16). However, rather than displaying a reduction in COs, analysis of the short arm of chromosome 11 revealed a more than threefold increase in COs in zep1 mutants (16). Like rice, barley is a member of the grass family (Poaceae), and in common with rice, RNAi knockdown lines of the TF protein HvZYP1 are defective in SC formation, but in contrast, CO formation is reduced to ∼25% of wild-type levels (15). In Arabidopsis, the TF protein, ZYP1, is encoded by functionally redundant duplicated genes, ZYP1a and ZYP1b, which share 93% homology and are encoded within 2 kb of each other on opposite strands of chromosome 1. Individual zyp1a and zyp1b mutants are fertile and possess only mild meiotic phenotypes, and as isolation of a double mutant has thus far proved intractable, functional analysis of ZYP1 has relied on RNAi knockdown lines (14). As expected, these lines failed to assemble an SC. Chiasma frequency was reduced by ∼20 to 30% and based on metaphase I bivalent shapes, they appeared to exhibit interference, but a proportion involved ectopic recombination with nonhomologs (14).Although existing studies imply that there may be some variation in the role of the SC in relation to CO control in plants, the studies in Arabidopsis and barley were based on RNAi knockdown lines rather than TF mutants. Hence to address this issue, we have generated CRISPR/Cas zyp1a/zyp1b mutants. This has enabled a detailed analysis of ZYP1 function in Arabidopsis, revealing that it is required for formation of the obligate CO and implementation of CO patterning. Loss of the protein also disrupts the normal program of homolog coalignment during prophase I.     

A general model of hippocampal and dorsal striatal learning and decision making

Humans and other animals use multiple strategies for making decisions. Reinforcement-learning theory distinguishes between stimulus–response (model-free; MF) learning and deliberative (model-based; MB) planning. The spatial-navigation literature presents a parallel dichotomy between navigation strategies. In “response learning,” associated with the dorsolateral striatum (DLS), decisions are anchored to an egocentric reference frame. In “place learning,” associated with the hippocampus, decisions are anchored to an allocentric reference frame. Emerging evidence suggests that the contribution of hippocampus to place learning may also underlie its contribution to MB learning by representing relational structure in a cognitive map. Here, we introduce a computational model in which hippocampus subserves place and MB learning by learning a “successor representation” of relational structure between states; DLS implements model-free response learning by learning associations between actions and egocentric representations of landmarks; and action values from either system are weighted by the reliability of its predictions. We show that this model reproduces a range of seemingly disparate behavioral findings in spatial and nonspatial decision tasks and explains the effects of lesions to DLS and hippocampus on these tasks. Furthermore, modeling place cells as driven by boundaries explains the observation that, unlike navigation guided by landmarks, navigation guided by boundaries is robust to “blocking” by prior state–reward associations due to learned associations between place cells. Our model, originally shaped by detailed constraints in the spatial literature, successfully characterizes the hippocampal–striatal system as a general system for decision making via adaptive combination of stimulus–response learning and the use of a cognitive map.

Behavioral and neuroscientific studies suggest that animals can apply multiple strategies to the problem of maximizing future reward, referred to as the reinforcement-learning (RL) problem (1, 2). One strategy is to build a model of the environment that can be used to simulate the future to plan optimal actions (3) and the past for episodic memory (4–6). An alternative, model-free (MF) approach uses trial and error to estimate a direct mapping from the animal’s state to its expected future reward, which the agent caches and looks up at decision time (7, 8), potentially supporting procedural memory (9). This computation is thought to be carried out in the brain through prediction errors signaled by phasic dopamine responses (10). These strategies are associated with different tradeoffs (2). The model-based (MB) approach is powerful and flexible, but computationally expensive and, therefore, slow at decision time. MF methods, in contrast, enable rapid action selection, but these methods learn slowly and adapt poorly to changing environments. In addition to MF and MB methods, there are intermediate solutions that rely on learning useful representations that reduce burdens on the downstream RL process (11–13).In the spatial-memory literature, a distinction has been observed between “response learning” and “place learning” (14–16). When navigating to a previously visited location, response learning involves learning a sequence of actions, each of which depends on the preceding action or sensory cue (expressed in egocentric terms). For example, one might remember a sequence of left and right turns starting from a specific landmark. An alternative place-learning strategy involves learning a flexible internal representation of the spatial layout of the environment (expressed in allocentric terms). This “cognitive map” is thought to be supported by the hippocampal formation, where there are neurons tuned to place and heading direction (17–19). Spatial navigation using this map is flexible because it can be used with arbitrary starting locations and destinations, which need not be marked by immediate sensory cues.We posit that the distinction between place and response learning is analogous to that between MB and MF RL (20). Under this view, associative reinforcement is supported by the DLS (21, 22). Indeed, there is evidence from both rodents (23–25) and humans (26, 27) that spatial-response learning relies on the same basal ganglia structures that support MF RL. Evidence also suggests an analogy between MB reasoning and hippocampus (HPC)-based place learning (28, 29). However, this equivalence is not completely straightforward. For example, in rodents, multiple hippocampal lesion and inactivation studies failed to elicit an effect on action-outcome learning, a hallmark of MB planning (30–35). Nevertheless, there are indications that HPC might contribute to a different aspect of MB RL: namely, the representation of relational structure. Tasks that require memory of the relationships between stimuli do show dependence on HPC (36–42).Here, we formalize the perspective that hippocampal contributions to MB learning and place learning are the same, as are the dorsolateral striatal contributions to MF and response learning. In our model, HPC supports flexible behavior by representing the relational structure among different allocentric states, while dorsolateral striatum (DLS) supports associative reinforcement over egocentric sensory features. The model arbitrates between the use of these systems by weighting each system’s action values by the reliability of the system, as measured by a recent average of prediction errors, following Wan Lee et al. (43). We show that HPC and DLS maintain these roles across multiple task domains, including a range of spatial and nonspatial tasks. Our model can quantitatively explain a range of seemingly disparate findings, including the choice between place and response strategies in spatial navigation (23, 44) and choices on nonspatial multistep decision tasks (45, 46). Furthermore, it explains the puzzling finding that landmark-guided navigation is sensitive to the blocking effect, whereas boundary-guided navigation is not (27), and that these are supported by the DLS and HPC, respectively (26). Thus, different RL strategies that manage competing tradeoffs can explain a longstanding body of spatial navigation and decision-making literature under a unified model.     

Environmental noise degrades hippocampus-related learning and memory

The neural mechanisms underlying the impacts of noise on nonauditory function, particularly learning and memory, remain largely unknown. Here, we demonstrate that rats exposed postnatally (between postnatal days 9 and 56) to structured noise delivered at a sound pressure level of ∼65 dB displayed significantly degraded hippocampus-related learning and memory abilities. Noise exposure also suppressed the induction of hippocampal long-term potentiation (LTP). In parallel, the total or phosphorylated levels of certain LTP-related key signaling molecules in the synapses of the hippocampus were down-regulated. However, no significant changes in stress-related processes were found for the noise-exposed rats. These results in a rodent model indicate that even moderate-level noise with little effect on stress status can substantially impair hippocampus-related learning and memory by altering the plasticity of synaptic transmission. They support the importance of more thoroughly defining the unappreciated hazards of moderately loud noise in modern human environments.

The noise pollution accompanying industrialization is a risk factor to human health. Earlier studies have extensively examined the deleterious impacts of noise in the auditory systems of both humans and animal models (1–6), showing that noise exposure either early or late in life can induce progressive hearing loss, change neural coding along the auditory pathway, and alter auditory-related perception and behavior.The auditory system, however, contains direct and indirect pathways to other systems and structures of the brain that are necessary for functional integration. For example, earlier studies found that the hippocampus, the core area of the brain associated with learning and memory processes, receives neuronal inputs from the auditory system through the lemniscal and nonlemniscal pathways (7–11). It is thus conceivable that noise-evoked activities might be transmitted via these connections to the hippocampus, thereby affecting learning and memory. Indeed, animal studies have shown that exposure to loud noise (e.g., above a sound pressure level [SPL] of 95 dB) that induces temporary or permanent shifts in the auditory threshold disrupts hippocampal histology, decreases neurogenesis in the hippocampus, and impairs hippocampus-related learning and memory abilities (12–16). In addition, epidemiological studies have demonstrated that environment noise has substantially negative effects on children’s learning outcomes and cognitive abilities (17–19). While the usual explanations for the origins of these noise-induced effects on nonauditory functions have relied on stress-related processes (15, 16, 20–23), the underlying neural mechanisms remain largely unknown.In this study, we exposed rat pups to structured noise delivered at ∼65 dB SPL for a 7-wk period. Exposure to a moderate level of modulated broad-spectrum noise more realistically models the noise environments people encountered in industrial workplaces and other modern acoustic settings (2, 4, 24–26). We then evaluated the behavioral consequences of noise exposure on hippocampus-related learning and memory for these noise-exposed rats. In addition, we explored the mechanisms underlying possible postexposure changes in learning and memory via physiological and molecular assessments of the hippocampus.     

Layer and rhythm specificity for predictive routing

In predictive coding, experience generates predictions that attenuate the feeding forward of predicted stimuli while passing forward unpredicted “errors.” Different models have suggested distinct cortical layers, and rhythms implement predictive coding. We recorded spikes and local field potentials from laminar electrodes in five cortical areas (visual area 4 [V4], lateral intraparietal [LIP], posterior parietal area 7A, frontal eye field [FEF], and prefrontal cortex [PFC]) while monkeys performed a task that modulated visual stimulus predictability. During predictable blocks, there was enhanced alpha (8 to 14 Hz) or beta (15 to 30 Hz) power in all areas during stimulus processing and prestimulus beta (15 to 30 Hz) functional connectivity in deep layers of PFC to the other areas. Unpredictable stimuli were associated with increases in spiking and in gamma-band (40 to 90 Hz) power/connectivity that fed forward up the cortical hierarchy via superficial-layer cortex. Power and spiking modulation by predictability was stimulus specific. Alpha/beta power in LIP, FEF, and PFC inhibited spiking in deep layers of V4. Area 7A uniquely showed increases in high-beta (∼22 to 28 Hz) power/connectivity to unpredictable stimuli. These results motivate a conceptual model, predictive routing. It suggests that predictive coding may be implemented via lower-frequency alpha/beta rhythms that “prepare” pathways processing-predicted inputs by inhibiting feedforward gamma rhythms and associated spiking.

The brain exploits predictability. It makes cortical processing more efficient. Visuomotor integration, visual/auditory speech perception, and visual perception all benefit when sensory inputs are predictable (1–3). The brain has an arsenal of mechanisms to tamp down and improve processing of familiar, repeated, or predictable inputs. One example is stimulus-specific adaptation. All over cortex, there is less spiking and smaller blood-oxygen-level-dependent (BOLD) responses when a stimulus is repeated (4–9). Responsiveness is recovered if the stimulus is changed or a pattern is violated (i.e., to “oddballs”) (10, 11). This can lead to fewer activated neurons but finer-tuned, more robust representations (8).But the brain does more than adapt to repeated inputs. A wide variety of evidence indicates that it makes mental models of the world that actively generate predictions, a process known as predictive coding (12–14). Moment-to-moment predictions are used to inhibit processing of expected inputs which, because they were expected, are not informative. Unexpected sensory inputs that deviate from a prediction, are “prediction errors” (PEs). They are informative and thus not inhibited, fed forward, processed, affect behavior, and are used to update the prediction models.Much of the work on the neural mechanisms of prediction and its violation has focused on spiking activity (2, 15–17). But there is mounting evidence that oscillatory dynamics play a role in regulating cortical processing and thus can also play a role, especially the gamma (40 to 90 Hz) and alpha/beta (10 to 30 Hz) bands (1, 18–25). A key observation is that, all across cortex, gamma power (>35 Hz)/spiking is higher during bottom-up sensory inputs. They are anticorrelated with alpha/beta (8 to 30 Hz) power (26–29), which is higher under conditions of top-down control (e.g., attention and response inhibition) (30–34). This suggests that top-down alpha/beta help regulate the processing of bottom-up inputs served by gamma and spiking. The idea is that alpha/beta carries the top-down predictions that inhibit the gamma/spiking that process expected inputs. This is consistent with gamma power being higher in the superficial, feedforward, cortical layers, and alpha/beta power being higher in the deep, feedback, cortical layers (26, 35–40). Indeed, superficial cortical layers have been hypothesized to be specialized for computing PEs and feeding PEs forward at gamma frequency (1, 19). In addition, computational modeling studies have shown the plausibility of superficial gamma circuits to engage in prediction error computations (38, 41, 42). Direct evidence for alpha/beta and gamma in predictive coding per se comes from observations of increased gamma power to stimuli that are prediction errors (22, 24, 25).How these rhythms (and their relation to spiking) differ with stimulus repetition/predictability as well as their stimulus specificity is not well known. Most neurophysiological studies of the effects of stimulus predictability have focused on spiking activity, often in a single area. And none of them to date have examined and compared activity in different cortical layers. We recorded local field potentials (LFPs) and spiking using multiarea, multilaminar recordings from a visual area (V4) and higher-order cortical areas (posterior parietal cortex and prefrontal cortex [PFC]) simultaneously. Area V4 was selected as previous studies have shown this area to be a target of top-down signals such as attention (43, 44). Frontoparietal cortex was targeted because of its well-established role in top-down attention and working memory, cognitive processes that are engaged in the task employed here (30, 31). We manipulated the predictability of objects used in a working memory task. This revealed layer and frequency-specific associations with stimulus repetition/predictability as well as evidence for the direction of flow of these signals. The findings suggest an update of neural models of prediction and predictive coding.