Maternal Immune Activation Sensitizes Male Offspring Rats to Lipopolysaccharide-Induced Microglial Deficits Involving the Dysfunction of CD200–CD200R and CX3CL1–CX3CR1 Systems

Round 1

Reviewer 1 Report

This is a compelling study, which correlates MIA together with a second hit (LPS, in adulthood) with behavioral abnormalities and inflammatory factors. Having said that, the authors should describe the results with more precision. For instance, they should plot their results showing all data points plus the Mean ± SEM. Moreover, the authors should be more explicit with the data regarding the CD200-CD200R molecular axis. It would be beneficial for this study to show the immunofluorescence of the molecular axis altered to have a better understanding of the alterations in situ. It is possible to correlate the microglia activated (morphological analysis of microglia) with the behavioral changes?

Although this is a correlative study, I consider that it will pave the way for a more mechanistic follow up of the link between prenatal stress, depression, and the immune response. In other words, how the two-hit challenge induced an inflammatory response that is leading to abnormal behavior?

Author Response

Response to Reviewer 1 Comments

First of all, we would like to express our sincere gratitude to the Reviewer for the constructive comments on the initial version of our manuscript. Please find our responses to the raised issues hereafter. All changes applied are coloured in green in the text of the article.

Point 1: “This is a compelling study, which correlates MIA together with a second hit (LPS, in adulthood) with behavioral abnormalities and inflammatory factors. Having said that, the authors should describe the results with more precision. For instance, they should plot their results showing all data points plus the Mean ± SEM. Moreover, the authors should be more explicit with the data regarding the CD200-CD200R molecular axis. It would be beneficial for this study to show the immunofluorescence of the molecular axis altered to have a better understanding of the alterations in situ. It is possible to correlate the microglia activated (morphological analysis of microglia) with the behavioral changes?”

Response 1: According to the Reviewer’s suggestion, in the revised version of our manuscript, we described the results with more precision. For this purpose, the figure illustrating the results of PPI test (Figure 5, panel A), showing the occurrence of two behavioural phenotypes in the offspring after MIA (responsive and non-responsive), was modified to the form with all individual data points plus the mean ± SEM. Due to the complexity and a high number of individual data, the other figures remained in the original version. We do believe that this approach allows the figures and, therefore, the data presented in them, to be readable and accessible for the reader to draw conclusions. Furthermore, we provided all precise data (the means ± SEM) in the Supplementary Materials. We added information about that fact in the revised version of manuscript as follows: “Supplementary Materials: All precise data presented as the means ± SEM are provided in the Supplementary Materials.”.

Thank you for the suggestion concerning the immunofluorescence analysis. We fully agree with the Reviewer’s comment, indicating that the implementation of immunofluorescence would allow for a better understanding of the observed changes in the CD200-CD200R axis in situ. We will consider this approach in our future experiments. Unfortunately, at this point, we are not able to perform this kind of analyses due to the lack of the experimental material from the same set of experiments to correlate them.

The possible correlation of the changes in microglial morphology with behavioural disturbances seems to be challenging. The contribution of dysfunctions in the CD200-CD200R axis to behavioural deficits, through alterations of activation and resolution of microglia, should also be reflected as changes in microglia morphology. In the brain, many studies describe the high heterogeneity of microglia activation (Colton, 2009, DOI: 10.1007/s11481-009-9164-4), which manifests differently depending not only on structures involved in the regulation of behavioural processes (e.g., hippocampus, frontal cortex, amygdala) but also on specific areas within a structure (e.g., DG, CA1, CA3). Diversified profile of activation does not always find clear confirmation in corresponding morphological changes, which makes the profile of microglia activation (anti-inflammatory or pro-inflammatory) difficult to define only based on morphological analysis. Moreover, behavioural deficits are the result of many processes, including the regulation of microglial activity and its morphology by neurotransmitters, glucocorticoids, etc. (Horchar and Wohleb, 2019, DOI: 10.1016/j.bbi.2019.06.030; Frank et al., 2019, DOI: 10.1016/j.bbi.2019.05.014). Taking into account this diversity, the correlation of the microglia morphology with behavioural changes seems to be somewhat difficult and should be undoubtedly supported by other studies, including biochemical ones, which provide the basis for assessing the phenotype of microglia changes in individual brain structures (Colton and Wilcock, 2010, DOI: 10.2174/187152710791012053).

Point 2: “Although this is a correlative study, I consider that it will pave the way for a more mechanistic follow up of the link between prenatal stress, depression, and the immune response. In other words, how the two-hit challenge induced an inflammatory response that is leading to abnormal behavior?”

Response 2: The “two-hit” hypothesis of schizophrenia assumes that a developmental insult (“first hit”, in our research – MIA) can “prime” immune cells for an event occurring later in life (“second hit”, in our research – systemic LPS injection). The proposed mechanisms by which MIA might lead to behavioural disturbances in the offspring include: 1) direct infection of the developing foetus and subsequent abnormal neural development, 2) the generation of autoantibodies by the mother that subsequently react with fetal neural tissue, and 3) alterations in cytokine production which may be an underlying component of all three mechanisms (Meyer et al., 2005, DOI: 10.1016/j.neubiorev.2004.10.012; Meyer et al., 2006; DOI: 10.1016/j.bbi.2005.11.003; Bilbo et al., 2012, DOI: 10.1016/j.yfrne.2012.08.006). Elevated levels of pro-inflammatory cytokines (IL-1β, IL-6, TNF-α), generated by both the maternal immune system (in response to subcutaneous LPS administration) as well as the foetal immune system, have been associated with abnormal foetal brain development and an increased risk of neurodevelopmental disorders. Among them increased level of IL-6 in amniotic fluid, the result of increased inflammation from endotoxin treatment (especially in late pregnancy) correlates with increased rates of brain injury. Thus, MIA, the foetal immune system and the placenta induces cytokine synthesis each of which has been linked to increased risk of schizophrenia in the offspring (Urakubo et al., 2001, DOI: 10.1016/s0920-9964(00)00032-3). Moreover, the role of long-lasting elevation in the pro-inflammatory cytokines (e.g., IL-6) during ontogenesis intensifies the neurodevelopmental brain injury. In the brain, microglia are candidate for inducing long-term changes, because they are dynamic cells that play crucial roles in the development, plasticity and immune response. Moreover, microglia have the capacity to become and remain sensitized or “primed”, so once microglia reach a final differentiated state in the brain, their replication is limited.  In response to MIA, microglia up-regulate some surface receptors but do they not over-produce proinflammatory mediators in the foetal brain. On the other hand, the “primed” inflammatory response to a subsequent challenge (“second hit”) is exaggerated. Thus primed microglia adopt a prolonged sensitized state following initial activation by MIA (Norden et al., 2013, DOI: 10.1111/j.1365-2990.2012.01306.x). Some data suggested that “primed” microglia express also changes in various function, which are at least in part regulated by neuron-microglia CX3CL1-CX3CR1 and CD200-CD200R axes. The “second hit” in adulthood induces an intensified inflammatory response (“sickness behavior”) not only in the periphery but also in the brain of adult offspring (e.g., IL-6, IL-1β) revealing changes caused by MIA, which, although not visible until now, manifest themselves in behavioural deficits, especially in those animals in which they were previously invisible. Thus, immune activation increases the risk of cognitive deficits indirectly via long-term programming of neuroimmune responses that subsequently interfere with the cellular processes of sensorimotor gating. Moreover, the combination of an underlying vulnerability (MIA) and a later-life precipitating impulse is required for the manifestation of behavioural disturbances. However, many factors (including genetic, neuroendocrine factors, neurotransmitters) determine the need for a “second hit”, which probably distinguishes in our research animals for more susceptible to deficits already after the “first hit” (responsive) and more resistant which require its action to behavioural deficits (non-responsive).

Once again, we would like to express our gratitude for the time spent in reviewing our manuscript and all the insightful remarks. We do hope that the corrections made in the revised version of the article, based on the suggestions of the Reviewers, improved the quality of our paper and that our responses clarified all of the question raised.

Reviewer 2 Report

The manuscript by Dr. Chamera et al. investigated the immune responses in the dual hit schizophrenia model. They used rats resembling schizophrenia-like symptoms by exposed to maternal immune activation and to an additional lipopolysaccharide (LPS) administration in adulthood. They demonstrated maternal immune activation rats did not reveal the alternations of M1/M2-like microglial phenotype. However, the second hit responsive rats showed upregulation of cortical IL-6 levels and behavioral disturbance. Possibly microglial priming induced behavioral disturbance. This manuscript is nicely and clearly written. Some concerns were raised regarding the experimental design, and the results.

Major Compulsory Revisions

1. The reviewer wonder whether this dual hit model by LPS is appropriate schizophrenia model. Dual hit hypothesis in schizophrenia is noticeable. Several authors induced schizophrenia symptoms by nicotine (Waterhouse et al., 2016 Disease models & mechanism) or amphetamine (Fortier J Psy Res 2004). The administration of LPS in the maternal period is one of strategy. Is additional LPS administration in adulthood common? The authors did not investigate the sufficient efficacy of this administration. This is just acute inflammation. Appropriate doses and timing of LPS administration is unknown to induce schizophrenic symptoms.

2. The authors presented many experiments performed in 3-10 times. Why did the authors perform variable number of experiments? That means that the authors tried to obtain significant differences. They also performed too much experiments, for example N =69. Too much numbers make inappropriate differences. This leads one to question the power of their observations in this study.

3. In the RT-PCR and WB experiments, the authors should present representative images of bands. In Figure 8, the authors showed Iba1 bands. However, the authors did not present other experiments. This is not fair.

Author Response

Response to Reviewer 2 Comments

First of all, we would like to express our sincere gratitude to the Reviewer for the constructive comments on the initial version of our manuscript. Please find our responses to the raised issues hereafter.

Point 1: “The reviewer wonder whether this dual hit model by LPS is appropriate schizophrenia model. Dual hit hypothesis in schizophrenia is noticeable. Several authors induced schizophrenia symptoms by nicotine (Waterhouse et al., 2016 Disease models & mechanism) or amphetamine (Fortier J Psy Res 2004). The administration of LPS in the maternal period is one of strategy. Is additional LPS administration in adulthood common? The authors did not investigate the sufficient efficacy of this administration. This is just acute inflammation. Appropriate doses and timing of LPS administration is unknown to induce schizophrenic symptoms.”

Response 1: Thank you for the comment on that issue. In the literature, many animal models have been introduced for exploring the basis of behavioural schizophrenia-like deficits as well as assessing the effectiveness of antipsychotic drugs. Among them, maternal immune activation (MIA) occupies a special place, as it resembles epidemiological observations (Estes and McAllister, 2016, DOI: 10.1126/science.aag3194; Brown and Meyer, 2018, DOI: 10.1176/appi.ajp.2018.17121311; Minakova and Warner, 2018, DOI: 10.1002/bdr2.1416). Most MIA research to date has been done in rodents and usually involved the use of LPS or poly I:C (polyinosinic:polycytidylic acid). Accordingly, the model of schizophrenia used in our study was the neurodevelopmental model based on MIA with LPS. We have shown that in Wistar rats, the subcutaneous administration of LPS to the pregnant dams starting from the 7^th day of pregnancy and repeating every second day until delivery, resulted in various immunological dysfunctions in offspring (Basta-Kaim et al., 2011, DOI: 10.1016/j.pbb.2010.12.026; 2011, DOI: 10.1016/j.ejphar.2010.09.083; 2012, DOI: 10.1016/s1734-1140(12)70937-4; 2015, DOI: 10.1016/j.neuroscience.2014.12.013), which not only preceded the occurrence of behavioural deficits but were also long-lasting and present in adulthood. The MIA model employed in our study (with some modifications) was previously investigated also by the group of Borrell (2002, DOI: 10.1016/S0893-133X(01)00360-8; Romero et al., 2007, DOI: 10.1038/sj.npp.1301292; Romero et al., 2010, DOI: 10.1038/mp.2008.44). On the other hand, it seems that MIA is not always sufficient to reveal schizophrenia-like deficits in animals. Therefore, according to the “two-hit” hypothesis, we used an additional/second stimulus, which, in the present study, was applied in the form of the acute, peripheral administration of LPS in adulthood. The treatment with LPS in adulthood as the “second hit” has been already widely presented in the literature, in the example in the research by Clark et al. (2018, DOI: 10.1016/j.pnpbp.2018.09.011) or Bilbo’s group (Williamson et al., 2011, DOI: 10.1523/JNEUROSCI.3688-11.2011; Bilbo and Schwarz, 2009, DOI: 10.3389/neuro.08.014.2009).  Both the dose and the route of administration of LPS used in our study was based on the literature data (e.g., Kupferschmid et al., 2018, DOI: 10.1177/1099800418759599). As noted by the Reviewer, LPS treatment in adulthood induces the acute inflammatory reaction (termed also as sickness behaviour), which itself is not considered as a schizophrenia model. Yet, the usage of this endotoxin as the “second hit” that triggers behavioural schizophrenia-like dysfunctions in animals predisposed by MIA seems to be justified and supported by the literature.

Point 2: “The authors presented many experiments performed in 3-10 times. Why did the authors perform variable number of experiments? That means that the authors tried to obtain significant differences. They also performed too much experiments, for example N =69. Too much numbers make inappropriate differences. This leads one to question the power of their observations in this study.”

Response 2: Thank you for bringing this aspect to our attention. We would like to explain, however, that in our manuscript, the “n” number (e.g., n = 3-10) stands for the total number of animals used in the experiment (in case of behavioural tests) or the total number of samples (in biochemical analyses) from which statistical analysis was calculated. The “n” does not represent the number of experiments performed nor the number of replicates of the experiments. In another example mentioned by the Reviewer, n = 69 does not mean that the experiment was carried out 69 times. The n = 69 stands for the total number of male offspring Sprague-Dawley rats that were used in the test investigating the exploratory activity of animals as well as in the light-dark box test, and, accordingly, for which the results are presented in the appropriate figures (Figure 2 and 3, respectively).

Point 3: “In the RT-PCR and WB experiments, the authors should present representative images of bands. In Figure 8, the authors showed Iba1 bands. However, the authors did not present other experiments. This is not fair.”

Response 3: In the research that is presented within our article, we used the Western blot technique to determine the level of IBA1 protein in the frontal cortices and hippocampi of adult male offspring, exclusively. The results obtained with the Western blot method were presented in Figure 8 and the representative immunoblots, showing IBA1 and β-actin bands, were provided in panels C and D of Figure 8. The Western blot technique was not used for any other experiment. All other protein levels were measured using ELISA kits, which do not provide the opportunity to present representative images of bands. Regarding the qRT-PCR analysis, it was used in our study to examine the mRNA expression of selected factors. Similarly to ELISA, the qRT-PCR technique does not produce images of bands. Accordingly, we applied the approach consistent with the standards adopted in many publications.

Once again, we would like to express our gratitude for the time spent in reviewing our manuscript and all the insightful remarks. We do hope that the corrections made in the revised version of the article, based on the suggestions of the Reviewers, improved the quality of our paper and that our responses clarified all of the question raised.

Reviewer 3 Report

The authors have done a tremendous amount of work to show the dysfunction of the CD200-CD200R and CX3CL1-CX3CR1 axes in microglia upon MIA and the second hit model in rats. The manuscript is nicely structured and well-written. The figures are clear and comprehensive. Also, the discussion of the data is of high quality with valuable comparisons to the literature. We have some minor and major comments to improve the manuscript. In our opinion, the major comments need to be addressed, by literature citations, additional experiments and/or statistical re-analyses, in order to accept the manuscript.

Minor comments:

• Line 24: ‘In part of the rats’
• The introduction is well-written and organized. I suggest adding one sentence with a general definition of microglial cells (around line 50).
• Line 64: You distinguish 2 microglial phenotypes in general (i.e. pro- and anti-inflammatory). Make sure that it is clear to the reader that there is actually a plethora of different microglial phenotypes.
• Line 107: The M1/M2 polarization of microglia is oversimplified and does not cover the heterogeneity of microglial activation states. The use of ‘pro- and anti-inflammatory’ microglial activation states is preferred over ‘M1/M2’. Please adapt this throughout the manuscript.
• Line 112: Remove ‘the’ in ‘the ad libitum...’
• Line 127: ‘Pregnancy every second day at 10:00 am until delivery’.
• Line 128: ‘Appropriate’ amount. Please specify the volume throughout the manuscript.
• Line 130-131: The difference between ‘behavioral experimentations’ and ‘behavioral verification’ is not clear. Both animal groups underwent different experiments on different time points, thus the second experimental set-up is not a validation of the first one.
• Line 133: Why were the behavioral experiments conducted in the ‘light-phase’ (=sleeping phase) of the rats? You discuss that your data are in contrast with other published data and refer to the species, strain and test conditions that can exert crucial effects on the outcome of behavioral data. Conducting behavioral experiments in the light-phase or in the dark-phase might indeed influence your outcomes and should be taken into account here.
• The FST is very stressful for rats. This should be taken into account when interpreting the data of the PPI test that was performed 5 days later.
• Fig 3: Distance traveled in the dark compartment is significantly decreased for the MIA group compared to the control group. This is in contrast to what is shown in figure 2. Line 560 gives a possible explanation for this contradictory result. Can you distinguish the distance traveled near the borders and in the center of the cage in the dataset of figure 2 (anxiety measure)? Were both tests performed in the light-phase?
• Line 588: A link towards involved brain regions would benefit the discussion and the overall manuscript.
• Figure 8: IBA1 bands are not clear enough to be representative. Text resolution in figure 8C and 8D should be increased.
• Please show the western blots of figure 7 in the manuscript or as supplementary figures.
• Figure 9 +10: The data are not microglia specific. Other brain-resident cells, such as astrocytes, also secrete cytokines such as IL6. Please adapt the conclusions based on these data. 

Major comments:

• The idea of studying the CD200-CD200R and CX3CL1-CX3CR1 axes in MIA is not entirely new as Bilbo’s lab studied these axes before (Williamson, 2011). It might be of interest to study this paper in particular and to discuss your data accordingly.
• Line 126: A concentration of 2mg/kg LPS was used to induce MIA. This seems a relatively high concentration compared to other papers inducing MIA in mice or rats using LPS (for review see Ronovsky (2016) or Smolders (2018)). Is there a reason/reference why this concentration was used? The same applies to the consecutive injections (every 2 days) with LPS to induce MIA starting at GD7.
• Did you test whether the pregnant dams were going through a systemic infection while injecting LPS (visual appearance, body weight, cytokine levels in serum)? This could explain the difference between MIA responsive and non-responsive offspring. If not, is it possible to do this analysis on gathered body weight data and/or serum samples? If not, can you test the immune response of the used LPS batch in naive rats? We consider this crucial for this study as it is an essential validation of the used method.
• Please check statistical data analyses. The sample size of some experiments is not high enough to perform parametric student’s t-tests or ANOVA (e.g. table 1 and 2). Two-way ANOVA is needed when a dataset with 2 independent variables is analyzed. Factorial ANOVA will increase the probability of Type I errors and is therefore not the preferred method.
• Starting from table 5: We are not convinced that the MIA group can be split into a group responsive to MIA and a group non-responsive to MIA based on the results of the PPI test. Which two response patterns were observed (figure 5)? The individual data points of this test must be shown in a scatter plot or histogram to clearly show two distinct groups. However, an underlying mechanism that distinguishes these groups and can be used to test their validity will still be lacking as no further biochemical experiments were performed on these PND100 rats. How can you be sure that the MIA was successful in the non-responsive group? Were the rats from the responsive and non-responsive groups littermates or do they come from different litters? If you pool the MIA responsive and non-responsive rats, we expect that there will be no differences anymore in PPI when you compare the control group and the complete MIA group (figure 5).
• Figure 5b: By injecting the vehicle (PND120), significant data presented in figure 5a (PND100) are lost. Do you assume that this is due to the age-dependency of the PPI test or due to the injection with the vehicle? Is the PPI test then ideal as a basis to distinguish the MIA responsive and non-responsive group?
• Figure 5b: Before-after correlation plots might clarify the PPI data and the 2 distinguished MIA groups. Positive/negative correlations between the PPI responses before and after the second hit might show a responsive and non-responsive MIA group.
• Figure 6+7: Based on these results, we are not convinced that MIA disrupts the CD200-CD200R system in the frontal cortex, as significant gene expression data are mostly found for MIA non-responsive rats.
• Conclusions on the causal relationship between the two studied axes and the sensitization of the rats can not be made based on the results presented in figure 6+7. A careful interpretation of the data is necessary. A causal link (title: through the dysfunction of..') can not be demonstrated without the use of a knockout model and/or inhibitors of the axes. 
• Figure 9+10: These data don’t support the direct involvement of only microglia. Conclusions on the involvement of microglia and the CD200-CD200R and CXCL1-CXCR1 axes should be weakened throughout the manuscript.

Author Response

Response to Reviewer 3 Comments

First of all, we would like to express our sincere gratitude to the Reviewer for the meticulous revision of our manuscript and very constructive comments. Please find our responses to the raised issues hereafter. All changes applied are coloured in green in the text of the article.

Point 1: “Line 24: ‘In part of the rats’”

Response 1: The sentence was changed and now is as follows: “The most intriguing finding was observed in the prepulse inhibition (PPI) test, where the deficit in the sensorimotor gating was age-dependent and present only in part of the rats.”

Point 2: “The introduction is well-written and organized. I suggest adding one sentence with a general definition of microglial cells (around line 50).”

Response 2: The sentence with a general definition of microglia was added in the introduction in the form: “Microglia are the immune-competent cells of mesodermal origin with the ability to transform their morphological phenotype and the capacity to migrate, proliferate and phagocytose [7].”

Point 3: “Line 64: You distinguish 2 microglial phenotypes in general (i.e. pro- and anti-inflammatory). Make sure that it is clear to the reader that there is actually a plethora of different microglial phenotypes.”

Response 3: According to the Reviewer’s suggestion, in the revised version of the manuscript, we added information about differentiation in microglial phenotypes (supported by appropriate literature references) in the form of the following sentence: “Although the heterogeneity of microglial activation states is documented [15–17], generally, two phenotypes have been highlighted: (…)”.

Point 4: “Line 107: The M1/M2 polarization of microglia is oversimplified and does not cover the heterogeneity of microglial activation states. The use of ‘pro- and anti-inflammatory’ microglial activation states is preferred over ‘M1/M2’. Please adapt this throughout the manuscript.”

Response 4: As suggested by the Reviewer, the “M1/M2” microglial activation states were replaced by “pro- and anti-inflammatory”, respectively, in the revised version of the manuscript.

Point 5: “Line 112: Remove ‘the’ in ‘the ad libitum...’”

Response 5: The sentence was changed and now is as follows: “Sprague-Dawley rats (Charles River, Sulzfeld, Germany) were maintained under standard conditions: room temperature of 23 °C, 12/12 hours light/dark cycle, lights on at 6:00 am, ad libitum access to water and food.”

Point 6: “Line 127: ‘Pregnancy every second day at 10:00 am until delivery’.”

Response 6: The sentence was rearranged and in the revised version of the manuscript is as follows: “LPS (from Escherichia coli 026:B6, Sigma-Aldrich, St. Louis, MO, USA) was dissolved in saline to obtain a concentration of 2 mg/kg in 1 ml and administrated subcutaneously to pregnant rats of the MIA group from the 7^th day (GD7) of pregnancy every second day at 10:00 am until delivery.”

Point 7: “Line 128: ‘Appropriate’ amount. Please specify the volume throughout the manuscript.”

Response 7: The volume of saline that was administrated to pregnant dams was clarified and the sentence containing that information is as follows: “Control pregnant animals were receiving the corresponding volume (1 ml/kg) of vehicle (saline).” Also, the volume of saline that was injected in adulthood was clarified and the sentence is as follows: “The control + vehicle, MIA responsive + vehicle and MIA non-responsive + vehicle groups received an intraperitoneal injection of vehicle (saline) in the corresponding volume (1 ml/kg).”

Point 8: “Line 130-131: The difference between ‘behavioral experimentations’ and ‘behavioral verification’ is not clear. Both animal groups underwent different experiments on different time points, thus the second experimental set-up is not a validation of the first one.”

Response 8: In the initial version of our manuscript the terms “behavioural experimentations” and “behavioural verification” were used as synonyms to avoid repeating “experimentations”. As pointed out by the Reviewer, this approach can be misleading, as “verification” has quite a different meaning than “experimentations” and rather should not be used synonymously. To avoid any doubts and misunderstandings, in the revised version of our manuscript, we replaced the term “verification” and now the sentence is as follows: “Before further experiments, the rats were divided into two cohorts: the 1^st (both the control and MIA groups) was used only for the behavioural examinations, while the 2^nd (both the control and MIA groups) – underwent the behavioural tests and biochemical analyses.”

Point 9: “Line 133: Why were the behavioral experiments conducted in the ‘light-phase’ (=sleeping phase) of the rats? You discuss that your data are in contrast with other published data and refer to the species, strain and test conditions that can exert crucial effects on the outcome of behavioral data. Conducting behavioral experiments in the light-phase or in the dark-phase might indeed influence your outcomes and should be taken into account here.”

Response 9: Thank you for bringing this issue to our attention. We are fully aware that the fact that behavioural experiments were conducted in the light phase could exert effects on the outcome of behavioural examinations, causing discrepancies between our data and the results presented by other authors. The main reason why the experiments were performed in that manner was the fact that we wanted to maintain uniform conditions with our previously published studies (e.g., Basta-Kaim et al., 2011, DOI: 10.1016/j.pbb.2010.12.026; Basta-Kaim et al., 2012, DOI: 10.1016/s1734-1140(12)70937-4) that were also carried out in the light phase. Besides, we encountered some practical difficulties, including the fact that in our animal facility, we had to share an animal housing room with other groups which worked also during the light phase. Additionally, we were not able to reverse the light/dark cycle due to space limitations. However, taking into account reports (Fonken et al., 2016, DOI: 10.1016/j.psyneuen.2016.01.006) indicating that stress-induced neuroinflammatory priming is a time of day dependent, this issue may be an interesting target for future research.

Point 10: “The FST is very stressful for rats. This should be taken into account when interpreting the data of the PPI test that was performed 5 days later.”

Response 10: We fully agree with the remark of the Reviewer about the FST as a stressful test for rats. Due to this fact, we performed the PPI test 5 days later to avoid any possible interference of the FST on the PPI results. Corticosterone is the main glucocorticoid involved in regulating immune reactions and stress responses. Oosterhof et al. (2016, DOI: 10.1016/j.brainres.2016.04.032) measured corticosterone levels directly before and in few time points after the FST and showed that corticosterone levels remained significantly elevated 60 minutes after the FST but already after 120 minutes – the level was normalized. This observation indicates that the effect of the FST seems to be rather of short duration. Additionally, it is worth mentioning that many authors do not consider the FST as very stressful for animals as they performed other behavioural tests in a short interval after the FST (Sideromenos et al., 2020, DOI: 10.3390/cells9041048; Ueno et al., 2020, DOI: 10.1038/s41598-020-60530-4; Jagadeesan et al., 2019, DOI: 10.15419/bmrat.v6i6.550).

Point 11: “Fig 3: Distance traveled in the dark compartment is significantly decreased for the MIA group compared to the control group. This is in contrast to what is shown in figure 2. Line 560 gives a possible explanation for this contradictory result. Can you distinguish the distance traveled near the borders and in the center of the cage in the dataset of figure 2 (anxiety measure)? Were both tests performed in the light-phase?”

Response 11: Indeed, in our study, we obtained various results concerning the distance travelled by the adult Sprague-Dawley offspring in the light-dark box test and the exploratory activity test. However, it should be taken into account that those tests differ in conditions (two-compartment cage vs. open arena) so the results showing the total distance travelled should rather not be directly compared. As pointed out by the Reviewer, another explanation of these discrepancies might be the conflict between the exploratory drive and risk avoidance of the animals. Unfortunately, at this point, we are not able to distinguish the distance travelled near the borders and in the centre of the experimental arena in the exploratory activity test (the software did not provide graphs/data representing the trajectory paths of animals). Accordingly, we are not in a position to consider the results from the exploratory activity test in relation to anxiety-like behaviour. All behavioural tests, described in our study, were performed in the light phase.

Point 12: “Line 588: A link towards involved brain regions would benefit the discussion and the overall manuscript.”

Response 12: We highly appreciate the Reviewer’s observation regarding the involvement of brain regions in the processes presented in our manuscript. To highlight this issue, in the revised version of the manuscript, we added the following sentences: “The fact that those behaviours can be mediated by different brain structures (e.g., hippocampus, frontal cortex, etc.) and inputs from more regions [80–84] should also be taken into account.”; “The functional basis of PPI is regulated by the brainstem, but it is highly modulated by cerebral [83] and hippocampal [84] inputs as well as dopamine [85] and serotonin [86,87] transmission.”

Point 13: “Figure 8: IBA1 bands are not clear enough to be representative. Text resolution in figure 8C and 8D should be increased.”

Response 13: Representative images from Western blot analyses were added as originally obtained from the membranes and were not improved in any way. Unfortunately, in the initial version of our manuscript, the quality of images in Figure 8 was lowered by the preparation process. Accordingly, in the revised version of the manuscript, both the resolution of captions in Figure 8 (panels C and D) and the quality of the images was improved by a different approach to the Figure’s preparation.

Point 14: “Please show the western blots of figure 7 in the manuscript or as supplementary figures.”

Response 14: In the research that is presented within our article, we used the Western blot technique to determine the level of IBA1 protein (Figure 8), exclusively. The Western blot technique was not used for any other experiment. All other protein levels, including those presented in Figure 7, were measured using ELISA kits.

Point 15: “Figure 9 +10: The data are not microglia specific. Other brain-resident cells, such as astrocytes, also secrete cytokines such as IL6. Please adapt the conclusions based on these data.”

Response 15: Thank you for drawing our attention to this issue. We adapt the conclusions based on the mentioned data, indicating that the reason for the observed changes could also be related to the response of other CNS cells (e.g., neurons, astrocytes, oligodendrocytes) on the acute LPS treatment in adulthood. The appropriate information was added in the form of the following sentence: “The IL-6 release is not specific to microglia exclusively [125], thus, it cannot be excluded that the increase in Il-6 level was related in part to the secretion of this cytokine by astrocytes.”

Point 16: “The idea of studying the CD200-CD200R and CX3CL1-CX3CR1 axes in MIA is not entirely new as Bilbo’s lab studied these axes before (Williamson, 2011). It might be of interest to study this paper in particular and to discuss your data accordingly.”

Response 16: Thank you for drawing our attention to the studies of the Bilbo’s group. The results obtained by these authors seem to be, indeed, particularly relevant to our research. However, as reported in the article by Williamson et al. (2011, DOI: 10.1523/JNEUROSCI.3688-11.2011), the experimental model used in this study was not related to MIA. The researchers applied neonatal bacterial infection with E. coli on PND4. Nevertheless, they investigated thoroughly the impact of neonatal infection and immune challenge with LPS (the “two-hit” hypothesis) at PND65-90 on the involvement of microglia activation in learning processes. Accordingly, we have compiled our results to the data presented by Williamson et al. (2011, DOI: 10.1523/JNEUROSCI.3688-11.2011) and we applied the appropriate changes in the part Discussion in the revised version of our manuscript. The following sentences were added in different paragraphs: “Also, the individual’s risk or resilience to neuroinflammatory disorders as a process dependent on early life experiences was suggested by Williamson et al. [94].”; “Comparatively, Williamson et al. [94] showed that early-life infection can change microglial function within the brain and exaggerate the response of microglia to a subsequent challenge, e.g., “second hit” with LPS.”.

Point 17: “Line 126: A concentration of 2mg/kg LPS was used to induce MIA. This seems a relatively high concentration compared to other papers inducing MIA in mice or rats using LPS (for review see Ronovsky (2016) or Smolders (2018)). Is there a reason/reference why this concentration was used? The same applies to the consecutive injections (every 2 days) with LPS to induce MIA starting at GD7.”

Response 17: The Borell’s group (Borell et al., 2002, DOI: 10.1016/S0893-133X(01)00360-8) was the first which used MIA model based on this paradigm of LPS injections. They administered LPS at the dose of 1 mg/kg to pregnant rats on alternate days during pregnancy. This model was later modified by Romero et al. (2006, DOI: 10.1038/sj.npp.1301292; 2008, DOI: 10.1038/mp.2008.44). The authors administered LPS subcutaneously to pregnant rats at a dose of 2 mg/kg daily beginning on the 1^st day of pregnancy and continuing until delivery. The scheme of injection with LPS every 2^nd day starting at GD7 to induce MIA was introduced to literature by our group in 2011. This model was validated many times as neurodevelopmental schizophrenia-like not only in terms of behavioural, biochemical but also pharmacological approaches (Basta-Kaim et al., 2011, DOI: 10.1016/j.ejphar.2010.09.083; 2011, DOI: 10.1016/j.pbb.2010.12.026; 2012, DOI: 10.1016/s1734-1140(12)70937-4; 2015, DOI: 10.1016/j.neuroscience.2014.12.013), which allowed its widespread acceptance in literature.

Point 18: “Did you test whether the pregnant dams were going through a systemic infection while injecting LPS (visual appearance, body weight, cytokine levels in serum)? This could explain the difference between MIA responsive and non-responsive offspring. If not, is it possible to do this analysis on gathered body weight data and/or serum samples? If not, can you test the immune response of the used LPS batch in naive rats? We consider this crucial for this study as it is an essential validation of the used method.”

Response 18: Thank you for raising this important aspect of our study. We did control the pregnant dams during the generation of MIA by repeated subcutaneous injections of LPS. We did not observe any differences in the visual appearance and the body weight in the group of females subjected to MIA compering to those parameters in the control dams. Both groups of animals were gaining weight in a standard manner. However, within this study, we did not examine the levels of cytokines in serum. The reason for that was the fact that this MIA model was previously used in our laboratory and a long-lasting immune response was showed (based on the increase of IL-1β and IL-6 levels that were measured in the serum) (data unpublished). Also, in the article of Borrell et al. (2002, DOI: 10.1016/S0893-133X(01)00360-8) acute subcutaneous administration of LPS to adult female rats significantly increased serum levels of IL-1β, IL-6 and IL-2. Additionally, a direct comparison between the reactivity of pregnant females and naïve rats after administration of LPS may be difficult due to the changed reactivity of females during pregnancy, including a shift from Th1 to Th2 responses and progressive thymus involutions. Unfortunately, at this point, we are not able to perform analyses of cytokines levels in serum of dams subjected to MIA, due to the lack of samples from those animals.

Point 19: “Please check statistical data analyses. The sample size of some experiments is not high enough to perform parametric student’s t-tests or ANOVA (e.g. table 1 and 2). Two-way ANOVA is needed when a dataset with 2 independent variables is analyzed. Factorial ANOVA will increase the probability of Type I errors and is therefore not the preferred method.”

Response 19: As mentioned by the Reviewer, the sample size in some experiments is relatively small. However, the main premise for performing a nonparametric test should not be the size of the group. To the best of our knowledge, the main rationale for choosing a nonparametric statistical test is the lack of homogeneity of variance or the abnormal distribution of the studied variable. Therefore, in our research, we used a comparison between groups based on the analyses applying Student's t-test. Regarding the one-way ANOVA analysis in our study, we used a specific type of one-way ANOVA – contrast analysis because we were not interested in all possible comparisons between groups (as in two-way ANOVA), but only those that were of biological significance to our experimental approach.

Point 20: “Starting from table 5: We are not convinced that the MIA group can be split into a group responsive to MIA and a group non-responsive to MIA based on the results of the PPI test. Which two response patterns were observed (figure 5)? The individual data points of this test must be shown in a scatter plot or histogram to clearly show two distinct groups. However, an underlying mechanism that distinguishes these groups and can be used to test their validity will still be lacking as no further biochemical experiments were performed on these PND100 rats. How can you be sure that the MIA was successful in the non-responsive group? Were the rats from the responsive and non-responsive groups littermates or do they come from different litters? If you pool the MIA responsive and non-responsive rats, we expect that there will be no differences anymore in PPI when you compare the control group and the complete MIA group (figure 5).”

Response 20: We do appreciate the Reviewer's comment on that important issue. PPI is a neurological phenomenon in which a prepulse (weaker prestimulus) inhibits the organism reaction to a subsequent strong reflex-eliciting stimulus (pulse). The reduction of the startle amplitude reflects the ability of the nervous system to temporarily adapt to a pulse when a preceding prepulse signal is given to warn the organism. Deficits of PPI reflect the inability to filter out unnecessary information, which is one of the clinical symptoms of schizophrenia. Importantly, PPI is detected in numerous species including rats and humans, which indicates its high translational properties from studies using animal models to clinical observations. In our study, we performed the PPI at PND30, PND60, PND100 and PND120 (after acute treatment with LPS). We obtained the most surprising results of the PPI at PND100. As pointed out by the Reviewer, when all animals were analysed together (responsive and non-responsive), we did not observe significant changes in the PPI when compared to the control offspring. However, when we examined the data thoroughly we were able to distinguish two kinds of responses, which were later categorized into responsive and non-responsive groups. To clearly present this phenomenon, in the revised version of our manuscript, Figure 5a was rearranged to the plot showing all individual data points. The only report on the partial response to MIA reflected in the PPI we found was in the article by Mattei et al. (2017, DOI: 10.1038/tp.2017.80), where the authors stated that out of 16 mice that were prenatally subjected to MIA, 7 showed a robust deficit in sensorimotor gating. Unfortunately, in the study, the researchers investigated only the animals with the deficit in the PPI. Contrary, in our experiments, we decided to examine the offspring from both groups. As another aim of our study was to analyse the “two-hit” hypothesis of schizophrenia, we did not perform additional biochemical experiments at PND100. After obtaining the results from the PPI test (which indicated that MIA differentiated the animal's response in terms of positive deficits), the interval of 20 days was kept (based on the scheme used in our previous studies using chronic administration of antipsychotic drugs), the “second hit” with LPS was applied and the second PPI test was performed.

In our study, all pregnant female rats were subjected to MIA in the same manner. Also, the dams from the control group were treated correspondingly. At PND21, the male offspring were separated from dams and transferred to new cages (no two animals from one litter were in the same cage). The animals were housed in groups of 5-6 per cage. Accordingly, we are sure that MIA was successful in both responsive and non-responsive groups, as the offspring in those groups came not from one but different litters. In our opinion, the two response patterns in the PPI were not a result of the experimental procedure, especially taking into account also the fact that in other behavioural tests, we did not observe such a phenomenon.

Point 21: “Figure 5b: By injecting the vehicle (PND120), significant data presented in figure 5a (PND100) are lost. Do you assume that this is due to the age-dependency of the PPI test or due to the injection with the vehicle? Is the PPI test then ideal as a basis to distinguish the MIA responsive and non-responsive group?”

Response 21: Regarding the data presented in Figure 5a and Figure 5b, we suppose that there is one main reason why the significance of the differences was lost in Figure 5b. It is important to remember that the groups at PND120 were obtained by dividing the groups at PND100 (e.g., from the group of MIA responsive offspring, we obtained two groups: MIA responsive + vehicle and MIA responsive + LPS). Accordingly, the number of animals in the groups at PND120 (Figure 5b) was lower than in the case of PND100 (Figure 5a). It is possible then that the number of animals in particular groups was not high enough to demonstrate statistical significance. However, referring to this limitation, we want to highlight that the effect of the “second hit” with LPS in adulthood on the non-responsive offspring was so pronounced that we were able to observe the changes in PPI even with the smaller number of animals in groups. We assumed that this observation was much more important from the biological point of view.

The PPI test is commonly used as an indicator of the disturbed sensorimotor gating which is considered as one of the core behavioural features observed in schizophrenia, both for patients (Braff et al., 1992, DOI: 10.1001/archpsyc.1992.01820030038005; Moriwaki et al., 2009, DOI: 10.1016/j.neures.2009.07.009) and animal models of the disease (Borrell et al., 2002, DOI: 10.1016/S0893-133X(01)00360-8; Khan and Powell, 2018, DOI: 10.1016/j.schres.2017.10.009). The other tests assessing schizophrenia-like deficits in animals are used to examine the different profile of changes, e.g., social interaction test allows to investigate negative symptoms, attention switching test explores cognitive characteristics, etc. Accordingly, the importance of the PPI test for studying positive symptoms of schizophrenia in animal models of this disease is important, yet, it needs to be taken into account that PPI deficits have been also reported in autism. Therefore, to date, no other test investigating sensorimotor gating abnormalities in animals is available.

Point 22: “Figure 5b: Before-after correlation plots might clarify the PPI data and the 2 distinguished MIA groups. Positive/negative correlations between the PPI responses before and after the second hit might show a responsive and non-responsive MIA group.”

Response 22: Considering the data presented in Figure 5B, calculating the correlation between the PPI before and after acute LPS administration in adulthood would not, in our opinion, bring new information. The reason behind this approach is the fact that the PPI threshold, based on which the division into responsive and non-responsive groups, was established arbitrarily. In more detail, the categorization was carried out referring to the average result obtained for the control group. Therefore, the PPI response is the measure of the impact of LPS treatment. Additionally, as the Reviewer noted earlier, the number of animals in groups is relatively small and calculating any correlations with such numbers would not bring new valuable information to our data.

Point 23: “Figure 6+7: Based on these results, we are not convinced that MIA disrupts the CD200-CD200R system in the frontal cortex, as significant gene expression data are mostly found for MIA non-responsive rats.”

Response 23: Thank you for pointing out this important detail. We demonstrated that MIA decreased the level of CD200R (which determines the biological effect exerted by CD200) in the frontal cortex of both responsive and non-responsive animals, while the effect of MIA on the expression of Cd200 and Cd200r was observed, as the Reviewer noted, only in the non-responsive offspring. Therefore, in the revised version of the manuscript, we clarified this observation both in the section Discussion [“We are the first to report that MIA disrupted the CD200-CD200R system in the frontal cortex of adult Sprague-Dawley offspring (mostly from the non-responsive group), while the changes in the CX3CL1-CX3CR1 proteins were less evident.”] as well as in the Abstract (“Simultaneously, based on the results of the biochemical studies, MIA disrupted mainly the CD200-CD200R system, while the changes of CX3CL1-CX3CR1 axis were less evident in the frontal cortex of adult non-responsive offspring.”).

Point 24: “Conclusions on the causal relationship between the two studied axes and the sensitization of the rats can not be made based on the results presented in figure 6+7. A careful interpretation of the data is necessary. A causal link (title: through the dysfunction of..') can not be demonstrated without the use of a knockout model and/or inhibitors of the axes.”

Response 24: We are aware that our research has some limitations and requires further studies to fully explain the discussed subject. The use of transgenic animal models can be considered, as well as conducting functional studies based especially on the modulation of CD200-CD200R axis in neurodevelopmental models of schizophrenia. However, models with full transgenic elimination from birth need to be interpreted with caution (Kopec et al., 2019, DOI: 10.1016/j.tins.2019.02.005). To the best of our knowledge, to date, there is no study on the effect of MIA in transgenic animals, reported in the literature. Moreover, so far only mouse models are available, and there are no models with double Cx3cr1 and Cd200r knockout, which can be important in mutual modulation of both axes. In the case of functional study, the question remains about the route and the age after MIA when the modulators of the tested axes should be applied. All these points are undoubtedly interesting but require further experiments. Nevertheless, due to the doubts about the causal relationship between examined axes and the sensitization of rats, the title was changed and now is as follows: “Maternal immune activation sensitizes male offspring rats to lipopolysaccharide-induced microglial deficits involving the dysfunction of CD200-CD200R and CX3CL1-CX3CR1 systems.”

Point 25: “Figure 9+10: These data don’t support the direct involvement of only microglia. Conclusions on the involvement of microglia and the CD200-CD200R and CXCL1-CXCR1 axes should be weakened throughout the manuscript.”

Response 25: Thank you for the comment concerning the conclusions of our study. The data presented in Figures 9 and 10 indicate the effect of MIA and additional acute challenge with LPS on the gene expression of pro- (e.g., MhcII, Cd40) and anti-inflammatory (e.g., Arg1) surface antigens, which are considered markers of microglial cells (Walker and Lue, 2015, DOI: 10.1186/s13195-015-0139-9; Tang and Le, 2016, DOI: 10.1007/s12035-014-9070-5). We agree that other cell types (neurons, astrocytes, microglia and endothelial cells) in the CNS can be an essential source of some of these factors, e.g., IL-6, upon proper stimuli, can be secreted vastly by astrocytes (Erta et al., 2012, DOI: 10.7150/ijbs.4679). However, in our opinion, those results combined with the data concerning the CD200-CD200R and CX3CL1-CX3CR1 axes might implicate that mostly microglial cells played a crucial role in changes observed in our research. The main reason for this conclusion is the fact that the receptors (CD200R and CX3CR1) are present on microglia and not on astrocytes. Yet, further experiments on the involvement of other cell types within the CNS could bring some new information.

Once again, thank you for the time spent in reviewing our manuscript and all the constructive comments. We do hope that the changes which we applied in the revised version of the article, based on the suggestions of the Reviewers, improved the quality of our paper and that our responses clarified all of the question raised.

Round 2

Reviewer 2 Report

The authors did not answer the reviewer's comments.

Author Response

Response to Reviewer 2 Comments

Once again, we are very grateful to the Reviewer for the initial feedback, concerning our manuscript. At this point, we applied changes and suggestions of the Reviewer from the first round of revision, as well as issues that were raised by other Reviewers. As we are fully aware of the positive impact of the Reviewer’s comments, we are willing to implement any further modifications, if necessary. However, to be able to do that, we would highly appreciate a more detailed comment from the Reviewer. We would be grateful for specifying which comment of the Reviewer we did not answer or to which issue our response was not addressed adequately.

Reviewer 3 Report

Most of our discussion points were addressed adequately in the ‘response to reviewer’s comments’. It will be beneficial to the reader to address the most important discussion points in the manuscript as well. For example, support your methodology by referring to other studies using a similar methodology.

For future studies, we highly recommend analyzing the cytokine levels in serum of dams subjected to MIA, as researchers have noticed that LPS-injection does not always induce immune activation. Unfortunately, this is not possible for this study due to the lack of serum samples. Please add the unpublished validation of your MIA model (the increase of IL-1β and IL-6 levels that were measured in serum) in supplementary.

Some of our most important concerns are not or not completely addressed. In our opinion, the manuscript cannot be published in Cells without addressing or elaborating upon these issues:

1. Although theoretically possible, testing the distribution of the studied variable is not reliable when the sample size is too small (for example when n = 3 or 4). If there are only three samples, it may be difficult to ensure that these are (not) normally distributed (See Kim and Park, 2019), and as a consequence, non-parametric statistics should be used.
2. The subcategorization of MIA responsive and MIA non-responsive rats is insufficiently supported to be the basis of this manuscript. Preferably, the method of how MIA rats are subdivided into a responsive and non-responsive group must be shown. We believe that a correlation study between the PPI at PND100 and the PPI at PND120 could reveal differences between both groups. Even if no statistical solidification for a correlation can be found, a graphical representation of the criteria to discriminate responsive and non-responsive cases will still be valuable information to the reader and justify the validity of the criteria. In case the cut-off value was arbitrarily (and without clear method) determined, this must be clearly explained in the manuscript. Otherwise, a risk/suspicion of data selection might be suspected, which is unacceptable for publication.

Author Response

Response to Reviewer 3 Comments

Once again, we are grateful to the Reviewer for the feedback, concerning our manuscript. Please find our responses to the raised issues hereafter. All changes applied are coloured in green in the text of the article.

Point 1: “For example, support your methodology by referring to other studies using a similar methodology.”

Response 1: According to the Reviewer’s suggestion, in the first round of revision, references to other studies using a similar methodology were added. In line 131, the references concerning the MIA model have been included (literature positions from 35 to 37). All behavioural tests already had specified references that had been taken into account in our study (e.g., exploratory activity test – previously reference 38, now 40; light-dark box test – previously reference 39, now 41, etc.). Nevertheless, we also added the appropriate references in the paragraph concerning an additional immune activation with LPS in adulthood (point 2.2.2., line 161, references 38 and 39). Other techniques (e.g., ELISA) were performed based on the procedures provided by the manufacturers of the kits and the information about that fact was already included in our manuscript (e.g., point 2.4.4., line 276).

Point 2: “For future studies, we highly recommend analyzing the cytokine levels in serum of dams subjected to MIA, as researchers have noticed that LPS-injection does not always induce immune activation. Unfortunately, this is not possible for this study due to the lack of serum samples. Please add the unpublished validation of your MIA model (the increase of IL-1β and IL-6 levels that were measured in serum) in supplementary.

Response 2: Thank you for the suggestion concerning the cytokine levels in serum of MIA-subjected dams. In our future research, we will surely implement this approach. At this point, as we wrote in the first response, our previous data showed an increase in the level of IL-1β and IL-6 in the serum of pregnant rat females treated with LPS (the MIA model). However, it should be emphasized that those measurements were made at the time of implementation of the model (2010-2012), so about 10 years ago. To perform those experiments, we used rats from the Institute of Pharmacology’s breeding facility. In the study reported in this manuscript, we used rats purchased at Charles River (Sulzfeld, Germany). Considering that the MIA model applied in our study was repeatedly positively validated in terms of behavioural (Basta-Kaim et al., 2011, DOI: 10.1016/j.pbb.2010.12.026; 2012, DOI: 10.1016/s1734-1140(12)70937-4; 2015, DOI: 10.1016/j.neuroscience.2014.12.013), biochemical (Basta-Kaim et al., 2012, DOI: 10.1016/s1734-1140(12)70937-4; 2011, DOI: 10.1016/j.pbb.2010.12.026; 2015, DOI: 10.1016/j.neuroscience.2014.12.013) and pharmacological (Basta-Kaim et al., 2011, DOI: 10.1016/j.ejphar.2010.09.083) aspects in the offspring, no additional evaluation of immunological parameters or maternal immune activation in females was performed, either in the CNS or the periphery (serum, spleen cells, thymocytes) as further validation of the model. We believe that adding the above-mentioned results, coming from different experiments performed some time ago and not on the same animals that those used in the study reported in our manuscript would not be a fully justified approach. Nevertheless, as we are aware of the importance of the MIA model’s verification in terms of cytokine levels in serum of dams, we added the appropriate sentence (line 696) in the discussion with the reference to the data published by Borrell et al. (2002, DOI: 10.1016/S0893-133X(01)00360-8), where serum levels of IL-1β, IL-6 and IL-2, as well as of corticosterone were significantly increased in a time course of two hours after subcutaneous administration of LPS (showing that the pregnant rats were submitted to the well-documented LPS-induced immune challenge).

Point 3: “Although theoretically possible, testing the distribution of the studied variable is not reliable when the sample size is too small (for example when n = 3 or 4). If there are only three samples, it may be difficult to ensure that these are (not) normally distributed (See Kim and Park, 2019), and as a consequence, non-parametric statistics should be used.”

Response 3: Thank you for the detailed explanation of the issue concerning statistics in our manuscript. In our study, only in the analysis of IBA1 protein level, using the Western Blot technique, the n value was relatively low for all examined groups (n = 4). Accordingly, as suggested by the Reviewer, in the revised version of the manuscript, we tested the results applying the non-parametric statistic, to be precise Kruskal-Wallis test (all information about that fact were applied in line 317). In any other analysis, the n value for most of the groups was higher, then, the contrast analysis was used (e.g., only one group had n = 3 and other groups had n = 7-9). Moreover, we would like to underline that our goal was to examine specific comparisons (e.g., control vehicle vs. MIA responsive vehicle), which, in our opinion, were crucial from the biological perspective. We did not make comparisons of all possible variants as some were with no relevance to our study (e.g., control vehicle vs. MIA responsive LPS). As far as we know, the contrast analysis is the only statistic allowing to compare exact groups. Additionally, often in the comparisons were the n value was relatively low the results were not significantly different, even using the contrast analysis. Regarding the assumptions of the (non)parametric tests, including the reliability of distribution in small-sized groups, Monte-Carlo simulation showed that failure to meet test’s assumptions has no impact on the test reliability. In the article of Haidous and Sawilowsky (2013, DOI: 10.12691/ajams-1-5-4), there was no evidence that power and the accuracy of classical parametric and nonparametric alternatives of the test were different when the assumption of normality was not met or the sample size was small. Although type I error and power properties of statistics can be calculated for data drawn from classical theoretical distributions for asymptotic conditions, real data often do not have such distributions.

Point 4: “The subcategorization of MIA responsive and MIA non-responsive rats is insufficiently supported to be the basis of this manuscript. Preferably, the method of how MIA rats are subdivided into a responsive and non-responsive group must be shown. We believe that a correlation study between the PPI at PND100 and the PPI at PND120 could reveal differences between both groups. Even if no statistical solidification for a correlation can be found, a graphical representation of the criteria to discriminate responsive and non-responsive cases will still be valuable information to the reader and justify the validity of the criteria. In case the cut-off value was arbitrarily (and without clear method) determined, this must be clearly explained in the manuscript. Otherwise, a risk/suspicion of data selection might be suspected, which is unacceptable for publication.”

Response 4: Following the Reviewer's suggestion, the detailed description of the subcategorization of the MIA offspring to the responsive and non-responsive groups was added in the section “Materials and methods” point 2.2.5. in the form of: “At PND100, the offspring from the MIA group were divided into two categories: MIA responsive (with the deficit in PPI) and MIA non-responsive (without the deficit). The subcategorization was done based on the PPI results calculated with the AVGs for the 75 dB prepulse. First, the mean response in PPI at 75 dB prepulse was calculated for the control group. Then, the MIA offspring were divided in such a way that all animals with %PPI lower than the average response of the control rats were categorized as “responsive” and all animals with %PPI higher than the mean for the control group were assigned to “non-responsive” group. Categories obtained for 75 dB prepulse were maintained for the remaining prepulse intensities (70 and 80 dB).”