M&B Inaugural Journal Club Topic

[Moonbear]

Okay everyone, I'm ready to get this puppy off the ground! Official discussion of this paper will begin Sat. Jan 28. Until then, here is the link to the full article, the citation, and a bit of introduction to the topic.

Messager S, Chatzidaki EE, Ma D, Hendrick AG, Zahn D, Dixon J, Thresher RR, Malinge I, Lomet D, Carlton MB, Colledge WH, Caraty A, Aparicio SA. Kisspeptin directly stimulates gonadotropin-releasing hormone release via G protein-coupled receptor 54. Proc Natl Acad Sci U S A. 2005 Feb 1;102(5):1761-6.

http://www.pnas.org/cgi/content/full/102/5/1761

Note: you can also download a PDF version from the site I've linked to if you prefer that format for reading.

By way of introduction to the topic, as many of you may be aware, my field of research is in reproductive neuroendocrinology...yes, it's a mouthful, and basically means I study how the brain controls the production and release of hormones involved in reproductive function. Over the past year or two, a recently discovered protein, known as kisspeptin, has become a hot topic in the field.

As the article you will read indicates, the current hypothesis is that kisspeptin, acting through its receptor, GPCR 54, is required for the release of gonadotropin-releasing hormone (GnRH) from terminals of neurons that are located in a structure at the base of the part of the hypothalamus, called the median eminence, and into the blood vessels that carry neuroendocrine hormones to the anterior pituitary (pituitary portal vessels).

A few things that should be noted with regard to this system, which I hope will help you understand what you're reading better:

Anatomy - GnRH neuronal perikarya (cell bodies) are scattered throughout the hypothalamus, which makes them more challenging to study than neurons located in restricted nuclei of the brain, because you cannot easily target the cell bodies with drugs, or to even lesion them. The axons, however, converge so that the terminals of these neurons are mainly located in the median eminence. Some scattered terminals are present in areas contacting the brain ventricles and more caudally (toward the tail), in parts of the brain involved in reproductive behavior, but we don't know if these terminals are important for reproductive behavior (yet).

Pattern of release - GnRH, during most of the estrous/menstrual cycle in females, and under normal physiological conditions in males, is released in a pulsatile pattern. The frequency at which these pulses are released is dependent on the concentrations of progesterone (more progesterone, slower pulses) in females, and testosterone in males, and the amplitude of these pulses is dependent on the concentrations of estradiol (more estradiol, lower amplitude) in both sexes. This is part of the normal, negative feedback control of the hypothalamo-pituitary-gonadal axis, where increased steroid hormones produced by the gonads regulate the secretory patterns of GnRH. GnRH in turn regulates the release of luteinizing hormone (LH) and follicle stimulating hormone (FSH) from the pituitary into peripheral blood circulation, which reaches the gonads. The pattern of LH and FSH secretion alters the rate of secretion of progesterone, estradiol and testosterone from the gonads, etc.

Prior to ovulation, the pattern of secretion of these hormones changes dramatically, and the mechanism remains elusive. Rather than pulses of GnRH and LH being the predominant pattern of release, and negative feedback of estradiol on GnRH and LH being the predominant feedback mechanism, there is a shift to what we call a positive feedback loop, where increasing concentrations of estradiol result in increasing release of GnRH, which increases LH, until we observe a sustained release of very high concentrations of GnRH known as the GnRH surge. During this time, there may still be underlying pulse release, but pulses are not detectable.

Okay, I think that's a bit of a start, and will give everyone something to chew on as they delve into the article.

 

[Monique]

Nice, I'm in!

 

[selfAdjoint]

I'm in too.

First of many questions. I'm guessing that gpr54-/- mice are without GPR54; is that right? What does the -/- suffix mean?

 

[neurocomp2003]

aaah soo much biology...

 

[Moonbear]

I'm in too.
First of many questions. I'm guessing that gpr54-/- mice are without GPR54; is that right? What does the -/- suffix mean?

Yes, they are knockout mice, meaning that gene has been deleted.

If anyone else has questions on details like that, which they need answered to understand the paper, feel free to ask those now, and I'll try to answer when I can (or if some of them aren't just details, I may defer them until I'm ready to get into the meat of the discussion). We'll get into the more in-depth discussion starting Saturday.

 

[Monique]

Just to elaborate a little further: the -/- means both copies of the gene have been deleted, +/+ would be wildtype and +/- would mean one functional copy is still present, the other deleted (or non-functional actually).

 

[DocToxyn]

...and to elaborate even further, you may run across the term "null" referring to the knockout, thus the mouse line in question could be referred to as the "gpr54-null" mouse line. Another set of terms that are somtimes used are "heterozygous" (or hetero, or het), which is the +/- mutation, and "homozygous" which can refer to the +/+ wildtype or the -/-knockout.

 

[selfAdjoint]

Thank you Moonbear, Monique, and DoxToxyn! I guessed that the -/- meant both copies of the gene were deleted and I am happy to see that I was right. I am continuing to study the paper.

 

[Moonbear]

Okay everyone, the topic is officially open for discussion. My schedule had an unexpected twist yesterday, so I'm not as ready to discuss this as I had hoped to be by today, and need to run into the lab for a bit today (a carryover of yesterday's insanity), so can't dedicate as much of my day to starting this as planned. But, I WILL be around by late afternoon/evening to get going. So, I'm just posting now to let you all know this topic is open if people have burning questions that can't wait for me to come back this evening to start up the discussion, and so you know why I won't be around to continue discussion until evening.

 

[Monique]

And I'm working in the lab all this weekend (with a cold) but I'll participate when I can

 

[Moonbear]

Feel better Monique!

Actually, Monique, being the geneticist, you can probably answer a quick question I had. In the introduction of the article, the authors state that the GPR54 protein deficient mice "display phenocopy syndromes..." I've never heard that term before, phenocopy syndromes. Is that just another way of saying they exhibit an abnormal phenotype, or does it have a more precise meaning?

Okay, that aside, let's delve in!

I'll start here with the overall objectives of the work in this paper and highlights in their introduction. Then I'll continue in a separate post with my thoughts on the methods and results. I'll hold off on their discussion and conclusions until we've discussed the methods and results for ourselves, because I view it as a better learning experience to form our own opinions on what the results mean first, then see if we agree with the authors' conclusions, and if we have any discrepancies, we can see if their discussion convinces us we've missed something, or if we have to agree to disagree with them.

This group has taken a 3-pronged approach to addressing the overall hypothesis that kisspeptin stimulates GnRH release via GPR54 (gotta love it when the hypothesis is clearly stated right in the title).

The first approach was anatomical. This was to first confirm that deficits in GnRH release in GPR54 knockout mice are not due to a defect in GnRH neuronal migration, and that GPR54 is present in GnRH neurons. The anatomical evidence presented is what I consider the weakest element of this paper, as I'll explain later.

The second approach was to use GPR54 knockout mice (with wild types as controls), to test whether kisspeptin-induced release of GnRH is blocked in mice lacking GPR54. This was a fairly straightforward approach, and the evidence presented is convincing.

The third approach was to use a different animal model, the sheep, to test the effects of kisspeptin on GnRH and LH release. Again, this is a fairly straightforward (although technically challenging) approach, with reasonably convincing evidence. There are a few details in the results that raise some questions, but the alternative approach to answer those questions is extremely challenging (although, the author on this paper who conducted that part of the experiment is the one person who actually has the expertise and lab set-up to do it, so it's somewhat disappointing to see he didn't take it the step further). Again, I'll get into the details later when I discuss the results.

As was presented within the introduction of this paper, GPR54 has a 45% homology with galanin receptors but does not seem to actually be bound by galanin. Galanin and galanin-like peptides have also been implicated as mediators of GnRH release, so this was an important thing for the authors to note for us so we know the effects observed are not due to an interaction with a previously identified neurotransmitter.

Mice and humans with deficiencies in the GPR54 protein have a syndrome known as hypogonadotropic hypogonadism. What that big mouthful of words means is that a deficiency of the hormones LH and FSH leaves them infertile. The previous studies cited by these authors also demonstrate that if you give exogenous GnRH, the pituitary can respond with normal production of LH and FSH, indicating that the source of the defect is at the hypothalamus, not the pituitary. This is the rationale for studying the role of kisspeptin and GPR54 at the level of the hypothalamus, and specifically on GnRH neurons.

 

[Moonbear]

Now on to the methods and results.

First, for the anatomical component, the methods used were fairly standard. The only thing that's a tad uncommon is that they used milk powder as the blocking agent when performing immunocytochemistry for GnRH. Usually this isn't quite as clean as using normal serum from the species your secondary antibody was generated in (their details were a bit skimpy there too, which given some of my concerns when looking at the figures, I'd like to have seen those details fully provided).

Okay, there are a few things I see in Fig 1 that cause me concern. First, do you notice that bright spot in panel A that's off-center? Whoever captured this image doesn't know how to use their microscope properly. None of the GnRH neurons looks clearly in focus (the antibody they used is a very robust one and well-characterized in numerous species, so there is no reason they shouldn't have gotten crisp, darkly-stained GnRH neurons). This is of course worsened by the improper adjustment of the condenser, which is what gives that bright spot and grainy background. I'm not convinced that the GnRH neurons are in the same focal plane as the sharper, more in-focus beta-galactosidase staining. Indeed, when you look at the higher magnification image in Panel C, the beta-gal staining looks like it's off to the right of the GnRH neuron, not at all colocalized with it. Another thing that caught my attention is that Panel B is just a higher power magnification of a part of Panel A, but the legend does not indicate this. Overall, this raises a lot of concerns for me that the examples shown do not demonstrate colocalization, and the lack of skill of the person doing the microscopy does not inspire a good deal of confidence that they correctly identified colocalization. My other big concern with this one is that within the results, the authors state that 20 GnRH cell bodies were clearly identified. They included EVERY section from the preoptic area through the median eminence, which means they should have had the majority of GnRH neurons in these mice present and accounted for. Usually, there are about 300 - 500 GnRH neurons in a mouse. So, either they have really done a poor job in processing this tissue, or there is a severe deficit in the numbers of GnRH neurons in these mice. There is no way to tell this from what is presented.

So...this is the point where I'd say, if I had been the reviewer of this article, this first part would have raised a big red flag, and it's disappointing that the reviewers did not catch this and demand that tissue be reanalyzed/reprocessed and at the least, that the figures show neurons that are in focus.

In figure 2, you see better examples of GnRH neurons that are actually in focus. However, since they are making the claim that the overall distribution does not differ between wild type and GPR54 -/- mice, it would have been much more convincing to see a series of lower magnification images that give a better overall view of the locations of the neurons, or in lieu of that, drawings mapping out their locations. This might have helped clear up questions of whether the small number of GnRH neurons identified in the colocalization study was a result of the processing, microscopy or an actual difference in the numbers of neurons between wild type and mutant mice.

When comparing panels C and D in figure 2, one thing to be aware of is that they don't look entirely the same because a bit of the pituitary, called the pars tuberalis, is still attached at the base of the median eminence in Panel D. This is not a problem of any kind, sometimes more of that stays attached, and sometimes less, when you remove the brain. I just wanted to point it out in case anyone is noticing that the GnRH fibers go all the way to the bottom of the median eminence in Panel C, but look like they are all located in the middle of it in Panel D. (Oh yeah, the median eminence is that bit stretched across the bottom, underneath the big space, which is the ventricle...I'll locate an image of my own to attach later and stick in some arrows to help with orientation).

Fortunately, the rest of the paper gets better, so while this may leave me uncertain whether GPR54 is really on GnRH neurons and not acting via some other indirect route, the pharmacological evidence is convincing that kisspeptin and GPR54 are important for GnRH release.

I think I'll keep this in bite-size chunks, so will discuss the pharmacological part of the study in a separate post.

 

[Moonbear]

Now, for the pharmacological evidence:

Mice, being the tiny things that they are, don't have a whole lot of blood. So, it's not possible to measure patterns of hormone release in them over time, because it would require more blood than they have to donate. So, you can only do a single time-point sample, which is what was done by these authors to measure LH and FSH in mutant and wild type mice that were injected with either vehicle (control) or kisspeptin (in the same vehicle...in this case, the vehicle was PBS). This is a very standard experimental design.

In Fig. 3, Panel A shows LH concentrations in the mice, and Panel B shows FSH concentrations. The first two bars in each graph are the mutant mice (the ones lacking GPR54). It's pretty clear that there is no significant difference between the vehicle controls and the kisspeptin treated mice. In contrast, the wild-type mice have a significant increase in LH after kisspeptin treatment (the wild type mice are two bars furthest to the right in the graphs, with the last bar being the kisspeptin treated ones). Something I noticed that was not mentioned by the authors is that it appears that the wild-type mice already have slightly higher FSH than the knockout mice, even when just given the vehicle injections. There's no way to know if this was a statistically significant difference without more information from the authors, but this actually adds an element of confirmation to their findings that the mice with a functioning receptor should have higher levels of these gonadotropins than the hypogonadotropic knockout mice.

Okay, I'm going to take a break from writing and will return to present the sheep results later. Feel free to begin commentary on the parts presented so far...if you agree, disagree, see things differently, feel hopelessly confused, etc.

 

[Monique]

Feel better Monique!

Thank you, I want to stay in bed all day, but I have to go out into the cold I'll stay in bed a few hours longer and read the article (I was just having a quick sip of tea in the morning).

Actually, Monique, being the geneticist, you can probably answer a quick question I had. In the introduction of the article, the authors state that the GPR54 protein deficient mice "display phenocopy syndromes..." I've never heard that term before, phenocopy syndromes. Is that just another way of saying they exhibit an abnormal phenotype, or does it have a more precise meaning?

Well, I did not read the article yet, but a phenocopy would be a phenotype produced by environmental factors, which cannot be distinguished from the genetic phenotype.

A simplified example would be a study into the genetic occurance of blonde hair in the Japanese population. You should get 100% black, but you don't. That's because of phenocopies: some of the Japanese dye their hair blonde.

They mean with "humans and mice deficient for GPR54 protein display phenocopy syndromes characterized as isolated hypogonadotrophic hypogonadism", that the genetic deficiency of GPR54 leads to the same syndrome as environmentally caused hypogonadotrophic hypogonadism?

 

[Moonbear]

They mean with "humans and mice deficient for GPR54 protein display phenocopy syndromes characterized as isolated hypogonadotrophic hypogonadism", that the genetic deficiency of GPR54 leads to the same syndrome as environmentally caused hypogonadotrophic hypogonadism?

That would be a strange thing to say, because, although there can be many reasons for hypogonadotropic hypogonadism, it's not usually thought of as environmentally caused (except in a situation like a head trauma that damages the connection between the brain and pituitary, but that's a pretty rare event). But based on your explanation, it may be that the deficiency has not been traced to a genetic deletion/mutation of the GPR54 gene in those humans and mice, but has only been confirmed at the protein expression level.

If that's the case, and it's a naturally occurring deficiency, I'd be interested in learning why the protein is not being expressed normally, despite the presence of the gene encoding it. There are two reasons this would be interesting to me:

1) IF this is a common cause of hypogonadotropic hypogonadism (big IF there...I'll check the references cited on that statement to find out if they say anything about the incidence of this deficiency) , understanding the mechanism behind this reduced/absent receptor expression may provide insight to a therapeutic approach for treating infertility in these patients.

2) It might lead to identifying regulators of normal receptor expression. Since GnRH is not released in a continuous fashion, but in pulses, it's possible that either GPR54 or kisspeptin expression patterns underlie that pulsatile secretion (I'd also be curious to know more about the time it takes for this particular receptor to be internalized and either recycled to the membrane, or for new receptor to reach the membrane; if the timing is similar to the interpulse interval, that could be an exciting mechanism for GnRH pulsatile release).

[As an aside: for those reading along who have wondered how scientists come up with ideas for new experiments, those questions I just asked in 1 and 2 are the beginning of the process. This is why it's good to discuss articles, because the process of answering each other's questions can bring up new questions that have not been answered. Of course, if I was interested in pursuing those specific questions, I would need to now go back to the literature with my questions and search with that focus in mind to determine if someone has already done it. It would also be wise to contact the author on this paper who I know well enough to entrust with such an idea to find out if they're already working on it so I don't waste time doing something that someone else has already begun work on. At this point, though, I know NOTHING about the regulation of the GPR54 receptor, so it would be a complete shot in the dark for me, so not a very productive use of resources for me to pursue, but I'll keep such questions in mind in case my research path intersects with that one eventually.]

I need to spend a few hours in the lab again, then will return to pick up the discussion where I left off.

 

[shruth]

There are a few things in the paper that i did not understand.
1. How is that B- galactosidase activity correlates spatially with GPR54 location?
2. Why do you think two mice yielded only 20 neurons? Why did they not use more samples? Is it because the knockout mice are very costly or something like that?

 

[Moonbear]

There are a few things in the paper that i did not understand.
1. How is that B- galactosidase activity correlates spatially with GPR54 location?

That's due to the way they create the knockout mouse. When they introduce the mutation that deletes the GPR54 gene, they insert a beta-galactosidase gene. Monique can probably explain in more detail how that's done.

2. Why do you think two mice yielded only 20 neurons? Why did they not use more samples? Is it because the knockout mice are very costly or something like that?

Knockout mice are very costly, but I also raised the same question regarding such a small number of neurons. This is a suspiciously low number for a mouse, and one of the reasons that particular part of the results seems weak to me.

 

[hypnagogue]

Thanks for the thorough treatment, Moonbear. I'm definitely getting a lot more out of this now than I did right after finishing the paper.

Some questions and comments on what you've written so far:

My other big concern with this one is that within the results, the authors state that 20 GnRH cell bodies were clearly identified. They included EVERY section from the preoptic area through the median eminence, which means they should have had the majority of GnRH neurons in these mice present and accounted for. Usually, there are about 300 - 500 GnRH neurons in a mouse. So, either they have really done a poor job in processing this tissue, or there is a severe deficit in the numbers of GnRH neurons in these mice. There is no way to tell this from what is presented.

In figure 2, you see better examples of GnRH neurons that are actually in focus. However, since they are making the claim that the overall distribution does not differ between wild type and GPR54 -/- mice, it would have been much more convincing to see a series of lower magnification images that give a better overall view of the locations of the neurons, or in lieu of that, drawings mapping out their locations. This might have helped clear up questions of whether the small number of GnRH neurons identified in the colocalization study was a result of the processing, microscopy or an actual difference in the numbers of neurons between wild type and mutant mice.

It does look as if, in panels C and D of figure 2, there are a comparable amount of GnRH neurons. (If anything, it looks as if the knockout mouse in panel D might have a slightly higher concentration of GnRH neurons in the median eminance than the wildtype in panel C.) And the number of GnRH neurons shown in panel D does look like a significant amount. Just by eyeballing it with my admittedly untrained eye, I'd guess that there are much more than 20 GnRH neurons there-- I'd guess it's more on the order of roughly 100. I concede that I could be way off base with that estimate though-- you would know better than I.

I agree that the authors should have been more careful in establishing that the wildtype and knockout mouse have comparable numbers of GnRH neurons so we wouldn't have to do guesswork like this. But judging by figure 2, my preliminary guess would be that only 20 GnRH cell bodies were "clearly identified" for the first study because a large number of GnRH neurons in the knockout mice were just not clearly identifiable for the experimenters. Perhaps they really did do a bad job of processing the tissue, or perhaps they had unusually high standards for what counts as "clearly identifiable" (maybe because of someone's inability to properly use a microscope? :tongue2:). Perhaps it was a bit of both.

Something I noticed that was not mentioned by the authors is that it appears that the wild-type mice already have slightly higher FSH than the knockout mice, even when just given the vehicle injections. There's no way to know if this was a statistically significant difference without more information from the authors, but this actually adds an element of confirmation to their findings that the mice with a functioning receptor should have higher levels of these gonadotropins than the hypogonadotropic knockout mice.

I noticed that too. Although the authors don't provide explicit information on this, I think that eyeballing the graph pretty strongly indicates that the FSH levels of the wildtypes that were administered PBS were significantly higher than those of the knockout mice who received the PBS injection. Try visualizing the FSH value you would get by adding two standard error bars to the -/- PBS bar, and compare that to what you would get by subtracting two standard error bars from the +/+ PBS bar. I'm fairly certain the resultant +/+ PBS bar would still be higher than the resultant -/- PBS bar, suggesting a significant difference.

 

[detta]

Just a few comments on the paper although some have already been answered.
Figure 1. Along with what Moonbear said about the number of neurons being low (which I never would have known), I was wondering why only 55% showed co-localization or any nearby beta-gal staining. It would be nice to have seen an explanation. Is it because the receptor isn't always expressed or because of low lacZ expression? Also some information about how lacZ was expressed in the mice would have been useful, even if they have already reported it in other papers.
At least they show that in some cells both proteins are expressed. I'm surprised they didn't try it in a normal mouse where both proteins actually are expressed. They could have used fluorescent tags to coexpress like GFP. I'm not sure if GFP works with receptors but expressing both proteins instead of using a marker would have made a cleaner experiment. For the rest of the paper, I thought they showed good evidence that, without gpr54, the mice can not release FSH and LH. It would have been nice to have GnRH release as well(Fig3). They prove that gpr54 is necessary for the release of LSH and FSH but I don't see the direct pathway that they claim; the evidence still seems very indirect.
If a viable knockout for GnRH has been made, a good in-vivo experiment would be to inject kisspeptin in the GnRH knockout mice and see if it still blocks hormone release from the pituitary. This could show that kisspeptin requires GnRH to signal the release of FSh and LH.
I would have like to seen some in-vitro work. Performing a colocalization experiment with the peptide and the receptor or plating the neurons out, adding kisspeptin, and then measuring the GnRH would have been two good experiments. I don't know if these types of experiments can be done for these particular cell types, but I've seen them done before for other types of neuronal cells. I think if you are going to claim any type of receptor interaction, in-vivo and in-vitro studies are really necessary for a complete picture.Finally, I had a hard time figuring out what was the purpose of this paper. In other words, how does this finding go beyond just another interaction paper? Why is it important in understanding the brain? I understand that its important for control of release of hormones from the pituitary but how does it fit in a bigger picture.

By the way, I think a phenocopy, when using mouse models, means that the symptoms of the mouse match the symptoms of the human disease even if they may not have the same genetic causes. For example, there are mice that are phenocopies for obesity but they don't necessarily have the same genetic cause for obesity as a human who is very obese.

 

[Moonbear]

It does look as if, in panels C and D of figure 2, there are a comparable amount of GnRH neurons. (If anything, it looks as if the knockout mouse in panel D might have a slightly higher concentration of GnRH neurons in the median eminance than the wildtype in panel C.) And the number of GnRH neurons shown in panel D does look like a significant amount. Just by eyeballing it with my admittedly untrained eye, I'd guess that there are much more than 20 GnRH neurons there-- I'd guess it's more on the order of roughly 100. I concede that I could be way off base with that estimate though-- you would know better than I.

Panels C and D aren't showing the cell bodies, just the fiber projections where the terminals contact the pituitary portal circulation. GnRH fibers look sort of like beads on a string. Panels A and B show the GnRH cell bodies.

I agree that the authors should have been more careful in establishing that the wildtype and knockout mouse have comparable numbers of GnRH neurons so we wouldn't have to do guesswork like this. But judging by figure 2, my preliminary guess would be that only 20 GnRH cell bodies were "clearly identified" for the first study because a large number of GnRH neurons in the knockout mice were just not clearly identifiable for the experimenters. Perhaps they really did do a bad job of processing the tissue, or perhaps they had unusually high standards for what counts as "clearly identifiable" (maybe because of someone's inability to properly use a microscope? :tongue2:). Perhaps it was a bit of both.

I'm inclined to say it's not "unusually high standards." There's really no reason not to get good staining for GnRH, especially with the antibody they used (it's the best one you can get your hands on, and you should be able to get really clear, unambiguous GnRH staining). I don't even know why they stained every section. Usually, people stain only every 4th section, and use the other alternate sections either for looking at other neurons, or as back-up in case they screw up the staining and need to do it again.

 

[Monique]

By the way, I think a phenocopy, when using mouse models, means that the symptoms of the mouse match the symptoms of the human disease even if they may not have the same genetic causes. For example, there are mice that are phenocopies for obesity but they don't necessarily have the same genetic cause for obesity as a human who is very obese.

Yes, I think you are right. It is used in a different way in mouse genetics where it simply means copying a phenotype. Every genetic glossery I looked up does use my definition that was first coined by the developmental geneticist Waddington, namely "A phenotype that is not genetically controlled but looks like a genetically controlled phenotype. An environmentally induced phenotype that resembles the phenotype produced by a mutation." So that's interesting for me to see.

 

[hypnagogue]

Panels C and D aren't showing the cell bodies, just the fiber projections where the terminals contact the pituitary portal circulation. GnRH fibers look sort of like beads on a string. Panels A and B show the GnRH cell bodies.

Just to make sure I'm understanding you-- by fiber projections, do you mean individual axons, or bundles thereof (or maybe something else)? If I'm interpreting you correctly and if GnRH fibers look like beads on a string, this would mean that a number of dots in panel C or D could correspond to the same neuron axon, right? If so, yeah, my previous argument would be invalidated. It does seem pretty surprising that they wouldn't even address why they got such a low number.

 

[Moonbear]

Figure 1. Along with what Moonbear said about the number of neurons being low (which I never would have known), I was wondering why only 55% showed co-localization or any nearby beta-gal staining. It would be nice to have seen an explanation. Is it because the receptor isn't always expressed or because of low lacZ expression?

I can only speculate as to a possible answer to this question. A few reasons I can think of are that there are always detection limits to any method, so it could simply be that all the neurons have beta-gal, but that the expression is too low in some to detect it using this method. Another possibility is that only a subpopulation of GnRH neurons are regulated using this pathway. There's been a fairly longstanding controversy as to whether or not there are two populations of GnRH neurons, one that is involved in pulsatile secretion, and the other that is involved in the pre-ovulatory surge secretion. So, it could be that kisspeptin is only acting on one part of that GnRH population.

Also some information about how lacZ was expressed in the mice would have been useful, even if they have already reported it in other papers.
At least they show that in some cells both proteins are expressed. I'm surprised they didn't try it in a normal mouse where both proteins actually are expressed. They could have used fluorescent tags to coexpress like GFP. I'm not sure if GFP works with receptors but expressing both proteins instead of using a marker would have made a cleaner experiment.

There is another paper contemporary to this one, published by another group, that has provided more convincing evidence, Irwig, et al., Neuroendocrinology 2004;80:264–272. They did dual-label in situ hybridization for GnRH and GPR54 and found 77% of neurons with GnRH mRNA had colocalization with GPR54 mRNA. So, what this one group lacked, others have picked up on.

For the rest of the paper, I thought they showed good evidence that, without gpr54, the mice can not release FSH and LH. It would have been nice to have GnRH release as well(Fig3). They prove that gpr54 is necessary for the release of LSH and FSH but I don't see the direct pathway that they claim; the evidence still seems very indirect.

There's no way to measure GnRH release in mice. That's why they switched to the sheep model in the final part of the study. That's where they show the direct effects on GnRH release (Figure 5). In this paper, they took the approach of measuring GnRH released into CSF. It's possible it's still acting indirectly, via some other interneurons, to effect GnRH release, but combined with the added evidence for colocalization in the other paper I cited above, the argument that it's a direct action is gathering strength. Also, in the note at the end of the article, they cite references demonstrating that GnRH receptor antagonists block the effects of kisspeptin on LH release, indicating that kisspeptin is not acting directly on the pituitary, but on GnRH release.

There is a way to further confirm this, again, in the sheep model. It's possible to surgically disconnect the hypothalamus from the pituitary. In that case, even if GnRH is released, it doesn't reach the pituitary, so does not stimulate LH or FSH release. If you treated sheep with this hypothalamo-pituitary disconnection with kisspeptin and saw LH or FSH release, that would demonstrate that there is a direct pituitary action distinct from the actions on GnRH release, and likewise, if you did not see any LH or FSH release, you'd know conclusively that there is no direct action on the pituitary, and all the pituitary effects are mediated through GnRH.

If a viable knockout for GnRH has been made, a good in-vivo experiment would be to inject kisspeptin in the GnRH knockout mice and see if it still blocks hormone release from the pituitary. This could show that kisspeptin requires GnRH to signal the release of FSh and LH.

There is no viable knockout for GnRH. But, the approach I described above would demonstrate the same thing. If there were a knockout, your approach would be equally good.

I would have like to seen some in-vitro work. Performing a colocalization experiment with the peptide and the receptor or plating the neurons out, adding kisspeptin, and then measuring the GnRH would have been two good experiments. I don't know if these types of experiments can be done for these particular cell types, but I've seen them done before for other types of neuronal cells. I think if you are going to claim any type of receptor interaction, in-vivo and in-vitro studies are really necessary for a complete picture.

I'm never very fond of the in vitro work in this system. The GnRH cell lines have some strange quirks that are not at all the same as what we see in vivo, such as having receptors that don't exist in GnRH neurons in vivo.

Finally, I had a hard time figuring out what was the purpose of this paper. In other words, how does this finding go beyond just another interaction paper? Why is it important in understanding the brain? I understand that its important for control of release of hormones from the pituitary but how does it fit in a bigger picture.

It's presenting a novel mechanism for regulation of GnRH neurons, not for release of pituitary hormones; those still require GnRH. The big picture is that these are important neurons for species survival since they are essential for reproduction. One of the mysteries about GnRH is how these scattered neurons can synchronize their release into discrete pulses, and this gives us new clues to help identify another player in a complex system.

By the way, I think a phenocopy, when using mouse models, means that the symptoms of the mouse match the symptoms of the human disease even if they may not have the same genetic causes. For example, there are mice that are phenocopies for obesity but they don't necessarily have the same genetic cause for obesity as a human who is very obese.

Thanks. That makes more sense in the context used in this article.

 

[Moonbear]

Just to make sure I'm understanding you-- by fiber projections, do you mean individual axons, or bundles thereof (or maybe something else)? If I'm interpreting you correctly and if GnRH fibers look like beads on a string, this would mean that a number of dots in panel C or D could correspond to the same neuron axon, right? If so, yeah, my previous argument would be invalidated. It does seem pretty surprising that they wouldn't even address why they got such a low number.

Yes, sorry, I'm talking about axons. And yes, a lot of those dots in panels C and D would all be part of the same axons (there are no cell bodies in that region of the mouse). The labeling there does not look abnormal other than that there appears to be a bit more background in Panel D (the diffuse brown you see; but individual variation or slight differences in fixation can lead to that too). Because a lot of GnRH is stored out in vesicles in the axons, those are detectable at lower antibody concentrations than cell bodies. It's possible something in their method was not optimized to detect the cell bodies, but they still got decent axonal labeling.

 

[detta]

Thanks for clearing some stuff Moonbear. My background is in cell biology and I know very little about neurobiology. Its really strange that the cells act so differently in-vitro. It makes this a tough system to study. Knowing that another paper has made similar conclusions does make this paper more convincing.

I was also wondering why you think some sheep had such a large increase in GnRH while others did not. All the sheep show an increase and then a constant expression but one went up very high. Is this normal variance? There doesn't seem to be an effect on LH levels, they all seem the same.

 

[DocToxyn]

Hey all...sorry I haven't chimed in one this one sooner. Weekends are tough for me with family and work stuff and to top it off, we all caught a rather nasty respiratory virus that we're still recovering from. Anyway, here are some of my thoughts.

The issue of the numbers of KOs used for the colocalization does seem weak. According to their methods they breed the KOs there so they probably aren't paying anything more than per diem maintenance costs. Now a lot of KO strains have reproductive problems so perhaps it's just difficult to get animals, but they should have waited to collect data from more than two subjects.

I can only speculate as to a possible answer to this question. A few reasons I can think of are that there are always detection limits to any method, so it could simply be that all the neurons have beta-gal, but that the expression is too low in some to detect it using this method.

They used the enzymatic mathod for detection of their beta-gal signal, which means they must have enough functional enzyme present in their sections and be able to deliver the appropriate reagents to those regions. This method is fairly common and I use it as a reporter gene to show activation of the aryl hydrocarbon receptor by exogenous or endogenous ligands. In brief, the enzyme will react with the x-gal in the dye solution and create an insoluble precipitate (dark blue) that remians in the tissue/cell. Penetration of the reagents into the tissue can be an issue, I use detergents like NP-40 and sodium deoxycholate to aid in semi-permeabilizing the tissues, sometimes this helps. The other issue is just having enough enzyme there. I have seen stained tissue blocks (whole fetuses) where you can easily see beta-gal induced staining with the naked eye, but once you slice it (even at 50 um like they did), there just isn't enough endproduct precipitate to visualize, even under a microscope.
It's quite possible that the beta-gal marker was there, they just couldn't detect/record it.

There is another paper contemporary to this one, published by another group, that has provided more convincing evidence, Irwig, et al., Neuroendocrinology 2004;80:264–272. They did dual-label in situ hybridization for GnRH and GPR54 and found 77% of neurons with GnRH mRNA had colocalization with GPR54 mRNA. So, what this one group lacked, others have picked up on.

I was going to suggest that they do in situ. It is generally more sensitive and can be semi-quantifiable, with enzymatic beta-gal all you can say is 'yeah it's blue'. They don't even need to change their system, just create the probe for a section of their beta-gal and look for colocalization with the IHC GnRH.

There is no viable knockout for GnRH. But, the approach I described above would demonstrate the same thing. If there were a knockout, your approach would be equally good.

What about a conditional knockout? Obviously these are much harder to come by or create, but if you really want to study GnRH funtions, it might be worth it.

I'm never very fond of the in vitro work in this system. The GnRH cell lines have some strange quirks that are not at all the same as what we see in vivo, such as having receptors that don't exist in GnRH neurons in vivo.

I've got to aree with Moonbear here. In vitro an be useful for investigating molecular or cellular mechanisms, but once you start getting into multiple regional interactions like one finds in the brain, reproducing that characteristic becomes difficult as does interpretation of the data.

 

[DocToxyn]

Overall it was a really interesting paper. I'm not trying to steer this tread away from the original paper, but... One point in their intro that I found fascinating was that the initial formation of the neurons and subsequent GnRH release begins in a late gestation/early postnatal period and most likely plays a role in mamaging developmental processes and solidifying appropriate connections. Then GnRH produciton shuts down until puberty and re-activates to intitate sexual maturity. Upon going to the Ebling and Cronin review that the authors cited I found some interesting related info. They make a statement in their text that "endogenous stimulation of GnRH release in the rate-limiting step for puberty". Strange how something so complicated, pivotal (and sometimes confusing) like puberty can be ruled by some 200 or so neurons buried in some tiny part of the brain.

They also brought up data that show that the decrease in age of onset of puberty correlates with the increase in prevalence of overweight adolescents. The causal event is believed to be body composition, mainly fat deposits, that tips the positive energy balance towards intiation of maturation and puberty. They go on to talk about leptin and environmental cues such as photoperiod, very interesting stuff. OK, you can go back to the original paper now.

 

[Moonbear]

I was also wondering why you think some sheep had such a large increase in GnRH while others did not. All the sheep show an increase and then a constant expression but one went up very high. Is this normal variance? There doesn't seem to be an effect on LH levels, they all seem the same.

The concentrations they reported were within the normal range of variance. Of course, the pattern observed is not normal looking at all, but that's not unexpected when you're inducing a response pharmacologically (i.e., with a constant infusion of kisspeptin without knowing what it's physiological pattern of secretion is, which could be very different).

Essentially, the concentrations are within the range you'd expect for pulses of GnRH. To help with that, I'm attaching a figure that shows a comparison of LH and GnRH release. This is adapted from Skinner DC, Caraty A, Malpaux B, Evans N. 1997 Simultaneous measurement of gonadotropin-releasing hormone in the third ventricular cerebrospinal fluid and hypophyseal portal blood of the ewe. Endocrinology 138: 4699-4704.

In the figure, the left-side panels show the patterns of pulsatile release in a single ewe, and the right-side panels show the patterns of release during a surge. Take note that the scales for the y-axis differ in each panel.

The top panels show LH release, which is measured from a catheter inserted into the jugular vein, just as was done in the paper presented, and the bottom panels are GnRH released into cerebrospinal fluid (CSF), again collected using the same methods of measurement as were done in the paper being discussed. The middle panel shows GnRH that was measured directly from pituitary portal blood as it's being released from the hypothalamus to act in the pituitary; this was not done in the currently discussed paper.

The panel of most interest is the bottom, left panel, that shows the pulsatile release of GnRH into cerebrospinal fluid. You'll see that not all pulses are the same size, and that the range for the amplitude of the pulses in a single animal is similar to the range of responses to kisspeptin in multiple animals.

 

[Moonbear]

Overall it was a really interesting paper. I'm not trying to steer this tread away from the original paper, but... One point in their intro that I found fascinating was that the initial formation of the neurons and subsequent GnRH release begins in a late gestation/early postnatal period and most likely plays a role in mamaging developmental processes and solidifying appropriate connections. Then GnRH produciton shuts down until puberty and re-activates to intitate sexual maturity. Upon going to the Ebling and Cronin review that the authors cited I found some interesting related info. They make a statement in their text that "endogenous stimulation of GnRH release in the rate-limiting step for puberty". Strange how something so complicated, pivotal (and sometimes confusing) like puberty can be ruled by some 200 or so neurons buried in some tiny part of the brain.

They also brought up data that show that the decrease in age of onset of puberty correlates with the increase in prevalence of overweight adolescents. The causal event is believed to be body composition, mainly fat deposits, that tips the positive energy balance towards intiation of maturation and puberty. They go on to talk about leptin and environmental cues such as photoperiod, very interesting stuff. OK, you can go back to the original paper now.

Yes, this is all very interesting stuff. Maybe when my turn rolls around again, I'll dig into the puberty literature, though I think a lot more of those studies that catch my interest are still pure endocrinology rather than neuroendocrinology, so I'll have to see if there's something new and with enough of a neuro focus to be interesting to this group. Of course, the photoperiod cues are relevant to a lot of species, but not so much with humans. I have a keen interest in photoperiodism, but mostly rely on it more as a tool to study mechanisms by which reproduction can turn off and on naturally, and to study more generally neural plasticity.

It's a little limiting to have to stick with papers that are freely available to everyone.

Oh, I'm just remembering that I promised a picture to clearly point out what they should be looking at to identify the median eminence. I should go do that before I go to sleep and forget again.

Edit: Okay, the image I thought I had that would be good isn't at the same level as the one in the paper, so won't help much. The others I have are all from sheep, which won't help much either, so I just drew a quick sketch to sort of help out. Where I wrote the text "median eminence" is the only part that is median eminence. Everything further toward the top of the figure is other parts of the brain.

 

[Moonbear]

Okay, being close enough to midnight, I'm going to unstick this thread and put hypnagogue's up for discussion. The formal discussion on this paper is ended, but that doesn't mean folks who are interested can't continue to discuss it "unofficially."