Pruning hypothesis comes of age

The idea that disrupted pruning of neuronal connections in the brain during adolescence is a cause of schizophrenia was proposed in 1983. This proved prescient, as subsequent imaging, genetic and molecular research has shown.
Matthew B. Johnson is at the F. M. Kirby Neurobiology Center, Boston Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115, USA, and at the Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard, Cambridge, Massachusetts.

Search for this author in:

Beth Stevens is at the F. M. Kirby Neurobiology Center, Boston Children’s Hospital and Harvard Medical School, Boston, Massachusetts 02115, USA, and at the Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard, Cambridge, Massachusetts.

Search for this author in:

It is common knowledge among neuroscientists today that the brain’s frontal lobe, which is involved in many complex human behaviours, continues to develop throughout adolescence and into early adulthood. But in 1983, the idea of such protracted brain development was, in the words of the psychiatrist Irwin Feinberg, “a newly emerging theme”1. That year, he published the hypothesis1 that schizophrenia is caused by defects in a particular aspect of adolescent brain development. Genomic studies are now beginning to shed long-awaited light on the mechanisms that might underlie such a process.

Among the earliest contributions to this emerging field was a landmark 1979 study by the neuroscientist Peter Huttenlocher. Using electron microscopy, Huttenlocher imaged and counted the connections, called synapses, between neurons in slices of the frontal lobe from individuals ranging in age from newborn to 90 years old. Unexpectedly, he found that a drastic reduction occurred in the number of synapses between infancy and adulthood2. At the time, Feinberg was studying sleep patterns during human development and ageing. The changes detailed in Huttenlocher’s report were strikingly similar to a phenomenon Feinberg had observed to occur during the second decade of life — a steep reduction in the amplitude of waves of brain activity during dreamless sleep, as measured using a technique called electroencephalography (EEG)3.

Taking this information together with the propensity for symptoms of schizophrenia to emerge in patients’ late teens and early twenties4, Feinberg boldly speculated that the three phenomena might be linked. According to his hypothesis, normal changes in synapse numbers or organization, determined by genetic programs of synaptic pruning in the adolescent brain, could reduce the amplitude of sleep EEG waveforms. Disturbances of these programs (causing either too much or too little pruning) might produce or contribute to the symptoms of schizophrenia.

Feinberg had been considering the relationship between brain structure and function since the early 1960s, and recalls sitting in the office of his then mentor, Jean Piaget, a pioneer in the field of cognitive development, and “raising the possibility that maybe part of [cognitive] development was based on biological changes in the brain. He went on with the conversation as though I had not said anything”. Twenty years later, neuroscientists reacted similarly to his hypothesis on pruning and schizophrenia, with Feinberg describing the initial response in an interview last year as “totally indifferent” (see

Indeed, a major hole in Huttenlocher’s analysis remained a sticking point for years. Owing to the difficulty of obtaining samples, and the painstaking analysis involved, his study included only one adolescent. Although data from primates revealed similar postnatal synaptic pruning (for a review, see ref. 5), it was not until 1997 that additional human data supporting this phenomenon became available from a handful of adolescent brains6. In 2011, 5 more samples from individuals between 10 and 22 years old were finally added to the literature7.

As researchers were working to fill this gap in the data on normal development, at least two groups were looking for — and, at the turn of the twenty-first century, found — signs of aberrant pruning in the brains of people with schizophrenia8,9 (Fig. 1). Using a staining technique that enabled them to count the number of neuronal structures called dendritic spines that receive certain synaptic inputs, these researchers provided evidence that synapse densities in brain regions involved in higher cognition are lower in affected than in unaffected people.

Figure 1 | Photomicrographs of neurons in the human brain. a, A neuron from a healthy brain is dotted with structures called dendritic spines (seen in the zoomed-in image). Spine numbers can be used as a proxy for the number of synaptic connections a neuron receives from other neurons (not visible). Scale bar in main image, 30 µm; in zoomed-in image, 20 µm. b, There are fewer spines, and so synapses, in the brain of a person who has schizophrenia. Scale bar, 15 µm. (Figure adapted from ref. 8.)

Subsequently, the development of non-invasive brain-imaging techniques made the analysis of both healthy brain development and psychiatric disorders much easier. One imaging study10 confirmed that the volume of grey matter (the brain tissue that generally contains synapses) in the temporal and frontal lobes increases during early childhood and then decreases over the course of adolescence. These changes are consistent with the pruning of neuronal processes and synapses in adolescence. Another study11 revealed that at-risk individuals who go on to develop psychosis exhibit an accelerated rate of grey-matter loss in the frontal lobe compared with those who do not. Evidence to support Feinberg’s hypothesis was gradually growing.

More recently, the hypothesis has been strengthened, thanks to the genomics revolution. Studies12,13 of tens of thousands of people with schizophrenia have identified genetic variants associated with an increased risk of the disease. By far the strongest signal has been found in a portion of chromosome 6 that contains hundreds of genes related to immunity, making it hard to pinpoint individual risk genes. But in 2016, analyses of brain and DNA samples from affected and unaffected individuals enabled geneticists to pick out variation in one gene in the region, that encoding complement component 4 (C4), as partly responsible for schizophrenia risk14.

This work revealed that different variants of the C4 gene produce different levels of the protein it encodes, and that higher C4 expression is correlated with a higher risk of schizophrenia. Independently of this work, our group and others have shown that the immune signalling cascade of which the C4 protein is a part, called the complement cascade, instructs the brain’s immune cells to prune synapses during development in mice (Fig. 2)15,16. How this pruning mechanism relates to schizophrenia risk or age of onset, or to normal adolescent changes in brain activity during sleep, is an exciting and active area of investigation. Precisely how and where in the human brain C4 is targeted to synapses, whether particular synapses or circuits are more or less vulnerable to complement-mediated pruning, and which of these mechanisms are most relevant to specific symptoms of schizophrenia, remain unknown.

Figure 2 | A growing understanding of brain rewiring in adolescence. a, In 1979, an analysis of synapse numbers in the frontal lobe at different ages revealed a dramatic decrease in synapses during adolescence2. Sleep brainwave amplitudes follow a similar pattern3. Furthermore, symptoms of schizophrenia typically emerge in adolescence4. In 1983, these facts together led the psychiatrist Irwin Feinberg, who was studying sleep, to propose1 that defects in adolescent pruning of synapses might be a cause of schizophrenia. b, In the past decade, molecular and genetic studies1216 have provided evidence that this pruning defect involves genes of the complement signalling cascade. During normal development, brain-resident immune cells called microglia engulf and prune synapses that are tagged by complement proteins. Higher expression of one complement gene, C4, is associated with a higher risk of schizophrenia, perhaps because of higher pruning levels.

Other gene variants associated with schizophrenia risk are scattered throughout the genome, implicating a variety of other synaptic proteins13. Thus, synaptic pruning is likely to be disrupted through different mechanisms in different patients. New tools are now enabling us to better understand this variability. For instance, methods for converting human stem or skin cells into brain cells and 3D ‘mini-brains’ allow us to probe synapse development and pruning using living patients’ own cells, or cells engineered to harbour risk variants. In addition, genome-editing tools such as CRISPR–Cas could be used to generate new primate models of frontal-lobe synapse development and psychiatric disorders.

One could charitably interpret the initial resistance to Feinberg’s pruning hypothesis as tacit recognition that, at that time, such hypotheses could only be pure, untestable speculation. Indeed, Feinberg himself wrote, “The probability of guessing correctly the brain mechanisms causing schizophrenia is, with our limited knowledge, vanishingly small”1. And yet, although we are just beginning to understand the molecular processes underlying his hypothesis, evidence is now accumulating that Feinberg might — at least in part — have done just that.

Nature 554, 438-439 (2018)

doi: 10.1038/d41586-018-02053-7


  1. 1.

    Feinberg, I. J. Psychiatr. Res. 17, 319–334 (1982–1983).

  2. 2.

    Huttenlocher, P. R. Brain Res. 163, 195–205 (1979).

  3. 3.

    Feinberg, I., Koresko, R. L. & Heller, N. J. Psychiatr. Res. 5, 107–144 (1967).

  4. 4.

    Sham, P. C., MacLean, C. J. & Kendler, K. S. Acta Psychiatr. Scand. 89, 135–141 (1994).

  5. 5.

    Elston, G. N. & Fujita, I. Front. Neuroanat. 8, 78 (2014).

  6. 6.

    Huttenlocher, P. R. & Dabholkar, A. S. J. Comp. Neurol. 387, 167–178 (1997).

  7. 7.

    Petanjek, Z. et al. Proc. Natl Acad. Sci. USA 108, 13281–13286 (2011).

  8. 8.

    Garey, L. J. et al. J. Neurol. Neurosurg. Psychiatr. 65, 446–453 (1998).

  9. 9.

    Glantz, L. A. & Lewis, D. A. Arch. Gen. Psychiatry 57, 65–73 (2000).

  10. 10.

    Gogtay, N. et al. Proc. Natl Acad. Sci. USA 101, 8174–8179 (2004).

  11. 11.

    Cannon, T. D. et al. Biol. Psychiatry 77, 147–157 (2015).

  12. 12.

    The International Schizophrenia Consortium. Nature 460, 748–752 (2009).

  13. 13.

    Schizophrenia Working Group of the Psychiatric Genomics Consortium. Nature 511, 421–427 (2014).

  14. 14.

    Sekar, A. et al. Nature 530, 177–183 (2016).

  15. 15.

    Stevens, B. et al. Cell 131, 1164–1178 (2007).

  16. 16.

    Schafer, D. P. et al. Neuron 74, 691–705 (2012).

Download references

Paid content

Nature Briefing

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.