Novel germline PALB2 truncating mutations in African-American breast cancer patients

Abstract

Background

It has been demonstrated that PALB2 acts as a bridging molecule between the BRCA1 and BRCA2 proteins and is responsible for facilitating BRCA2-mediated DNA repair. Truncating mutations in the PALB2 gene have been reported to be enriched in Fanconi anemia and breast cancer patients in various populations.

Methods

We evaluated the contribution of PALB2 germline mutations in 279 African-American breast cancer patients including 29 patients with a strong family history, 29 patients with a moderate family history, 75 patients with a weak family history, and 146 non-familial or sporadic breast cancer cases.

Results

After direct sequencing of all the coding exons, exon/intron boundaries, 5′UTR and 3′UTR of PALB2, three (1.08%; 3 in 279) novel monoallelic truncating mutations were identified: c.758dupT (exon4), c.1479delC (exon4) and c.3048delT (exon 10); together with 50 sequence variants, 27 of which are novel. None of the truncating mutations were found in 262 controls from the same population.

Conclusions

PALB2 mutations are present in both familial and non-familial breast cancer among African-Americans. Rare PALB2 mutations account for a small but substantial proportion of breast cancer patients.

INTRODUCTION

Breast cancer is a global health problem as about 1 million women are diagnosed with it annually and it is the second leading cause of cancer death among women in the United States. In 2010, more than 207,090 new cases of invasive breast cancer were diagnosed.¹ A deadly combination of genetic, as well as endogenous and exogenous factors, lead to an increased risk for breast cancer. Study of germline mutations in BRCA1 and BRCA2 over the past two decades has established their role in inherited breast and/or ovarian cancers. Mutation screening of genes functionally related to BRCA1 and/or BRCA2 has revealed mutations in genes such as CHEK2, ATM, BRIP1, and PALB2. As mutations in BRCA1 and BRCA2 have been suggested to be rare high-penetrance mutations, mutations in ATM, CHEK2, BRIP1 and PALB2 were classified as rare moderate-penetrance mutations, while single nucleotide polymorphisms (SNPs) identified by genome-wide association studies were characterized as common low-penetrance genetic variants. Although mutations/variants in the above-mentioned breast cancer susceptibility genes as well as TP53 and PTEN account for a significant fraction of breast cancer risk, more than 70% of the genetic component of breast cancer risk remains uncharacterized.^{2, 3}

As an essential component of the BRCA1-PALB2-BRCA2-RAD51 pathway, the protein PALB2 (partner and localizer of BRCA2) acts as a bridging molecule that directly connects BRCA1 and BRCA2 to form a “BRCA complex” and subsequently facilitates BRCA2-mediated DNA repair.⁴ It has been demonstrated that PALB2 plays a crucial role in maintenance of genomic integrity and stability by executing double-strand break repair via homologous recombination, and tumor suppression of Fanconi anemia and cancers. Monoallelic/heterozygous mutations in Fanconi anemia genes have been implicated in breast cancer.^{4, 5} The similarity of clinical phenotypes caused by biallelic/homozygous mutations in BRCA2/FANCD1 and PALB2/FANCN pointed out that Fanconi anemia and breast cancer might share some genetic risk factors. To date, deleterious mutations in PALB2 have been identified in Fanconi anemia,^{6, 7} breast cancer,^8-24 ovarian cancer,^{17, 25} pancreatic cancer,^{24, 26-28} and prostate cancer^{9, 29} patients in various populations (Figure 1). The spectrum, penetrance and clinical significance of PALB2 mutations in diverse populations remain to be further characterized.

Figure 1.

Deleterious mutations in the PALB2 gene

Chromosomal positions were based on Human Feb. 2009 assembly (GRCh37/hg19).

In this study, we have conducted mutational analysis and elucidated the mutation spectrum of the PALB2 gene in a cohort of 279 female African-American breast cancer patients unselected by family history, early age of onset, or BRCA1/2 mutation status.

MATERIAL AND METHODS

Study subjects

Two hundred and seventy-nine African-American women diagnosed with breast cancer were recruited at the Cancer Risk Clinic at the University of Chicago, Chicago, from 1993 to 2008. The physicians who participated in this project were asked to inquire about the family history of cancer in as much detail as possible and to obtain accurate patient information. In order to have a comprehensive understanding of the mutation rate in both familial and non-familial breast cancer patients, we did not select the participants based on positive family history, early age of onset, or BRCA1/2 mutation status. Our cohort contains 279 breast cancer patients, including 29 patients with strong family history (three or more first-degree relatives [FDR] and/or second-degree relatives [SDR] affected by breast and/or ovarian cancer, FDR + SDR n ≥ 3); 29 patients with moderate family history (FDR + SDR n = 2); 75 patients with weak family history (FDR + SDR n = 1); and 146 non-familial or sporadic breast cancer cases). 19 patients were previously tested positive for BRCA1/2, 51 patients were tested negative, whereas the mutation screening has not been conducted in the rest 209 patients. The characteristics of the breast cancer cases are listed in Table 1. A total of 262 cancer-free African-American female volunteers were recruited as normal controls to examine whether the deleterious mutations/variants identified occur in the normal population. Informed consents were obtained from all the individuals before the collection of blood samples. This study was approved by the Institutional Review Boards of the University of Chicago.

Table 1.

279 African-American breast cancer patients in the present study

Subject group	Family history of breast and/or ovarian caner	Number	Age of onset, mean ± SD (years)
BRCA1/2 mutation - positive	high	8	43.63 ± 12.52
	moderate	5	43.60 ± 4.39
	weak	6	37.17 ± 10.44
	sporadic	0	-
	total	19	41.58 ± 10.25
BRCA1/2 mutation - negative	high	11	43.19 ± 12.14
	moderate	9	49.00 ± 8.19
	weak	20	44.63 ± 11.75
	sporadic	11	38.27 ± 7.24
	total	51	43.68 ± 10.71
BRCA1/2 mutation - unknown	high	10	45.50 ± 11.30
	moderate	15	42.27 ± 6.08
	weak	49	44.73 ± 8.77
	sporadic	135	44.04 ± 10.06
	total	209	44.14 ± 9.55
All	high	29	44.07 ± 11.58
	moderate	29	44.59 ± 7.05
	weak	75	44.08 ± 9.83
	sporadic	146	43.59 ± 9.97
	total	279	43.88 ± 9.81

Mutation identification by sequencing

Genomic DNA was isolated from peripheral blood leukocytes and stored at −20°C. Oligonucleotide primers were designed using Primer3 v0.4.0³⁰ to cover all the thirteen coding exons, exon/intron boundaries (at least 50 bp), 5′UTR and 3′UTR of the PALB2 gene (Table 2). Polymerase chain reaction (PCR) reactions were performed in a final volume of 10 μl containing 10 ng DNA, 5 pmol of each primer, 25 nmol MgCl₂, 2 nmol of each dNTP, 1 μl of 10× PCR Buffer II, and 0.5U of AmpliTaq Gold® DNA Polymerase (Applied Biosystems, USA). The cycling profile included: an initial denaturation at 95°C for 5 min, followed by 20 touchdown cycles consisting of a denaturation step of 95°C for 30 s, primer annealing started from 64°C to 54°C (0.5°C decrease per cycle) for 30 s and an elongation step at 72°C for 1 min; then followed by 20 cycles of 95°C for 30 s, 58°C for 30 s and 72°C for 1 min; and a final extension step at 72°C for 10 min. PCR products were subsequently checked by agarose gel electrophoresis and cleaned by ExoSap-IT (USB, USA) prior to the cycle-sequencing reactions using BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, USA). Capillary electrophoretic sequencing was performed on an automated Applied Biosystems 3730xl DNA Analyzer according to the manufacturer’s instructions. SeqScape analysis software v2.6.0 (Applied Biosystems, USA) was used to conduct base calling, quality assessment, and assembly. The average sequencing success rate was 99.08% for all amplicons. The amplicons harboring truncating mutations and one predicted deleterious variant of the PALB2 gene (see below) were also analyzed in the normal control subjects with an average call rate of 99.50%. In addition, we confirmed those three truncating mutations by bi-direction sequencing of the same DNA samples as well as other batches of DNA extracted from bloods of the corresponding patients.

Table 2.

Primers for PALB2 sequencing

Primer	Primer sequence (5′->3′)	Amplicon size (bp)
PALB2_Ex1-F	ACAGCGCGGCTCTCCTTTAG	410
PALB2_Ex1-R	ATACTGCTGCCCTCGGACTG
PALB2_Ex2&3-F	GTAGATTGTTATGGACCAGTGCTACT	511
PALB2_Ex2&3-R	GTCTATTGCTAGTCATTATCTTCACAC
PALB2_Ex4A-F	TCTGCCTGAATGAAATGTCACTGA	645
PALB2_Ex4A-R	GGTCTTCTTAGGAATGTATCAACACC
PALB2_Ex4B-F	TGAAATAAGAACTCACCTTTTAAGTCT	600
PALB2_Ex4B-R	TTTTCTGCAGAAAGAGGAGAGGTT
PALB2_Ex4C-F	TGACACTCTTGATGGCAGGA	683
PALB2_Ex4C-R	AAGGAAGTGCCAGGCAAATA
PALB2_Ex5A-F	TTAAATCTAGGAGATCCTATTCTCTTTG	531
PALB2_Ex5A-R	GTATAAAGTAATATGGATGAAGAAAGGC
PALB2_Ex5B-F	GAGGGAAGCTGTATTTTTCCA	636
PALB2_Ex5B-R	CATGCTGTTTACATTCACTAAGGC
PALB2_Ex6-F	GCTGCTGTTATAAGAGGAAATAAAGACA	344
PALB2_Ex6-R	GGGAAAAATAACCAATCCAAATCTG
PALB2_Ex7-F	GCTCTTTCTTTTCACCTGCATAAGA	492
PALB2_Ex7-R	TGGGTATATGGGTGCTCACTATACA
PALB2_Ex8-F	CCTTGTACAGTGAGAATACAAAAGAATGTGA	271
PALB2_Ex8-R	TAGGTTATTACCTGCACTTAAAACCA
PALB2_Ex9-F	AAAAAAGTGAACCTAGTCCTTTAATATT	581
PALB2_Ex9-R	GCTCAAACTTCTGCCTTGGC
PALB2_Ex10-F	ATATTATGCAGTTCAACAATGCGG	386
PALB2_Ex10-R	ACCTGGGTGATAGGAGGAGACTC
PALB2_Ex11-F	ACCTCCTAAGACATGCTATGATGAATAA	379
PALB2_Ex11-R	GCCAGAAAATTTACCAAGCAATCA
PALB2_Ex12-F	CAGAGCCTATCGGTCATTGCTT	438
PALB2_Ex12-R	TCTGGGGTTTGACTCAAGTCCA
PALB2_Ex13-F	TTTAATTGTTTTTTGGATATGTAATCTGAA	678
PALB2_Ex13-R	TGAAATTTACACTTACTAGGCAAAGA

Assessment of pathogenicity

All variants detected were classified based on their variant type as frameshift/truncating, nonsynonymous, synonymous, untranslated-5′UTR, untranslated-3′UTR, intronic, and 3′ downstream. Five different computational algorithms, SIFT,³¹ PolyPhen,³² PMUT,³³ PANTHER,³⁴ and Align GVGD³⁵ were used to predict the functional impact of missense variants detected in this study. We constructed protein multiple sequence alignments (PMSAs) for use with Align GVGD by EXPRESSO(3DCoffee)::Regular³⁶ which takes protein structure into account when performing multiple sequence alignment. Sequences of thirty-six sequences species were extracted from ENSEMBL³⁷ for creating PMSAs file: Anolis carolinensis, Bos taurus, Callithrix jacchus, Cavia porcellus, Choloepus hoffmanni, Dasypus novemcinctus, Dipodomys ordii, Echinops telfairi, Equus caballus, Erinaceus europaeus, Felis catus, Gallus gallus, Gorilla gorilla, Homo sapiens, Loxodonta africana, Macaca mulatta, Macropus eugenii, Microcebus murinus, Monodelphis domestica, Mus musculus, Ochotona princeps, Ornithorhynchus anatinus, Oryctolagus cuniculus, Otolemur garnettii, Pan troglodytes, Pongo pygmaeus, Procavia capensis, Pteropus vampyrus, Rattus norvegicus, Sorex araneus, Spermophilus tridecemlineatus, Sus scrofa, Tarsius syrichta, Tupaia belangeri, Tursiops truncatus, Vicugna pacos.

Classification scores

Above prediction methods rely on PMSAs, and/or protein structural considerations, and/or evolutionary evaluation, additionally, their specificity and sensitivity vary. Hence, concordance between these in silico assessment programs is suggested to be an effective predictor to differentiate pathological missense variants from benign ones. We established a classification system based on the combined computational analysis for each nonsynonymous variant detected in this study. Firstly, a score of 0 (benign), 0.5 (possibly deleterious) and 1 (deleterious) was given to results of each of the five predictions: SIFT − tolerated = 0 and affect = 1; PolyPhen − benign = 0, possible damaging = 0.5 and probably damaging = 1; PMUT − neutral = 0 and pathologic =1; PANTHER − neutral = 0 and pathologic = 1; Align GVGD − C0 = 0, C15, C25, or C35 = 0.5, and, C45, C55, or C65 = 1. Secondly, for each variant, the scores for all five algorithms were summed and subsequently classified as follows: benign = 0, possibly benign = 0.5-2, possibly deleterious = 2.5-4, and deleterious = 4.5-5.

Nomenclature

The variants found in this study were named according to the GenBank PALB2 reference sequence NM_024675.3 and NP_078951, following the instructions provided by the Human Genome Variation Society.³⁸ In addition, we checked the nomenclature by Mutalyzer v2.0.³⁹

RESULTS

Variants identified

After direct sequencing of all the coding exons, exon/intron boundaries, 5′UTR and 3′UTR of the PALB2 gene, we identified a total of fifty-three sequence variants. Although neither nonsense mutations nor splicing variants which occur in canonical AG or GT splice acceptor or donor sites were detected, the nucleotide alterations found belong to frameshift-duplication, frameshift-deletion, nonsynonymous, synonymous, untranslated-5′UTR, untranslated-3′UTR, intronic, and 3′ downstream. Basically, all the sequence variants appear to be located throughout the PALB2 gene. Table 3 lists the truncating mutations and nonsynonymous variants identified in this study, and Table 4 lists all the novel synonymous and non-coding variants identified in the present study.

Table 3.

Truncating mutations and nonsynonymous variants of PALB2 discovered in the present study

Location	Nucleotide change	Protein change	rs# (SNP132)	Variant type	Computational prediction						Minor allele frequency (MAF)
					SIFT	PolyPhen	PMUT	PANTHER	Align GVGD	Re-score	BRCA1/2 mutation positive cases (n = 19)	BRCA1/2 mutation negative cases (n = 51)	BRCA1/2 mutation unknown cases (n = 209)	All cases (n = 279)	Controls (n = 262)
exon2	c.53A>G	p.Lys18Arg	-	nonsynonymous	tolerated (0.21)	benign (1.252)	neutral (0.046)	neutral (−1.770)	C0	benign (0)	0.000	0.000	0.005	0.004	-
exon4	c.400G>A	p.Asp134Asn	-	nonsynonymous	tolerated (0.41)	benign (0.979)	neutral (0.154)	neutral (−1.352)	C0	benign (0)	0.000	0.000	0.002	0.002	-
exon4	c.629C>T	p.Pro210Leu	rs57605939	nonsynonymous	tolerated (0.44)	benign (1.139)	pathological (0.708)	neutral (−0.666)	C0	possibly benign (1)	0.079	0.029	0.060	0.056	-
exon4	c.721A>G	p.Asn241Asp	rs113217267	nonsynonymous	tolerated (0.62)	benign (0.114)	neutral (0.032)	neutral (−0.597)	C0	benign (0)	0.000	0.030	0.014	0.016	0.004
exon4	c.758dupT	p.Ser254Ilefs*3	-	frameshift- duplication	truncating	truncating	truncating	truncating	truncating	truncating	0.000	0.000	0.002	0.002	0.000
exon4	c.925A>G	p.Ile309Val	rs3809683	nonsynonymous	tolerated (0.30)	benign (0.097)	neutral (0.045)	neutral (−0.266)	C0	benign (0)	0.056	0.010	0.012	0.015	0.035
exon4	c.1010T>C	p.Leu337Ser	rs45494092	nonsynonymous	tolerated (0.73)	possibly damaging (1.691)	pathological (0.734)	neutral (−2.282)	C0	possibly benign (1.5)	0.000	0.010	0.005	0.005	0.004
exon4	c.1479delC	p.Thr494Leufs*67	-	frameshift- deletion	truncating	truncating	truncating	truncating	truncating	truncating	0.000	0.000	0.002	0.002	0.000
exon4	c.1676A>G	p.Gln559Arg	rs152451	nonsynonymous	tolerated (0.56)	benign (0.175)	neutral (0.070)	neutral (−0.476)	C0	benign (0)	0.316	0.163	0.189	0.193	0.215
exon5	c.2014G>C	p.Glu672Gln	rs45532440	nonsynonymous	tolerated (0.44)	benign (1.348)	neutral (0.107)	neutral (−2.154)	C0	benign (0)	0.000	0.000	0.007	0.005	-
exon7	c.2590C>T	p.Pro864Ser	rs45568339	nonsynonymous	tolerated (0.46)	probably damaging (2.274)	pathological (0.819)	neutral (−1.5704)	C0	possibly benign (2)	0.000	0.020	0.002	0.005	-
exon9	c.2851T>C	p.Ser951Pro	-	nonsynonymous	tolerated (0.37)	benign (0.044)	pathological (0.652)	neutral (−1.955)	C0	possibly benign (1)	0.000	0.010	0.000	0.002	0.002
exon9	c.2993G>A	p.Gly998Glu	rs45551636	nonsynonymous	affect (0.01)	probably damaging (2.172)	pathological (0.840)	deleterious (− 4.581)	C65	deleterious (5)	0.000	0.000	0.007	0.005	0.004
exon10	c.3048delT	p.Phe1016Leufs*17	-	frameshift- deletion	truncating	truncating	truncating	truncating	truncating	truncating	0.000	0.000	0.002	0.002	0.000
exon10	c.3056T>C	p.Val1019Ala	-	nonsynonymous	affect (0.01)	possibly damaging (1.583)	pathological (0.577)	deleterious (− 3.190)	C25	possibly deleterious (4)	0.000	0.000	0.002	0.002	0.000

Three truncating mutations, p.Ser254Ilefs*3, p.Thr494Leufs*67, and p.Phe1016Leufs*17 identified in this study are shown in bold.

Table 4.

Novel synonymous and non-coding variants of PALB2 identified in the present study

Location	Nucleotide change	Variant type	BRCA1/2 mutation positive cases (n = 19)	BRCA1/2 mutation negative cases (n = 51)	BRCA1/2 mutation unknown cases (n = 209)	All cases (n = 279)	Controls (n = 260)
5′UTR	c.−194C>G	untranslated-5′UTR	0.000	0.000	0.002	0.002	-
exon4	c.909C>T	synonymous	0.000	0.000	0.005	0.004	0.002
exon4	c.1038A>G	synonymous	0.000	0.010	0.000	0.002	0.000
exon4	c.1606C>T	synonymous	0.000	0.031	0.007	0.011	0.002
exon5	c.1810C>T	synonymous	0.000	0.000	0.005	0.004	-
exon5	c.2256A>G	synonymous	0.000	0.020	0.007	0.009	-
exon5	c.2301C>A	synonymous	0.000	0.000	0.002	0.002	-
exon5	c.2365C>T	synonymous	0.000	0.000	0.002	0.002	-
intron5	c.2515−24A>G	intronic	0.000	0.010	0.000	0.002	-
intron7	c.2748+121T>C	intronic	0.000	0.010	0.005	0.005	-
intron8	c.2835−27C>T	intronic	0.000	0.010	0.000	0.002	0.002
intron9	c.2996+124C>T	intronic	0.000	0.010	0.000	0.002	0.002
intron9	c.2996+183delA	intronic	0.000	0.000	0.002	0.002	0.000
intron10	c.3113+131G>T	intronic	0.000	0.000	0.002	0.002	0.000
intron10	c.3114−40T>G	intronic	0.000	0.000	0.002	0.002	-
intron11	c.3201+96G>A	intronic	0.000	0.010	0.000	0.002	-
intron11	c.3201+112A>G	intronic	0.000	0.000	0.002	0.002	-
intron11	c.3201+115T>C	intronic	0.000	0.000	0.005	0.004	-
intron11	c.3202−45A>G	intronic	0.000	0.000	0.005	0.004	-
exon12	c.3252G>A	synonymous	0.000	0.000	0.002	0.002	-
intron12	c.3350+16T>G	intronic	0.000	0.010	0.000	0.002	-
3′UTR	c.*232G>T	untranslated-3′UTR	0.000	0.010	0.000	0.002	-
3′ downstream	c.*347A>C	3′ downstream	0.000	0.010	0.000	0.002	-

In 279 African-American women diagnosed with breast cancer, we found one deletion and two duplication mutations (totally 1.08%, 3 in 279) (Figure 1). They are all frameshift truncating mutations due to their possibility of altering the encoded protein, and turn out to be singleton mutations, namely, each of them was found only once in the present study. These three novel truncating mutations have not been deposited in The Single Nucleotide Polymorphism database (dbSNP) Build 132⁴⁰ and have not been reported in the literature. They were not detected in any of the 262 normal controls in our cohort. We did not detect any nonsense, frameshift truncating, or deleterious missense mutations in 19 BRCA1/2 mutation carriers.

We also uncovered twelve nonsynonymous/missense variants (four novel and eight known), thirteen synonymous/silent variants (eight novel and five known), two untranslated-5′UTR variants (one novel and one known), one novel untranslated-3′UTR variant, twenty-one intronic variants (twelve novel and nine known), and one novel 3′ downstream variant. Forty (75.5%) variants are very rare (minor allele frequency, MAF < 0.01), six (11.3%) variants are rare (0.01 ≤ MAF ≤ 0.05), and the remaining seven (13.2%) variants are common (MAF > 0.05).

Patients with truncating mutations

The c.758dupT (p.Ser254Ilefs*3) mutation was identified in a patient diagnosed at the age of 45 years with infiltrating ductal carcinoma of the breast (IDC) not otherwise specified (NOS). Immunohistochemistry of her grade II tumor showed the expression of estrogen receptor (ER) and progesterone receptor (PR), as well as borderline expression of human epidermal growth factor receptor 2 (HER2/neu). Her family history of breast cancer is weak as only a half-sister was reported as having breast cancer at the age of 50 years and her maternal aunt had colorectal cancer at the age of 65 years, with the remaining family members reported as cancer-free. No additional family members were available for testing.

The patient who carried c.1479delC (p.Thr494Leufs*67) has a strong family history. She was diagnosed with grade III IDC NOS (ER+, PR+, but HER2/neu status unknown) at the age of 36 years. Her daughter, siblings, and parents never had cancer. However, she had maternal and paternal history of breast cancer and ovarian cancer. Her maternal grandmother had four sisters: one was healthy; two of them had breast cancer at the age of 45 and 68 years, respectively; while another one was diagnosed with breast cancer at the age of 58 years and subsequent ovarian cancer fourteen years later. In addition, three of four sisters of the patient’s father were diagnosed with breast cancer at the age of 37, 45 and >50 years, respectively. Because of the lack of DNA from the family members of the index patient, we were not able to determine from which side the mutation was inherited.

The patient with c.3048delT (p.Phe1016Leufs*17) was diagnosed with grade II IDC NOS (ER-, PR-, but HER2/neu status unknown) at the age of 60 years (died four years later). A family history of breast and/or ovarian cancer was not reported, but, she had a sister who died of colorectal cancer at the age of 55 years, a mother who died from leukemia at the age of 33 years, and her maternal grandfather who had pancreatic cancer at the age of 52 years. It is very likely that the deleterious mutation was inherited from her maternal side; however, we could not confirm this due to the absence of DNA samples.

In silico analyses of nonsynonymous variants

Of the fifty-three variants identified in the breast cancer cases, twelve were nonsynonymous/missense and analyzed by the above-mentioned five computational algorithms: SIFT, PolyPhen, PMUT, PANTHER, and Align GVGD (Table 3). In order to obtain a better classification of the variants, a classification/re-score system was established based on the summed scores of these individual analyses to help assess the possible pathological effects of the nonsynonymous variants. Of the twelve missense variants, six were consistently (all the five programs) evaluated as benign, four as possibly benign (three of five, or four of five), one as possibly deleterious (four of five), and another as deleterious (all the five predictions).

Occurrences of variants in controls

We sequenced four amplicons containing all the three frameshift/truncating mutations, one predicted deleterious (c.2993G>A/p.Gly998Glu) and one possibly deleterious missense variant (c.3056T>C/p.Val1019Ala), in 262 cancer-free African-American women. None of the three truncating mutations were found in control subjects from the same population. The MAF of c.2993G>A was 0.005 in caseswhich is comparable to the MAF of 0.004 in 262 controls. The c.3056T>C was found with a MAF of 0.002 in cases but not in controls.

DISCUSSION

PALB2 as a breast cancer susceptibility gene

Efforts have been made to investigate the mutation spectrum of PALB2 in breast cancer patients, and deleterious mutations have been reported in various populations such as the British,^{8, 23} Irish and Scot,²³ Finnish,^{9, 15} Scottish-Canadian,¹⁰ French-Canadian,¹¹ Spanish,¹² Chinese,¹³ South African,¹⁴ Italian,^{16, 19, 22} Polish,¹⁷ Netherlander,¹⁸ French,²³ German,^{20, 23} Russian,²⁰ African-American,²³ European-American,²⁴ as well as Australian and New Zealander.^{21, 41} All the deleterious mutations associated with breast cancer found in PALB2 appeared to be singletons, except a few were recurrent mutations (c.3116delA, c.3113G>A, and c.3549C>G in British,⁸ c.751C>T in Chinese,¹³ c.509-510delGA in Polish,¹⁷ and c.3113G>A in Australian and New Zealander²¹), and founder mutations (c.1592delT in Finnish^{9, 15} and c.2323C>T in French-Canadian¹¹). Five PALB2 mutations found in the UK were estimated to be associated with a 2.3 fold increased risk.⁸ The recurrent mutation c.3113G>A detected in Australia and New Zealand was speculated to originate from the UK and conferred an age-specific cumulative risk of 49% (95% CI: 15-93) to age 50 years and 91% (95% CI: 44-100) to age 70 years.²¹ The founder mutation c.1592delT was estimated to contribute a 6-fold increased breast cancer risk (95% CI: 2-17, equivalent to a risk of 40% to age 70 years) in Finland.^{9, 15}

In spite of different study designs of the above mutational analyses, namely, full gene sequencing versus individual mutation screening, familial cases versus non-familial cases, selected versus unselected early age onset, female versus male patients, and the various sample size, the majority of mutations identified turned out to be rare with frequencies from 0.0011 (1/923, c.2982insT and c.2386G>T in British)⁸ to 0.0208 (1/48, c.697delG in South Africans).¹⁴ Mutation prevalence of PALB2 per study varied, from 0.56 to 2.65% in familial breast cancer patients, and < 1% in unselected cases. We found three truncating PALB2 mutations in our cohort of 279 female African-American breast cancer patients which consists of 133 familial and 146 non-familial or sporadic breast cancer cases. Of these three mutations, two (c.758dupT and c.1479delC) were found in individuals with family history of breast and/or ovarian cancer, and one (c.3048delT) was found in a non-familial patient. Therefore, the mutation prevalence in the present study are 1.08, 1.50 and 0.68% for overall, familial and non-familial, respectively. There is no significant discrepancy of mutation prevalence between our study and the literature. However, it is worth noting that, Ding et al. recently sequenced PALB2 in 139 African-American breast cancer cases (134 familial and 5 non-familial) and failed to detect any truncating mutation.⁴² Recently, Casadei et al. screened 1,144 familial breast cancer patients (172 Ashkenazi Jewish and 972 cases of multiple ancestries) and found two (out of 18) African-American patients carried PALB2 mutations.²³ Taking their results into account, mutation prevalence of PALB2 in African-American women with breast cancer turn out to be 1.15 (5/436), 1.40 (4/285) and 0.66% (1/151) for overall, familial and non-familial, respectively. Given the low or undetectable frequencies of PALB2 mutations in different racial/ethnic groups, study sample size would have to be large to capture rare mutations. Moreover, the majority of the studies employed criteria that would enrich for PALB2 mutation carriers, for instance, positive family history of breast and/or ovarian cancer and early age of onset. It is therefore difficult to apply these estimates to the general population. Future studies evaluating mutation spectrum and prevalence in non-familial patients are warranted. To date, there is paucity of data on PALB2 germline mutation screenings in breast cancer patients of African ancestry, additional studies are needed to fill in this gap.

Genotype-phenotype correlation

To date, all the PALB2 deleterious mutations found in breast cancer are frameshift deletion/insertion, nonsense mutations, and mutations affecting splicing (Figure 1). It seems that mutations causing breast and/or ovarian cancer tend to cluster at the 5′ of the PALB2 gene; mutations in Fanconi anemia spread throughout the gene, mutations in pancreatic cancer might be enriched at the 5′ of PALB2 (although few studies have been carried out), and only one mutation, c.1592delT, has been reported in prostate cancer.

PALB2 has a coiled-coil domain (p.9-p.43, corresponding to c.25-c.129) at the N terminus which binds to the coiled-coil pre-BRCT motif of BRCA1; WD40 motifs (p.848-p.1186, corresponding to c.2542-c.3558) enriched in the C-terminal of PALB2 interacts with the N-terminal of BRCA2; and a region (p.611-p.738, corresponding to c.1831-c.2214) approximately in the middle has been identified as a binding site of MRG15 which is a MRG domain-containing protein involved in chromatin remodeling complexes.⁴³ It appears that PALB2 mutations identified in Fanconi anemia and cancers do not affect the binding sites of BRCA1 and MRG15 directly (except c.72delG), but some are predicted to disrupt the BRCA2 binding site. Most of the mutations were mapped to exons 4 and 5 of the PALB2 gene, where the biological functions of the corresponding protein structure remain unknown. One possible explanation for the enrichment in exons 4 and 5 is that they are the largest two exons in PALB2. Thus, full-gene sequencing of PALB2 gene may be necessary, especially in populations in which recurrent and founder mutations have not yet been identified.

In addition, it has been proposed that, similar to BRCA1- but not BRCA2-related breast cancers, around 40% of PALB2-related breast cancer patients might be triple-negative (ER-, PR-, and HER2/neu-) tumors.⁴ In our African-American cohort, none of three mutation carriers had triple-negative tumor, which does not support the previous observation. A better understanding of PALB2-related breast cancer subtypes awaits future investigation in the same population. It is also worth noting that PALB2 mutations have been detected in <5% of familial pancreatic cancer patients, however, it remains to be determined whether screening for deleterious mutations of the PALB2 gene is warranted in families enriched for both breast cancer and pancreatic cancer cases.^{24, 26-28}

CONCLUSIONS

Accumulated evidence suggests that PALB2 is a breast cancer susceptibility gene. We performed direct full gene sequencing of PALB2 in 133 familial and 146 non-familial female African-American breast cancer cases. Besides twelve nonsynonymous, thirteen synonymous, two untranslated-5′UTR, one untranslated-3′UTR, one 3′ downstream, and twenty-one intronic variants, we were able to identify three novel truncating mutations. Interestingly, these deleterious mutations were detectable in both familial and non-familial groups. Our data suggest that PALB2 mutations may be linked to both familial and non-familial breast cancer among African-Americans. Taking into account the fact that approximately 1-2% of familial breast cancer and 3-5% of familial pancreatic cancer patients have been identified to carry germline PALB2 mutations, better characterizing the mutation spectrum in diverse populations will definitely benefit the evaluation and consideration of the routine clinical testing of the PALB2 gene for patients with breast and/or other cancers.